qemu-e2k/hw/net
Ding Hui 034d00d485 e1000: set RX descriptor status in a separate operation
The code of setting RX descriptor status field maybe work fine in
previously, however with the update of glibc version, it shows two
issues when guest using dpdk receive packets:

  1. The dpdk has a certain probability getting wrong buffer_addr

     this impact may be not obvious, such as lost a packet once in
     a while

  2. The dpdk may consume a packet twice when scan the RX desc queue
     over again

     this impact will lead a infinite wait in Qemu, since the RDT
     (tail pointer) be inscreased to equal to RDH by unexpected,
     which regard as the RX desc queue is full

Write a whole of RX desc with DD flag on is not quite correct, because
when the underlying implementation of memcpy using XMM registers to
copy e1000_rx_desc (when AVX or something else CPU feature is usable),
the bytes order of desc writing to memory is indeterminacy

We can use full-scale test case to reproduce the issue-2 by
https://github.com/BASM/qemu_dpdk_e1000_test (thanks to Leonid Myravjev)

I also write a POC test case at https://github.com/cdkey/e1000_poc
which can reproduce both of them, and easy to verify the patch effect.

The hw watchpoint also shows that, when Qemu using XMM related instructions
writing 16 bytes e1000_rx_desc, concurrent with DPDK using movb
writing 1 byte status, the final result of writing to memory will be one
of them, if it made by Qemu which DD flag is on, DPDK will consume it
again.

Setting DD status in a separate operation, can prevent the impact of
disorder memory writing by memcpy, also avoid unexpected data when
concurrent writing status by qemu and guest dpdk.

Links: https://lore.kernel.org/qemu-devel/20200102110504.GG121208@stefanha-x1.localdomain/T/

Reported-by: Leonid Myravjev <asm@asm.pp.ru>
Cc: Stefan Hajnoczi <stefanha@gmail.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: qemu-stable@nongnu.org
Tested-by: Jing Zhang <zhangjing@sangfor.com.cn>
Reviewed-by: Frank Lee <lifan38153@sangfor.com.cn>
Signed-off-by: Ding Hui <dinghui@sangfor.com.cn>
Signed-off-by: Jason Wang <jasowang@redhat.com>
2022-07-06 11:39:09 +08:00
..
can ptimer: Rename PTIMER_POLICY_DEFAULT to PTIMER_POLICY_LEGACY 2022-05-19 16:19:03 +01:00
fsl_etsec hw/net/fsl_etsec/etsec: Remove obsolete and unused etsec_create() 2022-06-11 11:44:50 +02:00
rocker Replace GCC_FMT_ATTR with G_GNUC_PRINTF 2022-03-22 14:40:51 +04:00
Kconfig
allwinner-sun8i-emac.c dma: Let dma_memory_read/write() take MemTxAttrs argument 2021-12-30 17:16:32 +01:00
allwinner_emac.c
cadence_gem.c cadence_gem: switch to use qemu_receive_packet() for loopback 2021-03-15 16:41:22 +08:00
dp8393x.c dp8393x: don't force 32-bit register access 2021-07-11 22:29:54 +02:00
e1000.c e1000: set RX descriptor status in a separate operation 2022-07-06 11:39:09 +08:00
e1000_regs.h
e1000e.c hw/net: e1000e: Correct the initial value of VET register 2021-08-02 12:19:18 +08:00
e1000e_core.c hw/net: e1000e: Clear ICR on read when using non MSI-X interrupts 2022-02-14 11:50:44 +08:00
e1000e_core.h
e1000x_common.c
e1000x_common.h
eepro100.c pci: Let ld*_pci_dma() propagate MemTxResult 2021-12-31 01:05:27 +01:00
etraxfs_eth.c
ftgmac100.c dma: Let dma_memory_read/write() take MemTxAttrs argument 2021-12-30 17:16:32 +01:00
i82596.c Do not include sysemu/sysemu.h if it's not really necessary 2021-05-02 17:24:50 +02:00
i82596.h
imx_fec.c Trivial: 3 char repeat typos 2022-06-28 11:06:02 +02:00
lan9118.c ptimer: Rename PTIMER_POLICY_DEFAULT to PTIMER_POLICY_LEGACY 2022-05-19 16:19:03 +01:00
lance.c
lasi_i82596.c Do not include sysemu/sysemu.h if it's not really necessary 2021-05-02 17:24:50 +02:00
mcf_fec.c
meson.build meson: use have_vhost_* variables to pick sources 2022-05-07 07:46:58 +02:00
mipsnet.c
msf2-emac.c Remove qemu-common.h include from most units 2022-04-06 14:31:55 +02:00
mv88w8618_eth.c hw/net: Move MV88W8618 network device out of hw/arm/ directory 2022-01-20 11:47:52 +00:00
ne2000-isa.c hw/isa: Inline and remove one-line isa_init_irq() 2022-03-08 19:38:17 +01:00
ne2000-pci.c
ne2000.c
ne2000.h
net_rx_pkt.c
net_rx_pkt.h
net_tx_pkt.c hw/net/net_tx_pkt: Fix crash detected by fuzzer 2021-07-19 09:33:39 +02:00
net_tx_pkt.h
npcm7xx_emc.c Remove qemu-common.h include from most units 2022-04-06 14:31:55 +02:00
opencores_eth.c
pcnet-pci.c
pcnet.c pcnet: switch to use qemu_receive_packet() for loopback 2021-03-15 16:41:22 +08:00
pcnet.h
rtl8139.c Replace GCC_FMT_ATTR with G_GNUC_PRINTF 2022-03-22 14:40:51 +04:00
smc91c111.c
spapr_llan.c Do not include cpu.h if it's not really necessary 2021-05-02 17:24:51 +02:00
stellaris_enet.c
sungem.c sungem: switch to use qemu_receive_packet() for loopback 2021-03-15 16:41:22 +08:00
sunhme.c
trace-events hw/net: e1000e: Clear ICR on read when using non MSI-X interrupts 2022-02-14 11:50:44 +08:00
trace.h
tulip.c tulip: Assign default MAC address if not specified 2022-05-17 16:48:23 +08:00
tulip.h
vhost_net-stub.c Revert "virtio-net: add support for configure interrupt" 2022-01-10 16:00:54 -05:00
vhost_net.c vhost-net: fix improper cleanup in vhost_net_start 2022-05-16 16:15:41 -04:00
virtio-net.c virtio-net: don't handle mq request in userspace handler for vhost-vdpa 2022-05-16 16:15:41 -04:00
vmware_utils.h
vmxnet3.c hw/net/vmxnet3: Log guest-triggerable errors using LOG_GUEST_ERROR 2022-02-14 11:50:44 +08:00
vmxnet3.h Replace config-time define HOST_WORDS_BIGENDIAN 2022-04-06 10:50:37 +02:00
vmxnet3_defs.h
vmxnet_debug.h
xen_nic.c Revert "net: Move NetClientState.info_str to dynamic allocations" 2021-04-08 17:33:59 +08:00
xgmac.c hw: Do not include qemu/log.h if it is not necessary 2021-05-02 17:24:50 +02:00
xilinx_axienet.c
xilinx_ethlite.c