github-mirror/linux
1
0
Fork 0
mirror of https://github.com/torvalds/linux.git synced 2026-11-13 05:04:29 +01:00
Code Issues Actions Packages Projects Releases Wiki Activity Graph
136360290f
linux/drivers
History
Matthew Brost 136360290f drm/xe: Order ring writes before ring tail updates 
The ring buffer and the LRC context image are both mapped WC, so the ring
tail update can become visible to the device before the ring contents it
is meant to publish.

The GuC CT send does contain an xe_device_wmb(), so sending the H2G would
flush the ring contents. The problem is that it comes too late:
xe_lrc_set_ring_tail() publishes the tail before the H2G is sent, and the
device samples the tail from the context image independently of it, either
at context switch-in or while the context is already resident. A submitter
which is interrupted between updating the tail and sending its H2G
therefore leaves the device free to observe the new tail while the ring
contents behind it are not yet visible:

  1. Thread A emits a job into the ring, sets the tail to T_A and sends
     the H2G, which flushes A's ring contents. The GuC starts scheduling
     the context in, but it is not executing yet.
  2. Thread B emits a job into ring[T_A..T_B]. Those writes are not yet
     visible to the device.
  3. Thread B updates the ring tail to T_B. That write targets a
     different page and becomes visible first.
  4. Thread B is interrupted before it sends its H2G, so the flush which
     would have published ring[T_A..T_B] has not happened yet.
  5. The context is switched in and samples the ring tail from the
     context image, picking up T_B rather than T_A.
  6. The GPU executes A's job, advances HEAD to T_A, and continues on to
     ring[T_A..T_B], which still holds the previous wrap's contents, so
     the CS parses stale commands.

The result is command stream corruption, which typically manifests as a
hang or a spurious pagefault rather than anything that points back at the
submission path.

Kernel jobs are by far the most likely to hit this. Kernel queues such as
the migration queue are shared and can be driven by many threads
concurrently, producing back-to-back submissions on an LRC which is
already executing. User queues are typically tied to a single submitting
thread, so the same interleaving is much harder to produce.

Add an xe_device_wmb() at the end of xe_lrc_write_ring() so that it covers
every ring tail publication site, and so the invariant is local: once
xe_lrc_write_ring() returns, the ring contents are visible to the device.

Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/work_items/8651
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/work_items/7810
Cc: stable@vger.kernel.org
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Assisted-by: GitHub_Copilot:claude-opus-5
Reviewed-by: Stuart Summers <stuart.summers@intel.com>
Link: https://patch.msgid.link/20260807171716.140475-1-matthew.brost@intel.com
2026-08-07 15:37:44 -07:00
..
accel accel/ivpu: Remove unnecessary min_t()/max_t() usage 2026-07-17 10:16:08 +02:00
accessibility
acpi ACPI support fixes for 7.2-rc2 2026-07-03 15:13:50 -10:00
amba
android
ata Split <linux/mod_devicetable.h> in per subsystem headers 2026-07-02 20:54:26 -10:00
atm atm: remove unused ATM PHY operations 2026-06-16 08:53:53 -07:00
auxdisplay Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
base regmap: Fix for v7.2 2026-06-26 10:47:26 -07:00
bcma
block Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
bluetooth Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
bus Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
cache Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
cdrom
cdx Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
char Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
clk Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
clocksource A series of updates for the VDSO: 2026-06-15 13:57:13 +05:30
comedi
connector
counter Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
cpufreq Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
cpuidle cpuidle: Allow exit latency to exceed target residency 2026-06-23 14:00:29 +02:00
crypto Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
cxl Merge branch 'for-7.2/cxl-type2-attach-region' into cxl-for-next 2026-06-12 13:47:53 -07:00
dax mm.git review status for mm-hotfixes-stable..mm-stable 2026-06-19 10:14:34 -07:00
dca
devfreq Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
dibs
dio
dma Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
dma-buf Merge drm/drm-next into drm-misc-next 2026-07-08 13:00:55 +02:00
dpll Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
edac Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
eisa
extcon Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
firewire Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
firmware drm-misc-next for 7.3: 2026-07-07 09:34:04 +10:00
fpga Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
fsi Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
fwctl
gnss
gpib Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
gpio gpio fixes for v7.2-rc2 2026-07-03 05:38:12 -10:00
gpu drm/xe: Order ring writes before ring tail updates 2026-08-07 15:37:44 -07:00
greybus
hid Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
hsi Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
hte Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
hv hyperv-next for v7.2-rc1 2026-06-22 08:06:13 -07:00
hwmon Split <linux/mod_devicetable.h> in per subsystem headers 2026-07-02 20:54:26 -10:00
hwspinlock
hwtracing Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
i2c Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
i3c i3c: mipi-i3c-hci: Use named initializers for platform_device_id's .driver_data 2026-06-16 23:19:45 +02:00
idle x86/msr updates for v7.2: 2026-06-15 15:08:14 +05:30
iio Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
infiniband RDMA v7.2 merge window 2026-06-18 08:16:21 -07:00
input Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
interconnect Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
iommu Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
ipack
irqchip Misc irqchip driver fixes: 2026-07-05 05:29:41 -10:00
leds Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
macintosh
mailbox Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
mcb
md block-7.2-20260625 2026-06-25 09:56:47 -07:00
media Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
memory Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
memstick
message
mfd Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
misc Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
mmc Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
most
mtd Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
mux Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
net Split <linux/mod_devicetable.h> in per subsystem headers 2026-07-02 20:54:26 -10:00
nfc Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
ntb NTB updates include an EPF bug fix to prevent an invalid unmap during 2026-06-28 07:46:12 -07:00
nubus
nvdimm Updates for the 7.2 release 2026-06-18 16:27:58 -07:00
nvme block-7.2-20260625 2026-06-25 09:56:47 -07:00
nvmem Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
of Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
opp
parisc
parport
pci Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
pcmcia
peci
perf Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
phy Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
pinctrl gpio fixes for v7.2-rc2 2026-07-03 05:38:12 -10:00
platform Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
pmdomain Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
pnp Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
power Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
powercap
pps Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
ps3
ptp Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
pwm Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
rapidio mm.git review status for mm-hotfixes-stable..mm-nonmm-stable 2026-06-21 13:20:19 -07:00
ras - Fix a malformed Kconfig default for the AMD Address Translation Library 2026-06-16 05:31:01 +05:30
regulator Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
remoteproc
resctrl arm64 updates for 7.2 2026-06-16 05:18:04 +05:30
reset Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
rpmsg rpmsg update for v7.2 2026-06-20 23:31:15 -07:00
rtc Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
s390 s390 updates for 7.2-rc2 2026-07-04 06:28:45 -10:00
sbus
scsi mm.git review status for mm-hotfixes-stable..mm-nonmm-stable 2026-06-21 13:20:19 -07:00
sh
siox Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
slimbus
soc Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
soundwire Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
spi spi: Fixes for v7.2 2026-07-05 05:24:06 -10:00
spmi Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
ssb
staging Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
target Merge branch 7.1/scsi-fixes into 7.2/scsi-staging 2026-06-15 21:01:30 -04:00
tc
tee soc: drivers for 7.2 2026-06-17 11:21:40 -07:00
thermal Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
thunderbolt USB/Thunderbolt driver changes for 7.2-rc1 2026-06-22 12:09:47 -07:00
tty Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
ufs SCSI misc on 20260621 2026-06-21 10:29:45 -07:00
uio Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
usb Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
vdpa Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
vfio vfio/pci: Expose latched module parameter policy in debugfs 2026-06-29 14:26:31 -06:00
vhost mm.git review status for mm-hotfixes-stable..mm-nonmm-stable 2026-06-21 13:20:19 -07:00
video Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
virt Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
virtio virtio,vhost,vdpa: features, fixes 2026-06-17 11:49:00 -07:00
w1 Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
watchdog Replace <linux/mod_devicetable.h> by more specific <linux/device-id/*.h> (c files) 2026-07-03 07:38:17 +02:00
xen xenbus: reject unterminated directory replies 2026-07-01 10:01:14 +02:00
zorro
Kconfig
Makefile