| Age | Commit message (Collapse) | Author |
|
We weren't checking whether the values provided in the private
data in kfd CRIU restore were within bounds.
For queue type, add a KFD_QUEUE_TYPE_MAX and ensure the provided
type is less than it.
For mqd_size, add new function mqd_size_from_queue_type and confirm
that the provided mqd_size matches expectations.
Reviewed-by: David Yat Sin <david.yatsin@amd.com>
Signed-off-by: David Francis <David.Francis@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit f19d8086f6644083c913d70bfdeee20e1b6f46a5)
Cc: stable@vger.kernel.org
|
|
The valid event ids go from 0 to KFD_SIGNAL_EVENT_LIMIT
allocate_event_notification_slot has an option to specify
an event id to allocate at, used by CRIU. We weren't checking
the bounds on that value.
Check them.
v2: Lower bounds check is unecessary because of idr_alloc
already rejecting negative numbers. Upper bounds check should
be KFD_SIGNAL_EVENT_LIMIT since the signal mode mappings might
not yet exist
Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: David Yat Sin <david.yatsin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 6853f1f6cbbeb3f53ebbbd7286536aeb2c5d5f50)
Cc: stable@vger.kernel.org
|
|
Function kfd_process_free_id() should skip over
the primary kfd process because its context id
is fixed assigned, not allocated through the ida table.
This function should only work on secondary contexts.
Fixes: fac682a1d1af ("amdkfd: identify a secondary kfd process by its id")
Signed-off-by: Zhu Lingshan <lingshan.zhu@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 8799ba6fb6a48438aea20c82e74c2f2a3d2b2e7a)
Cc: stable@vger.kernel.org
|
|
kfd_criu_restore_queue's call of kfd_queue_acquire_buffers was
failing for multiple reasons
- The ctl_stack_size set by the CRIU plugin doesn't match
what is expected by acquire_buffers
- The svm buffer cannot be acquired at this point because
CRIU may not have restored it, or may have restored it
to a different address.
The only reason acquire_buffers was necessary here was to
avoid a null ptr dereference in init_user_queue.
Just put in a check for that dereference; it doesn't appear to
come up in real use cases right now. That is, there is no
usage of CRIU with shared MES.
This is a partial revert of
commit 20a5e7ffdfec ("drm/amdkfd: Properly acquire queue buffers in CRIU restore")
Fixes: 20a5e7ffdfec ("drm/amdkfd: Properly acquire queue buffers in CRIU restore")
Reviewed-by: David Yat Sin <david.yatsin@amd.com>
Signed-off-by: David Francis <David.Francis@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 1cafa8b29e029eac3ddf64604f891b35dbf6262b)
Cc: stable@vger.kernel.org
|
|
The valid amdkfd event ids go from 0 to KFD_SIGNAL_EVENT_LIMIT - 1.
During CRIU restore, ensure that the provided event ids are
in that range.
v2: No need for lower bound check since idr_alloc rejects negative
inputs
v3: Also change error message to reflect new error condition
Reviewed-by: David Yat Sin <david.yatsin@amd.com>
Signed-off-by: David Francis <David.Francis@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 5c6c247992d4d9200e073b83f4ec6c703c096845)
|
|
Running RCCL unit tests on a system with a 64K PAGE_SIZE triggers
the following warning and causes the test to terminate on latest
upstream kernel:
WARNING: drivers/gpu/drm/amd/amdgpu/amdgpu_object.c:1335 at
amdgpu_bo_release_notify+0x1bc/0x280 [amdgpu],
CPU#18: rccl-UnitTests/33151
Call trace:
amdgpu_bo_release_notify
ttm_bo_release
amdgpu_gem_object_free
drm_gem_object_free
amdgpu_bo_unref
amdgpu_bo_create
amdgpu_bo_create_user
amdgpu_gem_object_create
amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu
kfd_ioctl_alloc_memory_of_gpu
kfd_ioctl
sys_ioctl
The warning is triggered because
amdgpu_ttm_next_clear_entity() returns NULL when a clear buffer
operation is requested. This happens because the GART window
allocation for the default_entity, clear_entity and move_entity
fails during initialization.
Commit [1] introduced separate GART windows for the
default_entity, clear_entity and move_entity of each SDMA
instance. Their sizes are derived from
AMDGPU_GTT_MAX_TRANSFER_SIZE, which is currently defined as 1024
pages. This implicitly assumes a 4K PAGE_SIZE, where 1024 pages
correspond to a 4MB transfer. On a 64K PAGE_SIZE system, however,
the same value expands to 64MB.
The default_entity and clear_entity each allocate one
AMDGPU_GTT_MAX_TRANSFER_SIZE GART window, while the move_entity
allocates two such windows. This results in 16MB of GART space
per SDMA instance on a 4K PAGE_SIZE system, but 256MB per SDMA
instance on a 64K PAGE_SIZE system.
On an MI210 system with five SDMA instances and a 512MB GART
aperture, the total GART space required becomes 1.25GB,
exceeding the available GART aperture. Consequently, GART window
allocation fails, amdgpu_ttm_next_clear_entity() returns NULL,
and the above warning is triggered.
Redefine AMDGPU_GTT_MAX_TRANSFER_SIZE in bytes instead of page
units. Where a page count is required, convert it using
PAGE_SHIFT. This preserves the existing 4MB transfer size across
all PAGE_SIZE configurations while keeping GART window
allocations within the available GART aperture.
[1] https://lore.kernel.org/all/20260408100327.1372-3-pierre-eric.pelloux-prayer@amd.com/#t
Closes: https://gitlab.freedesktop.org/drm/amd/-/work_items/5435
Fixes: 897ee11ec020 ("drm/amdgpu: create multiple clear/move ttm entities")
Signed-off-by: Donet Tom <donettom@linux.ibm.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 27213b776a666d3030de5acc3cd75278197b0494)
Cc: stable@vger.kernel.org
|
|
There were a few instances in kfd_chardev.c of kvzalloc being
used to allocate memory for an array.
Switch those to kvcalloc, which
- is the standard way of allocating a zero-initialized array
- does a check for the mul overflowing
Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 60b048c93f7a3add39757ad65fe2bb6e58eeae23)
Cc: stable@vger.kernel.org
|
|
Initialize GC IP 11_7_1
Signed-off-by: Granthali Vinodkumar Dhandar <granthali.vinodkumardhandar@amd.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit a928d8d81ec5cdb5a8944d08136720811efad0f6)
|
|
Initialize GC IP 11_7_0
Signed-off-by: Granthali Vinodkumar Dhandar <granthali.vinodkumardhandar@amd.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit cf591e67c095542a16475df293ec7bc9a118e4ee)
|
|
CRIU checkpoint copies the MQD control stack using cp_hqd_cntl_stack_size
from hardware without bounding it to the allocated BO region. If the HW
field is larger than the queue's control stack allocation, memcpy reads
past the BO into adjacent GTT memory and can leak kernel data to userspace.
Store the page-aligned control stack BO size in mqd_manager and clamp
checkpoint copies and reported checkpoint sizes to
min(cp_hqd_cntl_stack_size, mm->ctl_stack_size). Apply the same bound
for multi-XCC v9.4.3 checkpoint layout.
Signed-off-by: Yongqiang Sun <Yongqiang.Sun@amd.com>
Reviewed-by: David Francis <David.Francis@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 6c2abd0ec09e86c6323010673766f76050e28aa3)
Cc: stable@vger.kernel.org
|
|
To avoid wraparound if the value is 0.
Signed-off-by: Xiaogang Chen <xiaogang.chen@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit c0cae35661868af207077a4306bc42c7c972947c)
Cc: stable@vger.kernel.org
|
|
SVM ranges use inclusive page indices: prange->last is the last page in
the range. The split-remap logic introduced by commit 448ee45353ef
("drm/amdkfd: Use huge page size to check split svm range alignment")
uses ALIGN_DOWN(prange->last, 512) to determine whether the original
range can contain a 2MB huge-page mapping.
That aligns the last page itself down. Thus a range ending one page
before the next 2MB boundary is classified as if the final 2MB block did
not exist. When such a range is split inside that final block, the
split head or tail can be left off the remap list even though it was
derived from an original range that may have PMD mappings.
Use prange->last + 1 as the exclusive upper bound when computing the
original range's last 2MB-aligned boundary. Then use the actual split
boundary for the head and tail alignment checks: tail->start for a tail
split, and new_start for a head split. new_start is equivalent to
head->last + 1 and directly names the exclusive end of the split head.
Using head->last for the head-side check can both remap a head that ends
exactly one page before a 2MB boundary and miss a head whose split
boundary is one page after such a boundary. Philip Yang pointed out in
the review of the original change that this condition should use
head->last + 1 or new_start.
Xiaogang Chen identified the inclusive-last cause and posted the
candidate fix in the regression thread. With the culprit change active
and the local revert not applied, the unchanged C/HSA reproducer
completes 10/10 runs with this change on an RX 7600 XT.
Fixes: 448ee45353ef ("drm/amdkfd: Use huge page size to check split svm range alignment")
Closes: https://gitlab.freedesktop.org/drm/amd/-/work_items/4914
Link: https://lore.kernel.org/stable/IA1PR12MB85172F7FE9157C092EDA46A0E3112@IA1PR12MB8517.namprd12.prod.outlook.com/
Link: https://lore.kernel.org/all/32ce2b72-aa16-4202-9f99-92e3cd4408bc@amd.com/
Suggested-by: Xiaogang Chen <xiaogang.chen@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Gerhard Schwanzer <geschw@pm.me>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit a60ea15807126b148a328051636977a33ad0e9bb)
Cc: stable@vger.kernel.org
|
|
Several kfd ioctls need transfer array data from/to user space. Kfd driver
uses kmalloc_array with user provided size. That can oversize alloc or 32-bit
wrap with hostile value. Replace it by memdup_array_user that does overflow
checking and allocates through dedicated slab caches, also physical continuous
as kmalloc.
Signed-off-by: Xiaogang Chen <xiaogang.chen@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 4eca4742eb215951f9739ffe0122d179d545a7a4)
|
|
If inx from find_first_zero_bit is beyond range not need set doorbell_bitmap.
Signed-off-by: Xiaogang Chen <xiaogang.chen@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 2664ce9143d174651a793d96a6a2326050c4f45a)
|
|
amdkfd driver needs allocate buffer to return bo metadata to user space. The
buffer size is controlled by user currently. It is a potential security issue
that hostile value (e.g. 2 GiB) lets any render-group user trigger order-MAX
allocation/OOM in kernel context.
This patch first finds bo metadata size. If the size is smaller than user
provided value drive can safely allocate buffer in kernel space and copy to
user space buffer. If not, driver will let user know, not allocate and copy.
User will redo with new buffer in user space.
This patch lets driver decide buffer allocation size to avoid potential hostile
size from user space.
Signed-off-by: Xiaogang Chen <xiaogang.chen@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit f54ce9e8cbd3abe0eda3a285f54dc4f572fe589a)
|
|
Move kfd_process_remove_sysfs() earlier in kfd_process_wq_release() so
that all sysfs/procfs entries are removed before tearing down PDDs and
dropping lead_thread. The per-process sysfs attributes are backed by
struct kfd_process_device, and their show/store callbacks dereference
PDD fields. Since sysfs removal waits for active callbacks to complete,
removing these entries first closes a race where userspace reads sdma_*
and stats_* files after PDD teardown.
Previously this cleanup ran after kfd_process_destroy_pdds(), which
resets p->n_pdds to 0. This meant kfd_process_remove_sysfs() could no
longer walk the PDD array, so the per-PDD sysfs cleanup did not run as
intended.
This race caused NULL pointer dereferences observed in
kfd_sdma_activity_worker and kfd_procfs_stats_show.
Also harden kfd_process_remove_sysfs() against partially
initialized or already-freed objects:
- Check kobj_queues before removing PASID and deleting it
- Guard kobj_stats and kobj_counters before use
These checks prevent invalid dereferences during cleanup.
Cc: Felix Kuehling <Felix.Kuehling@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Geoffrey McRae <geoffrey.mcrae@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 674c692702341fed321720b4b92036c5934fb485)
|
|
The cleanup tail of kfd_criu_resume_svm() walks
svms->criu_svm_metadata_list and kfree()s each struct criu_svm_metadata
without removing it from the list. The list head is left pointing at
freed kmalloc-96 objects.
A second AMDKFD_IOC_CRIU_OP from the same process re-enters: list_empty()
reads the dangling ->next (use-after-free), the loop walks freed entries,
and each is kfree()'d again (double-free). This is reachable by an
unprivileged render-group user via /dev/kfd with no capabilities required.
Add list_del() before the kfree() so the list is properly emptied. The
list_for_each_entry_safe() iterator already caches the next pointer, so
unlinking during the walk is safe.
Fixes: 2a909ae71871 ("drm/amdkfd: CRIU resume shared virtual memory ranges")
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 6322d278a298e2c1430b9d2697743d3a04b788b1)
|
|
SMI events were reporting incorrect PIDs in containerized environments,
causing test failures where container processes expected to see their
namespace-local PIDs but instead received global host PIDs.
The issue had two root causes:
1. Event functions were called from kernel context (page fault handlers,
migration workers) where 'current' refers to the kernel worker thread,
not the userspace GPU process that triggered the event.
2. PID conversion used task_tgid_vnr() which returns the PID in the
caller's namespace (init namespace for kernel threads), not the task's
own namespace.
This patch updates the SMI event interface:
- Change 8 event function signatures to accept task_struct pointer
instead of pid_t, allowing proper namespace-aware PID conversion
- Convert PIDs using task_tgid_nr_ns(task, task_active_pid_ns(task))
which returns the PID as the process sees it via getpid()
- Update 10 call sites to pass p->lead_thread (the GPU process)
instead of p->lead_thread->pid or current (kernel worker)
This ensures SMI events report container-local PIDs, which is critical
for containerized GPU workloads to correctly correlate events with their
processes.
Tested-by: Andrew Martin <andmarti@amd.com>
Assisted-by: Claude:Sonnet 4-5
Signed-off-by: Andrew Martin <andrew.martin@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 60271ec06e04ba5d69d68714f3abdf637d86c257)
|
|
When kfd_queue_acquire_buffers() was split off from
set_queue_properties_from_user(), set_queue_properties_from_criu()
was missed. Thus, set_queue_properties_from_criu() is not
filling out the buffer fields of queue_properties, which
can come up when subsequent code expects them to be non-null.
Add the proper call to kfd_queue_acquire_buffers(), and also
use the right cast types in set_queue_properties_from_criu()
(which were missed at the same time)
Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Kent Russell <kent.russell@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 88ed96abbbe27b70193544fbc1ee06448c274714)
|
|
Need to restore any good queues even if the suspend_all
failed for some. Always run remove_queue as that will
schedule a GPU reset is removing the queue fails.
v2: move resume_all after remove
Fixes: eb067d65c33e ("drm/amdkfd: Update BadOpcode Interrupt handling with MES")
Reviewed-by: Amber Lin <Amber.Lin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
Malformed ACPI CRAT tables can advertise a zero or undersized subtype
length. The parser then fails to advance the cursor and loops forever
while the remaining image still looks large enough for a generic header.
Validate sub_type_hdr->length on each iteration before parsing or
advancing. Return -EINVAL and warn when length is zero or smaller than
the generic subtype header.
Signed-off-by: Yongqiang Sun <Yongqiang.Sun@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
sysfs_show_gen_prop() accumulated snprintf()'s return value into the
offset. snprintf() reports bytes that would have been written, not
bytes actually written, so a truncated sysfs show could over-report
its length. Use sysfs_emit_at(), which returns only the bytes written.
Signed-off-by: Yongqiang Sun <Yongqiang.Sun@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
If kfd_dbg_trap_enable() fails while copying runtime_info to userspace,
it had already activated the trap, set debug_trap_enabled, taken an extra
process reference, and opened the debug event file. Return -EFAULT without
unwinding that state, leaving inconsistent trap state and a refcount
imbalance that could break later DISABLE/ENABLE.
On copy_to_user failure, deactivate the trap and undo the rest of the
enable setup before returning.
Signed-off-by: Yongqiang Sun <Yongqiang.Sun@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
The kfd_wait_on_events ioctl passes a user-supplied num_events parameter
directly to alloc_event_waiters() which calls kcalloc() without validation.
This allows unprivileged users with /dev/kfd access to trigger large kernel
memory allocations, potentially causing memory exhaustion and denial of
service via the OOM killer.
Add a check to reject num_events values exceeding KFD_SIGNAL_EVENT_LIMIT
(4096), which is the maximum number of events a single process can create.
Signed-off-by: Sunday Clement <Sunday.Clement@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
kfd_smi_ev_enabled() skips the suser privilege check when pid=0.
PROCESS_START, PROCESS_END, and VMFAULT events are emitted with
pid=0 while carrying another process's PID and command name, so any
/dev/kfd user in the render group can monitor all GPU workloads.
Pass the target process PID into kfd_smi_event_add() for these events
so the existing per-client filter restricts delivery to the owning
process or CAP_SYS_ADMIN subscribers.
Signed-off-by: Yongqiang Sun <Yongqiang.Sun@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
A call chain at driver probe exists where profiler lock is used before it
is initialized:
[ 12.131440] kfd kfd: Allocated 3969056 bytes on gart
[ 12.131561] kfd kfd: Total number of KFD nodes to be created: 1
[ 12.132691] ------------[ cut here ]------------
[ 12.132703] DEBUG_LOCKS_WARN_ON(lock->magic != lock)
[ 12.132705] WARNING: kernel/locking/mutex.c:625 at __mutex_lock+0x616/0x1150, CPU#0: (udev-worker)/569
...
[ 12.133051] Call Trace:
[ 12.133055] <TASK>
[ 12.133059] ? mark_held_locks+0x40/0x70
[ 12.133068] ? init_mqd+0xe1/0x1b0 [amdgpu 5154987db73e842b9b4f761e2bd86e17c7ada65c]
[ 12.133671] ? _raw_spin_unlock_irqrestore+0x4c/0x60
[ 12.133683] ? init_mqd+0xe1/0x1b0 [amdgpu 5154987db73e842b9b4f761e2bd86e17c7ada65c]
[ 12.134235] init_mqd+0xe1/0x1b0 [amdgpu 5154987db73e842b9b4f761e2bd86e17c7ada65c]
[ 12.134781] init_mqd_hiq+0x12/0x30 [amdgpu 5154987db73e842b9b4f761e2bd86e17c7ada65c]
[ 12.135340] kq_initialize.constprop.0+0x309/0x400 [amdgpu 5154987db73e842b9b4f761e2bd86e17c7ada65c]
[ 12.135898] kernel_queue_init+0x44/0x80 [amdgpu 5154987db73e842b9b4f761e2bd86e17c7ada65c]
[ 12.136439] pm_init+0x70/0x100 [amdgpu 5154987db73e842b9b4f761e2bd86e17c7ada65c]
[ 12.136984] start_cpsch+0x1dc/0x280 [amdgpu 5154987db73e842b9b4f761e2bd86e17c7ada65c]
[ 12.137525] kgd2kfd_device_init+0x70f/0xd10 [amdgpu 5154987db73e842b9b4f761e2bd86e17c7ada65c]
[ 12.138070] amdgpu_amdkfd_device_init+0x172/0x230 [amdgpu 5154987db73e842b9b4f761e2bd86e17c7ada65c]
[ 12.138618] amdgpu_device_init+0x246a/0x2960 [amdgpu 5154987db73e842b9b4f761e2bd86e17c7ada65c]
The human readable call chain is:
kgd2kfd_device_init
kfd_init_node
kfd_resume
node->dqm->ops.start
Where start can be start_cpsch, which calls pm_init, etc, which ends up
calling kq->mqd_mgr->init_mqd, which takes the profiler lock:
init_mqd()
{
...
mutex_lock(&mm->dev->kfd->profiler_lock);
...
Fix it by initializing the mutext at the top of kgd2kfd_device_init().
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Fixes: a789761de305 ("amd/amdkfd: Add kfd_ioctl_profiler to contain profiler kernel driver changes")
Cc: Benjamin Welton <benjamin.welton@amd.com>
Cc: Perry Yuan <perry.yuan@amd.com>
Cc: Kent Russell <kent.russell@amd.com>
Cc: Yifan Zhang <yifan1.zhang@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
The v11 MQD manager incorrectly assigned the CP-compute variants of
checkpoint_mqd/restore_mqd for KFD_MQD_TYPE_SDMA queues. These functions
use sizeof(struct v11_compute_mqd) (2048 bytes) instead of sizeof(struct
v11_sdma_mqd) (512 bytes), causing a 1536-byte overflow.
During CRIU checkpoint of an SDMA queue on Navi3x:
- checkpoint_mqd() reads 2048 bytes from a 512-byte SDMA MQD buffer,
leaking 1536 bytes of adjacent GTT memory to userspace
During CRIU restore:
- restore_mqd() writes 2048 bytes into a 512-byte SDMA MQD buffer,
corrupting 1536 bytes of adjacent GTT memory (often the ring buffer
or neighboring MQDs)
This is a copy-paste regression unique to v11. All other ASIC backends
(cik, vi, v9, v10, v12) correctly use the SDMA-specific variants.
Add checkpoint_mqd_sdma() and restore_mqd_sdma() functions that properly
handle the smaller v11_sdma_mqd structure, matching the pattern used in
other MQD managers.
Fixes: cc009e613de6 ("drm/amdkfd: Add KFD support for soc21 v3")
Assisted-by: Claude:Sonnet 4-5
Signed-off-by: Andrew Martin <andrew.martin@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
When usr_queue_id_array is NULL and num_queues is non-zero,
get_queue_ids() returns NULL. The callers check only IS_ERR() on the
return value; since IS_ERR(NULL) == false the check passes, and
suspend_queues() calls q_array_invalidate() which immediately
dereferences NULL while iterating num_queues times.
Userspace can trigger this via kfd_ioctl_set_debug_trap() by supplying
num_queues > 0 with a zero queue_array_ptr, causing a kernel panic.
A NULL usr_queue_id_array with num_queues == 0 is a legitimate no-op
(q_array_invalidate never executes, and resume_queues already guards
all queue_ids dereferences behind a NULL check). Return ERR_PTR(-EINVAL)
only when num_queues is non-zero and the pointer is absent; both callers
already propagate IS_ERR() returns correctly to userspace.
Fixes: a70a93fa568b ("drm/amdkfd: add debug suspend and resume process queues operation")
Signed-off-by: Muhammad Bilal <meatuni001@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
This initializes SDMA IP version 6.4.0.
Signed-off-by: Pratik Vishwakarma <Pratik.Vishwakarma@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
Initialize GC IP 11_5_6
Signed-off-by: Pratik Vishwakarma <Pratik.Vishwakarma@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
wait_on_destroy_queue() drops locks to wait for queue resume, allowing
a concurrent destroy to free the queue. Use is_being_destroyed flag to
serialize destruction.
Reviewed-by: Amir Shetaia <Amir.Shetaia@amd.com>
Signed-off-by: Alysa Liu <Alysa.Liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
get_queue_ids() computes array_size = num_queues * sizeof(uint32_t),
which could overflow on 32-bit size_t build. using array_size()
instead, it saturates to SIZE_MAX on overflow.
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
CRIU restore ioctls are meant to be called by CRIU with no
existing drm file. There's an error path
for if the drm file unexpectedly exists. It was positioned so
it was missing a fput(drm_file).
Do that check earlier, as soon as we have the pdd.
Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
The process_info could be NULL if user doesn't call kfd_ioctl_acquire_vm
before calling kfd_ioctl_svm.
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
cpu_data().topo.apicid and kfd_fill_iolink_info_for_cpu() rely on
x86-specific structs not present on UML. The kfd_topology.c and
kfd_crat.c were guarded by CONFIG_X86_64 alone, causing build
failures when CONFIG_DRM_AMDGPU is selected on UML.
Update guards to '#if defined(CONFIG_X86_64) && !defined(CONFIG_UML)'
to ensure x86_64-only paths are excluded on UML builds.
Fixes: af3f2f5db265 ("drm/amdgpu: Remove UML build exclusion from Kconfig")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202605140506.TI8zPIBG-lkp@intel.com/
Cc: Harry Wentland <harry.wentland@amd.com>
Assisted-by: Copilot:Claude-Sonnet-4.6
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
reset
During Mode 1 reset, the ASIC undergoes a reset cycle and becomes temporarily
inaccessible via PCIe. Any attempt to access framebuffer or MMIO registers during
this window can result in uncompleted PCIe transactions, leading to NMI panics or
system hangs.
To prevent this, Unmap all of the applications mappings of the framebuffer
and doorbell BARs before mode1 reset. Also prevent new mappings from coming in
during the reset process.
v2: remove inode in kfd_dev (Christian)
v3: correct unmap offset (Felix), remove prevent new mappings part
to avoid deadlock (Christian)
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
allocate_sdma_queue has an option where the sdma queue id can be
specified (used by CRIU). We weren't bounds-checking that
value.
Confirm it's less than the maximum number of queues.
Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
allocated_doorbell has an option to set the doorbell id
to a specific value (used by CRIU). This value was not
bounds checked.
Check to confirm it's less than KFD_MAX_NUM_OF_QUEUES_PER_PROCESS.
Signed-off-by: David Francis <David.Francis@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
The get_wave_state() function for v9 trusts cp_hqd_cntl_stack_size and
cp_hqd_cntl_stack_offset values read directly from the MQD, which are
written by GPU microcode and fully attacker-controlled on the
CRIU-restore path (via AMDKFD_IOC_RESTORE_PROCESS with H3).
this leads to an unbounded copy_to_user() that can leak adjacent
GTT/kernel memory. If offset > size, integer underflow produces a ~4 GiB
read length, if size is set to 1 MiB against a 4 KiB allocation, we leak
1 MiB of adjacent kernel memory (other queues' MQDs, ring buffers, KASLR
pointers).
Fix by clamping both cp_hqd_cntl_stack_size to the actual allocated
buffer size (q->ctl_stack_size) and cp_hqd_cntl_stack_offset to the
clamped size before performing arithmetic and copy_to_user().
This ensures we never read beyond the allocated kernel BO regardless of
attacker-supplied MQD field values.
Signed-off-by: Sunday Clement <Sunday.Clement@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
After suspend/resume sdma_gang is supported on MES 12.1, SDMA queue reset
is supported too.
Signed-off-by: Amber Lin <Amber.Lin@amd.com>
Reviewed-by: Michael Chen<michael.chen@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
Downgrade unhalt_cpsch warning to dev_dbg when sched is already stopped
Signed-off-by: Perry Yuan <perry.yuan@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
Move amdgpu_amdkfd_stop/start_sched calls from kfd_ptl_control()
into amdgpu_ptl_perf_monitor_ctrl() so all PTL callers (KFD ioctl,
sysfs, GFX init) get consistent scheduling management.
Add amdgpu_amdkfd_stop/start_sched_all() wrappers to stop and
restart KFD scheduling on all nodes without assuming node ID ordering.
v3:
* call start/stop for PTL Set Only
v2:
* move the stop/start sched function to
amdgpu_ptl_perf_monitor_ctrl(Lijo)
* add wrapper amdgpu_amdkfd_stop_sched_all and
amdgpu_amdkfd_start_sched_all (Lijo)
Signed-off-by: Perry Yuan <perry.yuan@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
GC 9.4.4 uses SPI busy status for idle detection instead of GRBM GUI_ACTIVE.
Add version check to use SPI_BUSY for 9.4.4 while keeping GRBM_STATUS
GUI_ACTIVE check for other GC versions.
v2: move this check into amdgpu_ptl_perf_monitor_ctrl(Lijo)
Signed-off-by: Perry Yuan <perry.yuan@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
Ensure GFX engine is idle before switching PTL state to prevent
register access violations and CP hang. This addresses the race
condition where in-flight GPU commands could conflict with PTL
state changes.
Signed-off-by: Perry Yuan <perry.yuan@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
Use a bitmap to track PTL disable requests from sysfs and profiler.
PTL is only re-enabled once all sources have released their disable
requests, avoiding premature enablement.
Signed-off-by: Perry Yuan <perry.yuan@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
Stop the scheduler before releasing the PTL disable request to ensure
the GPU is quiescent during the PTL state transition. This prevents
potential queue preemption failures and GPU resets caused by modifying
PTL state while waves are executing
v1->v2:
only stop/start the scheduler when the PTL state actually needs to transition(Yifan)
Signed-off-by: Perry Yuan <perry.yuan@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
Introduce a new IOCTL option to allow userspace explicit control over
the Peak Tops Limiter (PTL) state for profiling
Link: https://github.com/ROCm/rocm-systems/tree/develop/projects/rocprofiler-sdk
Signed-off-by: Perry Yuan <perry.yuan@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
Introduce a Peak Tops Limiter (PTL) driver that dynamically caps
engine frequency to ensure delivered TOPS never exceeds a defined
TOPS_limit. This initial implementation provides core data structures
and kernel-space interfaces (set/get, enable/disable) to manage PTL state.
PTL performs a firmware handshake to initialize its state and update
predefined format types. It supports updating these format types at
runtime while user-space tools automatically switch PTL state, and
also allows explicitly switching PTL state via newly added commands.
Signed-off-by: Perry Yuan <perry.yuan@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
kfd_ioctl_profiler takes a similar approach to that of
kfd_ioctl_dbg_trap (which contains debugger related IOCTL
services) where kfd_ioctl_profiler will contain all profiler
related IOCTL services. The IOCTL is designed to be expanded
as needed to support additional profiler functionality.
The current functionality of the IOCTL is to allow for profilers
which need PMC counters from GPU devices to both signal to other
profilers that may be on the system that the device has active PMC
profiling taking place on it (multiple PMC profilers on the same
device can result in corrupted counter data) and to setup the device
to allow for the collection of SQ PMC data on all queues on the device.
For PMC data for the SQ block (such as SQ_WAVES) to be available
to a profiler, mmPERFCOUNT_ENABLE must be set on the queues. When
profiling a single process, the profiler can inject PM4 packets into
each queue to turn on PERFCOUNT_ENABLE. When profiling system wide,
the profiler does not have this option and must have a way to turn
on profiling for queues in which it cannot inject packets into directly.
Accomplishing this requires a few steps:
1. Checking if the user has the necessary permissions to profile system
wide on the device. This check uses the same check that linux perf
uses to determine if a user has the necessary permissions to profile
at this scope (primarily if the process has CAP_SYS_PERFMON or is root).
2. Locking the device for profiling. This is done by setting a lock bit
on the device struct and storing the process that locked the device.
3. Iterating all queues on the device and issuing an MQD Update to enable
perfcounting on the queues.
4. Actions to cleanup if the process exits or releases the lock.
The IOCTL also contains a link to the existing PC Sampling IOCTL as well.
This is per a suggestion that we should potentially remove the PC Sampling
IOCTL to have it be a part of the profiler IOCTL. This is a future change.
In addition, we do expect to expand the profiler IOCTL to include
additional profiler functionality in the future (which necessitates the
use of a version number).
v2: sqaush in proper IOCTL number
Proposed userpace support:
https://github.com/ROCm/rocm-systems/commit/40abc95a6463a61bb318a67efd6d9cc3e5ee8839
Signed-off-by: Benjamin Welton <benjamin.welton@amd.com>
Signed-off-by: Perry Yuan <perry.yuan@amd.com>
Acked-by: Kent Russell <kent.russell@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|
|
kfd_reserved_mem_mmap is only for mapping CWSR on APU in IOMMUv2
mode, which is no longer supported, and qpd->cwsr_base has been
set before calling kfd_process_init_cwsr_apu, which is the only
caller for KFD_MMAP_TYPE_RESERVED_MEM, so kfd_process_init_cwsr_apu
is not functional anymore, remove them together. On the other hand,
it will fix a vulnerability issue to abuse KFD_MMAP_TYPE_RESERVED_MEM
of kfd_mmap from userspace.
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
|