qemu-e2k/hw/net/e1000e_core.c

3513 lines
100 KiB
C
Raw Normal View History

net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
/*
* Core code for QEMU e1000e emulation
*
* Software developer's manuals:
* http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf
*
* Copyright (c) 2015 Ravello Systems LTD (http://ravellosystems.com)
* Developed by Daynix Computing LTD (http://www.daynix.com)
*
* Authors:
* Dmitry Fleytman <dmitry@daynix.com>
* Leonid Bloch <leonid@daynix.com>
* Yan Vugenfirer <yan@daynix.com>
*
* Based on work done by:
* Nir Peleg, Tutis Systems Ltd. for Qumranet Inc.
* Copyright (c) 2008 Qumranet
* Based on work done by:
* Copyright (c) 2007 Dan Aloni
* Copyright (c) 2004 Antony T Curtis
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, see <http://www.gnu.org/licenses/>.
*/
#include "qemu/osdep.h"
hw/net/e1000e: Do not abort() on invalid PSRCTL register value libFuzzer found using 'qemu-system-i386 -M q35': qemu: hardware error: e1000e: PSRCTL.BSIZE0 cannot be zero CPU #0: EAX=00000000 EBX=00000000 ECX=00000000 EDX=00000663 ESI=00000000 EDI=00000000 EBP=00000000 ESP=00000000 EIP=0000fff0 EFL=00000002 [-------] CPL=0 II=0 A20=1 SMM=0 HLT=0 ES =0000 00000000 0000ffff 00009300 CS =f000 ffff0000 0000ffff 00009b00 SS =0000 00000000 0000ffff 00009300 DS =0000 00000000 0000ffff 00009300 FS =0000 00000000 0000ffff 00009300 GS =0000 00000000 0000ffff 00009300 LDT=0000 00000000 0000ffff 00008200 TR =0000 00000000 0000ffff 00008b00 GDT= 00000000 0000ffff IDT= 00000000 0000ffff CR0=60000010 CR2=00000000 CR3=00000000 CR4=00000000 DR0=00000000 DR1=00000000 DR2=00000000 DR3=00000000 DR6=ffff0ff0 DR7=00000400 EFER=0000000000000000 FCW=037f FSW=0000 [ST=0] FTW=00 MXCSR=00001f80 FPR0=0000000000000000 0000 FPR1=0000000000000000 0000 FPR2=0000000000000000 0000 FPR3=0000000000000000 0000 FPR4=0000000000000000 0000 FPR5=0000000000000000 0000 FPR6=0000000000000000 0000 FPR7=0000000000000000 0000 XMM00=00000000000000000000000000000000 XMM01=00000000000000000000000000000000 XMM02=00000000000000000000000000000000 XMM03=00000000000000000000000000000000 XMM04=00000000000000000000000000000000 XMM05=00000000000000000000000000000000 XMM06=00000000000000000000000000000000 XMM07=00000000000000000000000000000000 ==1988== ERROR: libFuzzer: deadly signal #6 0x7fae4d3ea894 in __GI_abort (/lib64/libc.so.6+0x22894) #7 0x563f4cc59a1d in hw_error (qemu-fuzz-i386+0xe8ca1d) #8 0x563f4d7c93f2 in e1000e_set_psrctl (qemu-fuzz-i386+0x19fc3f2) #9 0x563f4d7b798f in e1000e_core_write (qemu-fuzz-i386+0x19ea98f) #10 0x563f4d7afc46 in e1000e_mmio_write (qemu-fuzz-i386+0x19e2c46) #11 0x563f4cc9a0a7 in memory_region_write_accessor (qemu-fuzz-i386+0xecd0a7) #12 0x563f4cc99c13 in access_with_adjusted_size (qemu-fuzz-i386+0xeccc13) #13 0x563f4cc987b4 in memory_region_dispatch_write (qemu-fuzz-i386+0xecb7b4) It simply sent the following 2 I/O command to the e1000e PCI BAR #2 I/O region: writew 0x0100 0x0c00 # RCTL = E1000_RCTL_DTYP_MASK writeb 0x2170 0x00 # PSRCTL = 0 2813 static void 2814 e1000e_set_psrctl(E1000ECore *core, int index, uint32_t val) 2815 { 2816 if (core->mac[RCTL] & E1000_RCTL_DTYP_MASK) { 2817 2818 if ((val & E1000_PSRCTL_BSIZE0_MASK) == 0) { 2819 hw_error("e1000e: PSRCTL.BSIZE0 cannot be zero"); 2820 } Instead of calling hw_error() which abort the process (it is meant for CPU fatal error condition, not for device logging), log the invalid request with qemu_log_mask(LOG_GUEST_ERROR) and return, ignoring the request. Cc: qemu-stable@nongnu.org Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Jason Wang <jasowang@redhat.com>
2020-05-25 14:23:30 +02:00
#include "qemu/log.h"
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
#include "net/net.h"
#include "net/tap.h"
#include "hw/pci/msi.h"
#include "hw/pci/msix.h"
#include "sysemu/runstate.h"
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
#include "net_tx_pkt.h"
#include "net_rx_pkt.h"
#include "e1000x_common.h"
#include "e1000e_core.h"
#include "trace.h"
#define E1000E_MIN_XITR (500) /* No more then 7813 interrupts per
second according to spec 10.2.4.2 */
#define E1000E_MAX_TX_FRAGS (64)
static inline void
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000e_set_interrupt_cause(E1000ECore *core, uint32_t val);
static inline void
e1000e_process_ts_option(E1000ECore *core, struct e1000_tx_desc *dp)
{
if (le32_to_cpu(dp->upper.data) & E1000_TXD_EXTCMD_TSTAMP) {
trace_e1000e_wrn_no_ts_support();
}
}
static inline void
e1000e_process_snap_option(E1000ECore *core, uint32_t cmd_and_length)
{
if (cmd_and_length & E1000_TXD_CMD_SNAP) {
trace_e1000e_wrn_no_snap_support();
}
}
static inline void
e1000e_raise_legacy_irq(E1000ECore *core)
{
trace_e1000e_irq_legacy_notify(true);
e1000x_inc_reg_if_not_full(core->mac, IAC);
pci_set_irq(core->owner, 1);
}
static inline void
e1000e_lower_legacy_irq(E1000ECore *core)
{
trace_e1000e_irq_legacy_notify(false);
pci_set_irq(core->owner, 0);
}
static inline void
e1000e_intrmgr_rearm_timer(E1000IntrDelayTimer *timer)
{
int64_t delay_ns = (int64_t) timer->core->mac[timer->delay_reg] *
timer->delay_resolution_ns;
trace_e1000e_irq_rearm_timer(timer->delay_reg << 2, delay_ns);
timer_mod(timer->timer, qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + delay_ns);
timer->running = true;
}
static void
e1000e_intmgr_timer_resume(E1000IntrDelayTimer *timer)
{
if (timer->running) {
e1000e_intrmgr_rearm_timer(timer);
}
}
static void
e1000e_intmgr_timer_pause(E1000IntrDelayTimer *timer)
{
if (timer->running) {
timer_del(timer->timer);
}
}
static inline void
e1000e_intrmgr_stop_timer(E1000IntrDelayTimer *timer)
{
if (timer->running) {
timer_del(timer->timer);
timer->running = false;
}
}
static inline void
e1000e_intrmgr_fire_delayed_interrupts(E1000ECore *core)
{
trace_e1000e_irq_fire_delayed_interrupts();
e1000e_set_interrupt_cause(core, 0);
}
static void
e1000e_intrmgr_on_timer(void *opaque)
{
E1000IntrDelayTimer *timer = opaque;
trace_e1000e_irq_throttling_timer(timer->delay_reg << 2);
timer->running = false;
e1000e_intrmgr_fire_delayed_interrupts(timer->core);
}
static void
e1000e_intrmgr_on_throttling_timer(void *opaque)
{
E1000IntrDelayTimer *timer = opaque;
assert(!msix_enabled(timer->core->owner));
timer->running = false;
if (!timer->core->itr_intr_pending) {
trace_e1000e_irq_throttling_no_pending_interrupts();
return;
}
if (msi_enabled(timer->core->owner)) {
trace_e1000e_irq_msi_notify_postponed();
e1000e_set_interrupt_cause(timer->core, 0);
} else {
trace_e1000e_irq_legacy_notify_postponed();
e1000e_set_interrupt_cause(timer->core, 0);
}
}
static void
e1000e_intrmgr_on_msix_throttling_timer(void *opaque)
{
E1000IntrDelayTimer *timer = opaque;
int idx = timer - &timer->core->eitr[0];
assert(msix_enabled(timer->core->owner));
timer->running = false;
if (!timer->core->eitr_intr_pending[idx]) {
trace_e1000e_irq_throttling_no_pending_vec(idx);
return;
}
trace_e1000e_irq_msix_notify_postponed_vec(idx);
msix_notify(timer->core->owner, idx);
}
static void
e1000e_intrmgr_initialize_all_timers(E1000ECore *core, bool create)
{
int i;
core->radv.delay_reg = RADV;
core->rdtr.delay_reg = RDTR;
core->raid.delay_reg = RAID;
core->tadv.delay_reg = TADV;
core->tidv.delay_reg = TIDV;
core->radv.delay_resolution_ns = E1000_INTR_DELAY_NS_RES;
core->rdtr.delay_resolution_ns = E1000_INTR_DELAY_NS_RES;
core->raid.delay_resolution_ns = E1000_INTR_DELAY_NS_RES;
core->tadv.delay_resolution_ns = E1000_INTR_DELAY_NS_RES;
core->tidv.delay_resolution_ns = E1000_INTR_DELAY_NS_RES;
core->radv.core = core;
core->rdtr.core = core;
core->raid.core = core;
core->tadv.core = core;
core->tidv.core = core;
core->itr.core = core;
core->itr.delay_reg = ITR;
core->itr.delay_resolution_ns = E1000_INTR_THROTTLING_NS_RES;
for (i = 0; i < E1000E_MSIX_VEC_NUM; i++) {
core->eitr[i].core = core;
core->eitr[i].delay_reg = EITR + i;
core->eitr[i].delay_resolution_ns = E1000_INTR_THROTTLING_NS_RES;
}
if (!create) {
return;
}
core->radv.timer =
timer_new_ns(QEMU_CLOCK_VIRTUAL, e1000e_intrmgr_on_timer, &core->radv);
core->rdtr.timer =
timer_new_ns(QEMU_CLOCK_VIRTUAL, e1000e_intrmgr_on_timer, &core->rdtr);
core->raid.timer =
timer_new_ns(QEMU_CLOCK_VIRTUAL, e1000e_intrmgr_on_timer, &core->raid);
core->tadv.timer =
timer_new_ns(QEMU_CLOCK_VIRTUAL, e1000e_intrmgr_on_timer, &core->tadv);
core->tidv.timer =
timer_new_ns(QEMU_CLOCK_VIRTUAL, e1000e_intrmgr_on_timer, &core->tidv);
core->itr.timer = timer_new_ns(QEMU_CLOCK_VIRTUAL,
e1000e_intrmgr_on_throttling_timer,
&core->itr);
for (i = 0; i < E1000E_MSIX_VEC_NUM; i++) {
core->eitr[i].timer =
timer_new_ns(QEMU_CLOCK_VIRTUAL,
e1000e_intrmgr_on_msix_throttling_timer,
&core->eitr[i]);
}
}
static inline void
e1000e_intrmgr_stop_delay_timers(E1000ECore *core)
{
e1000e_intrmgr_stop_timer(&core->radv);
e1000e_intrmgr_stop_timer(&core->rdtr);
e1000e_intrmgr_stop_timer(&core->raid);
e1000e_intrmgr_stop_timer(&core->tidv);
e1000e_intrmgr_stop_timer(&core->tadv);
}
static bool
e1000e_intrmgr_delay_rx_causes(E1000ECore *core, uint32_t *causes)
{
uint32_t delayable_causes;
uint32_t rdtr = core->mac[RDTR];
uint32_t radv = core->mac[RADV];
uint32_t raid = core->mac[RAID];
if (msix_enabled(core->owner)) {
return false;
}
delayable_causes = E1000_ICR_RXQ0 |
E1000_ICR_RXQ1 |
E1000_ICR_RXT0;
if (!(core->mac[RFCTL] & E1000_RFCTL_ACK_DIS)) {
delayable_causes |= E1000_ICR_ACK;
}
/* Clean up all causes that may be delayed */
core->delayed_causes |= *causes & delayable_causes;
*causes &= ~delayable_causes;
/* Check if delayed RX interrupts disabled by client
or if there are causes that cannot be delayed */
if ((rdtr == 0) || (*causes != 0)) {
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
return false;
}
/* Check if delayed RX ACK interrupts disabled by client
and there is an ACK packet received */
if ((raid == 0) && (core->delayed_causes & E1000_ICR_ACK)) {
return false;
}
/* All causes delayed */
e1000e_intrmgr_rearm_timer(&core->rdtr);
if (!core->radv.running && (radv != 0)) {
e1000e_intrmgr_rearm_timer(&core->radv);
}
if (!core->raid.running && (core->delayed_causes & E1000_ICR_ACK)) {
e1000e_intrmgr_rearm_timer(&core->raid);
}
return true;
}
static bool
e1000e_intrmgr_delay_tx_causes(E1000ECore *core, uint32_t *causes)
{
static const uint32_t delayable_causes = E1000_ICR_TXQ0 |
E1000_ICR_TXQ1 |
E1000_ICR_TXQE |
E1000_ICR_TXDW;
if (msix_enabled(core->owner)) {
return false;
}
/* Clean up all causes that may be delayed */
core->delayed_causes |= *causes & delayable_causes;
*causes &= ~delayable_causes;
/* If there are causes that cannot be delayed */
if (*causes != 0) {
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
return false;
}
/* All causes delayed */
e1000e_intrmgr_rearm_timer(&core->tidv);
if (!core->tadv.running && (core->mac[TADV] != 0)) {
e1000e_intrmgr_rearm_timer(&core->tadv);
}
return true;
}
static uint32_t
e1000e_intmgr_collect_delayed_causes(E1000ECore *core)
{
uint32_t res;
if (msix_enabled(core->owner)) {
assert(core->delayed_causes == 0);
return 0;
}
res = core->delayed_causes;
core->delayed_causes = 0;
e1000e_intrmgr_stop_delay_timers(core);
return res;
}
static void
e1000e_intrmgr_fire_all_timers(E1000ECore *core)
{
int i;
uint32_t val = e1000e_intmgr_collect_delayed_causes(core);
trace_e1000e_irq_adding_delayed_causes(val, core->mac[ICR]);
core->mac[ICR] |= val;
if (core->itr.running) {
timer_del(core->itr.timer);
e1000e_intrmgr_on_throttling_timer(&core->itr);
}
for (i = 0; i < E1000E_MSIX_VEC_NUM; i++) {
if (core->eitr[i].running) {
timer_del(core->eitr[i].timer);
e1000e_intrmgr_on_msix_throttling_timer(&core->eitr[i]);
}
}
}
static void
e1000e_intrmgr_resume(E1000ECore *core)
{
int i;
e1000e_intmgr_timer_resume(&core->radv);
e1000e_intmgr_timer_resume(&core->rdtr);
e1000e_intmgr_timer_resume(&core->raid);
e1000e_intmgr_timer_resume(&core->tidv);
e1000e_intmgr_timer_resume(&core->tadv);
e1000e_intmgr_timer_resume(&core->itr);
for (i = 0; i < E1000E_MSIX_VEC_NUM; i++) {
e1000e_intmgr_timer_resume(&core->eitr[i]);
}
}
static void
e1000e_intrmgr_pause(E1000ECore *core)
{
int i;
e1000e_intmgr_timer_pause(&core->radv);
e1000e_intmgr_timer_pause(&core->rdtr);
e1000e_intmgr_timer_pause(&core->raid);
e1000e_intmgr_timer_pause(&core->tidv);
e1000e_intmgr_timer_pause(&core->tadv);
e1000e_intmgr_timer_pause(&core->itr);
for (i = 0; i < E1000E_MSIX_VEC_NUM; i++) {
e1000e_intmgr_timer_pause(&core->eitr[i]);
}
}
static void
e1000e_intrmgr_reset(E1000ECore *core)
{
int i;
core->delayed_causes = 0;
e1000e_intrmgr_stop_delay_timers(core);
e1000e_intrmgr_stop_timer(&core->itr);
for (i = 0; i < E1000E_MSIX_VEC_NUM; i++) {
e1000e_intrmgr_stop_timer(&core->eitr[i]);
}
}
static void
e1000e_intrmgr_pci_unint(E1000ECore *core)
{
int i;
timer_free(core->radv.timer);
timer_free(core->rdtr.timer);
timer_free(core->raid.timer);
timer_free(core->tadv.timer);
timer_free(core->tidv.timer);
timer_free(core->itr.timer);
for (i = 0; i < E1000E_MSIX_VEC_NUM; i++) {
timer_free(core->eitr[i].timer);
}
}
static void
e1000e_intrmgr_pci_realize(E1000ECore *core)
{
e1000e_intrmgr_initialize_all_timers(core, true);
}
static inline bool
e1000e_rx_csum_enabled(E1000ECore *core)
{
return (core->mac[RXCSUM] & E1000_RXCSUM_PCSD) ? false : true;
}
static inline bool
e1000e_rx_use_legacy_descriptor(E1000ECore *core)
{
return (core->mac[RFCTL] & E1000_RFCTL_EXTEN) ? false : true;
}
static inline bool
e1000e_rx_use_ps_descriptor(E1000ECore *core)
{
return !e1000e_rx_use_legacy_descriptor(core) &&
(core->mac[RCTL] & E1000_RCTL_DTYP_PS);
}
static inline bool
e1000e_rss_enabled(E1000ECore *core)
{
return E1000_MRQC_ENABLED(core->mac[MRQC]) &&
!e1000e_rx_csum_enabled(core) &&
!e1000e_rx_use_legacy_descriptor(core);
}
typedef struct E1000E_RSSInfo_st {
bool enabled;
uint32_t hash;
uint32_t queue;
uint32_t type;
} E1000E_RSSInfo;
static uint32_t
e1000e_rss_get_hash_type(E1000ECore *core, struct NetRxPkt *pkt)
{
bool isip4, isip6, isudp, istcp;
assert(e1000e_rss_enabled(core));
net_rx_pkt_get_protocols(pkt, &isip4, &isip6, &isudp, &istcp);
if (isip4) {
bool fragment = net_rx_pkt_get_ip4_info(pkt)->fragment;
trace_e1000e_rx_rss_ip4(fragment, istcp, core->mac[MRQC],
E1000_MRQC_EN_TCPIPV4(core->mac[MRQC]),
E1000_MRQC_EN_IPV4(core->mac[MRQC]));
if (!fragment && istcp && E1000_MRQC_EN_TCPIPV4(core->mac[MRQC])) {
return E1000_MRQ_RSS_TYPE_IPV4TCP;
}
if (E1000_MRQC_EN_IPV4(core->mac[MRQC])) {
return E1000_MRQ_RSS_TYPE_IPV4;
}
} else if (isip6) {
eth_ip6_hdr_info *ip6info = net_rx_pkt_get_ip6_info(pkt);
bool ex_dis = core->mac[RFCTL] & E1000_RFCTL_IPV6_EX_DIS;
bool new_ex_dis = core->mac[RFCTL] & E1000_RFCTL_NEW_IPV6_EXT_DIS;
e1000e: Fix build with ust trace backend ust trace backend has limitation of maximum 10 arguments per event. Traces with more arguments cannot be compiled for this backend. Trace e1000e_rx_rss_ip6 introduced by previous commits has 11 arguments and fails to compile with ust trace backend. This patch fixes the problem by splitting this tracepoint into two successive tracepoints with smaller number of arguments. For more information see comment regarding TP_ARGS in lttng/tracepoint.h: /* * TP_ARGS takes tuples of type, argument separated by a comma. * It can take up to 10 tuples (which means that less than 10 tuples is * fine too). * Each tuple is also separated by a comma. */ Build log generated by this problem: In file included from ./trace/generated-tracers.h:9:0, from /home/travis/build/qemu/qemu/include/trace.h:4, from util/oslib-posix.c:36: ./trace/generated-ust-provider.h:16556:3: error: unknown type name ‘_TP_EXPROTO_Bool’ In file included from /home/travis/build/qemu/qemu/include/trace.h:4:0, from util/oslib-posix.c:36: ./trace/generated-tracers.h: In function ‘trace_e1000e_rx_rss_ip6’: ./trace/generated-tracers.h:8379:431: error: expected string literal before ‘_SDT_ASM_OPERANDS_ipv6_enabled’ ./trace/generated-tracers.h:8379:431: error: implicit declaration of function ‘__tracepoint_cb_qemu___e1000e_rx_rss_ip6’ [-Werror=implicit-function-declaration] ./trace/generated-tracers.h:8379:431: error: nested extern declaration of ‘__tracepoint_cb_qemu___e1000e_rx_rss_ip6’ [-Werror=nested-externs] cc1: all warnings being treated as errors make: *** [util/oslib-posix.o] Error 1 make: *** Waiting for unfinished jobs.... In file included from ./trace/generated-tracers.h:9:0, from /home/travis/build/qemu/qemu/include/trace.h:4, from util/hbitmap.c:16: ./trace/generated-ust-provider.h:16556:3: error: unknown type name ‘_TP_EXPROTO_Bool’ In file included from /home/travis/build/qemu/qemu/include/trace.h:4:0, from util/hbitmap.c:16: ./trace/generated-tracers.h: In function ‘trace_e1000e_rx_rss_ip6’: ./trace/generated-tracers.h:8379:431: error: expected string literal before ‘_SDT_ASM_OPERANDS_ipv6_enabled’ ./trace/generated-tracers.h:8379:431: error: implicit declaration of function ‘__tracepoint_cb_qemu___e1000e_rx_rss_ip6’ [-Werror=implicit-function-declaration] ./trace/generated-tracers.h:8379:431: error: nested extern declaration of ‘__tracepoint_cb_qemu___e1000e_rx_rss_ip6’ [-Werror=nested-externs] cc1: all warnings being treated as errors make: *** [util/hbitmap.o] Error 1 Signed-off-by: Dmitry Fleytman <dmitry@daynix.com> Message-id: 1464894748-27803-1-git-send-email-dmitry@daynix.com Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
2016-06-02 21:12:28 +02:00
/*
* Following two traces must not be combined because resulting
* event will have 11 arguments totally and some trace backends
* (at least "ust") have limitation of maximum 10 arguments per
* event. Events with more arguments fail to compile for
* backends like these.
*/
trace_e1000e_rx_rss_ip6_rfctl(core->mac[RFCTL]);
trace_e1000e_rx_rss_ip6(ex_dis, new_ex_dis, istcp,
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
ip6info->has_ext_hdrs,
ip6info->rss_ex_dst_valid,
ip6info->rss_ex_src_valid,
core->mac[MRQC],
E1000_MRQC_EN_TCPIPV6(core->mac[MRQC]),
E1000_MRQC_EN_IPV6EX(core->mac[MRQC]),
E1000_MRQC_EN_IPV6(core->mac[MRQC]));
if ((!ex_dis || !ip6info->has_ext_hdrs) &&
(!new_ex_dis || !(ip6info->rss_ex_dst_valid ||
ip6info->rss_ex_src_valid))) {
if (istcp && !ip6info->fragment &&
E1000_MRQC_EN_TCPIPV6(core->mac[MRQC])) {
return E1000_MRQ_RSS_TYPE_IPV6TCP;
}
if (E1000_MRQC_EN_IPV6EX(core->mac[MRQC])) {
return E1000_MRQ_RSS_TYPE_IPV6EX;
}
}
if (E1000_MRQC_EN_IPV6(core->mac[MRQC])) {
return E1000_MRQ_RSS_TYPE_IPV6;
}
}
return E1000_MRQ_RSS_TYPE_NONE;
}
static uint32_t
e1000e_rss_calc_hash(E1000ECore *core,
struct NetRxPkt *pkt,
E1000E_RSSInfo *info)
{
NetRxPktRssType type;
assert(e1000e_rss_enabled(core));
switch (info->type) {
case E1000_MRQ_RSS_TYPE_IPV4:
type = NetPktRssIpV4;
break;
case E1000_MRQ_RSS_TYPE_IPV4TCP:
type = NetPktRssIpV4Tcp;
break;
case E1000_MRQ_RSS_TYPE_IPV6TCP:
type = NetPktRssIpV6TcpEx;
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
break;
case E1000_MRQ_RSS_TYPE_IPV6:
type = NetPktRssIpV6;
break;
case E1000_MRQ_RSS_TYPE_IPV6EX:
type = NetPktRssIpV6Ex;
break;
default:
assert(false);
return 0;
}
return net_rx_pkt_calc_rss_hash(pkt, type, (uint8_t *) &core->mac[RSSRK]);
}
static void
e1000e_rss_parse_packet(E1000ECore *core,
struct NetRxPkt *pkt,
E1000E_RSSInfo *info)
{
trace_e1000e_rx_rss_started();
if (!e1000e_rss_enabled(core)) {
info->enabled = false;
info->hash = 0;
info->queue = 0;
info->type = 0;
trace_e1000e_rx_rss_disabled();
return;
}
info->enabled = true;
info->type = e1000e_rss_get_hash_type(core, pkt);
trace_e1000e_rx_rss_type(info->type);
if (info->type == E1000_MRQ_RSS_TYPE_NONE) {
info->hash = 0;
info->queue = 0;
return;
}
info->hash = e1000e_rss_calc_hash(core, pkt, info);
info->queue = E1000_RSS_QUEUE(&core->mac[RETA], info->hash);
}
static void
e1000e_setup_tx_offloads(E1000ECore *core, struct e1000e_tx *tx)
{
if (tx->props.tse && tx->cptse) {
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
net_tx_pkt_build_vheader(tx->tx_pkt, true, true, tx->props.mss);
net_tx_pkt_update_ip_checksums(tx->tx_pkt);
e1000x_inc_reg_if_not_full(core->mac, TSCTC);
return;
}
if (tx->sum_needed & E1000_TXD_POPTS_TXSM) {
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
net_tx_pkt_build_vheader(tx->tx_pkt, false, true, 0);
}
if (tx->sum_needed & E1000_TXD_POPTS_IXSM) {
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
net_tx_pkt_update_ip_hdr_checksum(tx->tx_pkt);
}
}
static bool
e1000e_tx_pkt_send(E1000ECore *core, struct e1000e_tx *tx, int queue_index)
{
int target_queue = MIN(core->max_queue_num, queue_index);
NetClientState *queue = qemu_get_subqueue(core->owner_nic, target_queue);
e1000e_setup_tx_offloads(core, tx);
net_tx_pkt_dump(tx->tx_pkt);
if ((core->phy[0][PHY_CTRL] & MII_CR_LOOPBACK) ||
((core->mac[RCTL] & E1000_RCTL_LBM_MAC) == E1000_RCTL_LBM_MAC)) {
return net_tx_pkt_send_loopback(tx->tx_pkt, queue);
} else {
return net_tx_pkt_send(tx->tx_pkt, queue);
}
}
static void
e1000e_on_tx_done_update_stats(E1000ECore *core, struct NetTxPkt *tx_pkt)
{
static const int PTCregs[6] = { PTC64, PTC127, PTC255, PTC511,
PTC1023, PTC1522 };
size_t tot_len = net_tx_pkt_get_total_len(tx_pkt);
e1000x_increase_size_stats(core->mac, PTCregs, tot_len);
e1000x_inc_reg_if_not_full(core->mac, TPT);
e1000x_grow_8reg_if_not_full(core->mac, TOTL, tot_len);
switch (net_tx_pkt_get_packet_type(tx_pkt)) {
case ETH_PKT_BCAST:
e1000x_inc_reg_if_not_full(core->mac, BPTC);
break;
case ETH_PKT_MCAST:
e1000x_inc_reg_if_not_full(core->mac, MPTC);
break;
case ETH_PKT_UCAST:
break;
default:
g_assert_not_reached();
}
core->mac[GPTC] = core->mac[TPT];
core->mac[GOTCL] = core->mac[TOTL];
core->mac[GOTCH] = core->mac[TOTH];
}
static void
e1000e_process_tx_desc(E1000ECore *core,
struct e1000e_tx *tx,
struct e1000_tx_desc *dp,
int queue_index)
{
uint32_t txd_lower = le32_to_cpu(dp->lower.data);
uint32_t dtype = txd_lower & (E1000_TXD_CMD_DEXT | E1000_TXD_DTYP_D);
unsigned int split_size = txd_lower & 0xffff;
uint64_t addr;
struct e1000_context_desc *xp = (struct e1000_context_desc *)dp;
bool eop = txd_lower & E1000_TXD_CMD_EOP;
if (dtype == E1000_TXD_CMD_DEXT) { /* context descriptor */
e1000x_read_tx_ctx_descr(xp, &tx->props);
e1000e_process_snap_option(core, le32_to_cpu(xp->cmd_and_length));
return;
} else if (dtype == (E1000_TXD_CMD_DEXT | E1000_TXD_DTYP_D)) {
/* data descriptor */
tx->sum_needed = le32_to_cpu(dp->upper.data) >> 8;
tx->cptse = (txd_lower & E1000_TXD_CMD_TSE) ? 1 : 0;
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000e_process_ts_option(core, dp);
} else {
/* legacy descriptor */
e1000e_process_ts_option(core, dp);
tx->cptse = 0;
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
}
addr = le64_to_cpu(dp->buffer_addr);
if (!tx->skip_cp) {
if (!net_tx_pkt_add_raw_fragment(tx->tx_pkt, addr, split_size)) {
tx->skip_cp = true;
}
}
if (eop) {
if (!tx->skip_cp && net_tx_pkt_parse(tx->tx_pkt)) {
if (e1000x_vlan_enabled(core->mac) &&
e1000x_is_vlan_txd(txd_lower)) {
net_tx_pkt_setup_vlan_header_ex(tx->tx_pkt,
le16_to_cpu(dp->upper.fields.special), core->mac[VET]);
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
}
if (e1000e_tx_pkt_send(core, tx, queue_index)) {
e1000e_on_tx_done_update_stats(core, tx->tx_pkt);
}
}
tx->skip_cp = false;
net_tx_pkt_reset(tx->tx_pkt);
tx->sum_needed = 0;
tx->cptse = 0;
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
}
}
static inline uint32_t
e1000e_tx_wb_interrupt_cause(E1000ECore *core, int queue_idx)
{
if (!msix_enabled(core->owner)) {
return E1000_ICR_TXDW;
}
return (queue_idx == 0) ? E1000_ICR_TXQ0 : E1000_ICR_TXQ1;
}
static inline uint32_t
e1000e_rx_wb_interrupt_cause(E1000ECore *core, int queue_idx,
bool min_threshold_hit)
{
if (!msix_enabled(core->owner)) {
return E1000_ICS_RXT0 | (min_threshold_hit ? E1000_ICS_RXDMT0 : 0);
}
return (queue_idx == 0) ? E1000_ICR_RXQ0 : E1000_ICR_RXQ1;
}
static uint32_t
e1000e_txdesc_writeback(E1000ECore *core, dma_addr_t base,
struct e1000_tx_desc *dp, bool *ide, int queue_idx)
{
uint32_t txd_upper, txd_lower = le32_to_cpu(dp->lower.data);
if (!(txd_lower & E1000_TXD_CMD_RS) &&
!(core->mac[IVAR] & E1000_IVAR_TX_INT_EVERY_WB)) {
return 0;
}
*ide = (txd_lower & E1000_TXD_CMD_IDE) ? true : false;
txd_upper = le32_to_cpu(dp->upper.data) | E1000_TXD_STAT_DD;
dp->upper.data = cpu_to_le32(txd_upper);
pci_dma_write(core->owner, base + ((char *)&dp->upper - (char *)dp),
&dp->upper, sizeof(dp->upper));
return e1000e_tx_wb_interrupt_cause(core, queue_idx);
}
typedef struct E1000E_RingInfo_st {
int dbah;
int dbal;
int dlen;
int dh;
int dt;
int idx;
} E1000E_RingInfo;
static inline bool
e1000e_ring_empty(E1000ECore *core, const E1000E_RingInfo *r)
{
return core->mac[r->dh] == core->mac[r->dt] ||
core->mac[r->dt] >= core->mac[r->dlen] / E1000_RING_DESC_LEN;
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
}
static inline uint64_t
e1000e_ring_base(E1000ECore *core, const E1000E_RingInfo *r)
{
uint64_t bah = core->mac[r->dbah];
uint64_t bal = core->mac[r->dbal];
return (bah << 32) + bal;
}
static inline uint64_t
e1000e_ring_head_descr(E1000ECore *core, const E1000E_RingInfo *r)
{
return e1000e_ring_base(core, r) + E1000_RING_DESC_LEN * core->mac[r->dh];
}
static inline void
e1000e_ring_advance(E1000ECore *core, const E1000E_RingInfo *r, uint32_t count)
{
core->mac[r->dh] += count;
if (core->mac[r->dh] * E1000_RING_DESC_LEN >= core->mac[r->dlen]) {
core->mac[r->dh] = 0;
}
}
static inline uint32_t
e1000e_ring_free_descr_num(E1000ECore *core, const E1000E_RingInfo *r)
{
trace_e1000e_ring_free_space(r->idx, core->mac[r->dlen],
core->mac[r->dh], core->mac[r->dt]);
if (core->mac[r->dh] <= core->mac[r->dt]) {
return core->mac[r->dt] - core->mac[r->dh];
}
if (core->mac[r->dh] > core->mac[r->dt]) {
return core->mac[r->dlen] / E1000_RING_DESC_LEN +
core->mac[r->dt] - core->mac[r->dh];
}
g_assert_not_reached();
return 0;
}
static inline bool
e1000e_ring_enabled(E1000ECore *core, const E1000E_RingInfo *r)
{
return core->mac[r->dlen] > 0;
}
static inline uint32_t
e1000e_ring_len(E1000ECore *core, const E1000E_RingInfo *r)
{
return core->mac[r->dlen];
}
typedef struct E1000E_TxRing_st {
const E1000E_RingInfo *i;
struct e1000e_tx *tx;
} E1000E_TxRing;
static inline int
e1000e_mq_queue_idx(int base_reg_idx, int reg_idx)
{
return (reg_idx - base_reg_idx) / (0x100 >> 2);
}
static inline void
e1000e_tx_ring_init(E1000ECore *core, E1000E_TxRing *txr, int idx)
{
static const E1000E_RingInfo i[E1000E_NUM_QUEUES] = {
{ TDBAH, TDBAL, TDLEN, TDH, TDT, 0 },
{ TDBAH1, TDBAL1, TDLEN1, TDH1, TDT1, 1 }
};
assert(idx < ARRAY_SIZE(i));
txr->i = &i[idx];
txr->tx = &core->tx[idx];
}
typedef struct E1000E_RxRing_st {
const E1000E_RingInfo *i;
} E1000E_RxRing;
static inline void
e1000e_rx_ring_init(E1000ECore *core, E1000E_RxRing *rxr, int idx)
{
static const E1000E_RingInfo i[E1000E_NUM_QUEUES] = {
{ RDBAH0, RDBAL0, RDLEN0, RDH0, RDT0, 0 },
{ RDBAH1, RDBAL1, RDLEN1, RDH1, RDT1, 1 }
};
assert(idx < ARRAY_SIZE(i));
rxr->i = &i[idx];
}
static void
e1000e_start_xmit(E1000ECore *core, const E1000E_TxRing *txr)
{
dma_addr_t base;
struct e1000_tx_desc desc;
bool ide = false;
const E1000E_RingInfo *txi = txr->i;
uint32_t cause = E1000_ICS_TXQE;
if (!(core->mac[TCTL] & E1000_TCTL_EN)) {
trace_e1000e_tx_disabled();
return;
}
while (!e1000e_ring_empty(core, txi)) {
base = e1000e_ring_head_descr(core, txi);
pci_dma_read(core->owner, base, &desc, sizeof(desc));
trace_e1000e_tx_descr((void *)(intptr_t)desc.buffer_addr,
desc.lower.data, desc.upper.data);
e1000e_process_tx_desc(core, txr->tx, &desc, txi->idx);
cause |= e1000e_txdesc_writeback(core, base, &desc, &ide, txi->idx);
e1000e_ring_advance(core, txi, 1);
}
if (!ide || !e1000e_intrmgr_delay_tx_causes(core, &cause)) {
e1000e_set_interrupt_cause(core, cause);
}
}
static bool
e1000e_has_rxbufs(E1000ECore *core, const E1000E_RingInfo *r,
size_t total_size)
{
uint32_t bufs = e1000e_ring_free_descr_num(core, r);
trace_e1000e_rx_has_buffers(r->idx, bufs, total_size,
core->rx_desc_buf_size);
return total_size <= bufs / (core->rx_desc_len / E1000_MIN_RX_DESC_LEN) *
core->rx_desc_buf_size;
}
void
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000e_start_recv(E1000ECore *core)
{
int i;
trace_e1000e_rx_start_recv();
for (i = 0; i <= core->max_queue_num; i++) {
qemu_flush_queued_packets(qemu_get_subqueue(core->owner_nic, i));
}
}
bool
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000e_can_receive(E1000ECore *core)
{
int i;
if (!e1000x_rx_ready(core->owner, core->mac)) {
return false;
}
for (i = 0; i < E1000E_NUM_QUEUES; i++) {
E1000E_RxRing rxr;
e1000e_rx_ring_init(core, &rxr, i);
if (e1000e_ring_enabled(core, rxr.i) &&
e1000e_has_rxbufs(core, rxr.i, 1)) {
trace_e1000e_rx_can_recv();
return true;
}
}
trace_e1000e_rx_can_recv_rings_full();
return false;
}
ssize_t
e1000e_receive(E1000ECore *core, const uint8_t *buf, size_t size)
{
const struct iovec iov = {
.iov_base = (uint8_t *)buf,
.iov_len = size
};
return e1000e_receive_iov(core, &iov, 1);
}
static inline bool
e1000e_rx_l3_cso_enabled(E1000ECore *core)
{
return !!(core->mac[RXCSUM] & E1000_RXCSUM_IPOFLD);
}
static inline bool
e1000e_rx_l4_cso_enabled(E1000ECore *core)
{
return !!(core->mac[RXCSUM] & E1000_RXCSUM_TUOFLD);
}
static bool
e1000e_receive_filter(E1000ECore *core, const uint8_t *buf, int size)
{
uint32_t rctl = core->mac[RCTL];
if (e1000x_is_vlan_packet(buf, core->mac[VET]) &&
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000x_vlan_rx_filter_enabled(core->mac)) {
uint16_t vid = lduw_be_p(buf + 14);
uint32_t vfta = ldl_le_p((uint32_t *)(core->mac + VFTA) +
((vid >> 5) & 0x7f));
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
if ((vfta & (1 << (vid & 0x1f))) == 0) {
trace_e1000e_rx_flt_vlan_mismatch(vid);
return false;
} else {
trace_e1000e_rx_flt_vlan_match(vid);
}
}
switch (net_rx_pkt_get_packet_type(core->rx_pkt)) {
case ETH_PKT_UCAST:
if (rctl & E1000_RCTL_UPE) {
return true; /* promiscuous ucast */
}
break;
case ETH_PKT_BCAST:
if (rctl & E1000_RCTL_BAM) {
return true; /* broadcast enabled */
}
break;
case ETH_PKT_MCAST:
if (rctl & E1000_RCTL_MPE) {
return true; /* promiscuous mcast */
}
break;
default:
g_assert_not_reached();
}
return e1000x_rx_group_filter(core->mac, buf);
}
static inline void
e1000e_read_lgcy_rx_descr(E1000ECore *core, uint8_t *desc, hwaddr *buff_addr)
{
struct e1000_rx_desc *d = (struct e1000_rx_desc *) desc;
*buff_addr = le64_to_cpu(d->buffer_addr);
}
static inline void
e1000e_read_ext_rx_descr(E1000ECore *core, uint8_t *desc, hwaddr *buff_addr)
{
union e1000_rx_desc_extended *d = (union e1000_rx_desc_extended *) desc;
*buff_addr = le64_to_cpu(d->read.buffer_addr);
}
static inline void
e1000e_read_ps_rx_descr(E1000ECore *core, uint8_t *desc,
hwaddr (*buff_addr)[MAX_PS_BUFFERS])
{
int i;
union e1000_rx_desc_packet_split *d =
(union e1000_rx_desc_packet_split *) desc;
for (i = 0; i < MAX_PS_BUFFERS; i++) {
(*buff_addr)[i] = le64_to_cpu(d->read.buffer_addr[i]);
}
trace_e1000e_rx_desc_ps_read((*buff_addr)[0], (*buff_addr)[1],
(*buff_addr)[2], (*buff_addr)[3]);
}
static inline void
e1000e_read_rx_descr(E1000ECore *core, uint8_t *desc,
hwaddr (*buff_addr)[MAX_PS_BUFFERS])
{
if (e1000e_rx_use_legacy_descriptor(core)) {
e1000e_read_lgcy_rx_descr(core, desc, &(*buff_addr)[0]);
(*buff_addr)[1] = (*buff_addr)[2] = (*buff_addr)[3] = 0;
} else {
if (core->mac[RCTL] & E1000_RCTL_DTYP_PS) {
e1000e_read_ps_rx_descr(core, desc, buff_addr);
} else {
e1000e_read_ext_rx_descr(core, desc, &(*buff_addr)[0]);
(*buff_addr)[1] = (*buff_addr)[2] = (*buff_addr)[3] = 0;
}
}
}
static void
e1000e_verify_csum_in_sw(E1000ECore *core,
struct NetRxPkt *pkt,
uint32_t *status_flags,
bool istcp, bool isudp)
{
bool csum_valid;
uint32_t csum_error;
if (e1000e_rx_l3_cso_enabled(core)) {
if (!net_rx_pkt_validate_l3_csum(pkt, &csum_valid)) {
trace_e1000e_rx_metadata_l3_csum_validation_failed();
} else {
csum_error = csum_valid ? 0 : E1000_RXDEXT_STATERR_IPE;
*status_flags |= E1000_RXD_STAT_IPCS | csum_error;
}
} else {
trace_e1000e_rx_metadata_l3_cso_disabled();
}
if (!e1000e_rx_l4_cso_enabled(core)) {
trace_e1000e_rx_metadata_l4_cso_disabled();
return;
}
if (!net_rx_pkt_validate_l4_csum(pkt, &csum_valid)) {
trace_e1000e_rx_metadata_l4_csum_validation_failed();
return;
}
csum_error = csum_valid ? 0 : E1000_RXDEXT_STATERR_TCPE;
if (istcp) {
*status_flags |= E1000_RXD_STAT_TCPCS |
csum_error;
} else if (isudp) {
*status_flags |= E1000_RXD_STAT_TCPCS |
E1000_RXD_STAT_UDPCS |
csum_error;
}
}
static inline bool
e1000e_is_tcp_ack(E1000ECore *core, struct NetRxPkt *rx_pkt)
{
if (!net_rx_pkt_is_tcp_ack(rx_pkt)) {
return false;
}
if (core->mac[RFCTL] & E1000_RFCTL_ACK_DATA_DIS) {
return !net_rx_pkt_has_tcp_data(rx_pkt);
}
return true;
}
static void
e1000e_build_rx_metadata(E1000ECore *core,
struct NetRxPkt *pkt,
bool is_eop,
const E1000E_RSSInfo *rss_info,
uint32_t *rss, uint32_t *mrq,
uint32_t *status_flags,
uint16_t *ip_id,
uint16_t *vlan_tag)
{
struct virtio_net_hdr *vhdr;
bool isip4, isip6, istcp, isudp;
uint32_t pkt_type;
*status_flags = E1000_RXD_STAT_DD;
/* No additional metadata needed for non-EOP descriptors */
if (!is_eop) {
goto func_exit;
}
*status_flags |= E1000_RXD_STAT_EOP;
net_rx_pkt_get_protocols(pkt, &isip4, &isip6, &isudp, &istcp);
trace_e1000e_rx_metadata_protocols(isip4, isip6, isudp, istcp);
/* VLAN state */
if (net_rx_pkt_is_vlan_stripped(pkt)) {
*status_flags |= E1000_RXD_STAT_VP;
*vlan_tag = cpu_to_le16(net_rx_pkt_get_vlan_tag(pkt));
trace_e1000e_rx_metadata_vlan(*vlan_tag);
}
/* Packet parsing results */
if ((core->mac[RXCSUM] & E1000_RXCSUM_PCSD) != 0) {
if (rss_info->enabled) {
*rss = cpu_to_le32(rss_info->hash);
*mrq = cpu_to_le32(rss_info->type | (rss_info->queue << 8));
trace_e1000e_rx_metadata_rss(*rss, *mrq);
}
} else if (isip4) {
*status_flags |= E1000_RXD_STAT_IPIDV;
*ip_id = cpu_to_le16(net_rx_pkt_get_ip_id(pkt));
trace_e1000e_rx_metadata_ip_id(*ip_id);
}
if (istcp && e1000e_is_tcp_ack(core, pkt)) {
*status_flags |= E1000_RXD_STAT_ACK;
trace_e1000e_rx_metadata_ack();
}
if (isip6 && (core->mac[RFCTL] & E1000_RFCTL_IPV6_DIS)) {
trace_e1000e_rx_metadata_ipv6_filtering_disabled();
pkt_type = E1000_RXD_PKT_MAC;
} else if (istcp || isudp) {
pkt_type = isip4 ? E1000_RXD_PKT_IP4_XDP : E1000_RXD_PKT_IP6_XDP;
} else if (isip4 || isip6) {
pkt_type = isip4 ? E1000_RXD_PKT_IP4 : E1000_RXD_PKT_IP6;
} else {
pkt_type = E1000_RXD_PKT_MAC;
}
*status_flags |= E1000_RXD_PKT_TYPE(pkt_type);
trace_e1000e_rx_metadata_pkt_type(pkt_type);
/* RX CSO information */
if (isip6 && (core->mac[RFCTL] & E1000_RFCTL_IPV6_XSUM_DIS)) {
trace_e1000e_rx_metadata_ipv6_sum_disabled();
goto func_exit;
}
if (!net_rx_pkt_has_virt_hdr(pkt)) {
trace_e1000e_rx_metadata_no_virthdr();
e1000e_verify_csum_in_sw(core, pkt, status_flags, istcp, isudp);
goto func_exit;
}
vhdr = net_rx_pkt_get_vhdr(pkt);
if (!(vhdr->flags & VIRTIO_NET_HDR_F_DATA_VALID) &&
!(vhdr->flags & VIRTIO_NET_HDR_F_NEEDS_CSUM)) {
trace_e1000e_rx_metadata_virthdr_no_csum_info();
e1000e_verify_csum_in_sw(core, pkt, status_flags, istcp, isudp);
goto func_exit;
}
if (e1000e_rx_l3_cso_enabled(core)) {
*status_flags |= isip4 ? E1000_RXD_STAT_IPCS : 0;
} else {
trace_e1000e_rx_metadata_l3_cso_disabled();
}
if (e1000e_rx_l4_cso_enabled(core)) {
if (istcp) {
*status_flags |= E1000_RXD_STAT_TCPCS;
} else if (isudp) {
*status_flags |= E1000_RXD_STAT_TCPCS | E1000_RXD_STAT_UDPCS;
}
} else {
trace_e1000e_rx_metadata_l4_cso_disabled();
}
trace_e1000e_rx_metadata_status_flags(*status_flags);
func_exit:
*status_flags = cpu_to_le32(*status_flags);
}
static inline void
e1000e_write_lgcy_rx_descr(E1000ECore *core, uint8_t *desc,
struct NetRxPkt *pkt,
const E1000E_RSSInfo *rss_info,
uint16_t length)
{
uint32_t status_flags, rss, mrq;
uint16_t ip_id;
struct e1000_rx_desc *d = (struct e1000_rx_desc *) desc;
assert(!rss_info->enabled);
d->length = cpu_to_le16(length);
d->csum = 0;
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000e_build_rx_metadata(core, pkt, pkt != NULL,
rss_info,
&rss, &mrq,
&status_flags, &ip_id,
&d->special);
d->errors = (uint8_t) (le32_to_cpu(status_flags) >> 24);
d->status = (uint8_t) le32_to_cpu(status_flags);
}
static inline void
e1000e_write_ext_rx_descr(E1000ECore *core, uint8_t *desc,
struct NetRxPkt *pkt,
const E1000E_RSSInfo *rss_info,
uint16_t length)
{
union e1000_rx_desc_extended *d = (union e1000_rx_desc_extended *) desc;
memset(&d->wb, 0, sizeof(d->wb));
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
d->wb.upper.length = cpu_to_le16(length);
e1000e_build_rx_metadata(core, pkt, pkt != NULL,
rss_info,
&d->wb.lower.hi_dword.rss,
&d->wb.lower.mrq,
&d->wb.upper.status_error,
&d->wb.lower.hi_dword.csum_ip.ip_id,
&d->wb.upper.vlan);
}
static inline void
e1000e_write_ps_rx_descr(E1000ECore *core, uint8_t *desc,
struct NetRxPkt *pkt,
const E1000E_RSSInfo *rss_info,
size_t ps_hdr_len,
uint16_t(*written)[MAX_PS_BUFFERS])
{
int i;
union e1000_rx_desc_packet_split *d =
(union e1000_rx_desc_packet_split *) desc;
memset(&d->wb, 0, sizeof(d->wb));
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
d->wb.middle.length0 = cpu_to_le16((*written)[0]);
for (i = 0; i < PS_PAGE_BUFFERS; i++) {
d->wb.upper.length[i] = cpu_to_le16((*written)[i + 1]);
}
e1000e_build_rx_metadata(core, pkt, pkt != NULL,
rss_info,
&d->wb.lower.hi_dword.rss,
&d->wb.lower.mrq,
&d->wb.middle.status_error,
&d->wb.lower.hi_dword.csum_ip.ip_id,
&d->wb.middle.vlan);
d->wb.upper.header_status =
cpu_to_le16(ps_hdr_len | (ps_hdr_len ? E1000_RXDPS_HDRSTAT_HDRSP : 0));
trace_e1000e_rx_desc_ps_write((*written)[0], (*written)[1],
(*written)[2], (*written)[3]);
}
static inline void
e1000e_write_rx_descr(E1000ECore *core, uint8_t *desc,
struct NetRxPkt *pkt, const E1000E_RSSInfo *rss_info,
size_t ps_hdr_len, uint16_t(*written)[MAX_PS_BUFFERS])
{
if (e1000e_rx_use_legacy_descriptor(core)) {
assert(ps_hdr_len == 0);
e1000e_write_lgcy_rx_descr(core, desc, pkt, rss_info, (*written)[0]);
} else {
if (core->mac[RCTL] & E1000_RCTL_DTYP_PS) {
e1000e_write_ps_rx_descr(core, desc, pkt, rss_info,
ps_hdr_len, written);
} else {
assert(ps_hdr_len == 0);
e1000e_write_ext_rx_descr(core, desc, pkt, rss_info,
(*written)[0]);
}
}
}
typedef struct e1000e_ba_state_st {
uint16_t written[MAX_PS_BUFFERS];
uint8_t cur_idx;
} e1000e_ba_state;
static inline void
e1000e_write_hdr_to_rx_buffers(E1000ECore *core,
hwaddr (*ba)[MAX_PS_BUFFERS],
e1000e_ba_state *bastate,
const char *data,
dma_addr_t data_len)
{
assert(data_len <= core->rxbuf_sizes[0] - bastate->written[0]);
pci_dma_write(core->owner, (*ba)[0] + bastate->written[0], data, data_len);
bastate->written[0] += data_len;
bastate->cur_idx = 1;
}
static void
e1000e_write_to_rx_buffers(E1000ECore *core,
hwaddr (*ba)[MAX_PS_BUFFERS],
e1000e_ba_state *bastate,
const char *data,
dma_addr_t data_len)
{
while (data_len > 0) {
uint32_t cur_buf_len = core->rxbuf_sizes[bastate->cur_idx];
uint32_t cur_buf_bytes_left = cur_buf_len -
bastate->written[bastate->cur_idx];
uint32_t bytes_to_write = MIN(data_len, cur_buf_bytes_left);
trace_e1000e_rx_desc_buff_write(bastate->cur_idx,
(*ba)[bastate->cur_idx],
bastate->written[bastate->cur_idx],
data,
bytes_to_write);
pci_dma_write(core->owner,
(*ba)[bastate->cur_idx] + bastate->written[bastate->cur_idx],
data, bytes_to_write);
bastate->written[bastate->cur_idx] += bytes_to_write;
data += bytes_to_write;
data_len -= bytes_to_write;
if (bastate->written[bastate->cur_idx] == cur_buf_len) {
bastate->cur_idx++;
}
assert(bastate->cur_idx < MAX_PS_BUFFERS);
}
}
static void
e1000e_update_rx_stats(E1000ECore *core,
size_t data_size,
size_t data_fcs_size)
{
e1000x_update_rx_total_stats(core->mac, data_size, data_fcs_size);
switch (net_rx_pkt_get_packet_type(core->rx_pkt)) {
case ETH_PKT_BCAST:
e1000x_inc_reg_if_not_full(core->mac, BPRC);
break;
case ETH_PKT_MCAST:
e1000x_inc_reg_if_not_full(core->mac, MPRC);
break;
default:
break;
}
}
static inline bool
e1000e_rx_descr_threshold_hit(E1000ECore *core, const E1000E_RingInfo *rxi)
{
return e1000e_ring_free_descr_num(core, rxi) ==
e1000e_ring_len(core, rxi) >> core->rxbuf_min_shift;
}
static bool
e1000e_do_ps(E1000ECore *core, struct NetRxPkt *pkt, size_t *hdr_len)
{
bool isip4, isip6, isudp, istcp;
bool fragment;
if (!e1000e_rx_use_ps_descriptor(core)) {
return false;
}
net_rx_pkt_get_protocols(pkt, &isip4, &isip6, &isudp, &istcp);
if (isip4) {
fragment = net_rx_pkt_get_ip4_info(pkt)->fragment;
} else if (isip6) {
fragment = net_rx_pkt_get_ip6_info(pkt)->fragment;
} else {
return false;
}
if (fragment && (core->mac[RFCTL] & E1000_RFCTL_IPFRSP_DIS)) {
return false;
}
if (!fragment && (isudp || istcp)) {
*hdr_len = net_rx_pkt_get_l5_hdr_offset(pkt);
} else {
*hdr_len = net_rx_pkt_get_l4_hdr_offset(pkt);
}
if ((*hdr_len > core->rxbuf_sizes[0]) ||
(*hdr_len > net_rx_pkt_get_total_len(pkt))) {
return false;
}
return true;
}
static void
e1000e_write_packet_to_guest(E1000ECore *core, struct NetRxPkt *pkt,
const E1000E_RxRing *rxr,
const E1000E_RSSInfo *rss_info)
{
PCIDevice *d = core->owner;
dma_addr_t base;
uint8_t desc[E1000_MAX_RX_DESC_LEN];
size_t desc_size;
size_t desc_offset = 0;
size_t iov_ofs = 0;
struct iovec *iov = net_rx_pkt_get_iovec(pkt);
size_t size = net_rx_pkt_get_total_len(pkt);
size_t total_size = size + e1000x_fcs_len(core->mac);
const E1000E_RingInfo *rxi;
size_t ps_hdr_len = 0;
bool do_ps = e1000e_do_ps(core, pkt, &ps_hdr_len);
bool is_first = true;
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
rxi = rxr->i;
do {
hwaddr ba[MAX_PS_BUFFERS];
e1000e_ba_state bastate = { { 0 } };
bool is_last = false;
desc_size = total_size - desc_offset;
if (desc_size > core->rx_desc_buf_size) {
desc_size = core->rx_desc_buf_size;
}
if (e1000e_ring_empty(core, rxi)) {
return;
}
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
base = e1000e_ring_head_descr(core, rxi);
pci_dma_read(d, base, &desc, core->rx_desc_len);
trace_e1000e_rx_descr(rxi->idx, base, core->rx_desc_len);
e1000e_read_rx_descr(core, desc, &ba);
if (ba[0]) {
if (desc_offset < size) {
static const uint32_t fcs_pad;
size_t iov_copy;
size_t copy_size = size - desc_offset;
if (copy_size > core->rx_desc_buf_size) {
copy_size = core->rx_desc_buf_size;
}
/* For PS mode copy the packet header first */
if (do_ps) {
if (is_first) {
size_t ps_hdr_copied = 0;
do {
iov_copy = MIN(ps_hdr_len - ps_hdr_copied,
iov->iov_len - iov_ofs);
e1000e_write_hdr_to_rx_buffers(core, &ba, &bastate,
iov->iov_base, iov_copy);
copy_size -= iov_copy;
ps_hdr_copied += iov_copy;
iov_ofs += iov_copy;
if (iov_ofs == iov->iov_len) {
iov++;
iov_ofs = 0;
}
} while (ps_hdr_copied < ps_hdr_len);
is_first = false;
} else {
/* Leave buffer 0 of each descriptor except first */
/* empty as per spec 7.1.5.1 */
e1000e_write_hdr_to_rx_buffers(core, &ba, &bastate,
NULL, 0);
}
}
/* Copy packet payload */
while (copy_size) {
iov_copy = MIN(copy_size, iov->iov_len - iov_ofs);
e1000e_write_to_rx_buffers(core, &ba, &bastate,
iov->iov_base + iov_ofs, iov_copy);
copy_size -= iov_copy;
iov_ofs += iov_copy;
if (iov_ofs == iov->iov_len) {
iov++;
iov_ofs = 0;
}
}
if (desc_offset + desc_size >= total_size) {
/* Simulate FCS checksum presence in the last descriptor */
e1000e_write_to_rx_buffers(core, &ba, &bastate,
(const char *) &fcs_pad, e1000x_fcs_len(core->mac));
}
}
} else { /* as per intel docs; skip descriptors with null buf addr */
trace_e1000e_rx_null_descriptor();
}
desc_offset += desc_size;
if (desc_offset >= total_size) {
is_last = true;
}
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000e_write_rx_descr(core, desc, is_last ? core->rx_pkt : NULL,
rss_info, do_ps ? ps_hdr_len : 0, &bastate.written);
pci_dma_write(d, base, &desc, core->rx_desc_len);
e1000e_ring_advance(core, rxi,
core->rx_desc_len / E1000_MIN_RX_DESC_LEN);
} while (desc_offset < total_size);
e1000e_update_rx_stats(core, size, total_size);
}
static inline void
e1000e_rx_fix_l4_csum(E1000ECore *core, struct NetRxPkt *pkt)
{
if (net_rx_pkt_has_virt_hdr(pkt)) {
struct virtio_net_hdr *vhdr = net_rx_pkt_get_vhdr(pkt);
if (vhdr->flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
net_rx_pkt_fix_l4_csum(pkt);
}
}
}
ssize_t
e1000e_receive_iov(E1000ECore *core, const struct iovec *iov, int iovcnt)
{
static const int maximum_ethernet_hdr_len = (14 + 4);
/* Min. octets in an ethernet frame sans FCS */
static const int min_buf_size = 60;
uint32_t n = 0;
uint8_t min_buf[min_buf_size];
struct iovec min_iov;
uint8_t *filter_buf;
size_t size, orig_size;
size_t iov_ofs = 0;
E1000E_RxRing rxr;
E1000E_RSSInfo rss_info;
size_t total_size;
ssize_t retval;
bool rdmts_hit;
trace_e1000e_rx_receive_iov(iovcnt);
if (!e1000x_hw_rx_enabled(core->mac)) {
return -1;
}
/* Pull virtio header in */
if (core->has_vnet) {
net_rx_pkt_set_vhdr_iovec(core->rx_pkt, iov, iovcnt);
iov_ofs = sizeof(struct virtio_net_hdr);
}
filter_buf = iov->iov_base + iov_ofs;
orig_size = iov_size(iov, iovcnt);
size = orig_size - iov_ofs;
/* Pad to minimum Ethernet frame length */
if (size < sizeof(min_buf)) {
iov_to_buf(iov, iovcnt, iov_ofs, min_buf, size);
memset(&min_buf[size], 0, sizeof(min_buf) - size);
e1000x_inc_reg_if_not_full(core->mac, RUC);
min_iov.iov_base = filter_buf = min_buf;
min_iov.iov_len = size = sizeof(min_buf);
iovcnt = 1;
iov = &min_iov;
iov_ofs = 0;
} else if (iov->iov_len < maximum_ethernet_hdr_len) {
/* This is very unlikely, but may happen. */
iov_to_buf(iov, iovcnt, iov_ofs, min_buf, maximum_ethernet_hdr_len);
filter_buf = min_buf;
}
/* Discard oversized packets if !LPE and !SBP. */
if (e1000x_is_oversized(core->mac, size)) {
return orig_size;
}
net_rx_pkt_set_packet_type(core->rx_pkt,
get_eth_packet_type(PKT_GET_ETH_HDR(filter_buf)));
if (!e1000e_receive_filter(core, filter_buf, size)) {
trace_e1000e_rx_flt_dropped();
return orig_size;
}
net_rx_pkt_attach_iovec_ex(core->rx_pkt, iov, iovcnt, iov_ofs,
e1000x_vlan_enabled(core->mac), core->mac[VET]);
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000e_rss_parse_packet(core, core->rx_pkt, &rss_info);
e1000e_rx_ring_init(core, &rxr, rss_info.queue);
trace_e1000e_rx_rss_dispatched_to_queue(rxr.i->idx);
total_size = net_rx_pkt_get_total_len(core->rx_pkt) +
e1000x_fcs_len(core->mac);
if (e1000e_has_rxbufs(core, rxr.i, total_size)) {
e1000e_rx_fix_l4_csum(core, core->rx_pkt);
e1000e_write_packet_to_guest(core, core->rx_pkt, &rxr, &rss_info);
retval = orig_size;
/* Perform small receive detection (RSRPD) */
if (total_size < core->mac[RSRPD]) {
n |= E1000_ICS_SRPD;
}
/* Perform ACK receive detection */
if (!(core->mac[RFCTL] & E1000_RFCTL_ACK_DIS) &&
(e1000e_is_tcp_ack(core, core->rx_pkt))) {
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
n |= E1000_ICS_ACK;
}
/* Check if receive descriptor minimum threshold hit */
rdmts_hit = e1000e_rx_descr_threshold_hit(core, rxr.i);
n |= e1000e_rx_wb_interrupt_cause(core, rxr.i->idx, rdmts_hit);
trace_e1000e_rx_written_to_guest(n);
} else {
n |= E1000_ICS_RXO;
retval = 0;
trace_e1000e_rx_not_written_to_guest(n);
}
if (!e1000e_intrmgr_delay_rx_causes(core, &n)) {
trace_e1000e_rx_interrupt_set(n);
e1000e_set_interrupt_cause(core, n);
} else {
trace_e1000e_rx_interrupt_delayed(n);
}
return retval;
}
static inline bool
e1000e_have_autoneg(E1000ECore *core)
{
return core->phy[0][PHY_CTRL] & MII_CR_AUTO_NEG_EN;
}
static void e1000e_update_flowctl_status(E1000ECore *core)
{
if (e1000e_have_autoneg(core) &&
core->phy[0][PHY_STATUS] & MII_SR_AUTONEG_COMPLETE) {
trace_e1000e_link_autoneg_flowctl(true);
core->mac[CTRL] |= E1000_CTRL_TFCE | E1000_CTRL_RFCE;
} else {
trace_e1000e_link_autoneg_flowctl(false);
}
}
static inline void
e1000e_link_down(E1000ECore *core)
{
e1000x_update_regs_on_link_down(core->mac, core->phy[0]);
e1000e_update_flowctl_status(core);
}
static inline void
e1000e_set_phy_ctrl(E1000ECore *core, int index, uint16_t val)
{
/* bits 0-5 reserved; MII_CR_[RESTART_AUTO_NEG,RESET] are self clearing */
core->phy[0][PHY_CTRL] = val & ~(0x3f |
MII_CR_RESET |
MII_CR_RESTART_AUTO_NEG);
if ((val & MII_CR_RESTART_AUTO_NEG) &&
e1000e_have_autoneg(core)) {
e1000x_restart_autoneg(core->mac, core->phy[0], core->autoneg_timer);
}
}
static void
e1000e_set_phy_oem_bits(E1000ECore *core, int index, uint16_t val)
{
core->phy[0][PHY_OEM_BITS] = val & ~BIT(10);
if (val & BIT(10)) {
e1000x_restart_autoneg(core->mac, core->phy[0], core->autoneg_timer);
}
}
static void
e1000e_set_phy_page(E1000ECore *core, int index, uint16_t val)
{
core->phy[0][PHY_PAGE] = val & PHY_PAGE_RW_MASK;
}
void
e1000e_core_set_link_status(E1000ECore *core)
{
NetClientState *nc = qemu_get_queue(core->owner_nic);
uint32_t old_status = core->mac[STATUS];
trace_e1000e_link_status_changed(nc->link_down ? false : true);
if (nc->link_down) {
e1000x_update_regs_on_link_down(core->mac, core->phy[0]);
} else {
if (e1000e_have_autoneg(core) &&
!(core->phy[0][PHY_STATUS] & MII_SR_AUTONEG_COMPLETE)) {
e1000x_restart_autoneg(core->mac, core->phy[0],
core->autoneg_timer);
} else {
e1000x_update_regs_on_link_up(core->mac, core->phy[0]);
e1000e_start_recv(core);
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
}
}
if (core->mac[STATUS] != old_status) {
e1000e_set_interrupt_cause(core, E1000_ICR_LSC);
}
}
static void
e1000e_set_ctrl(E1000ECore *core, int index, uint32_t val)
{
trace_e1000e_core_ctrl_write(index, val);
/* RST is self clearing */
core->mac[CTRL] = val & ~E1000_CTRL_RST;
core->mac[CTRL_DUP] = core->mac[CTRL];
trace_e1000e_link_set_params(
!!(val & E1000_CTRL_ASDE),
(val & E1000_CTRL_SPD_SEL) >> E1000_CTRL_SPD_SHIFT,
!!(val & E1000_CTRL_FRCSPD),
!!(val & E1000_CTRL_FRCDPX),
!!(val & E1000_CTRL_RFCE),
!!(val & E1000_CTRL_TFCE));
if (val & E1000_CTRL_RST) {
trace_e1000e_core_ctrl_sw_reset();
e1000x_reset_mac_addr(core->owner_nic, core->mac, core->permanent_mac);
}
if (val & E1000_CTRL_PHY_RST) {
trace_e1000e_core_ctrl_phy_reset();
core->mac[STATUS] |= E1000_STATUS_PHYRA;
}
}
static void
e1000e_set_rfctl(E1000ECore *core, int index, uint32_t val)
{
trace_e1000e_rx_set_rfctl(val);
if (!(val & E1000_RFCTL_ISCSI_DIS)) {
trace_e1000e_wrn_iscsi_filtering_not_supported();
}
if (!(val & E1000_RFCTL_NFSW_DIS)) {
trace_e1000e_wrn_nfsw_filtering_not_supported();
}
if (!(val & E1000_RFCTL_NFSR_DIS)) {
trace_e1000e_wrn_nfsr_filtering_not_supported();
}
core->mac[RFCTL] = val;
}
static void
e1000e_calc_per_desc_buf_size(E1000ECore *core)
{
int i;
core->rx_desc_buf_size = 0;
for (i = 0; i < ARRAY_SIZE(core->rxbuf_sizes); i++) {
core->rx_desc_buf_size += core->rxbuf_sizes[i];
}
}
static void
e1000e_parse_rxbufsize(E1000ECore *core)
{
uint32_t rctl = core->mac[RCTL];
memset(core->rxbuf_sizes, 0, sizeof(core->rxbuf_sizes));
if (rctl & E1000_RCTL_DTYP_MASK) {
uint32_t bsize;
bsize = core->mac[PSRCTL] & E1000_PSRCTL_BSIZE0_MASK;
core->rxbuf_sizes[0] = (bsize >> E1000_PSRCTL_BSIZE0_SHIFT) * 128;
bsize = core->mac[PSRCTL] & E1000_PSRCTL_BSIZE1_MASK;
core->rxbuf_sizes[1] = (bsize >> E1000_PSRCTL_BSIZE1_SHIFT) * 1024;
bsize = core->mac[PSRCTL] & E1000_PSRCTL_BSIZE2_MASK;
core->rxbuf_sizes[2] = (bsize >> E1000_PSRCTL_BSIZE2_SHIFT) * 1024;
bsize = core->mac[PSRCTL] & E1000_PSRCTL_BSIZE3_MASK;
core->rxbuf_sizes[3] = (bsize >> E1000_PSRCTL_BSIZE3_SHIFT) * 1024;
} else if (rctl & E1000_RCTL_FLXBUF_MASK) {
int flxbuf = rctl & E1000_RCTL_FLXBUF_MASK;
core->rxbuf_sizes[0] = (flxbuf >> E1000_RCTL_FLXBUF_SHIFT) * 1024;
} else {
core->rxbuf_sizes[0] = e1000x_rxbufsize(rctl);
}
trace_e1000e_rx_desc_buff_sizes(core->rxbuf_sizes[0], core->rxbuf_sizes[1],
core->rxbuf_sizes[2], core->rxbuf_sizes[3]);
e1000e_calc_per_desc_buf_size(core);
}
static void
e1000e_calc_rxdesclen(E1000ECore *core)
{
if (e1000e_rx_use_legacy_descriptor(core)) {
core->rx_desc_len = sizeof(struct e1000_rx_desc);
} else {
if (core->mac[RCTL] & E1000_RCTL_DTYP_PS) {
core->rx_desc_len = sizeof(union e1000_rx_desc_packet_split);
} else {
core->rx_desc_len = sizeof(union e1000_rx_desc_extended);
}
}
trace_e1000e_rx_desc_len(core->rx_desc_len);
}
static void
e1000e_set_rx_control(E1000ECore *core, int index, uint32_t val)
{
core->mac[RCTL] = val;
trace_e1000e_rx_set_rctl(core->mac[RCTL]);
if (val & E1000_RCTL_EN) {
e1000e_parse_rxbufsize(core);
e1000e_calc_rxdesclen(core);
core->rxbuf_min_shift = ((val / E1000_RCTL_RDMTS_QUAT) & 3) + 1 +
E1000_RING_DESC_LEN_SHIFT;
e1000e_start_recv(core);
}
}
static
void(*e1000e_phyreg_writeops[E1000E_PHY_PAGES][E1000E_PHY_PAGE_SIZE])
(E1000ECore *, int, uint16_t) = {
[0] = {
[PHY_CTRL] = e1000e_set_phy_ctrl,
[PHY_PAGE] = e1000e_set_phy_page,
[PHY_OEM_BITS] = e1000e_set_phy_oem_bits
}
};
static inline void
e1000e_clear_ims_bits(E1000ECore *core, uint32_t bits)
{
trace_e1000e_irq_clear_ims(bits, core->mac[IMS], core->mac[IMS] & ~bits);
core->mac[IMS] &= ~bits;
}
static inline bool
e1000e_postpone_interrupt(bool *interrupt_pending,
E1000IntrDelayTimer *timer)
{
if (timer->running) {
trace_e1000e_irq_postponed_by_xitr(timer->delay_reg << 2);
*interrupt_pending = true;
return true;
}
if (timer->core->mac[timer->delay_reg] != 0) {
e1000e_intrmgr_rearm_timer(timer);
}
return false;
}
static inline bool
e1000e_itr_should_postpone(E1000ECore *core)
{
return e1000e_postpone_interrupt(&core->itr_intr_pending, &core->itr);
}
static inline bool
e1000e_eitr_should_postpone(E1000ECore *core, int idx)
{
return e1000e_postpone_interrupt(&core->eitr_intr_pending[idx],
&core->eitr[idx]);
}
static void
e1000e_msix_notify_one(E1000ECore *core, uint32_t cause, uint32_t int_cfg)
{
uint32_t effective_eiac;
if (E1000_IVAR_ENTRY_VALID(int_cfg)) {
uint32_t vec = E1000_IVAR_ENTRY_VEC(int_cfg);
if (vec < E1000E_MSIX_VEC_NUM) {
if (!e1000e_eitr_should_postpone(core, vec)) {
trace_e1000e_irq_msix_notify_vec(vec);
msix_notify(core->owner, vec);
}
} else {
trace_e1000e_wrn_msix_vec_wrong(cause, int_cfg);
}
} else {
trace_e1000e_wrn_msix_invalid(cause, int_cfg);
}
if (core->mac[CTRL_EXT] & E1000_CTRL_EXT_EIAME) {
trace_e1000e_irq_iam_clear_eiame(core->mac[IAM], cause);
core->mac[IAM] &= ~cause;
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
}
trace_e1000e_irq_icr_clear_eiac(core->mac[ICR], core->mac[EIAC]);
effective_eiac = core->mac[EIAC] & cause;
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
core->mac[ICR] &= ~effective_eiac;
core->msi_causes_pending &= ~effective_eiac;
if (!(core->mac[CTRL_EXT] & E1000_CTRL_EXT_IAME)) {
core->mac[IMS] &= ~effective_eiac;
}
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
}
static void
e1000e_msix_notify(E1000ECore *core, uint32_t causes)
{
if (causes & E1000_ICR_RXQ0) {
e1000e_msix_notify_one(core, E1000_ICR_RXQ0,
E1000_IVAR_RXQ0(core->mac[IVAR]));
}
if (causes & E1000_ICR_RXQ1) {
e1000e_msix_notify_one(core, E1000_ICR_RXQ1,
E1000_IVAR_RXQ1(core->mac[IVAR]));
}
if (causes & E1000_ICR_TXQ0) {
e1000e_msix_notify_one(core, E1000_ICR_TXQ0,
E1000_IVAR_TXQ0(core->mac[IVAR]));
}
if (causes & E1000_ICR_TXQ1) {
e1000e_msix_notify_one(core, E1000_ICR_TXQ1,
E1000_IVAR_TXQ1(core->mac[IVAR]));
}
if (causes & E1000_ICR_OTHER) {
e1000e_msix_notify_one(core, E1000_ICR_OTHER,
E1000_IVAR_OTHER(core->mac[IVAR]));
}
}
static void
e1000e_msix_clear_one(E1000ECore *core, uint32_t cause, uint32_t int_cfg)
{
if (E1000_IVAR_ENTRY_VALID(int_cfg)) {
uint32_t vec = E1000_IVAR_ENTRY_VEC(int_cfg);
if (vec < E1000E_MSIX_VEC_NUM) {
trace_e1000e_irq_msix_pending_clearing(cause, int_cfg, vec);
msix_clr_pending(core->owner, vec);
} else {
trace_e1000e_wrn_msix_vec_wrong(cause, int_cfg);
}
} else {
trace_e1000e_wrn_msix_invalid(cause, int_cfg);
}
}
static void
e1000e_msix_clear(E1000ECore *core, uint32_t causes)
{
if (causes & E1000_ICR_RXQ0) {
e1000e_msix_clear_one(core, E1000_ICR_RXQ0,
E1000_IVAR_RXQ0(core->mac[IVAR]));
}
if (causes & E1000_ICR_RXQ1) {
e1000e_msix_clear_one(core, E1000_ICR_RXQ1,
E1000_IVAR_RXQ1(core->mac[IVAR]));
}
if (causes & E1000_ICR_TXQ0) {
e1000e_msix_clear_one(core, E1000_ICR_TXQ0,
E1000_IVAR_TXQ0(core->mac[IVAR]));
}
if (causes & E1000_ICR_TXQ1) {
e1000e_msix_clear_one(core, E1000_ICR_TXQ1,
E1000_IVAR_TXQ1(core->mac[IVAR]));
}
if (causes & E1000_ICR_OTHER) {
e1000e_msix_clear_one(core, E1000_ICR_OTHER,
E1000_IVAR_OTHER(core->mac[IVAR]));
}
}
static inline void
e1000e_fix_icr_asserted(E1000ECore *core)
{
core->mac[ICR] &= ~E1000_ICR_ASSERTED;
if (core->mac[ICR]) {
core->mac[ICR] |= E1000_ICR_ASSERTED;
}
trace_e1000e_irq_fix_icr_asserted(core->mac[ICR]);
}
static void
e1000e_send_msi(E1000ECore *core, bool msix)
{
uint32_t causes = core->mac[ICR] & core->mac[IMS] & ~E1000_ICR_ASSERTED;
core->msi_causes_pending &= causes;
causes ^= core->msi_causes_pending;
if (causes == 0) {
return;
}
core->msi_causes_pending |= causes;
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
if (msix) {
e1000e_msix_notify(core, causes);
} else {
if (!e1000e_itr_should_postpone(core)) {
trace_e1000e_irq_msi_notify(causes);
msi_notify(core->owner, 0);
}
}
}
static void
e1000e_update_interrupt_state(E1000ECore *core)
{
bool interrupts_pending;
bool is_msix = msix_enabled(core->owner);
/* Set ICR[OTHER] for MSI-X */
if (is_msix) {
if (core->mac[ICR] & E1000_ICR_OTHER_CAUSES) {
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
core->mac[ICR] |= E1000_ICR_OTHER;
trace_e1000e_irq_add_msi_other(core->mac[ICR]);
}
}
e1000e_fix_icr_asserted(core);
/*
* Make sure ICR and ICS registers have the same value.
* The spec says that the ICS register is write-only. However in practice,
* on real hardware ICS is readable, and for reads it has the same value as
* ICR (except that ICS does not have the clear on read behaviour of ICR).
*
* The VxWorks PRO/1000 driver uses this behaviour.
*/
core->mac[ICS] = core->mac[ICR];
interrupts_pending = (core->mac[IMS] & core->mac[ICR]) ? true : false;
if (!interrupts_pending) {
core->msi_causes_pending = 0;
}
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
trace_e1000e_irq_pending_interrupts(core->mac[ICR] & core->mac[IMS],
core->mac[ICR], core->mac[IMS]);
if (is_msix || msi_enabled(core->owner)) {
if (interrupts_pending) {
e1000e_send_msi(core, is_msix);
}
} else {
if (interrupts_pending) {
if (!e1000e_itr_should_postpone(core)) {
e1000e_raise_legacy_irq(core);
}
} else {
e1000e_lower_legacy_irq(core);
}
}
}
static void
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000e_set_interrupt_cause(E1000ECore *core, uint32_t val)
{
trace_e1000e_irq_set_cause_entry(val, core->mac[ICR]);
val |= e1000e_intmgr_collect_delayed_causes(core);
core->mac[ICR] |= val;
trace_e1000e_irq_set_cause_exit(val, core->mac[ICR]);
e1000e_update_interrupt_state(core);
}
static inline void
e1000e_autoneg_timer(void *opaque)
{
E1000ECore *core = opaque;
if (!qemu_get_queue(core->owner_nic)->link_down) {
e1000x_update_regs_on_autoneg_done(core->mac, core->phy[0]);
e1000e_start_recv(core);
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000e_update_flowctl_status(core);
/* signal link status change to the guest */
e1000e_set_interrupt_cause(core, E1000_ICR_LSC);
}
}
static inline uint16_t
e1000e_get_reg_index_with_offset(const uint16_t *mac_reg_access, hwaddr addr)
{
uint16_t index = (addr & 0x1ffff) >> 2;
return index + (mac_reg_access[index] & 0xfffe);
}
static const char e1000e_phy_regcap[E1000E_PHY_PAGES][0x20] = {
[0] = {
[PHY_CTRL] = PHY_ANYPAGE | PHY_RW,
[PHY_STATUS] = PHY_ANYPAGE | PHY_R,
[PHY_ID1] = PHY_ANYPAGE | PHY_R,
[PHY_ID2] = PHY_ANYPAGE | PHY_R,
[PHY_AUTONEG_ADV] = PHY_ANYPAGE | PHY_RW,
[PHY_LP_ABILITY] = PHY_ANYPAGE | PHY_R,
[PHY_AUTONEG_EXP] = PHY_ANYPAGE | PHY_R,
[PHY_NEXT_PAGE_TX] = PHY_ANYPAGE | PHY_RW,
[PHY_LP_NEXT_PAGE] = PHY_ANYPAGE | PHY_R,
[PHY_1000T_CTRL] = PHY_ANYPAGE | PHY_RW,
[PHY_1000T_STATUS] = PHY_ANYPAGE | PHY_R,
[PHY_EXT_STATUS] = PHY_ANYPAGE | PHY_R,
[PHY_PAGE] = PHY_ANYPAGE | PHY_RW,
[PHY_COPPER_CTRL1] = PHY_RW,
[PHY_COPPER_STAT1] = PHY_R,
[PHY_COPPER_CTRL3] = PHY_RW,
[PHY_RX_ERR_CNTR] = PHY_R,
[PHY_OEM_BITS] = PHY_RW,
[PHY_BIAS_1] = PHY_RW,
[PHY_BIAS_2] = PHY_RW,
[PHY_COPPER_INT_ENABLE] = PHY_RW,
[PHY_COPPER_STAT2] = PHY_R,
[PHY_COPPER_CTRL2] = PHY_RW
},
[2] = {
[PHY_MAC_CTRL1] = PHY_RW,
[PHY_MAC_INT_ENABLE] = PHY_RW,
[PHY_MAC_STAT] = PHY_R,
[PHY_MAC_CTRL2] = PHY_RW
},
[3] = {
[PHY_LED_03_FUNC_CTRL1] = PHY_RW,
[PHY_LED_03_POL_CTRL] = PHY_RW,
[PHY_LED_TIMER_CTRL] = PHY_RW,
[PHY_LED_45_CTRL] = PHY_RW
},
[5] = {
[PHY_1000T_SKEW] = PHY_R,
[PHY_1000T_SWAP] = PHY_R
},
[6] = {
[PHY_CRC_COUNTERS] = PHY_R
}
};
static bool
e1000e_phy_reg_check_cap(E1000ECore *core, uint32_t addr,
char cap, uint8_t *page)
{
*page =
(e1000e_phy_regcap[0][addr] & PHY_ANYPAGE) ? 0
: core->phy[0][PHY_PAGE];
if (*page >= E1000E_PHY_PAGES) {
return false;
}
return e1000e_phy_regcap[*page][addr] & cap;
}
static void
e1000e_phy_reg_write(E1000ECore *core, uint8_t page,
uint32_t addr, uint16_t data)
{
assert(page < E1000E_PHY_PAGES);
assert(addr < E1000E_PHY_PAGE_SIZE);
if (e1000e_phyreg_writeops[page][addr]) {
e1000e_phyreg_writeops[page][addr](core, addr, data);
} else {
core->phy[page][addr] = data;
}
}
static void
e1000e_set_mdic(E1000ECore *core, int index, uint32_t val)
{
uint32_t data = val & E1000_MDIC_DATA_MASK;
uint32_t addr = ((val & E1000_MDIC_REG_MASK) >> E1000_MDIC_REG_SHIFT);
uint8_t page;
if ((val & E1000_MDIC_PHY_MASK) >> E1000_MDIC_PHY_SHIFT != 1) { /* phy # */
val = core->mac[MDIC] | E1000_MDIC_ERROR;
} else if (val & E1000_MDIC_OP_READ) {
if (!e1000e_phy_reg_check_cap(core, addr, PHY_R, &page)) {
trace_e1000e_core_mdic_read_unhandled(page, addr);
val |= E1000_MDIC_ERROR;
} else {
val = (val ^ data) | core->phy[page][addr];
trace_e1000e_core_mdic_read(page, addr, val);
}
} else if (val & E1000_MDIC_OP_WRITE) {
if (!e1000e_phy_reg_check_cap(core, addr, PHY_W, &page)) {
trace_e1000e_core_mdic_write_unhandled(page, addr);
val |= E1000_MDIC_ERROR;
} else {
trace_e1000e_core_mdic_write(page, addr, data);
e1000e_phy_reg_write(core, page, addr, data);
}
}
core->mac[MDIC] = val | E1000_MDIC_READY;
if (val & E1000_MDIC_INT_EN) {
e1000e_set_interrupt_cause(core, E1000_ICR_MDAC);
}
}
static void
e1000e_set_rdt(E1000ECore *core, int index, uint32_t val)
{
core->mac[index] = val & 0xffff;
trace_e1000e_rx_set_rdt(e1000e_mq_queue_idx(RDT0, index), val);
e1000e_start_recv(core);
}
static void
e1000e_set_status(E1000ECore *core, int index, uint32_t val)
{
if ((val & E1000_STATUS_PHYRA) == 0) {
core->mac[index] &= ~E1000_STATUS_PHYRA;
}
}
static void
e1000e_set_ctrlext(E1000ECore *core, int index, uint32_t val)
{
trace_e1000e_link_set_ext_params(!!(val & E1000_CTRL_EXT_ASDCHK),
!!(val & E1000_CTRL_EXT_SPD_BYPS));
/* Zero self-clearing bits */
val &= ~(E1000_CTRL_EXT_ASDCHK | E1000_CTRL_EXT_EE_RST);
core->mac[CTRL_EXT] = val;
}
static void
e1000e_set_pbaclr(E1000ECore *core, int index, uint32_t val)
{
int i;
core->mac[PBACLR] = val & E1000_PBACLR_VALID_MASK;
if (!msix_enabled(core->owner)) {
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
return;
}
for (i = 0; i < E1000E_MSIX_VEC_NUM; i++) {
if (core->mac[PBACLR] & BIT(i)) {
msix_clr_pending(core->owner, i);
}
}
}
static void
e1000e_set_fcrth(E1000ECore *core, int index, uint32_t val)
{
core->mac[FCRTH] = val & 0xFFF8;
}
static void
e1000e_set_fcrtl(E1000ECore *core, int index, uint32_t val)
{
core->mac[FCRTL] = val & 0x8000FFF8;
}
static inline void
e1000e_set_16bit(E1000ECore *core, int index, uint32_t val)
{
core->mac[index] = val & 0xffff;
}
static void
e1000e_set_12bit(E1000ECore *core, int index, uint32_t val)
{
core->mac[index] = val & 0xfff;
}
static void
e1000e_set_vet(E1000ECore *core, int index, uint32_t val)
{
core->mac[VET] = val & 0xffff;
trace_e1000e_vlan_vet(core->mac[VET]);
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
}
static void
e1000e_set_dlen(E1000ECore *core, int index, uint32_t val)
{
core->mac[index] = val & E1000_XDLEN_MASK;
}
static void
e1000e_set_dbal(E1000ECore *core, int index, uint32_t val)
{
core->mac[index] = val & E1000_XDBAL_MASK;
}
static void
e1000e_set_tctl(E1000ECore *core, int index, uint32_t val)
{
E1000E_TxRing txr;
core->mac[index] = val;
if (core->mac[TARC0] & E1000_TARC_ENABLE) {
e1000e_tx_ring_init(core, &txr, 0);
e1000e_start_xmit(core, &txr);
}
if (core->mac[TARC1] & E1000_TARC_ENABLE) {
e1000e_tx_ring_init(core, &txr, 1);
e1000e_start_xmit(core, &txr);
}
}
static void
e1000e_set_tdt(E1000ECore *core, int index, uint32_t val)
{
E1000E_TxRing txr;
int qidx = e1000e_mq_queue_idx(TDT, index);
uint32_t tarc_reg = (qidx == 0) ? TARC0 : TARC1;
core->mac[index] = val & 0xffff;
if (core->mac[tarc_reg] & E1000_TARC_ENABLE) {
e1000e_tx_ring_init(core, &txr, qidx);
e1000e_start_xmit(core, &txr);
}
}
static void
e1000e_set_ics(E1000ECore *core, int index, uint32_t val)
{
trace_e1000e_irq_write_ics(val);
e1000e_set_interrupt_cause(core, val);
}
static void
e1000e_set_icr(E1000ECore *core, int index, uint32_t val)
{
e1000e: Fix ICR "Other" causes clear logic This commit fixes a bug which causes the guest to hang. The bug was observed upon a "receive overrun" (bit #6 of the ICR register) interrupt which could be triggered post migration in a heavy traffic environment. Even though the "receive overrun" bit (#6) is masked out by the IMS register (refer to the log below) the driver still receives an interrupt as the "receive overrun" bit (#6) causes the "Other" - bit #24 of the ICR register - bit to be set as documented below. The driver handles the interrupt and clears the "Other" bit (#24) but doesn't clear the "receive overrun" bit (#6) which leads to an infinite loop. Apparently the Windows driver expects that the "receive overrun" bit and other ones - documented below - to be cleared when the "Other" bit (#24) is cleared. So to sum that up: 1. Bit #6 of the ICR register is set by heavy traffic 2. As a results of setting bit #6, bit #24 is set 3. The driver receives an interrupt for bit 24 (it doesn't receieve an interrupt for bit #6 as it is masked out by IMS) 4. The driver handles and clears the interrupt of bit #24 5. Bit #6 is still set. 6. 2 happens all over again The Interrupt Cause Read - ICR register: The ICR has the "Other" bit - bit #24 - that is set when one or more of the following ICR register's bits are set: LSC - bit #2, RXO - bit #6, MDAC - bit #9, SRPD - bit #16, ACK - bit #17, MNG - bit #18 This bug can occur with any of these bits depending on the driver's behaviour and the way it configures the device. However, trying to reproduce it with any bit other than RX0 is challenging and came to failure as the drivers don't implement most of these bits, trying to reproduce it with LSC (Link Status Change - bit #2) bit didn't succeed too as it seems that Windows handles this bit differently. Log sample of the storm: 27563@1494850819.411877:e1000e_irq_pending_interrupts ICR PENDING: 0x1000000 (ICR: 0x815000c2, IMS: 0x1a00004) 27563@1494850819.411900:e1000e_irq_pending_interrupts ICR PENDING: 0x0 (ICR: 0x815000c2, IMS: 0xa00004) 27563@1494850819.411915:e1000e_irq_pending_interrupts ICR PENDING: 0x0 (ICR: 0x815000c2, IMS: 0xa00004) 27563@1494850819.412380:e1000e_irq_pending_interrupts ICR PENDING: 0x0 (ICR: 0x815000c2, IMS: 0xa00004) 27563@1494850819.412395:e1000e_irq_pending_interrupts ICR PENDING: 0x0 (ICR: 0x815000c2, IMS: 0xa00004) 27563@1494850819.412436:e1000e_irq_pending_interrupts ICR PENDING: 0x0 (ICR: 0x815000c2, IMS: 0xa00004) 27563@1494850819.412441:e1000e_irq_pending_interrupts ICR PENDING: 0x0 (ICR: 0x815000c2, IMS: 0xa00004) 27563@1494850819.412998:e1000e_irq_pending_interrupts ICR PENDING: 0x1000000 (ICR: 0x815000c2, IMS: 0x1a00004) * This bug behaviour wasn't observed with the Linux driver. This commit solves: https://bugzilla.redhat.com/show_bug.cgi?id=1447935 https://bugzilla.redhat.com/show_bug.cgi?id=1449490 Cc: qemu-stable@nongnu.org Signed-off-by: Sameeh Jubran <sjubran@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2017-05-22 13:26:22 +02:00
uint32_t icr = 0;
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
if ((core->mac[ICR] & E1000_ICR_ASSERTED) &&
(core->mac[CTRL_EXT] & E1000_CTRL_EXT_IAME)) {
trace_e1000e_irq_icr_process_iame();
e1000e_clear_ims_bits(core, core->mac[IAM]);
}
e1000e: Fix ICR "Other" causes clear logic This commit fixes a bug which causes the guest to hang. The bug was observed upon a "receive overrun" (bit #6 of the ICR register) interrupt which could be triggered post migration in a heavy traffic environment. Even though the "receive overrun" bit (#6) is masked out by the IMS register (refer to the log below) the driver still receives an interrupt as the "receive overrun" bit (#6) causes the "Other" - bit #24 of the ICR register - bit to be set as documented below. The driver handles the interrupt and clears the "Other" bit (#24) but doesn't clear the "receive overrun" bit (#6) which leads to an infinite loop. Apparently the Windows driver expects that the "receive overrun" bit and other ones - documented below - to be cleared when the "Other" bit (#24) is cleared. So to sum that up: 1. Bit #6 of the ICR register is set by heavy traffic 2. As a results of setting bit #6, bit #24 is set 3. The driver receives an interrupt for bit 24 (it doesn't receieve an interrupt for bit #6 as it is masked out by IMS) 4. The driver handles and clears the interrupt of bit #24 5. Bit #6 is still set. 6. 2 happens all over again The Interrupt Cause Read - ICR register: The ICR has the "Other" bit - bit #24 - that is set when one or more of the following ICR register's bits are set: LSC - bit #2, RXO - bit #6, MDAC - bit #9, SRPD - bit #16, ACK - bit #17, MNG - bit #18 This bug can occur with any of these bits depending on the driver's behaviour and the way it configures the device. However, trying to reproduce it with any bit other than RX0 is challenging and came to failure as the drivers don't implement most of these bits, trying to reproduce it with LSC (Link Status Change - bit #2) bit didn't succeed too as it seems that Windows handles this bit differently. Log sample of the storm: 27563@1494850819.411877:e1000e_irq_pending_interrupts ICR PENDING: 0x1000000 (ICR: 0x815000c2, IMS: 0x1a00004) 27563@1494850819.411900:e1000e_irq_pending_interrupts ICR PENDING: 0x0 (ICR: 0x815000c2, IMS: 0xa00004) 27563@1494850819.411915:e1000e_irq_pending_interrupts ICR PENDING: 0x0 (ICR: 0x815000c2, IMS: 0xa00004) 27563@1494850819.412380:e1000e_irq_pending_interrupts ICR PENDING: 0x0 (ICR: 0x815000c2, IMS: 0xa00004) 27563@1494850819.412395:e1000e_irq_pending_interrupts ICR PENDING: 0x0 (ICR: 0x815000c2, IMS: 0xa00004) 27563@1494850819.412436:e1000e_irq_pending_interrupts ICR PENDING: 0x0 (ICR: 0x815000c2, IMS: 0xa00004) 27563@1494850819.412441:e1000e_irq_pending_interrupts ICR PENDING: 0x0 (ICR: 0x815000c2, IMS: 0xa00004) 27563@1494850819.412998:e1000e_irq_pending_interrupts ICR PENDING: 0x1000000 (ICR: 0x815000c2, IMS: 0x1a00004) * This bug behaviour wasn't observed with the Linux driver. This commit solves: https://bugzilla.redhat.com/show_bug.cgi?id=1447935 https://bugzilla.redhat.com/show_bug.cgi?id=1449490 Cc: qemu-stable@nongnu.org Signed-off-by: Sameeh Jubran <sjubran@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2017-05-22 13:26:22 +02:00
icr = core->mac[ICR] & ~val;
/* Windows driver expects that the "receive overrun" bit and other
* ones to be cleared when the "Other" bit (#24) is cleared.
*/
icr = (val & E1000_ICR_OTHER) ? (icr & ~E1000_ICR_OTHER_CAUSES) : icr;
trace_e1000e_irq_icr_write(val, core->mac[ICR], icr);
core->mac[ICR] = icr;
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000e_update_interrupt_state(core);
}
static void
e1000e_set_imc(E1000ECore *core, int index, uint32_t val)
{
trace_e1000e_irq_ims_clear_set_imc(val);
e1000e_clear_ims_bits(core, val);
e1000e_update_interrupt_state(core);
}
static void
e1000e_set_ims(E1000ECore *core, int index, uint32_t val)
{
static const uint32_t ims_ext_mask =
E1000_IMS_RXQ0 | E1000_IMS_RXQ1 |
E1000_IMS_TXQ0 | E1000_IMS_TXQ1 |
E1000_IMS_OTHER;
static const uint32_t ims_valid_mask =
E1000_IMS_TXDW | E1000_IMS_TXQE | E1000_IMS_LSC |
E1000_IMS_RXDMT0 | E1000_IMS_RXO | E1000_IMS_RXT0 |
E1000_IMS_MDAC | E1000_IMS_TXD_LOW | E1000_IMS_SRPD |
E1000_IMS_ACK | E1000_IMS_MNG | E1000_IMS_RXQ0 |
E1000_IMS_RXQ1 | E1000_IMS_TXQ0 | E1000_IMS_TXQ1 |
E1000_IMS_OTHER;
uint32_t valid_val = val & ims_valid_mask;
trace_e1000e_irq_set_ims(val, core->mac[IMS], core->mac[IMS] | valid_val);
core->mac[IMS] |= valid_val;
if ((valid_val & ims_ext_mask) &&
(core->mac[CTRL_EXT] & E1000_CTRL_EXT_PBA_CLR) &&
msix_enabled(core->owner)) {
e1000e_msix_clear(core, valid_val);
}
if ((valid_val == ims_valid_mask) &&
(core->mac[CTRL_EXT] & E1000_CTRL_EXT_INT_TIMERS_CLEAR_ENA)) {
trace_e1000e_irq_fire_all_timers(val);
e1000e_intrmgr_fire_all_timers(core);
}
e1000e_update_interrupt_state(core);
}
static void
e1000e_set_rdtr(E1000ECore *core, int index, uint32_t val)
{
e1000e_set_16bit(core, index, val);
if ((val & E1000_RDTR_FPD) && (core->rdtr.running)) {
trace_e1000e_irq_rdtr_fpd_running();
e1000e_intrmgr_fire_delayed_interrupts(core);
} else {
trace_e1000e_irq_rdtr_fpd_not_running();
}
}
static void
e1000e_set_tidv(E1000ECore *core, int index, uint32_t val)
{
e1000e_set_16bit(core, index, val);
if ((val & E1000_TIDV_FPD) && (core->tidv.running)) {
trace_e1000e_irq_tidv_fpd_running();
e1000e_intrmgr_fire_delayed_interrupts(core);
} else {
trace_e1000e_irq_tidv_fpd_not_running();
}
}
static uint32_t
e1000e_mac_readreg(E1000ECore *core, int index)
{
return core->mac[index];
}
static uint32_t
e1000e_mac_ics_read(E1000ECore *core, int index)
{
trace_e1000e_irq_read_ics(core->mac[ICS]);
return core->mac[ICS];
}
static uint32_t
e1000e_mac_ims_read(E1000ECore *core, int index)
{
trace_e1000e_irq_read_ims(core->mac[IMS]);
return core->mac[IMS];
}
#define E1000E_LOW_BITS_READ_FUNC(num) \
static uint32_t \
e1000e_mac_low##num##_read(E1000ECore *core, int index) \
{ \
return core->mac[index] & (BIT(num) - 1); \
} \
#define E1000E_LOW_BITS_READ(num) \
e1000e_mac_low##num##_read
E1000E_LOW_BITS_READ_FUNC(4);
E1000E_LOW_BITS_READ_FUNC(6);
E1000E_LOW_BITS_READ_FUNC(11);
E1000E_LOW_BITS_READ_FUNC(13);
E1000E_LOW_BITS_READ_FUNC(16);
static uint32_t
e1000e_mac_swsm_read(E1000ECore *core, int index)
{
uint32_t val = core->mac[SWSM];
core->mac[SWSM] = val | 1;
return val;
}
static uint32_t
e1000e_mac_itr_read(E1000ECore *core, int index)
{
return core->itr_guest_value;
}
static uint32_t
e1000e_mac_eitr_read(E1000ECore *core, int index)
{
return core->eitr_guest_value[index - EITR];
}
static uint32_t
e1000e_mac_icr_read(E1000ECore *core, int index)
{
uint32_t ret = core->mac[ICR];
trace_e1000e_irq_icr_read_entry(ret);
if (core->mac[IMS] == 0) {
trace_e1000e_irq_icr_clear_zero_ims();
core->mac[ICR] = 0;
}
if (!msix_enabled(core->owner)) {
trace_e1000e_irq_icr_clear_nonmsix_icr_read();
core->mac[ICR] = 0;
}
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
if ((core->mac[ICR] & E1000_ICR_ASSERTED) &&
(core->mac[CTRL_EXT] & E1000_CTRL_EXT_IAME)) {
trace_e1000e_irq_icr_clear_iame();
core->mac[ICR] = 0;
trace_e1000e_irq_icr_process_iame();
e1000e_clear_ims_bits(core, core->mac[IAM]);
}
trace_e1000e_irq_icr_read_exit(core->mac[ICR]);
e1000e_update_interrupt_state(core);
return ret;
}
static uint32_t
e1000e_mac_read_clr4(E1000ECore *core, int index)
{
uint32_t ret = core->mac[index];
core->mac[index] = 0;
return ret;
}
static uint32_t
e1000e_mac_read_clr8(E1000ECore *core, int index)
{
uint32_t ret = core->mac[index];
core->mac[index] = 0;
core->mac[index - 1] = 0;
return ret;
}
static uint32_t
e1000e_get_ctrl(E1000ECore *core, int index)
{
uint32_t val = core->mac[CTRL];
trace_e1000e_link_read_params(
!!(val & E1000_CTRL_ASDE),
(val & E1000_CTRL_SPD_SEL) >> E1000_CTRL_SPD_SHIFT,
!!(val & E1000_CTRL_FRCSPD),
!!(val & E1000_CTRL_FRCDPX),
!!(val & E1000_CTRL_RFCE),
!!(val & E1000_CTRL_TFCE));
return val;
}
static uint32_t
e1000e_get_status(E1000ECore *core, int index)
{
uint32_t res = core->mac[STATUS];
if (!(core->mac[CTRL] & E1000_CTRL_GIO_MASTER_DISABLE)) {
res |= E1000_STATUS_GIO_MASTER_ENABLE;
}
if (core->mac[CTRL] & E1000_CTRL_FRCDPX) {
res |= (core->mac[CTRL] & E1000_CTRL_FD) ? E1000_STATUS_FD : 0;
} else {
res |= E1000_STATUS_FD;
}
if ((core->mac[CTRL] & E1000_CTRL_FRCSPD) ||
(core->mac[CTRL_EXT] & E1000_CTRL_EXT_SPD_BYPS)) {
switch (core->mac[CTRL] & E1000_CTRL_SPD_SEL) {
case E1000_CTRL_SPD_10:
res |= E1000_STATUS_SPEED_10;
break;
case E1000_CTRL_SPD_100:
res |= E1000_STATUS_SPEED_100;
break;
case E1000_CTRL_SPD_1000:
default:
res |= E1000_STATUS_SPEED_1000;
break;
}
} else {
res |= E1000_STATUS_SPEED_1000;
}
trace_e1000e_link_status(
!!(res & E1000_STATUS_LU),
!!(res & E1000_STATUS_FD),
(res & E1000_STATUS_SPEED_MASK) >> E1000_STATUS_SPEED_SHIFT,
(res & E1000_STATUS_ASDV) >> E1000_STATUS_ASDV_SHIFT);
return res;
}
static uint32_t
e1000e_get_tarc(E1000ECore *core, int index)
{
return core->mac[index] & ((BIT(11) - 1) |
BIT(27) |
BIT(28) |
BIT(29) |
BIT(30));
}
static void
e1000e_mac_writereg(E1000ECore *core, int index, uint32_t val)
{
core->mac[index] = val;
}
static void
e1000e_mac_setmacaddr(E1000ECore *core, int index, uint32_t val)
{
uint32_t macaddr[2];
core->mac[index] = val;
macaddr[0] = cpu_to_le32(core->mac[RA]);
macaddr[1] = cpu_to_le32(core->mac[RA + 1]);
qemu_format_nic_info_str(qemu_get_queue(core->owner_nic),
(uint8_t *) macaddr);
trace_e1000e_mac_set_sw(MAC_ARG(macaddr));
}
static void
e1000e_set_eecd(E1000ECore *core, int index, uint32_t val)
{
static const uint32_t ro_bits = E1000_EECD_PRES |
E1000_EECD_AUTO_RD |
E1000_EECD_SIZE_EX_MASK;
core->mac[EECD] = (core->mac[EECD] & ro_bits) | (val & ~ro_bits);
}
static void
e1000e_set_eerd(E1000ECore *core, int index, uint32_t val)
{
uint32_t addr = (val >> E1000_EERW_ADDR_SHIFT) & E1000_EERW_ADDR_MASK;
uint32_t flags = 0;
uint32_t data = 0;
if ((addr < E1000E_EEPROM_SIZE) && (val & E1000_EERW_START)) {
data = core->eeprom[addr];
flags = E1000_EERW_DONE;
}
core->mac[EERD] = flags |
(addr << E1000_EERW_ADDR_SHIFT) |
(data << E1000_EERW_DATA_SHIFT);
}
static void
e1000e_set_eewr(E1000ECore *core, int index, uint32_t val)
{
uint32_t addr = (val >> E1000_EERW_ADDR_SHIFT) & E1000_EERW_ADDR_MASK;
uint32_t data = (val >> E1000_EERW_DATA_SHIFT) & E1000_EERW_DATA_MASK;
uint32_t flags = 0;
if ((addr < E1000E_EEPROM_SIZE) && (val & E1000_EERW_START)) {
core->eeprom[addr] = data;
flags = E1000_EERW_DONE;
}
core->mac[EERD] = flags |
(addr << E1000_EERW_ADDR_SHIFT) |
(data << E1000_EERW_DATA_SHIFT);
}
static void
e1000e_set_rxdctl(E1000ECore *core, int index, uint32_t val)
{
core->mac[RXDCTL] = core->mac[RXDCTL1] = val;
}
static void
e1000e_set_itr(E1000ECore *core, int index, uint32_t val)
{
uint32_t interval = val & 0xffff;
trace_e1000e_irq_itr_set(val);
core->itr_guest_value = interval;
core->mac[index] = MAX(interval, E1000E_MIN_XITR);
}
static void
e1000e_set_eitr(E1000ECore *core, int index, uint32_t val)
{
uint32_t interval = val & 0xffff;
uint32_t eitr_num = index - EITR;
trace_e1000e_irq_eitr_set(eitr_num, val);
core->eitr_guest_value[eitr_num] = interval;
core->mac[index] = MAX(interval, E1000E_MIN_XITR);
}
static void
e1000e_set_psrctl(E1000ECore *core, int index, uint32_t val)
{
if (core->mac[RCTL] & E1000_RCTL_DTYP_MASK) {
if ((val & E1000_PSRCTL_BSIZE0_MASK) == 0) {
hw/net/e1000e: Do not abort() on invalid PSRCTL register value libFuzzer found using 'qemu-system-i386 -M q35': qemu: hardware error: e1000e: PSRCTL.BSIZE0 cannot be zero CPU #0: EAX=00000000 EBX=00000000 ECX=00000000 EDX=00000663 ESI=00000000 EDI=00000000 EBP=00000000 ESP=00000000 EIP=0000fff0 EFL=00000002 [-------] CPL=0 II=0 A20=1 SMM=0 HLT=0 ES =0000 00000000 0000ffff 00009300 CS =f000 ffff0000 0000ffff 00009b00 SS =0000 00000000 0000ffff 00009300 DS =0000 00000000 0000ffff 00009300 FS =0000 00000000 0000ffff 00009300 GS =0000 00000000 0000ffff 00009300 LDT=0000 00000000 0000ffff 00008200 TR =0000 00000000 0000ffff 00008b00 GDT= 00000000 0000ffff IDT= 00000000 0000ffff CR0=60000010 CR2=00000000 CR3=00000000 CR4=00000000 DR0=00000000 DR1=00000000 DR2=00000000 DR3=00000000 DR6=ffff0ff0 DR7=00000400 EFER=0000000000000000 FCW=037f FSW=0000 [ST=0] FTW=00 MXCSR=00001f80 FPR0=0000000000000000 0000 FPR1=0000000000000000 0000 FPR2=0000000000000000 0000 FPR3=0000000000000000 0000 FPR4=0000000000000000 0000 FPR5=0000000000000000 0000 FPR6=0000000000000000 0000 FPR7=0000000000000000 0000 XMM00=00000000000000000000000000000000 XMM01=00000000000000000000000000000000 XMM02=00000000000000000000000000000000 XMM03=00000000000000000000000000000000 XMM04=00000000000000000000000000000000 XMM05=00000000000000000000000000000000 XMM06=00000000000000000000000000000000 XMM07=00000000000000000000000000000000 ==1988== ERROR: libFuzzer: deadly signal #6 0x7fae4d3ea894 in __GI_abort (/lib64/libc.so.6+0x22894) #7 0x563f4cc59a1d in hw_error (qemu-fuzz-i386+0xe8ca1d) #8 0x563f4d7c93f2 in e1000e_set_psrctl (qemu-fuzz-i386+0x19fc3f2) #9 0x563f4d7b798f in e1000e_core_write (qemu-fuzz-i386+0x19ea98f) #10 0x563f4d7afc46 in e1000e_mmio_write (qemu-fuzz-i386+0x19e2c46) #11 0x563f4cc9a0a7 in memory_region_write_accessor (qemu-fuzz-i386+0xecd0a7) #12 0x563f4cc99c13 in access_with_adjusted_size (qemu-fuzz-i386+0xeccc13) #13 0x563f4cc987b4 in memory_region_dispatch_write (qemu-fuzz-i386+0xecb7b4) It simply sent the following 2 I/O command to the e1000e PCI BAR #2 I/O region: writew 0x0100 0x0c00 # RCTL = E1000_RCTL_DTYP_MASK writeb 0x2170 0x00 # PSRCTL = 0 2813 static void 2814 e1000e_set_psrctl(E1000ECore *core, int index, uint32_t val) 2815 { 2816 if (core->mac[RCTL] & E1000_RCTL_DTYP_MASK) { 2817 2818 if ((val & E1000_PSRCTL_BSIZE0_MASK) == 0) { 2819 hw_error("e1000e: PSRCTL.BSIZE0 cannot be zero"); 2820 } Instead of calling hw_error() which abort the process (it is meant for CPU fatal error condition, not for device logging), log the invalid request with qemu_log_mask(LOG_GUEST_ERROR) and return, ignoring the request. Cc: qemu-stable@nongnu.org Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Jason Wang <jasowang@redhat.com>
2020-05-25 14:23:30 +02:00
qemu_log_mask(LOG_GUEST_ERROR,
"e1000e: PSRCTL.BSIZE0 cannot be zero");
return;
}
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
if ((val & E1000_PSRCTL_BSIZE1_MASK) == 0) {
hw/net/e1000e: Do not abort() on invalid PSRCTL register value libFuzzer found using 'qemu-system-i386 -M q35': qemu: hardware error: e1000e: PSRCTL.BSIZE0 cannot be zero CPU #0: EAX=00000000 EBX=00000000 ECX=00000000 EDX=00000663 ESI=00000000 EDI=00000000 EBP=00000000 ESP=00000000 EIP=0000fff0 EFL=00000002 [-------] CPL=0 II=0 A20=1 SMM=0 HLT=0 ES =0000 00000000 0000ffff 00009300 CS =f000 ffff0000 0000ffff 00009b00 SS =0000 00000000 0000ffff 00009300 DS =0000 00000000 0000ffff 00009300 FS =0000 00000000 0000ffff 00009300 GS =0000 00000000 0000ffff 00009300 LDT=0000 00000000 0000ffff 00008200 TR =0000 00000000 0000ffff 00008b00 GDT= 00000000 0000ffff IDT= 00000000 0000ffff CR0=60000010 CR2=00000000 CR3=00000000 CR4=00000000 DR0=00000000 DR1=00000000 DR2=00000000 DR3=00000000 DR6=ffff0ff0 DR7=00000400 EFER=0000000000000000 FCW=037f FSW=0000 [ST=0] FTW=00 MXCSR=00001f80 FPR0=0000000000000000 0000 FPR1=0000000000000000 0000 FPR2=0000000000000000 0000 FPR3=0000000000000000 0000 FPR4=0000000000000000 0000 FPR5=0000000000000000 0000 FPR6=0000000000000000 0000 FPR7=0000000000000000 0000 XMM00=00000000000000000000000000000000 XMM01=00000000000000000000000000000000 XMM02=00000000000000000000000000000000 XMM03=00000000000000000000000000000000 XMM04=00000000000000000000000000000000 XMM05=00000000000000000000000000000000 XMM06=00000000000000000000000000000000 XMM07=00000000000000000000000000000000 ==1988== ERROR: libFuzzer: deadly signal #6 0x7fae4d3ea894 in __GI_abort (/lib64/libc.so.6+0x22894) #7 0x563f4cc59a1d in hw_error (qemu-fuzz-i386+0xe8ca1d) #8 0x563f4d7c93f2 in e1000e_set_psrctl (qemu-fuzz-i386+0x19fc3f2) #9 0x563f4d7b798f in e1000e_core_write (qemu-fuzz-i386+0x19ea98f) #10 0x563f4d7afc46 in e1000e_mmio_write (qemu-fuzz-i386+0x19e2c46) #11 0x563f4cc9a0a7 in memory_region_write_accessor (qemu-fuzz-i386+0xecd0a7) #12 0x563f4cc99c13 in access_with_adjusted_size (qemu-fuzz-i386+0xeccc13) #13 0x563f4cc987b4 in memory_region_dispatch_write (qemu-fuzz-i386+0xecb7b4) It simply sent the following 2 I/O command to the e1000e PCI BAR #2 I/O region: writew 0x0100 0x0c00 # RCTL = E1000_RCTL_DTYP_MASK writeb 0x2170 0x00 # PSRCTL = 0 2813 static void 2814 e1000e_set_psrctl(E1000ECore *core, int index, uint32_t val) 2815 { 2816 if (core->mac[RCTL] & E1000_RCTL_DTYP_MASK) { 2817 2818 if ((val & E1000_PSRCTL_BSIZE0_MASK) == 0) { 2819 hw_error("e1000e: PSRCTL.BSIZE0 cannot be zero"); 2820 } Instead of calling hw_error() which abort the process (it is meant for CPU fatal error condition, not for device logging), log the invalid request with qemu_log_mask(LOG_GUEST_ERROR) and return, ignoring the request. Cc: qemu-stable@nongnu.org Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org> Signed-off-by: Jason Wang <jasowang@redhat.com>
2020-05-25 14:23:30 +02:00
qemu_log_mask(LOG_GUEST_ERROR,
"e1000e: PSRCTL.BSIZE1 cannot be zero");
return;
}
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
}
core->mac[PSRCTL] = val;
}
static void
e1000e_update_rx_offloads(E1000ECore *core)
{
int cso_state = e1000e_rx_l4_cso_enabled(core);
trace_e1000e_rx_set_cso(cso_state);
if (core->has_vnet) {
qemu_set_offload(qemu_get_queue(core->owner_nic)->peer,
cso_state, 0, 0, 0, 0);
}
}
static void
e1000e_set_rxcsum(E1000ECore *core, int index, uint32_t val)
{
core->mac[RXCSUM] = val;
e1000e_update_rx_offloads(core);
}
static void
e1000e_set_gcr(E1000ECore *core, int index, uint32_t val)
{
uint32_t ro_bits = core->mac[GCR] & E1000_GCR_RO_BITS;
core->mac[GCR] = (val & ~E1000_GCR_RO_BITS) | ro_bits;
}
#define e1000e_getreg(x) [x] = e1000e_mac_readreg
typedef uint32_t (*readops)(E1000ECore *, int);
static const readops e1000e_macreg_readops[] = {
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000e_getreg(PBA),
e1000e_getreg(WUFC),
e1000e_getreg(MANC),
e1000e_getreg(TOTL),
e1000e_getreg(RDT0),
e1000e_getreg(RDBAH0),
e1000e_getreg(TDBAL1),
e1000e_getreg(RDLEN0),
e1000e_getreg(RDH1),
e1000e_getreg(LATECOL),
e1000e_getreg(SEQEC),
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000e_getreg(XONTXC),
e1000e_getreg(WUS),
e1000e_getreg(GORCL),
e1000e_getreg(MGTPRC),
e1000e_getreg(EERD),
e1000e_getreg(EIAC),
e1000e_getreg(PSRCTL),
e1000e_getreg(MANC2H),
e1000e_getreg(RXCSUM),
e1000e_getreg(GSCL_3),
e1000e_getreg(GSCN_2),
e1000e_getreg(RSRPD),
e1000e_getreg(RDBAL1),
e1000e_getreg(FCAH),
e1000e_getreg(FCRTH),
e1000e_getreg(FLOP),
e1000e_getreg(FLASHT),
e1000e_getreg(RXSTMPH),
e1000e_getreg(TXSTMPL),
e1000e_getreg(TIMADJL),
e1000e_getreg(TXDCTL),
e1000e_getreg(RDH0),
e1000e_getreg(TDT1),
e1000e_getreg(TNCRS),
e1000e_getreg(RJC),
e1000e_getreg(IAM),
e1000e_getreg(GSCL_2),
e1000e_getreg(RDBAH1),
e1000e_getreg(FLSWDATA),
e1000e_getreg(RXSATRH),
e1000e_getreg(TIPG),
e1000e_getreg(FLMNGCTL),
e1000e_getreg(FLMNGCNT),
e1000e_getreg(TSYNCTXCTL),
e1000e_getreg(EXTCNF_SIZE),
e1000e_getreg(EXTCNF_CTRL),
e1000e_getreg(EEMNGDATA),
e1000e_getreg(CTRL_EXT),
e1000e_getreg(SYSTIMH),
e1000e_getreg(EEMNGCTL),
e1000e_getreg(FLMNGDATA),
e1000e_getreg(TSYNCRXCTL),
e1000e_getreg(TDH),
e1000e_getreg(LEDCTL),
e1000e_getreg(TCTL),
e1000e_getreg(TDBAL),
e1000e_getreg(TDLEN),
e1000e_getreg(TDH1),
e1000e_getreg(RADV),
e1000e_getreg(ECOL),
e1000e_getreg(DC),
e1000e_getreg(RLEC),
e1000e_getreg(XOFFTXC),
e1000e_getreg(RFC),
e1000e_getreg(RNBC),
e1000e_getreg(MGTPTC),
e1000e_getreg(TIMINCA),
e1000e_getreg(RXCFGL),
e1000e_getreg(MFUTP01),
e1000e_getreg(FACTPS),
e1000e_getreg(GSCL_1),
e1000e_getreg(GSCN_0),
e1000e_getreg(GCR2),
e1000e_getreg(RDT1),
e1000e_getreg(PBACLR),
e1000e_getreg(FCTTV),
e1000e_getreg(EEWR),
e1000e_getreg(FLSWCTL),
e1000e_getreg(RXDCTL1),
e1000e_getreg(RXSATRL),
e1000e_getreg(SYSTIML),
e1000e_getreg(RXUDP),
e1000e_getreg(TORL),
e1000e_getreg(TDLEN1),
e1000e_getreg(MCC),
e1000e_getreg(WUC),
e1000e_getreg(EECD),
e1000e_getreg(MFUTP23),
e1000e_getreg(RAID),
e1000e_getreg(FCRTV),
e1000e_getreg(TXDCTL1),
e1000e_getreg(RCTL),
e1000e_getreg(TDT),
e1000e_getreg(MDIC),
e1000e_getreg(FCRUC),
e1000e_getreg(VET),
e1000e_getreg(RDBAL0),
e1000e_getreg(TDBAH1),
e1000e_getreg(RDTR),
e1000e_getreg(SCC),
e1000e_getreg(COLC),
e1000e_getreg(CEXTERR),
e1000e_getreg(XOFFRXC),
e1000e_getreg(IPAV),
e1000e_getreg(GOTCL),
e1000e_getreg(MGTPDC),
e1000e_getreg(GCR),
e1000e_getreg(IVAR),
e1000e_getreg(POEMB),
e1000e_getreg(MFVAL),
e1000e_getreg(FUNCTAG),
e1000e_getreg(GSCL_4),
e1000e_getreg(GSCN_3),
e1000e_getreg(MRQC),
e1000e_getreg(RDLEN1),
e1000e_getreg(FCT),
e1000e_getreg(FLA),
e1000e_getreg(FLOL),
e1000e_getreg(RXDCTL),
e1000e_getreg(RXSTMPL),
e1000e_getreg(TXSTMPH),
e1000e_getreg(TIMADJH),
e1000e_getreg(FCRTL),
e1000e_getreg(TDBAH),
e1000e_getreg(TADV),
e1000e_getreg(XONRXC),
e1000e_getreg(TSCTFC),
e1000e_getreg(RFCTL),
e1000e_getreg(GSCN_1),
e1000e_getreg(FCAL),
e1000e_getreg(FLSWCNT),
[TOTH] = e1000e_mac_read_clr8,
[GOTCH] = e1000e_mac_read_clr8,
[PRC64] = e1000e_mac_read_clr4,
[PRC255] = e1000e_mac_read_clr4,
[PRC1023] = e1000e_mac_read_clr4,
[PTC64] = e1000e_mac_read_clr4,
[PTC255] = e1000e_mac_read_clr4,
[PTC1023] = e1000e_mac_read_clr4,
[GPRC] = e1000e_mac_read_clr4,
[TPT] = e1000e_mac_read_clr4,
[RUC] = e1000e_mac_read_clr4,
[BPRC] = e1000e_mac_read_clr4,
[MPTC] = e1000e_mac_read_clr4,
[IAC] = e1000e_mac_read_clr4,
[ICR] = e1000e_mac_icr_read,
[RDFH] = E1000E_LOW_BITS_READ(13),
[RDFHS] = E1000E_LOW_BITS_READ(13),
[RDFPC] = E1000E_LOW_BITS_READ(13),
[TDFH] = E1000E_LOW_BITS_READ(13),
[TDFHS] = E1000E_LOW_BITS_READ(13),
[STATUS] = e1000e_get_status,
[TARC0] = e1000e_get_tarc,
[PBS] = E1000E_LOW_BITS_READ(6),
[ICS] = e1000e_mac_ics_read,
[AIT] = E1000E_LOW_BITS_READ(16),
[TORH] = e1000e_mac_read_clr8,
[GORCH] = e1000e_mac_read_clr8,
[PRC127] = e1000e_mac_read_clr4,
[PRC511] = e1000e_mac_read_clr4,
[PRC1522] = e1000e_mac_read_clr4,
[PTC127] = e1000e_mac_read_clr4,
[PTC511] = e1000e_mac_read_clr4,
[PTC1522] = e1000e_mac_read_clr4,
[GPTC] = e1000e_mac_read_clr4,
[TPR] = e1000e_mac_read_clr4,
[ROC] = e1000e_mac_read_clr4,
[MPRC] = e1000e_mac_read_clr4,
[BPTC] = e1000e_mac_read_clr4,
[TSCTC] = e1000e_mac_read_clr4,
[ITR] = e1000e_mac_itr_read,
[RDFT] = E1000E_LOW_BITS_READ(13),
[RDFTS] = E1000E_LOW_BITS_READ(13),
[TDFPC] = E1000E_LOW_BITS_READ(13),
[TDFT] = E1000E_LOW_BITS_READ(13),
[TDFTS] = E1000E_LOW_BITS_READ(13),
[CTRL] = e1000e_get_ctrl,
[TARC1] = e1000e_get_tarc,
[SWSM] = e1000e_mac_swsm_read,
[IMS] = e1000e_mac_ims_read,
[CRCERRS ... MPC] = e1000e_mac_readreg,
[IP6AT ... IP6AT + 3] = e1000e_mac_readreg,
[IP4AT ... IP4AT + 6] = e1000e_mac_readreg,
[RA ... RA + 31] = e1000e_mac_readreg,
[WUPM ... WUPM + 31] = e1000e_mac_readreg,
[MTA ... MTA + 127] = e1000e_mac_readreg,
[VFTA ... VFTA + 127] = e1000e_mac_readreg,
[FFMT ... FFMT + 254] = E1000E_LOW_BITS_READ(4),
[FFVT ... FFVT + 254] = e1000e_mac_readreg,
[MDEF ... MDEF + 7] = e1000e_mac_readreg,
[FFLT ... FFLT + 10] = E1000E_LOW_BITS_READ(11),
[FTFT ... FTFT + 254] = e1000e_mac_readreg,
[PBM ... PBM + 10239] = e1000e_mac_readreg,
[RETA ... RETA + 31] = e1000e_mac_readreg,
[RSSRK ... RSSRK + 31] = e1000e_mac_readreg,
[MAVTV0 ... MAVTV3] = e1000e_mac_readreg,
[EITR...EITR + E1000E_MSIX_VEC_NUM - 1] = e1000e_mac_eitr_read
};
enum { E1000E_NREADOPS = ARRAY_SIZE(e1000e_macreg_readops) };
#define e1000e_putreg(x) [x] = e1000e_mac_writereg
typedef void (*writeops)(E1000ECore *, int, uint32_t);
static const writeops e1000e_macreg_writeops[] = {
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
e1000e_putreg(PBA),
e1000e_putreg(SWSM),
e1000e_putreg(WUFC),
e1000e_putreg(RDBAH1),
e1000e_putreg(TDBAH),
e1000e_putreg(TXDCTL),
e1000e_putreg(RDBAH0),
e1000e_putreg(LEDCTL),
e1000e_putreg(FCAL),
e1000e_putreg(FCRUC),
e1000e_putreg(AIT),
e1000e_putreg(TDFH),
e1000e_putreg(TDFT),
e1000e_putreg(TDFHS),
e1000e_putreg(TDFTS),
e1000e_putreg(TDFPC),
e1000e_putreg(WUC),
e1000e_putreg(WUS),
e1000e_putreg(RDFH),
e1000e_putreg(RDFT),
e1000e_putreg(RDFHS),
e1000e_putreg(RDFTS),
e1000e_putreg(RDFPC),
e1000e_putreg(IPAV),
e1000e_putreg(TDBAH1),
e1000e_putreg(TIMINCA),
e1000e_putreg(IAM),
e1000e_putreg(EIAC),
e1000e_putreg(IVAR),
e1000e_putreg(TARC0),
e1000e_putreg(TARC1),
e1000e_putreg(FLSWDATA),
e1000e_putreg(POEMB),
e1000e_putreg(PBS),
e1000e_putreg(MFUTP01),
e1000e_putreg(MFUTP23),
e1000e_putreg(MANC),
e1000e_putreg(MANC2H),
e1000e_putreg(MFVAL),
e1000e_putreg(EXTCNF_CTRL),
e1000e_putreg(FACTPS),
e1000e_putreg(FUNCTAG),
e1000e_putreg(GSCL_1),
e1000e_putreg(GSCL_2),
e1000e_putreg(GSCL_3),
e1000e_putreg(GSCL_4),
e1000e_putreg(GSCN_0),
e1000e_putreg(GSCN_1),
e1000e_putreg(GSCN_2),
e1000e_putreg(GSCN_3),
e1000e_putreg(GCR2),
e1000e_putreg(MRQC),
e1000e_putreg(FLOP),
e1000e_putreg(FLOL),
e1000e_putreg(FLSWCTL),
e1000e_putreg(FLSWCNT),
e1000e_putreg(FLA),
e1000e_putreg(RXDCTL1),
e1000e_putreg(TXDCTL1),
e1000e_putreg(TIPG),
e1000e_putreg(RXSTMPH),
e1000e_putreg(RXSTMPL),
e1000e_putreg(RXSATRL),
e1000e_putreg(RXSATRH),
e1000e_putreg(TXSTMPL),
e1000e_putreg(TXSTMPH),
e1000e_putreg(SYSTIML),
e1000e_putreg(SYSTIMH),
e1000e_putreg(TIMADJL),
e1000e_putreg(TIMADJH),
e1000e_putreg(RXUDP),
e1000e_putreg(RXCFGL),
e1000e_putreg(TSYNCRXCTL),
e1000e_putreg(TSYNCTXCTL),
e1000e_putreg(EXTCNF_SIZE),
e1000e_putreg(EEMNGCTL),
e1000e_putreg(RA),
[TDH1] = e1000e_set_16bit,
[TDT1] = e1000e_set_tdt,
[TCTL] = e1000e_set_tctl,
[TDT] = e1000e_set_tdt,
[MDIC] = e1000e_set_mdic,
[ICS] = e1000e_set_ics,
[TDH] = e1000e_set_16bit,
[RDH0] = e1000e_set_16bit,
[RDT0] = e1000e_set_rdt,
[IMC] = e1000e_set_imc,
[IMS] = e1000e_set_ims,
[ICR] = e1000e_set_icr,
[EECD] = e1000e_set_eecd,
[RCTL] = e1000e_set_rx_control,
[CTRL] = e1000e_set_ctrl,
[RDTR] = e1000e_set_rdtr,
[RADV] = e1000e_set_16bit,
[TADV] = e1000e_set_16bit,
[ITR] = e1000e_set_itr,
[EERD] = e1000e_set_eerd,
[GCR] = e1000e_set_gcr,
[PSRCTL] = e1000e_set_psrctl,
[RXCSUM] = e1000e_set_rxcsum,
[RAID] = e1000e_set_16bit,
[RSRPD] = e1000e_set_12bit,
[TIDV] = e1000e_set_tidv,
[TDLEN1] = e1000e_set_dlen,
[TDLEN] = e1000e_set_dlen,
[RDLEN0] = e1000e_set_dlen,
[RDLEN1] = e1000e_set_dlen,
[TDBAL] = e1000e_set_dbal,
[TDBAL1] = e1000e_set_dbal,
[RDBAL0] = e1000e_set_dbal,
[RDBAL1] = e1000e_set_dbal,
[RDH1] = e1000e_set_16bit,
[RDT1] = e1000e_set_rdt,
[STATUS] = e1000e_set_status,
[PBACLR] = e1000e_set_pbaclr,
[CTRL_EXT] = e1000e_set_ctrlext,
[FCAH] = e1000e_set_16bit,
[FCT] = e1000e_set_16bit,
[FCTTV] = e1000e_set_16bit,
[FCRTV] = e1000e_set_16bit,
[FCRTH] = e1000e_set_fcrth,
[FCRTL] = e1000e_set_fcrtl,
[VET] = e1000e_set_vet,
[RXDCTL] = e1000e_set_rxdctl,
[FLASHT] = e1000e_set_16bit,
[EEWR] = e1000e_set_eewr,
[CTRL_DUP] = e1000e_set_ctrl,
[RFCTL] = e1000e_set_rfctl,
[RA + 1] = e1000e_mac_setmacaddr,
[IP6AT ... IP6AT + 3] = e1000e_mac_writereg,
[IP4AT ... IP4AT + 6] = e1000e_mac_writereg,
[RA + 2 ... RA + 31] = e1000e_mac_writereg,
[WUPM ... WUPM + 31] = e1000e_mac_writereg,
[MTA ... MTA + 127] = e1000e_mac_writereg,
[VFTA ... VFTA + 127] = e1000e_mac_writereg,
[FFMT ... FFMT + 254] = e1000e_mac_writereg,
[FFVT ... FFVT + 254] = e1000e_mac_writereg,
[PBM ... PBM + 10239] = e1000e_mac_writereg,
[MDEF ... MDEF + 7] = e1000e_mac_writereg,
[FFLT ... FFLT + 10] = e1000e_mac_writereg,
[FTFT ... FTFT + 254] = e1000e_mac_writereg,
[RETA ... RETA + 31] = e1000e_mac_writereg,
[RSSRK ... RSSRK + 31] = e1000e_mac_writereg,
[MAVTV0 ... MAVTV3] = e1000e_mac_writereg,
[EITR...EITR + E1000E_MSIX_VEC_NUM - 1] = e1000e_set_eitr
};
enum { E1000E_NWRITEOPS = ARRAY_SIZE(e1000e_macreg_writeops) };
enum { MAC_ACCESS_PARTIAL = 1 };
/* The array below combines alias offsets of the index values for the
* MAC registers that have aliases, with the indication of not fully
* implemented registers (lowest bit). This combination is possible
* because all of the offsets are even. */
static const uint16_t mac_reg_access[E1000E_MAC_SIZE] = {
/* Alias index offsets */
[FCRTL_A] = 0x07fe, [FCRTH_A] = 0x0802,
[RDH0_A] = 0x09bc, [RDT0_A] = 0x09bc, [RDTR_A] = 0x09c6,
[RDFH_A] = 0xe904, [RDFT_A] = 0xe904,
[TDH_A] = 0x0cf8, [TDT_A] = 0x0cf8, [TIDV_A] = 0x0cf8,
[TDFH_A] = 0xed00, [TDFT_A] = 0xed00,
[RA_A ... RA_A + 31] = 0x14f0,
[VFTA_A ... VFTA_A + 127] = 0x1400,
[RDBAL0_A ... RDLEN0_A] = 0x09bc,
[TDBAL_A ... TDLEN_A] = 0x0cf8,
/* Access options */
[RDFH] = MAC_ACCESS_PARTIAL, [RDFT] = MAC_ACCESS_PARTIAL,
[RDFHS] = MAC_ACCESS_PARTIAL, [RDFTS] = MAC_ACCESS_PARTIAL,
[RDFPC] = MAC_ACCESS_PARTIAL,
[TDFH] = MAC_ACCESS_PARTIAL, [TDFT] = MAC_ACCESS_PARTIAL,
[TDFHS] = MAC_ACCESS_PARTIAL, [TDFTS] = MAC_ACCESS_PARTIAL,
[TDFPC] = MAC_ACCESS_PARTIAL, [EECD] = MAC_ACCESS_PARTIAL,
[PBM] = MAC_ACCESS_PARTIAL, [FLA] = MAC_ACCESS_PARTIAL,
[FCAL] = MAC_ACCESS_PARTIAL, [FCAH] = MAC_ACCESS_PARTIAL,
[FCT] = MAC_ACCESS_PARTIAL, [FCTTV] = MAC_ACCESS_PARTIAL,
[FCRTV] = MAC_ACCESS_PARTIAL, [FCRTL] = MAC_ACCESS_PARTIAL,
[FCRTH] = MAC_ACCESS_PARTIAL, [TXDCTL] = MAC_ACCESS_PARTIAL,
[TXDCTL1] = MAC_ACCESS_PARTIAL,
[MAVTV0 ... MAVTV3] = MAC_ACCESS_PARTIAL
};
void
e1000e_core_write(E1000ECore *core, hwaddr addr, uint64_t val, unsigned size)
{
uint16_t index = e1000e_get_reg_index_with_offset(mac_reg_access, addr);
if (index < E1000E_NWRITEOPS && e1000e_macreg_writeops[index]) {
if (mac_reg_access[index] & MAC_ACCESS_PARTIAL) {
trace_e1000e_wrn_regs_write_trivial(index << 2);
}
trace_e1000e_core_write(index << 2, size, val);
e1000e_macreg_writeops[index](core, index, val);
} else if (index < E1000E_NREADOPS && e1000e_macreg_readops[index]) {
trace_e1000e_wrn_regs_write_ro(index << 2, size, val);
} else {
trace_e1000e_wrn_regs_write_unknown(index << 2, size, val);
}
}
uint64_t
e1000e_core_read(E1000ECore *core, hwaddr addr, unsigned size)
{
uint64_t val;
uint16_t index = e1000e_get_reg_index_with_offset(mac_reg_access, addr);
if (index < E1000E_NREADOPS && e1000e_macreg_readops[index]) {
if (mac_reg_access[index] & MAC_ACCESS_PARTIAL) {
trace_e1000e_wrn_regs_read_trivial(index << 2);
}
val = e1000e_macreg_readops[index](core, index);
trace_e1000e_core_read(index << 2, size, val);
return val;
} else {
trace_e1000e_wrn_regs_read_unknown(index << 2, size);
}
return 0;
}
static inline void
e1000e_autoneg_pause(E1000ECore *core)
{
timer_del(core->autoneg_timer);
}
static void
e1000e_autoneg_resume(E1000ECore *core)
{
if (e1000e_have_autoneg(core) &&
!(core->phy[0][PHY_STATUS] & MII_SR_AUTONEG_COMPLETE)) {
qemu_get_queue(core->owner_nic)->link_down = false;
timer_mod(core->autoneg_timer,
qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL) + 500);
}
}
static void
e1000e_vm_state_change(void *opaque, bool running, RunState state)
net: Introduce e1000e device emulation This patch introduces emulation for the Intel 82574 adapter, AKA e1000e. This implementation is derived from the e1000 emulation code, and utilizes the TX/RX packet abstractions that were initially developed for the vmxnet3 device. Although some parts of the introduced code may be shared with e1000, the differences are substantial enough so that the only shared resources for the two devices are the definitions in hw/net/e1000_regs.h. Similarly to vmxnet3, the new device uses virtio headers for task offloads (for backends that support virtio extensions). Usage of virtio headers may be forcibly disabled via a boolean device property "vnet" (which is enabled by default). In such case task offloads will be performed in software, in the same way it is done on backends that do not support virtio headers. The device code is split into two parts: 1. hw/net/e1000e.c: QEMU-specific code for a network device; 2. hw/net/e1000e_core.[hc]: Device emulation according to the spec. The new device name is e1000e. Intel specifications for the 82574 controller are available at: http://www.intel.com/content/dam/doc/datasheet/82574l-gbe-controller-datasheet.pdf Throughput measurement results (iperf2): Fedora 22 guest, TCP, RX 4 ++------------------------------------------+ | | | X X X X X 3.5 ++ X X X X | | X | | | 3 ++ | G | X | b | | / 2.5 ++ | s | | | | 2 ++ | | | | | 1.5 X+ | | | + + + + + + + + + + + + 1 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, TCP, TX 18 ++-------------------------------------------+ | X | 16 ++ X X X X X | X | 14 ++ | | | 12 ++ | G | X | b 10 ++ | / | | s 8 ++ | | | 6 ++ X | | | 4 ++ | | X | 2 ++ X | X + + + + + + + + + + + 0 ++--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Fedora 22 guest, UDP, RX 3 ++------------------------------------------+ | X | | 2.5 ++ | | | | | 2 ++ X | G | | b | | / 1.5 ++ | s | X | | | 1 ++ | | | | X | 0.5 ++ | | X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Fedora 22 guest, UDP, TX 1 ++------------------------------------------+ | X 0.9 ++ | | | 0.8 ++ | 0.7 ++ | | | G 0.6 ++ | b | | / 0.5 ++ | s | X | 0.4 ++ | | | 0.3 ++ | 0.2 ++ X | | | 0.1 ++ X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, TCP, RX 3.2 ++------------------------------------------+ | X | 3 ++ | | | 2.8 ++ | | | 2.6 ++ X | G | X X X X X b 2.4 ++ X X | / | | s 2.2 ++ | | | 2 ++ | | X X | 1.8 ++ | | | 1.6 X+ | + + + + + + + + + + + + 1.4 ++--+---+---+---+---+---+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, TCP, TX 14 ++-------------------------------------------+ | | | X X 12 ++ | | | 10 ++ | | | G | | b 8 ++ | / | X | s 6 ++ | | | | | 4 ++ X | | | 2 ++ | | X X X | + X X + + X X + + + + + 0 X+--+---+---+---+---+----+---+---+---+---+---+ 32 64 128 256 512 1 2 4 8 16 32 64 B B B B B KB KB KB KB KB KB KB Buffer size Windows 2012R2 guest, UDP, RX 1.6 ++------------------------------------------X | | 1.4 ++ | | | 1.2 ++ | | X | | | G 1 ++ | b | | / 0.8 ++ | s | | 0.6 ++ X | | | 0.4 ++ | | X | | | 0.2 ++ X | X + + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Windows 2012R2 guest, UDP, TX 0.6 ++------------------------------------------+ | X | | 0.5 ++ | | | | | 0.4 ++ | G | | b | | / 0.3 ++ X | s | | | | 0.2 ++ | | | | X | 0.1 ++ | | X | X X + + + + 0 ++-------+--------+-------+--------+--------+ 32 64 128 256 512 1 B B B B B KB Datagram size Signed-off-by: Dmitry Fleytman <dmitry.fleytman@ravellosystems.com> Signed-off-by: Leonid Bloch <leonid.bloch@ravellosystems.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Jason Wang <jasowang@redhat.com>
2016-06-01 10:23:45 +02:00
{
E1000ECore *core = opaque;
if (running) {
trace_e1000e_vm_state_running();
e1000e_intrmgr_resume(core);
e1000e_autoneg_resume(core);
} else {
trace_e1000e_vm_state_stopped();
e1000e_autoneg_pause(core);
e1000e_intrmgr_pause(core);
}
}
void
e1000e_core_pci_realize(E1000ECore *core,
const uint16_t *eeprom_templ,
uint32_t eeprom_size,
const uint8_t *macaddr)
{
int i;
core->autoneg_timer = timer_new_ms(QEMU_CLOCK_VIRTUAL,
e1000e_autoneg_timer, core);
e1000e_intrmgr_pci_realize(core);
core->vmstate =
qemu_add_vm_change_state_handler(e1000e_vm_state_change, core);
for (i = 0; i < E1000E_NUM_QUEUES; i++) {
net_tx_pkt_init(&core->tx[i].tx_pkt, core->owner,
E1000E_MAX_TX_FRAGS, core->has_vnet);
}
net_rx_pkt_init(&core->rx_pkt, core->has_vnet);
e1000x_core_prepare_eeprom(core->eeprom,
eeprom_templ,
eeprom_size,
PCI_DEVICE_GET_CLASS(core->owner)->device_id,
macaddr);
e1000e_update_rx_offloads(core);
}
void
e1000e_core_pci_uninit(E1000ECore *core)
{
int i;
timer_free(core->autoneg_timer);
e1000e_intrmgr_pci_unint(core);
qemu_del_vm_change_state_handler(core->vmstate);
for (i = 0; i < E1000E_NUM_QUEUES; i++) {
net_tx_pkt_reset(core->tx[i].tx_pkt);
net_tx_pkt_uninit(core->tx[i].tx_pkt);
}
net_rx_pkt_uninit(core->rx_pkt);
}
static const uint16_t
e1000e_phy_reg_init[E1000E_PHY_PAGES][E1000E_PHY_PAGE_SIZE] = {
[0] = {
[PHY_CTRL] = MII_CR_SPEED_SELECT_MSB |
MII_CR_FULL_DUPLEX |
MII_CR_AUTO_NEG_EN,
[PHY_STATUS] = MII_SR_EXTENDED_CAPS |
MII_SR_LINK_STATUS |
MII_SR_AUTONEG_CAPS |
MII_SR_PREAMBLE_SUPPRESS |
MII_SR_EXTENDED_STATUS |
MII_SR_10T_HD_CAPS |
MII_SR_10T_FD_CAPS |
MII_SR_100X_HD_CAPS |
MII_SR_100X_FD_CAPS,
[PHY_ID1] = 0x141,
[PHY_ID2] = E1000_PHY_ID2_82574x,
[PHY_AUTONEG_ADV] = 0xde1,
[PHY_LP_ABILITY] = 0x7e0,
[PHY_AUTONEG_EXP] = BIT(2),
[PHY_NEXT_PAGE_TX] = BIT(0) | BIT(13),
[PHY_1000T_CTRL] = BIT(8) | BIT(9) | BIT(10) | BIT(11),
[PHY_1000T_STATUS] = 0x3c00,
[PHY_EXT_STATUS] = BIT(12) | BIT(13),
[PHY_COPPER_CTRL1] = BIT(5) | BIT(6) | BIT(8) | BIT(9) |
BIT(12) | BIT(13),
[PHY_COPPER_STAT1] = BIT(3) | BIT(10) | BIT(11) | BIT(13) | BIT(15)
},
[2] = {
[PHY_MAC_CTRL1] = BIT(3) | BIT(7),
[PHY_MAC_CTRL2] = BIT(1) | BIT(2) | BIT(6) | BIT(12)
},
[3] = {
[PHY_LED_TIMER_CTRL] = BIT(0) | BIT(2) | BIT(14)
}
};
static const uint32_t e1000e_mac_reg_init[] = {
[PBA] = 0x00140014,
[LEDCTL] = BIT(1) | BIT(8) | BIT(9) | BIT(15) | BIT(17) | BIT(18),
[EXTCNF_CTRL] = BIT(3),
[EEMNGCTL] = BIT(31),
[FLASHT] = 0x2,
[FLSWCTL] = BIT(30) | BIT(31),
[FLOL] = BIT(0),
[RXDCTL] = BIT(16),
[RXDCTL1] = BIT(16),
[TIPG] = 0x8 | (0x8 << 10) | (0x6 << 20),
[RXCFGL] = 0x88F7,
[RXUDP] = 0x319,
[CTRL] = E1000_CTRL_FD | E1000_CTRL_SWDPIN2 | E1000_CTRL_SWDPIN0 |
E1000_CTRL_SPD_1000 | E1000_CTRL_SLU |
E1000_CTRL_ADVD3WUC,
[STATUS] = E1000_STATUS_ASDV_1000 | E1000_STATUS_LU,
[PSRCTL] = (2 << E1000_PSRCTL_BSIZE0_SHIFT) |
(4 << E1000_PSRCTL_BSIZE1_SHIFT) |
(4 << E1000_PSRCTL_BSIZE2_SHIFT),
[TARC0] = 0x3 | E1000_TARC_ENABLE,
[TARC1] = 0x3 | E1000_TARC_ENABLE,
[EECD] = E1000_EECD_AUTO_RD | E1000_EECD_PRES,
[EERD] = E1000_EERW_DONE,
[EEWR] = E1000_EERW_DONE,
[GCR] = E1000_L0S_ADJUST |
E1000_L1_ENTRY_LATENCY_MSB |
E1000_L1_ENTRY_LATENCY_LSB,
[TDFH] = 0x600,
[TDFT] = 0x600,
[TDFHS] = 0x600,
[TDFTS] = 0x600,
[POEMB] = 0x30D,
[PBS] = 0x028,
[MANC] = E1000_MANC_DIS_IP_CHK_ARP,
[FACTPS] = E1000_FACTPS_LAN0_ON | 0x20000000,
[SWSM] = 1,
[RXCSUM] = E1000_RXCSUM_IPOFLD | E1000_RXCSUM_TUOFLD,
[ITR] = E1000E_MIN_XITR,
[EITR...EITR + E1000E_MSIX_VEC_NUM - 1] = E1000E_MIN_XITR,
};
void
e1000e_core_reset(E1000ECore *core)
{
int i;
timer_del(core->autoneg_timer);
e1000e_intrmgr_reset(core);
memset(core->phy, 0, sizeof core->phy);
memmove(core->phy, e1000e_phy_reg_init, sizeof e1000e_phy_reg_init);
memset(core->mac, 0, sizeof core->mac);
memmove(core->mac, e1000e_mac_reg_init, sizeof e1000e_mac_reg_init);
core->rxbuf_min_shift = 1 + E1000_RING_DESC_LEN_SHIFT;
if (qemu_get_queue(core->owner_nic)->link_down) {
e1000e_link_down(core);
}
e1000x_reset_mac_addr(core->owner_nic, core->mac, core->permanent_mac);
for (i = 0; i < ARRAY_SIZE(core->tx); i++) {
net_tx_pkt_reset(core->tx[i].tx_pkt);
memset(&core->tx[i].props, 0, sizeof(core->tx[i].props));
core->tx[i].skip_cp = false;
}
}
void e1000e_core_pre_save(E1000ECore *core)
{
int i;
NetClientState *nc = qemu_get_queue(core->owner_nic);
/*
* If link is down and auto-negotiation is supported and ongoing,
* complete auto-negotiation immediately. This allows us to look
* at MII_SR_AUTONEG_COMPLETE to infer link status on load.
*/
if (nc->link_down && e1000e_have_autoneg(core)) {
core->phy[0][PHY_STATUS] |= MII_SR_AUTONEG_COMPLETE;
e1000e_update_flowctl_status(core);
}
for (i = 0; i < ARRAY_SIZE(core->tx); i++) {
if (net_tx_pkt_has_fragments(core->tx[i].tx_pkt)) {
core->tx[i].skip_cp = true;
}
}
}
int
e1000e_core_post_load(E1000ECore *core)
{
NetClientState *nc = qemu_get_queue(core->owner_nic);
/* nc.link_down can't be migrated, so infer link_down according
* to link status bit in core.mac[STATUS].
*/
nc->link_down = (core->mac[STATUS] & E1000_STATUS_LU) == 0;
return 0;
}