Gui-Dong Han
301c668d5e
drm/amdgpu: use atomic functions with memory barriers for vm fault info ...
[ Upstream commit 6df8e84aa6b5b1812cc2cacd6b3f5ccbb18cda2b ]
The atomic variable vm_fault_info_updated is used to synchronize access to
adev->gmc.vm_fault_info between the interrupt handler and
get_vm_fault_info().
The default atomic functions like atomic_set() and atomic_read() do not
provide memory barriers. This allows for CPU instruction reordering,
meaning the memory accesses to vm_fault_info and the vm_fault_info_updated
flag are not guaranteed to occur in the intended order. This creates a
race condition that can lead to inconsistent or stale data being used.
The previous implementation, which used an explicit mb(), was incomplete
and inefficient. It failed to account for all potential CPU reorderings,
such as the access of vm_fault_info being reordered before the atomic_read
of the flag. This approach is also more verbose and less performant than
using the proper atomic functions with acquire/release semantics.
Fix this by switching to atomic_set_release() and atomic_read_acquire().
These functions provide the necessary acquire and release semantics,
which act as memory barriers to ensure the correct order of operations.
It is also more efficient and idiomatic than using explicit full memory
barriers.
Fixes: b97dfa27efstable@vger.kernel.org
Signed-off-by: Gui-Dong Han <hanguidong02@gmail.com>
Signed-off-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[ kept kgd_dev parameter and adev cast in amdgpu_amdkfd_gpuvm_get_vm_fault_info ]
Signed-off-by: Sasha Levin <sashal@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2025-10-29 14:00:01 +01:00
ACPI: debug: fix signedness issues in read/write helpers
2025-10-29 13:59:52 +01:00
binder: remove "invalid inc weak" check
2025-10-29 14:00:00 +01:00
ata: Fix SATA_MOBILE_LPM_POLICY description in Kconfig
2025-08-28 16:21:35 +02:00
atm: atmtcp: Prevent arbitrary write in atmtcp_recv_control().
2025-09-04 14:05:54 +02:00
drivers/base/node: fix double free in register_one_node()
2025-10-29 13:59:49 +01:00
drbd: add missing kref_get in handle_write_conflicts
2025-08-28 16:21:24 +02:00
Revert "bus: ti-sysc: Probe for l4_wkup and l4_cfg interconnect devices first"
2025-06-27 11:02:55 +01:00
tpm_tis: Fix incorrect arguments in tpm_tis_probe_irq_single
2025-10-29 13:59:52 +01:00
clk: nxp: Fix pll0 rate check condition in LPC18xx CGU driver
2025-10-29 13:59:51 +01:00
clocksource/drivers/clps711x: Fix resource leaks in error paths
2025-10-29 13:59:50 +01:00
cpufreq: intel_pstate: Fix object lifecycle issue in update_qos_request()
2025-10-29 13:59:52 +01:00
Revert "cpuidle: menu: Avoid discarding useful information"
2025-10-29 13:59:59 +01:00
crypto: atmel - Fix dma_unmap_sg() direction
2025-10-29 13:59:52 +01:00
dmaengine: qcom: bam_dma: Fix DT error handling for num-channels/ees
2025-10-02 13:34:29 +02:00
dma-buf: fix timeout handling in dma_resv_wait_timeout v2
2025-07-17 18:25:04 +02:00
EDAC/altera: Delete an inappropriate dma_free_coherent() call
2025-10-02 13:34:27 +02:00
firmware: meson_sm: fix device leak at probe
2025-10-29 13:59:52 +01:00
fpga: zynq_fpga: Fix the wrong usage of dma_map_sgtable()
2025-08-28 16:21:32 +02:00
gpio: pca953x: fix IRQ storm on system wake up
2025-09-09 18:44:00 +02:00
drm/amdgpu: use atomic functions with memory barriers for vm fault info
2025-10-29 14:00:01 +01:00
HID: hid-ntrig: fix unable to handle page fault in ntrig_report_version()
2025-09-04 14:05:56 +02:00
hwmon: (occ) fix unaligned accesses
2025-06-27 11:02:57 +01:00
i2c: designware: Add disabling clocks when probe fails
2025-10-29 13:59:47 +01:00
i3c: don't fail if GETHDRCAP is unsupported
2025-08-28 16:21:29 +02:00
iio: frequency: adf4350: Fix prescaler usage.
2025-10-29 13:59:53 +01:00
RDMA/siw: Always report immediate post SQ errors
2025-10-29 13:59:49 +01:00
Input: uinput - zero-initialize uinput_ff_upload_compat to avoid info leak
2025-10-29 13:59:50 +01:00
mISDN: Fix memory leak in dsp_hwec_enable()
2025-09-09 18:43:57 +02:00
mailbox: zynqmp-ipi: Remove dev.parent check in zynqmp_ipi_free_mboxes
2025-10-29 13:59:52 +01:00
dm: fix NULL pointer dereference in __dm_suspend()
2025-10-29 13:59:55 +01:00
media: pci: ivtv: Add missing check after DMA map
2025-10-29 13:59:56 +01:00
memory: samsung: exynos-srom: Fix of_iomap leak in exynos_srom_probe
2025-10-29 14:00:00 +01:00
memstick: Fix deadlock by moving removing flag earlier
2025-08-28 16:21:32 +02:00
mfd: intel_soc_pmic_chtdc_ti: Set use_single_read regmap_config flag
2025-10-29 13:59:55 +01:00
misc: genwqe: Fix incorrect cmd field being reported in error
2025-10-29 13:59:48 +01:00
mmc: core: SPI mode remove cmd7
2025-10-29 13:59:53 +01:00
spi: cadence-quadspi: Flush posted register writes before DAC access
2025-10-29 14:00:00 +01:00
net: ravb: Ensure memory write completes before ringing TX doorbell
2025-10-29 13:59:59 +01:00
PCI: keystone: Use devm_request_irq() to free "ks-pcie-error-irq" on exit
2025-10-29 13:59:54 +01:00
pcmcia: omap_cf: Mark driver struct with __refdata to prevent section mismatch
2025-10-02 13:34:30 +02:00
perf: arm_spe: Prevent overflow in PERF_IDX2OFF()
2025-10-29 13:59:46 +01:00
phy: ti-pipe3: fix device leak at unbind
2025-10-02 13:34:30 +02:00
pinctrl: check the return value of pinmux_ops::get_function_name()
2025-10-29 13:59:50 +01:00
platform/x86: thinkpad_acpi: Handle KCOV __init vs inline mismatches
2025-08-28 16:21:25 +02:00
power: supply: bq27xxx: restrict no-battery detection to bq27000
2025-10-02 13:34:31 +02:00
pps: fix warning in pps_register_cdev when register device fail
2025-10-29 13:59:48 +01:00
pwm: berlin: Fix wrong register in suspend/resume
2025-10-29 13:59:56 +01:00
drivers/rapidio/rio_cm.c: prevent possible heap overwrite
2025-06-27 11:02:56 +01:00
regulator: core: fix NULL dereference on unbind due to stale coupling data
2025-08-28 16:21:17 +02:00
remoteproc: qcom: q6v5: Avoid disabling handover IRQ twice
2025-10-29 13:59:49 +01:00
rtc: interface: Fix long-standing race when setting alarm
2025-10-29 13:59:53 +01:00
scsi: s390: zfcp: Ensure synchronous unit_add
2025-06-27 11:02:56 +01:00
scsi: hpsa: Fix potential memory leak in hpsa_big_passthru_ioctl()
2025-10-29 13:59:53 +01:00
soc: qcom: rpmh-rsc: Unconditionally clear _TRIGGER bit for TCS
2025-10-29 13:59:46 +01:00
soundwire: stream: restore params when prepare ports fail
2025-08-28 16:21:21 +02:00
spi: spi-fsl-lpspi: Reset FIFO and disable module on transfer abort
2025-09-09 18:44:01 +02:00
comedi: fix divide-by-zero in comedi_buf_munge()
2025-10-29 14:00:00 +01:00
scsi: target: target_core_configfs: Add length check to avoid buffer overflow
2025-10-29 13:59:44 +01:00
tee: Prevent size calculation wraparound on 32-bit kernels
2025-06-27 11:02:55 +01:00
thermal: sysfs: Return ENODATA instead of EAGAIN for reads
2025-08-28 16:21:25 +02:00
thunderbolt: Fix copy+paste error in match_service_id()
2025-08-28 16:21:30 +02:00
serial: max310x: Add error checking in probe()
2025-10-29 13:59:47 +01:00
uio_hv_generic: Let userspace take care of interrupt mask
2025-10-29 13:59:50 +01:00
xhci: dbc: enable back DbC in resume if it was enabled before suspend
2025-10-29 14:00:00 +01:00
vhost/net: Protect ubufs with rcu read lock in vhost_net_ubuf_put()
2025-09-04 14:05:53 +02:00
fbcon: Fix OOB access in font allocation
2025-10-02 13:34:34 +02:00
watchdog: mpc8xxx_wdt: Reload the watchdog timer when enabling the watchdog
2025-10-29 13:59:48 +01:00
xen/events: Update virq_to_irq on migration
2025-10-29 13:59:56 +01:00
301c668d5e