OpenE2K
/
qemu-e2k
13
4
Fork 0
Code Issues 4 Pull Requests Projects 2 Releases Activity

vfio/pci: Split quirks to a separate file 

Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
This commit is contained in:
Alex Williamson 2015-09-23 13:04:45 -06:00
parent 78f33d2bfd
commit c00d61d8fa
4 changed files with 908 additions and 882 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
hw/vfio/Makefile.objs
View File

@ -1,6 +1,6 @@
ifeq ($(CONFIG_LINUX), y)
obj-$(CONFIG_SOFTMMU) += common.o
obj-$(CONFIG_PCI) += pci.o
obj-$(CONFIG_PCI) += pci.o pci-quirks.o
obj-$(CONFIG_SOFTMMU) += platform.o
obj-$(CONFIG_SOFTMMU) += calxeda-xgmac.o
endif

887
hw/vfio/pci-quirks.c Normal file
View File

@ -0,0 +1,887 @@
/*
* device quirks for PCI devices
*
* Copyright Red Hat, Inc. 2012-2015
*
* Authors:
* Alex Williamson <alex.williamson@redhat.com>
*
* This work is licensed under the terms of the GNU GPL, version 2. See
* the COPYING file in the top-level directory.
*/
#include "pci.h"
#include "trace.h"
#include "qemu/range.h"
/*
* List of device ids/vendor ids for which to disable
* option rom loading. This avoids the guest hangs during rom
* execution as noticed with the BCM 57810 card for lack of a
* more better way to handle such issues.
* The user can still override by specifying a romfile or
* rombar=1.
* Please see https://bugs.launchpad.net/qemu/+bug/1284874
* for an analysis of the 57810 card hang. When adding
* a new vendor id/device id combination below, please also add
* your card/environment details and information that could
* help in debugging to the bug tracking this issue
*/
static const VFIORomBlacklistEntry romblacklist[] = {
/* Broadcom BCM 57810 */
{ 0x14e4, 0x168e }
};
bool vfio_blacklist_opt_rom(VFIOPCIDevice *vdev)
{
PCIDevice *pdev = &vdev->pdev;
uint16_t vendor_id, device_id;
int count = 0;
vendor_id = pci_get_word(pdev->config + PCI_VENDOR_ID);
device_id = pci_get_word(pdev->config + PCI_DEVICE_ID);
while (count < ARRAY_SIZE(romblacklist)) {
if (romblacklist[count].vendor_id == vendor_id &&
romblacklist[count].device_id == device_id) {
return true;
}
count++;
}
return false;
}
/*
* Device specific quirks
*/
/* Is range1 fully contained within range2? */
static bool vfio_range_contained(uint64_t first1, uint64_t len1,
uint64_t first2, uint64_t len2) {
return (first1 >= first2 && first1 + len1 <= first2 + len2);
}
static bool vfio_flags_enabled(uint8_t flags, uint8_t mask)
{
return (mask && (flags & mask) == mask);
}
static uint64_t vfio_generic_window_quirk_read(void *opaque,
hwaddr addr, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
uint64_t data;
if (vfio_flags_enabled(quirk->data.flags, quirk->data.read_flags) &&
ranges_overlap(addr, size,
quirk->data.data_offset, quirk->data.data_size)) {
hwaddr offset = addr - quirk->data.data_offset;
if (!vfio_range_contained(addr, size, quirk->data.data_offset,
quirk->data.data_size)) {
hw_error("%s: window data read not fully contained: %s",
__func__, memory_region_name(&quirk->mem));
}
data = vfio_pci_read_config(&vdev->pdev,
quirk->data.address_val + offset, size);
trace_vfio_generic_window_quirk_read(memory_region_name(&quirk->mem),
vdev->vbasedev.name,
quirk->data.bar,
addr, size, data);
} else {
data = vfio_region_read(&vdev->bars[quirk->data.bar].region,
addr + quirk->data.base_offset, size);
}
return data;
}
static void vfio_generic_window_quirk_write(void *opaque, hwaddr addr,
uint64_t data, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
if (ranges_overlap(addr, size,
quirk->data.address_offset, quirk->data.address_size)) {
if (addr != quirk->data.address_offset) {
hw_error("%s: offset write into address window: %s",
__func__, memory_region_name(&quirk->mem));
}
if ((data & ~quirk->data.address_mask) == quirk->data.address_match) {
quirk->data.flags |= quirk->data.write_flags |
quirk->data.read_flags;
quirk->data.address_val = data & quirk->data.address_mask;
} else {
quirk->data.flags &= ~(quirk->data.write_flags |
quirk->data.read_flags);
}
}
if (vfio_flags_enabled(quirk->data.flags, quirk->data.write_flags) &&
ranges_overlap(addr, size,
quirk->data.data_offset, quirk->data.data_size)) {
hwaddr offset = addr - quirk->data.data_offset;
if (!vfio_range_contained(addr, size, quirk->data.data_offset,
quirk->data.data_size)) {
hw_error("%s: window data write not fully contained: %s",
__func__, memory_region_name(&quirk->mem));
}
vfio_pci_write_config(&vdev->pdev,
quirk->data.address_val + offset, data, size);
trace_vfio_generic_window_quirk_write(memory_region_name(&quirk->mem),
vdev->vbasedev.name,
quirk->data.bar,
addr, data, size);
return;
}
vfio_region_write(&vdev->bars[quirk->data.bar].region,
addr + quirk->data.base_offset, data, size);
}
static const MemoryRegionOps vfio_generic_window_quirk = {
.read = vfio_generic_window_quirk_read,
.write = vfio_generic_window_quirk_write,
.endianness = DEVICE_LITTLE_ENDIAN,
};
static uint64_t vfio_generic_quirk_read(void *opaque,
hwaddr addr, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
hwaddr base = quirk->data.address_match & TARGET_PAGE_MASK;
hwaddr offset = quirk->data.address_match & ~TARGET_PAGE_MASK;
uint64_t data;
if (vfio_flags_enabled(quirk->data.flags, quirk->data.read_flags) &&
ranges_overlap(addr, size, offset, quirk->data.address_mask + 1)) {
if (!vfio_range_contained(addr, size, offset,
quirk->data.address_mask + 1)) {
hw_error("%s: read not fully contained: %s",
__func__, memory_region_name(&quirk->mem));
}
data = vfio_pci_read_config(&vdev->pdev, addr - offset, size);
trace_vfio_generic_quirk_read(memory_region_name(&quirk->mem),
vdev->vbasedev.name, quirk->data.bar,
addr + base, size, data);
} else {
data = vfio_region_read(&vdev->bars[quirk->data.bar].region,
addr + base, size);
}
return data;
}
static void vfio_generic_quirk_write(void *opaque, hwaddr addr,
uint64_t data, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
hwaddr base = quirk->data.address_match & TARGET_PAGE_MASK;
hwaddr offset = quirk->data.address_match & ~TARGET_PAGE_MASK;
if (vfio_flags_enabled(quirk->data.flags, quirk->data.write_flags) &&
ranges_overlap(addr, size, offset, quirk->data.address_mask + 1)) {
if (!vfio_range_contained(addr, size, offset,
quirk->data.address_mask + 1)) {
hw_error("%s: write not fully contained: %s",
__func__, memory_region_name(&quirk->mem));
}
vfio_pci_write_config(&vdev->pdev, addr - offset, data, size);
trace_vfio_generic_quirk_write(memory_region_name(&quirk->mem),
vdev->vbasedev.name, quirk->data.bar,
addr + base, data, size);
} else {
vfio_region_write(&vdev->bars[quirk->data.bar].region,
addr + base, data, size);
}
}
static const MemoryRegionOps vfio_generic_quirk = {
.read = vfio_generic_quirk_read,
.write = vfio_generic_quirk_write,
.endianness = DEVICE_LITTLE_ENDIAN,
};
#define PCI_VENDOR_ID_ATI 0x1002
/*
* Radeon HD cards (HD5450 & HD7850) report the upper byte of the I/O port BAR
* through VGA register 0x3c3. On newer cards, the I/O port BAR is always
* BAR4 (older cards like the X550 used BAR1, but we don't care to support
* those). Note that on bare metal, a read of 0x3c3 doesn't always return the
* I/O port BAR address. Originally this was coded to return the virtual BAR
* address only if the physical register read returns the actual BAR address,
* but users have reported greater success if we return the virtual address
* unconditionally.
*/
static uint64_t vfio_ati_3c3_quirk_read(void *opaque,
hwaddr addr, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
uint64_t data = vfio_pci_read_config(&vdev->pdev,
PCI_BASE_ADDRESS_0 + (4 * 4) + 1,
size);
trace_vfio_ati_3c3_quirk_read(data);
return data;
}
static const MemoryRegionOps vfio_ati_3c3_quirk = {
.read = vfio_ati_3c3_quirk_read,
.endianness = DEVICE_LITTLE_ENDIAN,
};
static void vfio_vga_probe_ati_3c3_quirk(VFIOPCIDevice *vdev)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
if (pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_ATI) {
return;
}
/*
* As long as the BAR is >= 256 bytes it will be aligned such that the
* lower byte is always zero. Filter out anything else, if it exists.
*/
if (!vdev->bars[4].ioport || vdev->bars[4].region.size < 256) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_ati_3c3_quirk, quirk,
"vfio-ati-3c3-quirk", 1);
memory_region_add_subregion(&vdev->vga.region[QEMU_PCI_VGA_IO_HI].mem,
3 /* offset 3 bytes from 0x3c0 */, &quirk->mem);
QLIST_INSERT_HEAD(&vdev->vga.region[QEMU_PCI_VGA_IO_HI].quirks,
quirk, next);
trace_vfio_vga_probe_ati_3c3_quirk(vdev->vbasedev.name);
}
/*
* Newer ATI/AMD devices, including HD5450 and HD7850, have a window to PCI
* config space through MMIO BAR2 at offset 0x4000. Nothing seems to access
* the MMIO space directly, but a window to this space is provided through
* I/O port BAR4. Offset 0x0 is the address register and offset 0x4 is the
* data register. When the address is programmed to a range of 0x4000-0x4fff
* PCI configuration space is available. Experimentation seems to indicate
* that only read-only access is provided, but we drop writes when the window
* is enabled to config space nonetheless.
*/
static void vfio_probe_ati_bar4_window_quirk(VFIOPCIDevice *vdev, int nr)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
if (!vdev->has_vga || nr != 4 ||
pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_ATI) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.address_size = 4;
quirk->data.data_offset = 4;
quirk->data.data_size = 4;
quirk->data.address_match = 0x4000;
quirk->data.address_mask = PCIE_CONFIG_SPACE_SIZE - 1;
quirk->data.bar = nr;
quirk->data.read_flags = quirk->data.write_flags = 1;
memory_region_init_io(&quirk->mem, OBJECT(vdev),
&vfio_generic_window_quirk, quirk,
"vfio-ati-bar4-window-quirk", 8);
memory_region_add_subregion_overlap(&vdev->bars[nr].region.mem,
quirk->data.base_offset, &quirk->mem, 1);
QLIST_INSERT_HEAD(&vdev->bars[nr].quirks, quirk, next);
trace_vfio_probe_ati_bar4_window_quirk(vdev->vbasedev.name);
}
/*
* Trap the BAR2 MMIO window to config space as well.
*/
static void vfio_probe_ati_bar2_4000_quirk(VFIOPCIDevice *vdev, int nr)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
/* Only enable on newer devices where BAR2 is 64bit */
if (!vdev->has_vga || nr != 2 || !vdev->bars[2].mem64 ||
pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_ATI) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.flags = quirk->data.read_flags = quirk->data.write_flags = 1;
quirk->data.address_match = 0x4000;
quirk->data.address_mask = PCIE_CONFIG_SPACE_SIZE - 1;
quirk->data.bar = nr;
memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_generic_quirk, quirk,
"vfio-ati-bar2-4000-quirk",
TARGET_PAGE_ALIGN(quirk->data.address_mask + 1));
memory_region_add_subregion_overlap(&vdev->bars[nr].region.mem,
quirk->data.address_match & TARGET_PAGE_MASK,
&quirk->mem, 1);
QLIST_INSERT_HEAD(&vdev->bars[nr].quirks, quirk, next);
trace_vfio_probe_ati_bar2_4000_quirk(vdev->vbasedev.name);
}
/*
* Older ATI/AMD cards like the X550 have a similar window to that above.
* I/O port BAR1 provides a window to a mirror of PCI config space located
* in BAR2 at offset 0xf00. We don't care to support such older cards, but
* note it for future reference.
*/
#define PCI_VENDOR_ID_NVIDIA 0x10de
/*
* Nvidia has several different methods to get to config space, the
* nouveu project has several of these documented here:
* https://github.com/pathscale/envytools/tree/master/hwdocs
*
* The first quirk is actually not documented in envytools and is found
* on 10de:01d1 (NVIDIA Corporation G72 [GeForce 7300 LE]). This is an
* NV46 chipset. The backdoor uses the legacy VGA I/O ports to access
* the mirror of PCI config space found at BAR0 offset 0x1800. The access
* sequence first writes 0x338 to I/O port 0x3d4. The target offset is
* then written to 0x3d0. Finally 0x538 is written for a read and 0x738
* is written for a write to 0x3d4. The BAR0 offset is then accessible
* through 0x3d0. This quirk doesn't seem to be necessary on newer cards
* that use the I/O port BAR5 window but it doesn't hurt to leave it.
*/
enum {
NV_3D0_NONE = 0,
NV_3D0_SELECT,
NV_3D0_WINDOW,
NV_3D0_READ,
NV_3D0_WRITE,
};
static uint64_t vfio_nvidia_3d0_quirk_read(void *opaque,
hwaddr addr, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
PCIDevice *pdev = &vdev->pdev;
uint64_t data = vfio_vga_read(&vdev->vga.region[QEMU_PCI_VGA_IO_HI],
addr + quirk->data.base_offset, size);
if (quirk->data.flags == NV_3D0_READ && addr == quirk->data.data_offset) {
data = vfio_pci_read_config(pdev, quirk->data.address_val, size);
trace_vfio_nvidia_3d0_quirk_read(size, data);
}
quirk->data.flags = NV_3D0_NONE;
return data;
}
static void vfio_nvidia_3d0_quirk_write(void *opaque, hwaddr addr,
uint64_t data, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
PCIDevice *pdev = &vdev->pdev;
switch (quirk->data.flags) {
case NV_3D0_NONE:
if (addr == quirk->data.address_offset && data == 0x338) {
quirk->data.flags = NV_3D0_SELECT;
}
break;
case NV_3D0_SELECT:
quirk->data.flags = NV_3D0_NONE;
if (addr == quirk->data.data_offset &&
(data & ~quirk->data.address_mask) == quirk->data.address_match) {
quirk->data.flags = NV_3D0_WINDOW;
quirk->data.address_val = data & quirk->data.address_mask;
}
break;
case NV_3D0_WINDOW:
quirk->data.flags = NV_3D0_NONE;
if (addr == quirk->data.address_offset) {
if (data == 0x538) {
quirk->data.flags = NV_3D0_READ;
} else if (data == 0x738) {
quirk->data.flags = NV_3D0_WRITE;
}
}
break;
case NV_3D0_WRITE:
quirk->data.flags = NV_3D0_NONE;
if (addr == quirk->data.data_offset) {
vfio_pci_write_config(pdev, quirk->data.address_val, data, size);
trace_vfio_nvidia_3d0_quirk_write(data, size);
return;
}
break;
}
vfio_vga_write(&vdev->vga.region[QEMU_PCI_VGA_IO_HI],
addr + quirk->data.base_offset, data, size);
}
static const MemoryRegionOps vfio_nvidia_3d0_quirk = {
.read = vfio_nvidia_3d0_quirk_read,
.write = vfio_nvidia_3d0_quirk_write,
.endianness = DEVICE_LITTLE_ENDIAN,
};
static void vfio_vga_probe_nvidia_3d0_quirk(VFIOPCIDevice *vdev)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
if (pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_NVIDIA ||
!vdev->bars[1].region.size) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.base_offset = 0x10;
quirk->data.address_offset = 4;
quirk->data.address_size = 2;
quirk->data.address_match = 0x1800;
quirk->data.address_mask = PCI_CONFIG_SPACE_SIZE - 1;
quirk->data.data_offset = 0;
quirk->data.data_size = 4;
memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_nvidia_3d0_quirk,
quirk, "vfio-nvidia-3d0-quirk", 6);
memory_region_add_subregion(&vdev->vga.region[QEMU_PCI_VGA_IO_HI].mem,
quirk->data.base_offset, &quirk->mem);
QLIST_INSERT_HEAD(&vdev->vga.region[QEMU_PCI_VGA_IO_HI].quirks,
quirk, next);
trace_vfio_vga_probe_nvidia_3d0_quirk(vdev->vbasedev.name);
}
/*
* The second quirk is documented in envytools. The I/O port BAR5 is just
* a set of address/data ports to the MMIO BARs. The BAR we care about is
* again BAR0. This backdoor is apparently a bit newer than the one above
* so we need to not only trap 256 bytes @0x1800, but all of PCI config
* space, including extended space is available at the 4k @0x88000.
*/
enum {
NV_BAR5_ADDRESS = 0x1,
NV_BAR5_ENABLE = 0x2,
NV_BAR5_MASTER = 0x4,
NV_BAR5_VALID = 0x7,
};
static void vfio_nvidia_bar5_window_quirk_write(void *opaque, hwaddr addr,
uint64_t data, unsigned size)
{
VFIOQuirk *quirk = opaque;
switch (addr) {
case 0x0:
if (data & 0x1) {
quirk->data.flags |= NV_BAR5_MASTER;
} else {
quirk->data.flags &= ~NV_BAR5_MASTER;
}
break;
case 0x4:
if (data & 0x1) {
quirk->data.flags |= NV_BAR5_ENABLE;
} else {
quirk->data.flags &= ~NV_BAR5_ENABLE;
}
break;
case 0x8:
if (quirk->data.flags & NV_BAR5_MASTER) {
if ((data & ~0xfff) == 0x88000) {
quirk->data.flags |= NV_BAR5_ADDRESS;
quirk->data.address_val = data & 0xfff;
} else if ((data & ~0xff) == 0x1800) {
quirk->data.flags |= NV_BAR5_ADDRESS;
quirk->data.address_val = data & 0xff;
} else {
quirk->data.flags &= ~NV_BAR5_ADDRESS;
}
}
break;
}
vfio_generic_window_quirk_write(opaque, addr, data, size);
}
static const MemoryRegionOps vfio_nvidia_bar5_window_quirk = {
.read = vfio_generic_window_quirk_read,
.write = vfio_nvidia_bar5_window_quirk_write,
.valid.min_access_size = 4,
.endianness = DEVICE_LITTLE_ENDIAN,
};
static void vfio_probe_nvidia_bar5_window_quirk(VFIOPCIDevice *vdev, int nr)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
if (!vdev->has_vga || nr != 5 ||
pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_NVIDIA) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.read_flags = quirk->data.write_flags = NV_BAR5_VALID;
quirk->data.address_offset = 0x8;
quirk->data.address_size = 0; /* actually 4, but avoids generic code */
quirk->data.data_offset = 0xc;
quirk->data.data_size = 4;
quirk->data.bar = nr;
memory_region_init_io(&quirk->mem, OBJECT(vdev),
&vfio_nvidia_bar5_window_quirk, quirk,
"vfio-nvidia-bar5-window-quirk", 16);
memory_region_add_subregion_overlap(&vdev->bars[nr].region.mem,
0, &quirk->mem, 1);
QLIST_INSERT_HEAD(&vdev->bars[nr].quirks, quirk, next);
trace_vfio_probe_nvidia_bar5_window_quirk(vdev->vbasedev.name);
}
static void vfio_nvidia_88000_quirk_write(void *opaque, hwaddr addr,
uint64_t data, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
PCIDevice *pdev = &vdev->pdev;
hwaddr base = quirk->data.address_match & TARGET_PAGE_MASK;
vfio_generic_quirk_write(opaque, addr, data, size);
/*
* Nvidia seems to acknowledge MSI interrupts by writing 0xff to the
* MSI capability ID register. Both the ID and next register are
* read-only, so we allow writes covering either of those to real hw.
* NB - only fixed for the 0x88000 MMIO window.
*/
if ((pdev->cap_present & QEMU_PCI_CAP_MSI) &&
vfio_range_contained(addr, size, pdev->msi_cap, PCI_MSI_FLAGS)) {
vfio_region_write(&vdev->bars[quirk->data.bar].region,
addr + base, data, size);
}
}
static const MemoryRegionOps vfio_nvidia_88000_quirk = {
.read = vfio_generic_quirk_read,
.write = vfio_nvidia_88000_quirk_write,
.endianness = DEVICE_LITTLE_ENDIAN,
};
/*
* Finally, BAR0 itself. We want to redirect any accesses to either
* 0x1800 or 0x88000 through the PCI config space access functions.
*
* NB - quirk at a page granularity or else they don't seem to work when
* BARs are mmap'd
*
* Here's offset 0x88000...
*/
static void vfio_probe_nvidia_bar0_88000_quirk(VFIOPCIDevice *vdev, int nr)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
uint16_t vendor, class;
vendor = pci_get_word(pdev->config + PCI_VENDOR_ID);
class = pci_get_word(pdev->config + PCI_CLASS_DEVICE);
if (nr != 0 || vendor != PCI_VENDOR_ID_NVIDIA ||
class != PCI_CLASS_DISPLAY_VGA) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.flags = quirk->data.read_flags = quirk->data.write_flags = 1;
quirk->data.address_match = 0x88000;
quirk->data.address_mask = PCIE_CONFIG_SPACE_SIZE - 1;
quirk->data.bar = nr;
memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_nvidia_88000_quirk,
quirk, "vfio-nvidia-bar0-88000-quirk",
TARGET_PAGE_ALIGN(quirk->data.address_mask + 1));
memory_region_add_subregion_overlap(&vdev->bars[nr].region.mem,
quirk->data.address_match & TARGET_PAGE_MASK,
&quirk->mem, 1);
QLIST_INSERT_HEAD(&vdev->bars[nr].quirks, quirk, next);
trace_vfio_probe_nvidia_bar0_88000_quirk(vdev->vbasedev.name);
}
/*
* And here's the same for BAR0 offset 0x1800...
*/
static void vfio_probe_nvidia_bar0_1800_quirk(VFIOPCIDevice *vdev, int nr)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
if (!vdev->has_vga || nr != 0 ||
pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_NVIDIA) {
return;
}
/* Log the chipset ID */
trace_vfio_probe_nvidia_bar0_1800_quirk_id(
(unsigned int)(vfio_region_read(&vdev->bars[0].region, 0, 4) >> 20)
& 0xff);
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.flags = quirk->data.read_flags = quirk->data.write_flags = 1;
quirk->data.address_match = 0x1800;
quirk->data.address_mask = PCI_CONFIG_SPACE_SIZE - 1;
quirk->data.bar = nr;
memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_generic_quirk, quirk,
"vfio-nvidia-bar0-1800-quirk",
TARGET_PAGE_ALIGN(quirk->data.address_mask + 1));
memory_region_add_subregion_overlap(&vdev->bars[nr].region.mem,
quirk->data.address_match & TARGET_PAGE_MASK,
&quirk->mem, 1);
QLIST_INSERT_HEAD(&vdev->bars[nr].quirks, quirk, next);
trace_vfio_probe_nvidia_bar0_1800_quirk(vdev->vbasedev.name);
}
/*
* TODO - Some Nvidia devices provide config access to their companion HDA
* device and even to their parent bridge via these config space mirrors.
* Add quirks for those regions.
*/
#define PCI_VENDOR_ID_REALTEK 0x10ec
/*
* RTL8168 devices have a backdoor that can access the MSI-X table. At BAR2
* offset 0x70 there is a dword data register, offset 0x74 is a dword address
* register. According to the Linux r8169 driver, the MSI-X table is addressed
* when the "type" portion of the address register is set to 0x1. This appears
* to be bits 16:30. Bit 31 is both a write indicator and some sort of
* "address latched" indicator. Bits 12:15 are a mask field, which we can
* ignore because the MSI-X table should always be accessed as a dword (full
* mask). Bits 0:11 is offset within the type.
*
* Example trace:
*
* Read from MSI-X table offset 0
* vfio: vfio_bar_write(0000:05:00.0:BAR2+0x74, 0x1f000, 4) // store read addr
* vfio: vfio_bar_read(0000:05:00.0:BAR2+0x74, 4) = 0x8001f000 // latch
* vfio: vfio_bar_read(0000:05:00.0:BAR2+0x70, 4) = 0xfee00398 // read data
*
* Write 0xfee00000 to MSI-X table offset 0
* vfio: vfio_bar_write(0000:05:00.0:BAR2+0x70, 0xfee00000, 4) // write data
* vfio: vfio_bar_write(0000:05:00.0:BAR2+0x74, 0x8001f000, 4) // do write
* vfio: vfio_bar_read(0000:05:00.0:BAR2+0x74, 4) = 0x1f000 // complete
*/
static uint64_t vfio_rtl8168_window_quirk_read(void *opaque,
hwaddr addr, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
uint64_t val = 0;
if (!quirk->data.flags) { /* Non-MSI-X table access */
return vfio_region_read(&vdev->bars[quirk->data.bar].region,
addr + 0x70, size);
}
switch (addr) {
case 4: /* address */
val = quirk->data.address_match ^ 0x80000000U; /* latch/complete */
break;
case 0: /* data */
if ((vdev->pdev.cap_present & QEMU_PCI_CAP_MSIX)) {
memory_region_dispatch_read(&vdev->pdev.msix_table_mmio,
(hwaddr)(quirk->data.address_match & 0xfff),
&val, size, MEMTXATTRS_UNSPECIFIED);
}
break;
}
trace_vfio_rtl8168_quirk_read(vdev->vbasedev.name,
addr ? "address" : "data", val);
return val;
}
static void vfio_rtl8168_window_quirk_write(void *opaque, hwaddr addr,
uint64_t data, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
switch (addr) {
case 4: /* address */
if ((data & 0x7fff0000) == 0x10000) { /* MSI-X table */
quirk->data.flags = 1; /* Activate reads */
quirk->data.address_match = data;
trace_vfio_rtl8168_quirk_write(vdev->vbasedev.name, data);
if (data & 0x80000000U) { /* Do write */
if (vdev->pdev.cap_present & QEMU_PCI_CAP_MSIX) {
hwaddr offset = data & 0xfff;
uint64_t val = quirk->data.address_mask;
trace_vfio_rtl8168_quirk_msix(vdev->vbasedev.name,
(uint16_t)offset, val);
/* Write to the proper guest MSI-X table instead */
memory_region_dispatch_write(&vdev->pdev.msix_table_mmio,
offset, val, size,
MEMTXATTRS_UNSPECIFIED);
}
return; /* Do not write guest MSI-X data to hardware */
}
} else {
quirk->data.flags = 0; /* De-activate reads, non-MSI-X */
}
break;
case 0: /* data */
quirk->data.address_mask = data;
break;
}
vfio_region_write(&vdev->bars[quirk->data.bar].region,
addr + 0x70, data, size);
}
static const MemoryRegionOps vfio_rtl8168_window_quirk = {
.read = vfio_rtl8168_window_quirk_read,
.write = vfio_rtl8168_window_quirk_write,
.valid = {
.min_access_size = 4,
.max_access_size = 4,
.unaligned = false,
},
.endianness = DEVICE_LITTLE_ENDIAN,
};
static void vfio_probe_rtl8168_bar2_window_quirk(VFIOPCIDevice *vdev, int nr)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
if (pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_REALTEK ||
pci_get_word(pdev->config + PCI_DEVICE_ID) != 0x8168 || nr != 2) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.bar = nr;
memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_rtl8168_window_quirk,
quirk, "vfio-rtl8168-window-quirk", 8);
memory_region_add_subregion_overlap(&vdev->bars[nr].region.mem,
0x70, &quirk->mem, 1);
QLIST_INSERT_HEAD(&vdev->bars[nr].quirks, quirk, next);
trace_vfio_rtl8168_quirk_enable(vdev->vbasedev.name);
}
/*
* Common quirk probe entry points.
*/
void vfio_vga_quirk_setup(VFIOPCIDevice *vdev)
{
vfio_vga_probe_ati_3c3_quirk(vdev);
vfio_vga_probe_nvidia_3d0_quirk(vdev);
}
void vfio_vga_quirk_teardown(VFIOPCIDevice *vdev)
{
VFIOQuirk *quirk;
int i;
for (i = 0; i < ARRAY_SIZE(vdev->vga.region); i++) {
QLIST_FOREACH(quirk, &vdev->vga.region[i].quirks, next) {
memory_region_del_subregion(&vdev->vga.region[i].mem, &quirk->mem);
}
}
}
void vfio_vga_quirk_free(VFIOPCIDevice *vdev)
{
int i;
for (i = 0; i < ARRAY_SIZE(vdev->vga.region); i++) {
while (!QLIST_EMPTY(&vdev->vga.region[i].quirks)) {
VFIOQuirk *quirk = QLIST_FIRST(&vdev->vga.region[i].quirks);
object_unparent(OBJECT(&quirk->mem));
QLIST_REMOVE(quirk, next);
g_free(quirk);
}
}
}
void vfio_bar_quirk_setup(VFIOPCIDevice *vdev, int nr)
{
vfio_probe_ati_bar4_window_quirk(vdev, nr);
vfio_probe_ati_bar2_4000_quirk(vdev, nr);
vfio_probe_nvidia_bar5_window_quirk(vdev, nr);
vfio_probe_nvidia_bar0_88000_quirk(vdev, nr);
vfio_probe_nvidia_bar0_1800_quirk(vdev, nr);
vfio_probe_rtl8168_bar2_window_quirk(vdev, nr);
}
void vfio_bar_quirk_teardown(VFIOPCIDevice *vdev, int nr)
{
VFIOBAR *bar = &vdev->bars[nr];
VFIOQuirk *quirk;
QLIST_FOREACH(quirk, &bar->quirks, next) {
memory_region_del_subregion(&bar->region.mem, &quirk->mem);
}
}
void vfio_bar_quirk_free(VFIOPCIDevice *vdev, int nr)
{
VFIOBAR *bar = &vdev->bars[nr];
while (!QLIST_EMPTY(&bar->quirks)) {
VFIOQuirk *quirk = QLIST_FIRST(&bar->quirks);
object_unparent(OBJECT(&quirk->mem));
QLIST_REMOVE(quirk, next);
g_free(quirk);
}
}

886
hw/vfio/pci.c
View File

@ -35,30 +35,9 @@
#include "pci.h"
#include "trace.h"
/*
* List of device ids/vendor ids for which to disable
* option rom loading. This avoids the guest hangs during rom
* execution as noticed with the BCM 57810 card for lack of a
* more better way to handle such issues.
* The user can still override by specifying a romfile or
* rombar=1.
* Please see https://bugs.launchpad.net/qemu/+bug/1284874
* for an analysis of the 57810 card hang. When adding
* a new vendor id/device id combination below, please also add
* your card/environment details and information that could
* help in debugging to the bug tracking this issue
*/
static const VFIORomBlacklistEntry romblacklist[] = {
/* Broadcom BCM 57810 */
{ 0x14e4, 0x168e }
};
#define MSIX_CAP_LENGTH 12
static void vfio_disable_interrupts(VFIOPCIDevice *vdev);
static uint32_t vfio_pci_read_config(PCIDevice *pdev, uint32_t addr, int len);
static void vfio_pci_write_config(PCIDevice *pdev, uint32_t addr,
uint32_t val, int len);
static void vfio_mmap_set_enabled(VFIOPCIDevice *vdev, bool enabled);
/*
@ -888,26 +867,6 @@ static const MemoryRegionOps vfio_rom_ops = {
.endianness = DEVICE_LITTLE_ENDIAN,
};
static bool vfio_blacklist_opt_rom(VFIOPCIDevice *vdev)
{
PCIDevice *pdev = &vdev->pdev;
uint16_t vendor_id, device_id;
int count = 0;
vendor_id = pci_get_word(pdev->config + PCI_VENDOR_ID);
device_id = pci_get_word(pdev->config + PCI_DEVICE_ID);
while (count < ARRAY_SIZE(romblacklist)) {
if (romblacklist[count].vendor_id == vendor_id &&
romblacklist[count].device_id == device_id) {
return true;
}
count++;
}
return false;
}
static void vfio_pci_size_rom(VFIOPCIDevice *vdev)
{
uint32_t orig, size = cpu_to_le32((uint32_t)PCI_ROM_ADDRESS_MASK);
@ -985,7 +944,7 @@ static void vfio_pci_size_rom(VFIOPCIDevice *vdev)
vdev->rom_read_failed = false;
}
static void vfio_vga_write(void *opaque, hwaddr addr,
void vfio_vga_write(void *opaque, hwaddr addr,
uint64_t data, unsigned size)
{
VFIOVGARegion *region = opaque;
@ -1021,7 +980,7 @@ static void vfio_vga_write(void *opaque, hwaddr addr,
trace_vfio_vga_write(region->offset + addr, data, size);
}
static uint64_t vfio_vga_read(void *opaque, hwaddr addr, unsigned size)
uint64_t vfio_vga_read(void *opaque, hwaddr addr, unsigned size)
{
VFIOVGARegion *region = opaque;
VFIOVGA *vga = container_of(region, VFIOVGA, region[region->nr]);
@ -1066,845 +1025,10 @@ static const MemoryRegionOps vfio_vga_ops = {
.endianness = DEVICE_LITTLE_ENDIAN,
};
/*
* Device specific quirks
*/
/* Is range1 fully contained within range2? */
static bool vfio_range_contained(uint64_t first1, uint64_t len1,
uint64_t first2, uint64_t len2) {
return (first1 >= first2 && first1 + len1 <= first2 + len2);
}
static bool vfio_flags_enabled(uint8_t flags, uint8_t mask)
{
return (mask && (flags & mask) == mask);
}
static uint64_t vfio_generic_window_quirk_read(void *opaque,
hwaddr addr, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
uint64_t data;
if (vfio_flags_enabled(quirk->data.flags, quirk->data.read_flags) &&
ranges_overlap(addr, size,
quirk->data.data_offset, quirk->data.data_size)) {
hwaddr offset = addr - quirk->data.data_offset;
if (!vfio_range_contained(addr, size, quirk->data.data_offset,
quirk->data.data_size)) {
hw_error("%s: window data read not fully contained: %s",
__func__, memory_region_name(&quirk->mem));
}
data = vfio_pci_read_config(&vdev->pdev,
quirk->data.address_val + offset, size);
trace_vfio_generic_window_quirk_read(memory_region_name(&quirk->mem),
vdev->vbasedev.name,
quirk->data.bar,
addr, size, data);
} else {
data = vfio_region_read(&vdev->bars[quirk->data.bar].region,
addr + quirk->data.base_offset, size);
}
return data;
}
static void vfio_generic_window_quirk_write(void *opaque, hwaddr addr,
uint64_t data, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
if (ranges_overlap(addr, size,
quirk->data.address_offset, quirk->data.address_size)) {
if (addr != quirk->data.address_offset) {
hw_error("%s: offset write into address window: %s",
__func__, memory_region_name(&quirk->mem));
}
if ((data & ~quirk->data.address_mask) == quirk->data.address_match) {
quirk->data.flags |= quirk->data.write_flags |
quirk->data.read_flags;
quirk->data.address_val = data & quirk->data.address_mask;
} else {
quirk->data.flags &= ~(quirk->data.write_flags |
quirk->data.read_flags);
}
}
if (vfio_flags_enabled(quirk->data.flags, quirk->data.write_flags) &&
ranges_overlap(addr, size,
quirk->data.data_offset, quirk->data.data_size)) {
hwaddr offset = addr - quirk->data.data_offset;
if (!vfio_range_contained(addr, size, quirk->data.data_offset,
quirk->data.data_size)) {
hw_error("%s: window data write not fully contained: %s",
__func__, memory_region_name(&quirk->mem));
}
vfio_pci_write_config(&vdev->pdev,
quirk->data.address_val + offset, data, size);
trace_vfio_generic_window_quirk_write(memory_region_name(&quirk->mem),
vdev->vbasedev.name,
quirk->data.bar,
addr, data, size);
return;
}
vfio_region_write(&vdev->bars[quirk->data.bar].region,
addr + quirk->data.base_offset, data, size);
}
static const MemoryRegionOps vfio_generic_window_quirk = {
.read = vfio_generic_window_quirk_read,
.write = vfio_generic_window_quirk_write,
.endianness = DEVICE_LITTLE_ENDIAN,
};
static uint64_t vfio_generic_quirk_read(void *opaque,
hwaddr addr, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
hwaddr base = quirk->data.address_match & TARGET_PAGE_MASK;
hwaddr offset = quirk->data.address_match & ~TARGET_PAGE_MASK;
uint64_t data;
if (vfio_flags_enabled(quirk->data.flags, quirk->data.read_flags) &&
ranges_overlap(addr, size, offset, quirk->data.address_mask + 1)) {
if (!vfio_range_contained(addr, size, offset,
quirk->data.address_mask + 1)) {
hw_error("%s: read not fully contained: %s",
__func__, memory_region_name(&quirk->mem));
}
data = vfio_pci_read_config(&vdev->pdev, addr - offset, size);
trace_vfio_generic_quirk_read(memory_region_name(&quirk->mem),
vdev->vbasedev.name, quirk->data.bar,
addr + base, size, data);
} else {
data = vfio_region_read(&vdev->bars[quirk->data.bar].region,
addr + base, size);
}
return data;
}
static void vfio_generic_quirk_write(void *opaque, hwaddr addr,
uint64_t data, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
hwaddr base = quirk->data.address_match & TARGET_PAGE_MASK;
hwaddr offset = quirk->data.address_match & ~TARGET_PAGE_MASK;
if (vfio_flags_enabled(quirk->data.flags, quirk->data.write_flags) &&
ranges_overlap(addr, size, offset, quirk->data.address_mask + 1)) {
if (!vfio_range_contained(addr, size, offset,
quirk->data.address_mask + 1)) {
hw_error("%s: write not fully contained: %s",
__func__, memory_region_name(&quirk->mem));
}
vfio_pci_write_config(&vdev->pdev, addr - offset, data, size);
trace_vfio_generic_quirk_write(memory_region_name(&quirk->mem),
vdev->vbasedev.name, quirk->data.bar,
addr + base, data, size);
} else {
vfio_region_write(&vdev->bars[quirk->data.bar].region,
addr + base, data, size);
}
}
static const MemoryRegionOps vfio_generic_quirk = {
.read = vfio_generic_quirk_read,
.write = vfio_generic_quirk_write,
.endianness = DEVICE_LITTLE_ENDIAN,
};
#define PCI_VENDOR_ID_ATI 0x1002
/*
* Radeon HD cards (HD5450 & HD7850) report the upper byte of the I/O port BAR
* through VGA register 0x3c3. On newer cards, the I/O port BAR is always
* BAR4 (older cards like the X550 used BAR1, but we don't care to support
* those). Note that on bare metal, a read of 0x3c3 doesn't always return the
* I/O port BAR address. Originally this was coded to return the virtual BAR
* address only if the physical register read returns the actual BAR address,
* but users have reported greater success if we return the virtual address
* unconditionally.
*/
static uint64_t vfio_ati_3c3_quirk_read(void *opaque,
hwaddr addr, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
uint64_t data = vfio_pci_read_config(&vdev->pdev,
PCI_BASE_ADDRESS_0 + (4 * 4) + 1,
size);
trace_vfio_ati_3c3_quirk_read(data);
return data;
}
static const MemoryRegionOps vfio_ati_3c3_quirk = {
.read = vfio_ati_3c3_quirk_read,
.endianness = DEVICE_LITTLE_ENDIAN,
};
static void vfio_vga_probe_ati_3c3_quirk(VFIOPCIDevice *vdev)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
if (pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_ATI) {
return;
}
/*
* As long as the BAR is >= 256 bytes it will be aligned such that the
* lower byte is always zero. Filter out anything else, if it exists.
*/
if (!vdev->bars[4].ioport || vdev->bars[4].region.size < 256) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_ati_3c3_quirk, quirk,
"vfio-ati-3c3-quirk", 1);
memory_region_add_subregion(&vdev->vga.region[QEMU_PCI_VGA_IO_HI].mem,
3 /* offset 3 bytes from 0x3c0 */, &quirk->mem);
QLIST_INSERT_HEAD(&vdev->vga.region[QEMU_PCI_VGA_IO_HI].quirks,
quirk, next);
trace_vfio_vga_probe_ati_3c3_quirk(vdev->vbasedev.name);
}
/*
* Newer ATI/AMD devices, including HD5450 and HD7850, have a window to PCI
* config space through MMIO BAR2 at offset 0x4000. Nothing seems to access
* the MMIO space directly, but a window to this space is provided through
* I/O port BAR4. Offset 0x0 is the address register and offset 0x4 is the
* data register. When the address is programmed to a range of 0x4000-0x4fff
* PCI configuration space is available. Experimentation seems to indicate
* that only read-only access is provided, but we drop writes when the window
* is enabled to config space nonetheless.
*/
static void vfio_probe_ati_bar4_window_quirk(VFIOPCIDevice *vdev, int nr)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
if (!vdev->has_vga || nr != 4 ||
pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_ATI) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.address_size = 4;
quirk->data.data_offset = 4;
quirk->data.data_size = 4;
quirk->data.address_match = 0x4000;
quirk->data.address_mask = PCIE_CONFIG_SPACE_SIZE - 1;
quirk->data.bar = nr;
quirk->data.read_flags = quirk->data.write_flags = 1;
memory_region_init_io(&quirk->mem, OBJECT(vdev),
&vfio_generic_window_quirk, quirk,
"vfio-ati-bar4-window-quirk", 8);
memory_region_add_subregion_overlap(&vdev->bars[nr].region.mem,
quirk->data.base_offset, &quirk->mem, 1);
QLIST_INSERT_HEAD(&vdev->bars[nr].quirks, quirk, next);
trace_vfio_probe_ati_bar4_window_quirk(vdev->vbasedev.name);
}
#define PCI_VENDOR_ID_REALTEK 0x10ec
/*
* RTL8168 devices have a backdoor that can access the MSI-X table. At BAR2
* offset 0x70 there is a dword data register, offset 0x74 is a dword address
* register. According to the Linux r8169 driver, the MSI-X table is addressed
* when the "type" portion of the address register is set to 0x1. This appears
* to be bits 16:30. Bit 31 is both a write indicator and some sort of
* "address latched" indicator. Bits 12:15 are a mask field, which we can
* ignore because the MSI-X table should always be accessed as a dword (full
* mask). Bits 0:11 is offset within the type.
*
* Example trace:
*
* Read from MSI-X table offset 0
* vfio: vfio_bar_write(0000:05:00.0:BAR2+0x74, 0x1f000, 4) // store read addr
* vfio: vfio_bar_read(0000:05:00.0:BAR2+0x74, 4) = 0x8001f000 // latch
* vfio: vfio_bar_read(0000:05:00.0:BAR2+0x70, 4) = 0xfee00398 // read data
*
* Write 0xfee00000 to MSI-X table offset 0
* vfio: vfio_bar_write(0000:05:00.0:BAR2+0x70, 0xfee00000, 4) // write data
* vfio: vfio_bar_write(0000:05:00.0:BAR2+0x74, 0x8001f000, 4) // do write
* vfio: vfio_bar_read(0000:05:00.0:BAR2+0x74, 4) = 0x1f000 // complete
*/
static uint64_t vfio_rtl8168_window_quirk_read(void *opaque,
hwaddr addr, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
uint64_t val = 0;
if (!quirk->data.flags) { /* Non-MSI-X table access */
return vfio_region_read(&vdev->bars[quirk->data.bar].region,
addr + 0x70, size);
}
switch (addr) {
case 4: /* address */
val = quirk->data.address_match ^ 0x80000000U; /* latch/complete */
break;
case 0: /* data */
if ((vdev->pdev.cap_present & QEMU_PCI_CAP_MSIX)) {
memory_region_dispatch_read(&vdev->pdev.msix_table_mmio,
(hwaddr)(quirk->data.address_match & 0xfff),
&val, size, MEMTXATTRS_UNSPECIFIED);
}
break;
}
trace_vfio_rtl8168_quirk_read(vdev->vbasedev.name,
addr ? "address" : "data", val);
return val;
}
static void vfio_rtl8168_window_quirk_write(void *opaque, hwaddr addr,
uint64_t data, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
switch (addr) {
case 4: /* address */
if ((data & 0x7fff0000) == 0x10000) { /* MSI-X table */
quirk->data.flags = 1; /* Activate reads */
quirk->data.address_match = data;
trace_vfio_rtl8168_quirk_write(vdev->vbasedev.name, data);
if (data & 0x80000000U) { /* Do write */
if (vdev->pdev.cap_present & QEMU_PCI_CAP_MSIX) {
hwaddr offset = data & 0xfff;
uint64_t val = quirk->data.address_mask;
trace_vfio_rtl8168_quirk_msix(vdev->vbasedev.name,
(uint16_t)offset, val);
/* Write to the proper guest MSI-X table instead */
memory_region_dispatch_write(&vdev->pdev.msix_table_mmio,
offset, val, size,
MEMTXATTRS_UNSPECIFIED);
}
return; /* Do not write guest MSI-X data to hardware */
}
} else {
quirk->data.flags = 0; /* De-activate reads, non-MSI-X */
}
break;
case 0: /* data */
quirk->data.address_mask = data;
break;
}
vfio_region_write(&vdev->bars[quirk->data.bar].region,
addr + 0x70, data, size);
}
static const MemoryRegionOps vfio_rtl8168_window_quirk = {
.read = vfio_rtl8168_window_quirk_read,
.write = vfio_rtl8168_window_quirk_write,
.valid = {
.min_access_size = 4,
.max_access_size = 4,
.unaligned = false,
},
.endianness = DEVICE_LITTLE_ENDIAN,
};
static void vfio_probe_rtl8168_bar2_window_quirk(VFIOPCIDevice *vdev, int nr)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
if (pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_REALTEK ||
pci_get_word(pdev->config + PCI_DEVICE_ID) != 0x8168 || nr != 2) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.bar = nr;
memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_rtl8168_window_quirk,
quirk, "vfio-rtl8168-window-quirk", 8);
memory_region_add_subregion_overlap(&vdev->bars[nr].region.mem,
0x70, &quirk->mem, 1);
QLIST_INSERT_HEAD(&vdev->bars[nr].quirks, quirk, next);
trace_vfio_rtl8168_quirk_enable(vdev->vbasedev.name);
}
/*
* Trap the BAR2 MMIO window to config space as well.
*/
static void vfio_probe_ati_bar2_4000_quirk(VFIOPCIDevice *vdev, int nr)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
/* Only enable on newer devices where BAR2 is 64bit */
if (!vdev->has_vga || nr != 2 || !vdev->bars[2].mem64 ||
pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_ATI) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.flags = quirk->data.read_flags = quirk->data.write_flags = 1;
quirk->data.address_match = 0x4000;
quirk->data.address_mask = PCIE_CONFIG_SPACE_SIZE - 1;
quirk->data.bar = nr;
memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_generic_quirk, quirk,
"vfio-ati-bar2-4000-quirk",
TARGET_PAGE_ALIGN(quirk->data.address_mask + 1));
memory_region_add_subregion_overlap(&vdev->bars[nr].region.mem,
quirk->data.address_match & TARGET_PAGE_MASK,
&quirk->mem, 1);
QLIST_INSERT_HEAD(&vdev->bars[nr].quirks, quirk, next);
trace_vfio_probe_ati_bar2_4000_quirk(vdev->vbasedev.name);
}
/*
* Older ATI/AMD cards like the X550 have a similar window to that above.
* I/O port BAR1 provides a window to a mirror of PCI config space located
* in BAR2 at offset 0xf00. We don't care to support such older cards, but
* note it for future reference.
*/
#define PCI_VENDOR_ID_NVIDIA 0x10de
/*
* Nvidia has several different methods to get to config space, the
* nouveu project has several of these documented here:
* https://github.com/pathscale/envytools/tree/master/hwdocs
*
* The first quirk is actually not documented in envytools and is found
* on 10de:01d1 (NVIDIA Corporation G72 [GeForce 7300 LE]). This is an
* NV46 chipset. The backdoor uses the legacy VGA I/O ports to access
* the mirror of PCI config space found at BAR0 offset 0x1800. The access
* sequence first writes 0x338 to I/O port 0x3d4. The target offset is
* then written to 0x3d0. Finally 0x538 is written for a read and 0x738
* is written for a write to 0x3d4. The BAR0 offset is then accessible
* through 0x3d0. This quirk doesn't seem to be necessary on newer cards
* that use the I/O port BAR5 window but it doesn't hurt to leave it.
*/
enum {
NV_3D0_NONE = 0,
NV_3D0_SELECT,
NV_3D0_WINDOW,
NV_3D0_READ,
NV_3D0_WRITE,
};
static uint64_t vfio_nvidia_3d0_quirk_read(void *opaque,
hwaddr addr, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
PCIDevice *pdev = &vdev->pdev;
uint64_t data = vfio_vga_read(&vdev->vga.region[QEMU_PCI_VGA_IO_HI],
addr + quirk->data.base_offset, size);
if (quirk->data.flags == NV_3D0_READ && addr == quirk->data.data_offset) {
data = vfio_pci_read_config(pdev, quirk->data.address_val, size);
trace_vfio_nvidia_3d0_quirk_read(size, data);
}
quirk->data.flags = NV_3D0_NONE;
return data;
}
static void vfio_nvidia_3d0_quirk_write(void *opaque, hwaddr addr,
uint64_t data, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
PCIDevice *pdev = &vdev->pdev;
switch (quirk->data.flags) {
case NV_3D0_NONE:
if (addr == quirk->data.address_offset && data == 0x338) {
quirk->data.flags = NV_3D0_SELECT;
}
break;
case NV_3D0_SELECT:
quirk->data.flags = NV_3D0_NONE;
if (addr == quirk->data.data_offset &&
(data & ~quirk->data.address_mask) == quirk->data.address_match) {
quirk->data.flags = NV_3D0_WINDOW;
quirk->data.address_val = data & quirk->data.address_mask;
}
break;
case NV_3D0_WINDOW:
quirk->data.flags = NV_3D0_NONE;
if (addr == quirk->data.address_offset) {
if (data == 0x538) {
quirk->data.flags = NV_3D0_READ;
} else if (data == 0x738) {
quirk->data.flags = NV_3D0_WRITE;
}
}
break;
case NV_3D0_WRITE:
quirk->data.flags = NV_3D0_NONE;
if (addr == quirk->data.data_offset) {
vfio_pci_write_config(pdev, quirk->data.address_val, data, size);
trace_vfio_nvidia_3d0_quirk_write(data, size);
return;
}
break;
}
vfio_vga_write(&vdev->vga.region[QEMU_PCI_VGA_IO_HI],
addr + quirk->data.base_offset, data, size);
}
static const MemoryRegionOps vfio_nvidia_3d0_quirk = {
.read = vfio_nvidia_3d0_quirk_read,
.write = vfio_nvidia_3d0_quirk_write,
.endianness = DEVICE_LITTLE_ENDIAN,
};
static void vfio_vga_probe_nvidia_3d0_quirk(VFIOPCIDevice *vdev)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
if (pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_NVIDIA ||
!vdev->bars[1].region.size) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.base_offset = 0x10;
quirk->data.address_offset = 4;
quirk->data.address_size = 2;
quirk->data.address_match = 0x1800;
quirk->data.address_mask = PCI_CONFIG_SPACE_SIZE - 1;
quirk->data.data_offset = 0;
quirk->data.data_size = 4;
memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_nvidia_3d0_quirk,
quirk, "vfio-nvidia-3d0-quirk", 6);
memory_region_add_subregion(&vdev->vga.region[QEMU_PCI_VGA_IO_HI].mem,
quirk->data.base_offset, &quirk->mem);
QLIST_INSERT_HEAD(&vdev->vga.region[QEMU_PCI_VGA_IO_HI].quirks,
quirk, next);
trace_vfio_vga_probe_nvidia_3d0_quirk(vdev->vbasedev.name);
}
/*
* The second quirk is documented in envytools. The I/O port BAR5 is just
* a set of address/data ports to the MMIO BARs. The BAR we care about is
* again BAR0. This backdoor is apparently a bit newer than the one above
* so we need to not only trap 256 bytes @0x1800, but all of PCI config
* space, including extended space is available at the 4k @0x88000.
*/
enum {
NV_BAR5_ADDRESS = 0x1,
NV_BAR5_ENABLE = 0x2,
NV_BAR5_MASTER = 0x4,
NV_BAR5_VALID = 0x7,
};
static void vfio_nvidia_bar5_window_quirk_write(void *opaque, hwaddr addr,
uint64_t data, unsigned size)
{
VFIOQuirk *quirk = opaque;
switch (addr) {
case 0x0:
if (data & 0x1) {
quirk->data.flags |= NV_BAR5_MASTER;
} else {
quirk->data.flags &= ~NV_BAR5_MASTER;
}
break;
case 0x4:
if (data & 0x1) {
quirk->data.flags |= NV_BAR5_ENABLE;
} else {
quirk->data.flags &= ~NV_BAR5_ENABLE;
}
break;
case 0x8:
if (quirk->data.flags & NV_BAR5_MASTER) {
if ((data & ~0xfff) == 0x88000) {
quirk->data.flags |= NV_BAR5_ADDRESS;
quirk->data.address_val = data & 0xfff;
} else if ((data & ~0xff) == 0x1800) {
quirk->data.flags |= NV_BAR5_ADDRESS;
quirk->data.address_val = data & 0xff;
} else {
quirk->data.flags &= ~NV_BAR5_ADDRESS;
}
}
break;
}
vfio_generic_window_quirk_write(opaque, addr, data, size);
}
static const MemoryRegionOps vfio_nvidia_bar5_window_quirk = {
.read = vfio_generic_window_quirk_read,
.write = vfio_nvidia_bar5_window_quirk_write,
.valid.min_access_size = 4,
.endianness = DEVICE_LITTLE_ENDIAN,
};
static void vfio_probe_nvidia_bar5_window_quirk(VFIOPCIDevice *vdev, int nr)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
if (!vdev->has_vga || nr != 5 ||
pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_NVIDIA) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.read_flags = quirk->data.write_flags = NV_BAR5_VALID;
quirk->data.address_offset = 0x8;
quirk->data.address_size = 0; /* actually 4, but avoids generic code */
quirk->data.data_offset = 0xc;
quirk->data.data_size = 4;
quirk->data.bar = nr;
memory_region_init_io(&quirk->mem, OBJECT(vdev),
&vfio_nvidia_bar5_window_quirk, quirk,
"vfio-nvidia-bar5-window-quirk", 16);
memory_region_add_subregion_overlap(&vdev->bars[nr].region.mem,
0, &quirk->mem, 1);
QLIST_INSERT_HEAD(&vdev->bars[nr].quirks, quirk, next);
trace_vfio_probe_nvidia_bar5_window_quirk(vdev->vbasedev.name);
}
static void vfio_nvidia_88000_quirk_write(void *opaque, hwaddr addr,
uint64_t data, unsigned size)
{
VFIOQuirk *quirk = opaque;
VFIOPCIDevice *vdev = quirk->vdev;
PCIDevice *pdev = &vdev->pdev;
hwaddr base = quirk->data.address_match & TARGET_PAGE_MASK;
vfio_generic_quirk_write(opaque, addr, data, size);
/*
* Nvidia seems to acknowledge MSI interrupts by writing 0xff to the
* MSI capability ID register. Both the ID and next register are
* read-only, so we allow writes covering either of those to real hw.
* NB - only fixed for the 0x88000 MMIO window.
*/
if ((pdev->cap_present & QEMU_PCI_CAP_MSI) &&
vfio_range_contained(addr, size, pdev->msi_cap, PCI_MSI_FLAGS)) {
vfio_region_write(&vdev->bars[quirk->data.bar].region,
addr + base, data, size);
}
}
static const MemoryRegionOps vfio_nvidia_88000_quirk = {
.read = vfio_generic_quirk_read,
.write = vfio_nvidia_88000_quirk_write,
.endianness = DEVICE_LITTLE_ENDIAN,
};
/*
* Finally, BAR0 itself. We want to redirect any accesses to either
* 0x1800 or 0x88000 through the PCI config space access functions.
*
* NB - quirk at a page granularity or else they don't seem to work when
* BARs are mmap'd
*
* Here's offset 0x88000...
*/
static void vfio_probe_nvidia_bar0_88000_quirk(VFIOPCIDevice *vdev, int nr)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
uint16_t vendor, class;
vendor = pci_get_word(pdev->config + PCI_VENDOR_ID);
class = pci_get_word(pdev->config + PCI_CLASS_DEVICE);
if (nr != 0 || vendor != PCI_VENDOR_ID_NVIDIA ||
class != PCI_CLASS_DISPLAY_VGA) {
return;
}
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.flags = quirk->data.read_flags = quirk->data.write_flags = 1;
quirk->data.address_match = 0x88000;
quirk->data.address_mask = PCIE_CONFIG_SPACE_SIZE - 1;
quirk->data.bar = nr;
memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_nvidia_88000_quirk,
quirk, "vfio-nvidia-bar0-88000-quirk",
TARGET_PAGE_ALIGN(quirk->data.address_mask + 1));
memory_region_add_subregion_overlap(&vdev->bars[nr].region.mem,
quirk->data.address_match & TARGET_PAGE_MASK,
&quirk->mem, 1);
QLIST_INSERT_HEAD(&vdev->bars[nr].quirks, quirk, next);
trace_vfio_probe_nvidia_bar0_88000_quirk(vdev->vbasedev.name);
}
/*
* And here's the same for BAR0 offset 0x1800...
*/
static void vfio_probe_nvidia_bar0_1800_quirk(VFIOPCIDevice *vdev, int nr)
{
PCIDevice *pdev = &vdev->pdev;
VFIOQuirk *quirk;
if (!vdev->has_vga || nr != 0 ||
pci_get_word(pdev->config + PCI_VENDOR_ID) != PCI_VENDOR_ID_NVIDIA) {
return;
}
/* Log the chipset ID */
trace_vfio_probe_nvidia_bar0_1800_quirk_id(
(unsigned int)(vfio_region_read(&vdev->bars[0].region, 0, 4) >> 20)
& 0xff);
quirk = g_malloc0(sizeof(*quirk));
quirk->vdev = vdev;
quirk->data.flags = quirk->data.read_flags = quirk->data.write_flags = 1;
quirk->data.address_match = 0x1800;
quirk->data.address_mask = PCI_CONFIG_SPACE_SIZE - 1;
quirk->data.bar = nr;
memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_generic_quirk, quirk,
"vfio-nvidia-bar0-1800-quirk",
TARGET_PAGE_ALIGN(quirk->data.address_mask + 1));
memory_region_add_subregion_overlap(&vdev->bars[nr].region.mem,
quirk->data.address_match & TARGET_PAGE_MASK,
&quirk->mem, 1);
QLIST_INSERT_HEAD(&vdev->bars[nr].quirks, quirk, next);
trace_vfio_probe_nvidia_bar0_1800_quirk(vdev->vbasedev.name);
}
/*
* TODO - Some Nvidia devices provide config access to their companion HDA
* device and even to their parent bridge via these config space mirrors.
* Add quirks for those regions.
*/
/*
* Common quirk probe entry points.
*/
static void vfio_vga_quirk_setup(VFIOPCIDevice *vdev)
{
vfio_vga_probe_ati_3c3_quirk(vdev);
vfio_vga_probe_nvidia_3d0_quirk(vdev);
}
static void vfio_vga_quirk_teardown(VFIOPCIDevice *vdev)
{
VFIOQuirk *quirk;
int i;
for (i = 0; i < ARRAY_SIZE(vdev->vga.region); i++) {
QLIST_FOREACH(quirk, &vdev->vga.region[i].quirks, next) {
memory_region_del_subregion(&vdev->vga.region[i].mem, &quirk->mem);
}
}
}
static void vfio_vga_quirk_free(VFIOPCIDevice *vdev)
{
int i;
for (i = 0; i < ARRAY_SIZE(vdev->vga.region); i++) {
while (!QLIST_EMPTY(&vdev->vga.region[i].quirks)) {
VFIOQuirk *quirk = QLIST_FIRST(&vdev->vga.region[i].quirks);
object_unparent(OBJECT(&quirk->mem));
QLIST_REMOVE(quirk, next);
g_free(quirk);
}
}
}
static void vfio_bar_quirk_setup(VFIOPCIDevice *vdev, int nr)
{
vfio_probe_ati_bar4_window_quirk(vdev, nr);
vfio_probe_ati_bar2_4000_quirk(vdev, nr);
vfio_probe_nvidia_bar5_window_quirk(vdev, nr);
vfio_probe_nvidia_bar0_88000_quirk(vdev, nr);
vfio_probe_nvidia_bar0_1800_quirk(vdev, nr);
vfio_probe_rtl8168_bar2_window_quirk(vdev, nr);
}
static void vfio_bar_quirk_teardown(VFIOPCIDevice *vdev, int nr)
{
VFIOBAR *bar = &vdev->bars[nr];
VFIOQuirk *quirk;
QLIST_FOREACH(quirk, &bar->quirks, next) {
memory_region_del_subregion(&bar->region.mem, &quirk->mem);
}
}
static void vfio_bar_quirk_free(VFIOPCIDevice *vdev, int nr)
{
VFIOBAR *bar = &vdev->bars[nr];
while (!QLIST_EMPTY(&bar->quirks)) {
VFIOQuirk *quirk = QLIST_FIRST(&bar->quirks);
object_unparent(OBJECT(&quirk->mem));
QLIST_REMOVE(quirk, next);
g_free(quirk);
}
}
/*
* PCI config space
*/
static uint32_t vfio_pci_read_config(PCIDevice *pdev, uint32_t addr, int len)
uint32_t vfio_pci_read_config(PCIDevice *pdev, uint32_t addr, int len)
{
VFIOPCIDevice *vdev = DO_UPCAST(VFIOPCIDevice, pdev, pdev);
uint32_t emu_bits = 0, emu_val = 0, phys_val = 0, val;
@ -1937,8 +1061,8 @@ static uint32_t vfio_pci_read_config(PCIDevice *pdev, uint32_t addr, int len)
return val;
}
static void vfio_pci_write_config(PCIDevice *pdev, uint32_t addr,
uint32_t val, int len)
void vfio_pci_write_config(PCIDevice *pdev,
uint32_t addr, uint32_t val, int len)
{
VFIOPCIDevice *vdev = DO_UPCAST(VFIOPCIDevice, pdev, pdev);
uint32_t val_le = cpu_to_le32(val);

15
hw/vfio/pci.h
View File

@ -155,4 +155,19 @@ typedef struct VFIORomBlacklistEntry {
uint16_t device_id;
} VFIORomBlacklistEntry;
uint32_t vfio_pci_read_config(PCIDevice *pdev, uint32_t addr, int len);
void vfio_pci_write_config(PCIDevice *pdev,
uint32_t addr, uint32_t val, int len);
uint64_t vfio_vga_read(void *opaque, hwaddr addr, unsigned size);
void vfio_vga_write(void *opaque, hwaddr addr, uint64_t data, unsigned size);
bool vfio_blacklist_opt_rom(VFIOPCIDevice *vdev);
void vfio_vga_quirk_setup(VFIOPCIDevice *vdev);
void vfio_vga_quirk_teardown(VFIOPCIDevice *vdev);
void vfio_vga_quirk_free(VFIOPCIDevice *vdev);
void vfio_bar_quirk_setup(VFIOPCIDevice *vdev, int nr);
void vfio_bar_quirk_teardown(VFIOPCIDevice *vdev, int nr);
void vfio_bar_quirk_free(VFIOPCIDevice *vdev, int nr);
#endif /* HW_VFIO_VFIO_PCI_H */