summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorVictor Zhao <Victor.Zhao@amd.com>2024-10-17 16:20:40 +0800
committerAlex Deucher <alexander.deucher@amd.com>2024-10-22 17:50:39 -0400
commit8834456163a1b372a85891751e51cafbf443a2d8 (patch)
tree63ac8cf417a54e24f6252c76a55de2c64e3b0411
parent9343b904e7198e4804685133327dece7fe709bc1 (diff)
downloadlinux-8834456163a1b372a85891751e51cafbf443a2d8.tar.gz
linux-8834456163a1b372a85891751e51cafbf443a2d8.tar.bz2
linux-8834456163a1b372a85891751e51cafbf443a2d8.zip
drm/amdkfd: fix the hang caused by the write reorder to fence_addr
make sure KFD_FENCE_INIT write to fence_addr before pm_send_query_status called, to avoid qcm fence timeout caused by incorrect ordering. Signed-off-by: Victor Zhao <Victor.Zhao@amd.com> Reviewed-by: Philip Yang <Philip.Yang@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
-rw-r--r--drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c3
-rw-r--r--drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h2
2 files changed, 3 insertions, 2 deletions
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index b2b16a812e73..5a318376203c 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -2048,7 +2048,7 @@ int amdkfd_fence_wait_timeout(struct device_queue_manager *dqm,
{
unsigned long end_jiffies = msecs_to_jiffies(timeout_ms) + jiffies;
struct device *dev = dqm->dev->adev->dev;
- uint64_t *fence_addr = dqm->fence_addr;
+ volatile uint64_t *fence_addr = dqm->fence_addr;
while (*fence_addr != fence_value) {
/* Fatal err detected, this response won't come */
@@ -2254,6 +2254,7 @@ static int unmap_queues_cpsch(struct device_queue_manager *dqm,
goto out;
*dqm->fence_addr = KFD_FENCE_INIT;
+ mb();
pm_send_query_status(&dqm->packet_mgr, dqm->fence_gpu_addr,
KFD_FENCE_COMPLETED);
/* should be timed out */
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
index 09ab36f8e8c6..bddb169bb301 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
@@ -260,7 +260,7 @@ struct device_queue_manager {
uint16_t vmid_pasid[VMID_NUM];
uint64_t pipelines_addr;
uint64_t fence_gpu_addr;
- uint64_t *fence_addr;
+ volatile uint64_t *fence_addr;
struct kfd_mem_obj *fence_mem;
bool active_runlist;
int sched_policy;