drm/amdkfd: Don't touch the hardware in pre_reset callback
Subsystems (3)
Felix Kuehling
maintainerPATCH 1/1 drm/amdkfd: Don't touch the hardware in pre_reset callback
The intention of the pre_reset callback is to update the driver state to reflect that all user mode queues are preempted and the HIQ is destroyed. However we should not actually preempt any queues or otherwise touch the hardware because it is presumably hanging. The impending reset will take care of actually stopping all queues.
This should prevent KIQ register access hanging on SRIOV function level reset (FLR). It should also speed up the reset by avoiding unnecessary timeouts from a potentially hanging GPU scheduler.
--- drivers/gpu/drm/amd/amdkfd/kfd_device.c | 24 ++++++++++------- .../drm/amd/amdkfd/kfd_device_queue_manager.c | 27 ++++++++++++------- .../drm/amd/amdkfd/kfd_device_queue_manager.h | 5 ++-- drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c | 8 +++--- .../gpu/drm/amd/amdkfd/kfd_packet_manager.c | 4 +-- drivers/gpu/drm/amd/amdkfd/kfd_priv.h | 8 +++--- drivers/gpu/drm/amd/amdkfd/kfd_process.c | 11 ++++---- .../amd/amdkfd/kfd_process_queue_manager.c | 2 +- 8 files changed, 53 insertions(+), 36 deletions(-)
drivers/gpu/drm/amd/amdkfd/kfd_device.c +15 -9
@@ -486,6 +486,7 @@ static int kfd_gtt_sa_init(struct kfd_dev *kfd, unsigned int buf_size, unsigned int chunk_size);static void kfd_gtt_sa_fini (struct kfd_dev *kfd); +static void kfd_suspend (struct kfd_dev *kfd,bool pre_reset);static int kfd_resume (struct kfd_dev *kfd); struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd,
@@ -728,7 +729,7 @@ int kgd2kfd_pre_reset(struct kfd_dev *kfd) {if (!kfd->init_complete)return 0 ; -kgd2kfd_suspend (kfd); +kfd_suspend (kfd, true);kfd_signal_reset_event (kfd);return 0 ;
@@ -767,13 +768,7 @@ void kgd2kfd_suspend(struct kfd_dev *kfd)if (!kfd->init_complete)return ; -/* For first KFD device suspend all the KFD processes */ -if (atomic_inc_return (&kfd_locked) ==1 ) -kfd_suspend_all_processes (); - - kfd->dqm->ops.stop (kfd->dqm); - -kfd_iommu_suspend (kfd); +kfd_suspend (kfd, false); }int kgd2kfd_resume (struct kfd_dev *kfd)
@@ -795,6 +790,17 @@ int kgd2kfd_resume(struct kfd_dev *kfd)return ret; } +static void kfd_suspend (struct kfd_dev *kfd,bool pre_reset) +{ +/* For first KFD device suspend all the KFD processes */ +if (atomic_inc_return (&kfd_locked) ==1 ) +kfd_suspend_all_processes (pre_reset); + + kfd->dqm->ops.stop (kfd->dqm, pre_reset); + +kfd_iommu_suspend (kfd); +} +static int kfd_resume (struct kfd_dev *kfd) {int err =0 ;
@@ -877,7 +883,7 @@ int kgd2kfd_quiesce_mm(struct mm_struct *mm)if (!p)return -ESRCH ; - r =kfd_process_evict_queues (p); + r =kfd_process_evict_queues (p, false);kfd_unref_process (p);return r;
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +18 -9
@@ -592,7 +592,8 @@ static int update_queue(struct device_queue_manager *dqm, struct queue *q) }static int evict_process_queues_nocpsch (struct device_queue_manager *dqm, -struct qcm_process_device *qpd) +struct qcm_process_device *qpd, +bool pre_reset) {struct queue *q;struct mqd_manager *mqd_mgr;
@@ -622,6 +623,8 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm,if (WARN_ONCE (!dqm->sched_running,"Evict when stopped\n" ))continue ; +if (pre_reset) +continue ; retval = mqd_mgr->destroy_mqd(mqd_mgr, q->mqd, KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN,
@@ -639,7 +642,8 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm, }static int evict_process_queues_cpsch (struct device_queue_manager *dqm, -struct qcm_process_device *qpd) +struct qcm_process_device *qpd, +bool pre_reset) {struct queue *q;struct kfd_process_device *pdd;
@@ -664,6 +668,10 @@ static int evict_process_queues_cpsch(struct device_queue_manager *dqm, q->properties.is_active = false; dqm->queue_count--; } + +if (pre_reset) + goto out; + retval = execute_queues_cpsch(dqm, qpd->is_debug ? KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES :
@@ -944,10 +952,10 @@ static int start_nocpsch(struct device_queue_manager *dqm)return 0 ; } -static int stop_nocpsch (struct device_queue_manager *dqm) +static int stop_nocpsch (struct device_queue_manager *dqm,bool pre_reset) {if (dqm->dev->device_info->asic_family ==CHIP_HAWAII ) -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, pre_reset); dqm->sched_running = false;return 0 ;
@@ -1107,20 +1115,21 @@ static int start_cpsch(struct device_queue_manager *dqm)return 0 ; fail_allocate_vidmem: fail_set_sched_resources: -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, false); fail_packet_manager_init:return retval; } -static int stop_cpsch (struct device_queue_manager *dqm) +static int stop_cpsch (struct device_queue_manager *dqm,bool pre_reset) {dqm_lock (dqm); -unmap_queues_cpsch (dqm,KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES ,0 ); +if (!pre_reset) +unmap_queues_cpsch (dqm,KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES ,0 ); dqm->sched_running = false;dqm_unlock (dqm);kfd_gtt_sa_free (dqm->dev, dqm->fence_mem); -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, pre_reset);return 0 ; }
@@ -1891,7 +1900,7 @@ int kfd_process_vm_fault(struct device_queue_manager *dqm,return -EINVAL ; pdd =kfd_get_process_device_data (dqm->dev, p);if (pdd) - ret = dqm->ops.evict_process_queues (dqm, &pdd->qpd); + ret = dqm->ops.evict_process_queues (dqm, &pdd->qpd, false);kfd_unref_process (p);return ret;
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h +3 -2
@@ -103,7 +103,7 @@ struct device_queue_manager_ops {int (*initialize)(struct device_queue_manager *dqm);int (*start)(struct device_queue_manager *dqm); -int (*stop)(struct device_queue_manager *dqm); +int (*stop)(struct device_queue_manager *dqm,bool pre_reset);void (*uninitialize)(struct device_queue_manager *dqm); int (*create_kernel_queue)(struct device_queue_manager *dqm, struct kernel_queue *kq,
@@ -129,7 +129,8 @@ struct device_queue_manager_ops { struct qcm_process_device *qpd);int (*evict_process_queues)(struct device_queue_manager *dqm, -struct qcm_process_device *qpd); +struct qcm_process_device *qpd, +bool pre_reset);int (*restore_process_queues)(struct device_queue_manager *dqm,struct qcm_process_device *qpd);
drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c +4 -4
@@ -195,9 +195,9 @@ static bool kq_initialize(struct kernel_queue *kq, struct kfd_dev *dev, }/* Uninitialize a kernel queue and free all its memory usages. */ -static void kq_uninitialize (struct kernel_queue *kq) +static void kq_uninitialize (struct kernel_queue *kq,bool pre_reset) { -if (kq->queue->properties.type ==KFD_QUEUE_TYPE_HIQ ) +if (kq->queue->properties.type ==KFD_QUEUE_TYPE_HIQ && !pre_reset) kq->mqd_mgr->destroy_mqd(kq->mqd_mgr, kq->queue->mqd, KFD_PREEMPT_TYPE_WAVEFRONT_RESET,
@@ -337,9 +337,9 @@ struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,return NULL ; } -void kernel_queue_uninit (struct kernel_queue *kq) +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset) { -kq_uninitialize (kq); +kq_uninitialize (kq, pre_reset);kfree (kq); }
drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c +2 -2
@@ -264,10 +264,10 @@ int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm)return 0 ; } -void pm_uninit (struct packet_manager *pm) +void pm_uninit (struct packet_manager *pm,bool pre_reset) {mutex_destroy (&pm->lock); -kernel_queue_uninit (pm->priv_queue); +kernel_queue_uninit (pm->priv_queue, pre_reset); } int pm_send_set_resources(struct packet_manager *pm,
drivers/gpu/drm/amd/amdkfd/kfd_priv.h +4 -4
@@ -762,9 +762,9 @@ struct kfd_process *kfd_get_process(const struct task_struct *);struct kfd_process *kfd_lookup_process_by_pasid (unsigned int pasid);struct kfd_process *kfd_lookup_process_by_mm (const struct mm_struct *mm);void kfd_unref_process (struct kfd_process *p); -int kfd_process_evict_queues (struct kfd_process *p); +int kfd_process_evict_queues (struct kfd_process *p,bool pre_reset);int kfd_process_restore_queues (struct kfd_process *p); -void kfd_suspend_all_processes (void ); +void kfd_suspend_all_processes (bool pre_reset);int kfd_resume_all_processes (void ); int kfd_process_device_init_vm(struct kfd_process_device *pdd,
@@ -883,7 +883,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev);void device_queue_manager_uninit (struct device_queue_manager *dqm);struct kernel_queue *kernel_queue_init (struct kfd_dev *dev,enum kfd_queue_type type); -void kernel_queue_uninit (struct kernel_queue *kq); +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset);int kfd_process_vm_fault (struct device_queue_manager *dqm,unsigned int pasid);/* Process Queue Manager */
@@ -974,7 +974,7 @@ extern const struct packet_manager_funcs kfd_vi_pm_funcs;extern const struct packet_manager_funcs kfd_v9_pm_funcs;int pm_init (struct packet_manager *pm,struct device_queue_manager *dqm); -void pm_uninit (struct packet_manager *pm); +void pm_uninit (struct packet_manager *pm,bool pre_reset);int pm_send_set_resources (struct packet_manager *pm,struct scheduling_resources *res);int pm_send_runlist (struct packet_manager *pm,struct list_head *dqm_queues);
drivers/gpu/drm/amd/amdkfd/kfd_process.c +6 -5
@@ -948,7 +948,7 @@ struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm)* Eviction is reference-counted per process-device. This means multiple * evictions from different sources can be nested safely. */ -int kfd_process_evict_queues (struct kfd_process *p) +int kfd_process_evict_queues (struct kfd_process *p,bool pre_reset) {struct kfd_process_device *pdd;int r =0 ;
@@ -956,7 +956,8 @@ int kfd_process_evict_queues(struct kfd_process *p)list_for_each_entry (pdd, &p->per_device_data, per_device_list) { r = pdd->dev->dqm->ops.evict_process_queues (pdd->dev->dqm, - &pdd->qpd); + &pdd->qpd, + pre_reset);if (r) {pr_err ("Failed to evict process queues\n" ); goto fail;
@@ -1026,7 +1027,7 @@ static void evict_process_worker(struct work_struct *work)flush_delayed_work (&p->restore_work);pr_debug ("Started evicting pasid 0x%x\n" , p->pasid); - ret =kfd_process_evict_queues (p); + ret =kfd_process_evict_queues (p, false);if (!ret) {dma_fence_signal (p->ef);dma_fence_put (p->ef);
@@ -1082,7 +1083,7 @@ static void restore_process_worker(struct work_struct *work)pr_err ("Failed to restore queues of pasid 0x%x\n" , p->pasid); } -void kfd_suspend_all_processes (void ) +void kfd_suspend_all_processes (bool pre_reset) {struct kfd_process *p;unsigned int temp;
@@ -1092,7 +1093,7 @@ void kfd_suspend_all_processes(void)cancel_delayed_work_sync (&p->eviction_work);cancel_delayed_work_sync (&p->restore_work); -if (kfd_process_evict_queues (p)) +if (kfd_process_evict_queues (p, pre_reset))pr_err ("Failed to suspend process 0x%x\n" , p->pasid);dma_fence_signal (p->ef);dma_fence_put (p->ef);
drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +1 -1
@@ -374,7 +374,7 @@ int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid)/* destroy kernel queue (DIQ) */ dqm = pqn->kq->dev->dqm; dqm->ops.destroy_kernel_queue (dqm, pqn->kq, &pdd->qpd); -kernel_queue_uninit (pqn->kq); +kernel_queue_uninit (pqn->kq, false); }if (pqn->q) {
Signature · 6 lines
2.24.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Felix Kuehling
maintainer[+Andrey]
Hi Shaoyun, Monk, and Andrey,
I tested this on my bare-metal Vega10 system. GPU reset (using BACO) is flaky on this system with and without this patch. The first reset seems to work OK, the second one fails in different ways. In theory this change should be an improvement as it eliminated at least one failure mode (hanging before the actual reset). A least I don't think I'm making anything worse.
My test is to run KFDTest and trigger a GPU reset while it's busy with the eviction test. This test runs multiple KFD processes and also performs graphics memory allocations and command submissions to force TTM to evict KFD processes. I trigger the reset by hanging the HWS through a debugfs hook:
# cat /sys/class/kfd/kfd/topology/nodes/1/gpu_id > /sys/kernel/debug/kfd/hang_hws
Please give this patch a try and reinstate the pre_reset call before the FLR for SRIOV.
Regards,
Felix
Quoted 336 lines · Felix Kuehling
The intention of the
pre_resetcallback is to update the driver state to reflect that all user mode queues are preempted and the HIQ is destroyed. However we should not actually preempt any queues or otherwise touch the hardware because it is presumably hanging. The impending reset will take care of actually stopping all queues.This should prevent KIQ register access hanging on SRIOV function level reset (FLR). It should also speed up the reset by avoiding unnecessary timeouts from a potentially hanging GPU scheduler.
CC shaoyunl <[email protected]>CC Liu Monk <[email protected]>Signed-off-by Felix Kuehling <[email protected]>--- drivers/gpu/drm/amd/amdkfd/kfd_device.c | 24 ++++++++++------- .../drm/amd/amdkfd/kfd_device_queue_manager.c | 27 ++++++++++++------- .../drm/amd/amdkfd/kfd_device_queue_manager.h | 5 ++-- drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c | 8 +++--- .../gpu/drm/amd/amdkfd/kfd_packet_manager.c | 4 +-- drivers/gpu/drm/amd/amdkfd/kfd_priv.h | 8 +++--- drivers/gpu/drm/amd/amdkfd/kfd_process.c | 11 ++++---- .../amd/amdkfd/kfd_process_queue_manager.c | 2 +- 8 files changed, 53 insertions(+), 36 deletions(-)drivers/gpu/drm/amd/amdkfd/kfd_device.c +15
-9@@ -486,6 +486,7 @@ static int kfd_gtt_sa_init(struct kfd_dev *kfd, unsigned int buf_size, unsigned int chunk_size);static void kfd_gtt_sa_fini (struct kfd_dev *kfd); +static void kfd_suspend (struct kfd_dev *kfd,bool pre_reset);static int kfd_resume (struct kfd_dev *kfd); struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd,@@ -728,7 +729,7 @@ int kgd2kfd_pre_reset(struct kfd_dev *kfd) {if (!kfd->init_complete)return 0 ; -kgd2kfd_suspend (kfd); +kfd_suspend (kfd, true);kfd_signal_reset_event (kfd);return 0 ;@@ -767,13 +768,7 @@ void kgd2kfd_suspend(struct kfd_dev *kfd)if (!kfd->init_complete)return ; -/* For first KFD device suspend all the KFD processes */ -if (atomic_inc_return (&kfd_locked) ==1 ) -kfd_suspend_all_processes (); - - kfd->dqm->ops.stop (kfd->dqm); - -kfd_iommu_suspend (kfd); +kfd_suspend (kfd, false); }int kgd2kfd_resume (struct kfd_dev *kfd)@@ -795,6 +790,17 @@ int kgd2kfd_resume(struct kfd_dev *kfd)return ret; } +static void kfd_suspend (struct kfd_dev *kfd,bool pre_reset) +{ +/* For first KFD device suspend all the KFD processes */ +if (atomic_inc_return (&kfd_locked) ==1 ) +kfd_suspend_all_processes (pre_reset); + + kfd->dqm->ops.stop (kfd->dqm, pre_reset); + +kfd_iommu_suspend (kfd); +} +static int kfd_resume (struct kfd_dev *kfd) {int err =0 ;@@ -877,7 +883,7 @@ int kgd2kfd_quiesce_mm(struct mm_struct *mm)if (!p)return -ESRCH ; - r =kfd_process_evict_queues (p); + r =kfd_process_evict_queues (p, false);kfd_unref_process (p);return r;drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +18
-9@@ -592,7 +592,8 @@ static int update_queue(struct device_queue_manager *dqm, struct queue *q) }static int evict_process_queues_nocpsch (struct device_queue_manager *dqm, -struct qcm_process_device *qpd) +struct qcm_process_device *qpd, +bool pre_reset) {struct queue *q;struct mqd_manager *mqd_mgr;@@ -622,6 +623,8 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm,if (WARN_ONCE (!dqm->sched_running,"Evict when stopped\n" ))continue ; +if (pre_reset) +continue ; retval = mqd_mgr->destroy_mqd(mqd_mgr, q->mqd, KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN,@@ -639,7 +642,8 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm, }static int evict_process_queues_cpsch (struct device_queue_manager *dqm, -struct qcm_process_device *qpd) +struct qcm_process_device *qpd, +bool pre_reset) {struct queue *q;struct kfd_process_device *pdd;@@ -664,6 +668,10 @@ static int evict_process_queues_cpsch(struct device_queue_manager *dqm, q->properties.is_active = false; dqm->queue_count--; } + +if (pre_reset) + goto out; + retval = execute_queues_cpsch(dqm, qpd->is_debug ? KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES :@@ -944,10 +952,10 @@ static int start_nocpsch(struct device_queue_manager *dqm)return 0 ; } -static int stop_nocpsch (struct device_queue_manager *dqm) +static int stop_nocpsch (struct device_queue_manager *dqm,bool pre_reset) {if (dqm->dev->device_info->asic_family ==CHIP_HAWAII ) -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, pre_reset); dqm->sched_running = false;return 0 ;@@ -1107,20 +1115,21 @@ static int start_cpsch(struct device_queue_manager *dqm)return 0 ; fail_allocate_vidmem: fail_set_sched_resources: -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, false); fail_packet_manager_init:return retval; } -static int stop_cpsch (struct device_queue_manager *dqm) +static int stop_cpsch (struct device_queue_manager *dqm,bool pre_reset) {dqm_lock (dqm); -unmap_queues_cpsch (dqm,KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES ,0 ); +if (!pre_reset) +unmap_queues_cpsch (dqm,KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES ,0 ); dqm->sched_running = false;dqm_unlock (dqm);kfd_gtt_sa_free (dqm->dev, dqm->fence_mem); -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, pre_reset);return 0 ; }@@ -1891,7 +1900,7 @@ int kfd_process_vm_fault(struct device_queue_manager *dqm,return -EINVAL ; pdd =kfd_get_process_device_data (dqm->dev, p);if (pdd) - ret = dqm->ops.evict_process_queues (dqm, &pdd->qpd); + ret = dqm->ops.evict_process_queues (dqm, &pdd->qpd, false);kfd_unref_process (p);return ret;drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h +3
-2@@ -103,7 +103,7 @@ struct device_queue_manager_ops {int (*initialize)(struct device_queue_manager *dqm);int (*start)(struct device_queue_manager *dqm); -int (*stop)(struct device_queue_manager *dqm); +int (*stop)(struct device_queue_manager *dqm,bool pre_reset);void (*uninitialize)(struct device_queue_manager *dqm); int (*create_kernel_queue)(struct device_queue_manager *dqm, struct kernel_queue *kq,@@ -129,7 +129,8 @@ struct device_queue_manager_ops { struct qcm_process_device *qpd);int (*evict_process_queues)(struct device_queue_manager *dqm, -struct qcm_process_device *qpd); +struct qcm_process_device *qpd, +bool pre_reset);int (*restore_process_queues)(struct device_queue_manager *dqm,struct qcm_process_device *qpd);drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c +4
-4@@ -195,9 +195,9 @@ static bool kq_initialize(struct kernel_queue *kq, struct kfd_dev *dev, }/* Uninitialize a kernel queue and free all its memory usages. */ -static void kq_uninitialize (struct kernel_queue *kq) +static void kq_uninitialize (struct kernel_queue *kq,bool pre_reset) { -if (kq->queue->properties.type ==KFD_QUEUE_TYPE_HIQ ) +if (kq->queue->properties.type ==KFD_QUEUE_TYPE_HIQ && !pre_reset) kq->mqd_mgr->destroy_mqd(kq->mqd_mgr, kq->queue->mqd, KFD_PREEMPT_TYPE_WAVEFRONT_RESET,@@ -337,9 +337,9 @@ struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,return NULL ; } -void kernel_queue_uninit (struct kernel_queue *kq) +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset) { -kq_uninitialize (kq); +kq_uninitialize (kq, pre_reset);kfree (kq); }drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c +2
-2@@ -264,10 +264,10 @@ int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm)return 0 ; } -void pm_uninit (struct packet_manager *pm) +void pm_uninit (struct packet_manager *pm,bool pre_reset) {mutex_destroy (&pm->lock); -kernel_queue_uninit (pm->priv_queue); +kernel_queue_uninit (pm->priv_queue, pre_reset); } int pm_send_set_resources(struct packet_manager *pm,drivers/gpu/drm/amd/amdkfd/kfd_priv.h +4
-4@@ -762,9 +762,9 @@ struct kfd_process *kfd_get_process(const struct task_struct *);struct kfd_process *kfd_lookup_process_by_pasid (unsigned int pasid);struct kfd_process *kfd_lookup_process_by_mm (const struct mm_struct *mm);void kfd_unref_process (struct kfd_process *p); -int kfd_process_evict_queues (struct kfd_process *p); +int kfd_process_evict_queues (struct kfd_process *p,bool pre_reset);int kfd_process_restore_queues (struct kfd_process *p); -void kfd_suspend_all_processes (void ); +void kfd_suspend_all_processes (bool pre_reset);int kfd_resume_all_processes (void ); int kfd_process_device_init_vm(struct kfd_process_device *pdd,@@ -883,7 +883,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev);void device_queue_manager_uninit (struct device_queue_manager *dqm);struct kernel_queue *kernel_queue_init (struct kfd_dev *dev,enum kfd_queue_type type); -void kernel_queue_uninit (struct kernel_queue *kq); +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset);int kfd_process_vm_fault (struct device_queue_manager *dqm,unsigned int pasid);/* Process Queue Manager */ @@ -974,7 +974,7 @@ extern const struct packet_manager_funcs kfd_vi_pm_funcs;extern const struct packet_manager_funcs kfd_v9_pm_funcs;int pm_init (struct packet_manager *pm,struct device_queue_manager *dqm); -void pm_uninit (struct packet_manager *pm); +void pm_uninit (struct packet_manager *pm,bool pre_reset);int pm_send_set_resources (struct packet_manager *pm,struct scheduling_resources *res);int pm_send_runlist (struct packet_manager *pm,struct list_head *dqm_queues);drivers/gpu/drm/amd/amdkfd/kfd_process.c +6
-5@@ -948,7 +948,7 @@ struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm)* Eviction is reference-counted per process-device. This means multiple * evictions from different sources can be nested safely. */ -int kfd_process_evict_queues (struct kfd_process *p) +int kfd_process_evict_queues (struct kfd_process *p,bool pre_reset) {struct kfd_process_device *pdd;int r =0 ;@@ -956,7 +956,8 @@ int kfd_process_evict_queues(struct kfd_process *p)list_for_each_entry (pdd, &p->per_device_data, per_device_list) { r = pdd->dev->dqm->ops.evict_process_queues (pdd->dev->dqm, - &pdd->qpd); + &pdd->qpd, + pre_reset);if (r) {pr_err ("Failed to evict process queues\n" ); goto fail;@@ -1026,7 +1027,7 @@ static void evict_process_worker(struct work_struct *work)flush_delayed_work (&p->restore_work);pr_debug ("Started evicting pasid 0x%x\n" , p->pasid); - ret =kfd_process_evict_queues (p); + ret =kfd_process_evict_queues (p, false);if (!ret) {dma_fence_signal (p->ef);dma_fence_put (p->ef);@@ -1082,7 +1083,7 @@ static void restore_process_worker(struct work_struct *work)pr_err ("Failed to restore queues of pasid 0x%x\n" , p->pasid); } -void kfd_suspend_all_processes (void ) +void kfd_suspend_all_processes (bool pre_reset) {struct kfd_process *p;unsigned int temp;@@ -1092,7 +1093,7 @@ void kfd_suspend_all_processes(void)cancel_delayed_work_sync (&p->eviction_work);cancel_delayed_work_sync (&p->restore_work); -if (kfd_process_evict_queues (p)) +if (kfd_process_evict_queues (p, pre_reset))pr_err ("Failed to suspend process 0x%x\n" , p->pasid);dma_fence_signal (p->ef);dma_fence_put (p->ef);drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +0
-0
drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid) · line 380
Quoted 8 lines
drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +1
-1@@ -374,7 +374,7 @@ int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid)/* destroy kernel queue (DIQ) */ dqm = pqn->kq->dev->dqm; dqm->ops.destroy_kernel_queue (dqm, pqn->kq, &pdd->qpd); -kernel_queue_uninit (pqn->kq); +kernel_queue_uninit (pqn->kq, false); }if (pqn->q) {
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Will it looks cleaner if we keep a pre_reset flag in per device structure and check it in the function when talk to hw?
Regards
Shaoyun.liu
________________________________
From: Kuehling, Felix <[email protected]>
Sent: December 19, 2019 9:21:08 PM
To: [email protected] <[email protected]>; Liu, Monk <[email protected]>; Liu, Shaoyun <[email protected]>; Grodzovsky, Andrey <[email protected]>
Subject: Re: [PATCH 1/1] drm/amdkfd: Don't touch the hardware in pre_reset callback
[+Andrey]
Hi Shaoyun, Monk, and Andrey,
I tested this on my bare-metal Vega10 system. GPU reset (using BACO) is flaky on this system with and without this patch. The first reset seems to work OK, the second one fails in different ways. In theory this change should be an improvement as it eliminated at least one failure mode (hanging before the actual reset). A least I don't think I'm making anything worse.
My test is to run KFDTest and trigger a GPU reset while it's busy with the eviction test. This test runs multiple KFD processes and also performs graphics memory allocations and command submissions to force TTM to evict KFD processes. I trigger the reset by hanging the HWS through a debugfs hook:
# cat /sys/class/kfd/kfd/topology/nodes/1/gpu_id > /sys/kernel/debug/kfd/hang_hws
Please give this patch a try and reinstate the pre_reset call before the FLR for SRIOV.
Regards,
Felix
Quoted 336 lines · Felix Kuehling
The intention of the
pre_resetcallback is to update the driver state to reflect that all user mode queues are preempted and the HIQ is destroyed. However we should not actually preempt any queues or otherwise touch the hardware because it is presumably hanging. The impending reset will take care of actually stopping all queues.This should prevent KIQ register access hanging on SRIOV function level reset (FLR). It should also speed up the reset by avoiding unnecessary timeouts from a potentially hanging GPU scheduler.
CC shaoyunl <[email protected]>CC Liu Monk <[email protected]>Signed-off-by Felix Kuehling <[email protected]>--- drivers/gpu/drm/amd/amdkfd/kfd_device.c | 24 ++++++++++------- .../drm/amd/amdkfd/kfd_device_queue_manager.c | 27 ++++++++++++------- .../drm/amd/amdkfd/kfd_device_queue_manager.h | 5 ++-- drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c | 8 +++--- .../gpu/drm/amd/amdkfd/kfd_packet_manager.c | 4 +-- drivers/gpu/drm/amd/amdkfd/kfd_priv.h | 8 +++--- drivers/gpu/drm/amd/amdkfd/kfd_process.c | 11 ++++---- .../amd/amdkfd/kfd_process_queue_manager.c | 2 +- 8 files changed, 53 insertions(+), 36 deletions(-)drivers/gpu/drm/amd/amdkfd/kfd_device.c +15
-9@@ -486,6 +486,7 @@ static int kfd_gtt_sa_init(struct kfd_dev *kfd, unsigned int buf_size, unsigned int chunk_size);static void kfd_gtt_sa_fini (struct kfd_dev *kfd); +static void kfd_suspend (struct kfd_dev *kfd,bool pre_reset);static int kfd_resume (struct kfd_dev *kfd); struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd,@@ -728,7 +729,7 @@ int kgd2kfd_pre_reset(struct kfd_dev *kfd) {if (!kfd->init_complete)return 0 ; -kgd2kfd_suspend (kfd); +kfd_suspend (kfd, true);kfd_signal_reset_event (kfd);return 0 ;@@ -767,13 +768,7 @@ void kgd2kfd_suspend(struct kfd_dev *kfd)if (!kfd->init_complete)return ; -/* For first KFD device suspend all the KFD processes */ -if (atomic_inc_return (&kfd_locked) ==1 ) -kfd_suspend_all_processes (); - - kfd->dqm->ops.stop (kfd->dqm); - -kfd_iommu_suspend (kfd); +kfd_suspend (kfd, false); }int kgd2kfd_resume (struct kfd_dev *kfd)@@ -795,6 +790,17 @@ int kgd2kfd_resume(struct kfd_dev *kfd)return ret; } +static void kfd_suspend (struct kfd_dev *kfd,bool pre_reset) +{ +/* For first KFD device suspend all the KFD processes */ +if (atomic_inc_return (&kfd_locked) ==1 ) +kfd_suspend_all_processes (pre_reset); + + kfd->dqm->ops.stop (kfd->dqm, pre_reset); + +kfd_iommu_suspend (kfd); +} +static int kfd_resume (struct kfd_dev *kfd) {int err =0 ;@@ -877,7 +883,7 @@ int kgd2kfd_quiesce_mm(struct mm_struct *mm)if (!p)return -ESRCH ; - r =kfd_process_evict_queues (p); + r =kfd_process_evict_queues (p, false);kfd_unref_process (p);return r;drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +18
-9@@ -592,7 +592,8 @@ static int update_queue(struct device_queue_manager *dqm, struct queue *q) }static int evict_process_queues_nocpsch (struct device_queue_manager *dqm, -struct qcm_process_device *qpd) +struct qcm_process_device *qpd, +bool pre_reset) {struct queue *q;struct mqd_manager *mqd_mgr;@@ -622,6 +623,8 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm,if (WARN_ONCE (!dqm->sched_running,"Evict when stopped\n" ))continue ; +if (pre_reset) +continue ; retval = mqd_mgr->destroy_mqd(mqd_mgr, q->mqd, KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN,@@ -639,7 +642,8 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm, }static int evict_process_queues_cpsch (struct device_queue_manager *dqm, -struct qcm_process_device *qpd) +struct qcm_process_device *qpd, +bool pre_reset) {struct queue *q;struct kfd_process_device *pdd;@@ -664,6 +668,10 @@ static int evict_process_queues_cpsch(struct device_queue_manager *dqm, q->properties.is_active = false; dqm->queue_count--; } + +if (pre_reset) + goto out; + retval = execute_queues_cpsch(dqm, qpd->is_debug ? KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES :@@ -944,10 +952,10 @@ static int start_nocpsch(struct device_queue_manager *dqm)return 0 ; } -static int stop_nocpsch (struct device_queue_manager *dqm) +static int stop_nocpsch (struct device_queue_manager *dqm,bool pre_reset) {if (dqm->dev->device_info->asic_family ==CHIP_HAWAII ) -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, pre_reset); dqm->sched_running = false;return 0 ;@@ -1107,20 +1115,21 @@ static int start_cpsch(struct device_queue_manager *dqm)return 0 ; fail_allocate_vidmem: fail_set_sched_resources: -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, false); fail_packet_manager_init:return retval; } -static int stop_cpsch (struct device_queue_manager *dqm) +static int stop_cpsch (struct device_queue_manager *dqm,bool pre_reset) {dqm_lock (dqm); -unmap_queues_cpsch (dqm,KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES ,0 ); +if (!pre_reset) +unmap_queues_cpsch (dqm,KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES ,0 ); dqm->sched_running = false;dqm_unlock (dqm);kfd_gtt_sa_free (dqm->dev, dqm->fence_mem); -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, pre_reset);return 0 ; }@@ -1891,7 +1900,7 @@ int kfd_process_vm_fault(struct device_queue_manager *dqm,return -EINVAL ; pdd =kfd_get_process_device_data (dqm->dev, p);if (pdd) - ret = dqm->ops.evict_process_queues (dqm, &pdd->qpd); + ret = dqm->ops.evict_process_queues (dqm, &pdd->qpd, false);kfd_unref_process (p);return ret;drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h +3
-2@@ -103,7 +103,7 @@ struct device_queue_manager_ops {int (*initialize)(struct device_queue_manager *dqm);int (*start)(struct device_queue_manager *dqm); -int (*stop)(struct device_queue_manager *dqm); +int (*stop)(struct device_queue_manager *dqm,bool pre_reset);void (*uninitialize)(struct device_queue_manager *dqm); int (*create_kernel_queue)(struct device_queue_manager *dqm, struct kernel_queue *kq,@@ -129,7 +129,8 @@ struct device_queue_manager_ops { struct qcm_process_device *qpd);int (*evict_process_queues)(struct device_queue_manager *dqm, -struct qcm_process_device *qpd); +struct qcm_process_device *qpd, +bool pre_reset);int (*restore_process_queues)(struct device_queue_manager *dqm,struct qcm_process_device *qpd);drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c +4
-4@@ -195,9 +195,9 @@ static bool kq_initialize(struct kernel_queue *kq, struct kfd_dev *dev, }/* Uninitialize a kernel queue and free all its memory usages. */ -static void kq_uninitialize (struct kernel_queue *kq) +static void kq_uninitialize (struct kernel_queue *kq,bool pre_reset) { -if (kq->queue->properties.type ==KFD_QUEUE_TYPE_HIQ ) +if (kq->queue->properties.type ==KFD_QUEUE_TYPE_HIQ && !pre_reset) kq->mqd_mgr->destroy_mqd(kq->mqd_mgr, kq->queue->mqd, KFD_PREEMPT_TYPE_WAVEFRONT_RESET,@@ -337,9 +337,9 @@ struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,return NULL ; } -void kernel_queue_uninit (struct kernel_queue *kq) +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset) { -kq_uninitialize (kq); +kq_uninitialize (kq, pre_reset);kfree (kq); }drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c +2
-2@@ -264,10 +264,10 @@ int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm)return 0 ; } -void pm_uninit (struct packet_manager *pm) +void pm_uninit (struct packet_manager *pm,bool pre_reset) {mutex_destroy (&pm->lock); -kernel_queue_uninit (pm->priv_queue); +kernel_queue_uninit (pm->priv_queue, pre_reset); } int pm_send_set_resources(struct packet_manager *pm,drivers/gpu/drm/amd/amdkfd/kfd_priv.h +4
-4@@ -762,9 +762,9 @@ struct kfd_process *kfd_get_process(const struct task_struct *);struct kfd_process *kfd_lookup_process_by_pasid (unsigned int pasid);struct kfd_process *kfd_lookup_process_by_mm (const struct mm_struct *mm);void kfd_unref_process (struct kfd_process *p); -int kfd_process_evict_queues (struct kfd_process *p); +int kfd_process_evict_queues (struct kfd_process *p,bool pre_reset);int kfd_process_restore_queues (struct kfd_process *p); -void kfd_suspend_all_processes (void ); +void kfd_suspend_all_processes (bool pre_reset);int kfd_resume_all_processes (void ); int kfd_process_device_init_vm(struct kfd_process_device *pdd,@@ -883,7 +883,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev);void device_queue_manager_uninit (struct device_queue_manager *dqm);struct kernel_queue *kernel_queue_init (struct kfd_dev *dev,enum kfd_queue_type type); -void kernel_queue_uninit (struct kernel_queue *kq); +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset);int kfd_process_vm_fault (struct device_queue_manager *dqm,unsigned int pasid);/* Process Queue Manager */ @@ -974,7 +974,7 @@ extern const struct packet_manager_funcs kfd_vi_pm_funcs;extern const struct packet_manager_funcs kfd_v9_pm_funcs;int pm_init (struct packet_manager *pm,struct device_queue_manager *dqm); -void pm_uninit (struct packet_manager *pm); +void pm_uninit (struct packet_manager *pm,bool pre_reset);int pm_send_set_resources (struct packet_manager *pm,struct scheduling_resources *res);int pm_send_runlist (struct packet_manager *pm,struct list_head *dqm_queues);drivers/gpu/drm/amd/amdkfd/kfd_process.c +6
-5@@ -948,7 +948,7 @@ struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm)* Eviction is reference-counted per process-device. This means multiple * evictions from different sources can be nested safely. */ -int kfd_process_evict_queues (struct kfd_process *p) +int kfd_process_evict_queues (struct kfd_process *p,bool pre_reset) {struct kfd_process_device *pdd;int r =0 ;@@ -956,7 +956,8 @@ int kfd_process_evict_queues(struct kfd_process *p)list_for_each_entry (pdd, &p->per_device_data, per_device_list) { r = pdd->dev->dqm->ops.evict_process_queues (pdd->dev->dqm, - &pdd->qpd); + &pdd->qpd, + pre_reset);if (r) {pr_err ("Failed to evict process queues\n" ); goto fail;@@ -1026,7 +1027,7 @@ static void evict_process_worker(struct work_struct *work)flush_delayed_work (&p->restore_work);pr_debug ("Started evicting pasid 0x%x\n" , p->pasid); - ret =kfd_process_evict_queues (p); + ret =kfd_process_evict_queues (p, false);if (!ret) {dma_fence_signal (p->ef);dma_fence_put (p->ef);@@ -1082,7 +1083,7 @@ static void restore_process_worker(struct work_struct *work)pr_err ("Failed to restore queues of pasid 0x%x\n" , p->pasid); } -void kfd_suspend_all_processes (void ) +void kfd_suspend_all_processes (bool pre_reset) {struct kfd_process *p;unsigned int temp;@@ -1092,7 +1093,7 @@ void kfd_suspend_all_processes(void)cancel_delayed_work_sync (&p->eviction_work);cancel_delayed_work_sync (&p->restore_work); -if (kfd_process_evict_queues (p)) +if (kfd_process_evict_queues (p, pre_reset))pr_err ("Failed to suspend process 0x%x\n" , p->pasid);dma_fence_signal (p->ef);dma_fence_put (p->ef);drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +0
-0
drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid) · line 380
Quoted 8 lines
drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +1
-1@@ -374,7 +374,7 @@ int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid)/* destroy kernel queue (DIQ) */ dqm = pqn->kq->dev->dqm; dqm->ops.destroy_kernel_queue (dqm, pqn->kq, &pdd->qpd); -kernel_queue_uninit (pqn->kq); +kernel_queue_uninit (pqn->kq, false); }if (pqn->q) {
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Felix Kuehling
maintainerQuoted 1 line · Shaoyun
Will it looks cleaner if we keep a
pre_resetflag in per device structure and check it in the function when talk to hw?
I was briefly considering that when I saw how many function needed that pre_reset flag. But this could lead to race conditions with other threads doing evictions concurrently (e.g. MMU notifiers). I think we want concurrent evictions to hang until the GPU reset is done, not just skip the actual eviction quietly.
Hmm, I need to think more about the interaction between evictions and resets.
Regards,
Felix
Quoted 407 lines
Regards
Shaoyun.liu
------------------------------------------------------------------------
*From:* Kuehling, Felix <[email protected]>
*Sent:* December 19, 2019 9:21:08 PM
*To:* [email protected] <[email protected]>; Liu, Monk <[email protected]>; Liu, Shaoyun <[email protected]>; Grodzovsky, Andrey <[email protected]> *Subject:* Re: [PATCH 1/1] drm/amdkfd: Don't touch the hardware inpre_resetcallback
[+Andrey]Hi Shaoyun, Monk, and Andrey,
I tested this on my bare-metal Vega10 system. GPU reset (using BACO) is flaky on this system with and without this patch. The first reset seems to work OK, the second one fails in different ways. In theory this change should be an improvement as it eliminated at least one failure mode (hanging before the actual reset). A least I don't think I'm making anything worse.
My test is to run KFDTest and trigger a GPU reset while it's busy with the eviction test. This test runs multiple KFD processes and also performs graphics memory allocations and command submissions to force TTM to evict KFD processes. I trigger the reset by hanging the HWS through a debugfs hook:
# cat /sys/class/kfd/kfd/topology/nodes/1/gpu_id > /sys/kernel/debug/kfd/hang_hwsPlease give this patch a try and reinstate the
pre_resetcall before the FLR for SRIOV.Regards,
FelixQuoted 18 lines · Felix Kuehling
The intention of the
pre_resetcallback is to update the driver state to reflect that all user mode queues are preempted and the HIQ is destroyed. However we should not actually preempt any queues or otherwise touch the hardware because it is presumably hanging. The impending reset will take care of actually stopping all queues.This should prevent KIQ register access hanging on SRIOV function level reset (FLR). It should also speed up the reset by avoiding unnecessary timeouts from a potentially hanging GPU scheduler.
CC shaoyunl <[email protected]>CC Liu Monk <[email protected]>Signed-off-by Felix Kuehling <[email protected]>---
drivers/gpu/drm/amd/amdkfd/kfd_device.c | 24 ++++++++++-------.../drm/amd/amdkfd/kfd_device_queue_manager.c | 27 ++++++++++++------- .../drm/amd/amdkfd/kfd_device_queue_manager.h | 5 ++-- drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c | 8 +++---Review
Quoted 7 lines
.../gpu/drm/amd/amdkfd/kfd_packet_manager.c | 4 +-- drivers/gpu/drm/amd/amdkfd/kfd_priv.h | 8 +++--- drivers/gpu/drm/amd/amdkfd/kfd_process.c | 11 ++++---- .../amd/amdkfd/kfd_process_queue_manager.c | 2 +-
8 files changed, 53 insertions(+), 36 deletions(-)(quoted diff) +0
-0
b/drivers/gpu/drm/amd/amdkfd/kfd_device.cdrivers/gpu/drm/amd/amdkfd/kfd_device.c static int kfd_gtt_sa_init(struct kfd_dev *kfd,
Quoted 4 lines
index
c6b6901bbda3..796996a0d926100644drivers/gpu/drm/amd/amdkfd/kfd_device.c +0
-0@@ -486,6 +486,7 @@ static int kfd_gtt_sa_init(struct kfd_dev *kfd,unsigned int
buf_size,Review
Quoted 59 lines
drivers/gpu/drm/amd/amdkfd/kfd_device.c +15
-9unsigned int chunk_size); static void kfd_gtt_sa_fini(struct kfd_dev *kfd); +static void kfd_suspend (struct kfd_dev *kfd,bool pre_reset); static int kfd_resume(struct kfd_dev *kfd); struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd,@@ -728,7 +729,7 @@ int kgd2kfd_pre_reset(struct kfd_dev *kfd) { if (!kfd->init_complete) return 0; - kgd2kfd_suspend(kfd); + kfd_suspend(kfd, true); kfd_signal_reset_event(kfd); return 0;@@ -767,13 +768,7 @@ void kgd2kfd_suspend(struct kfd_dev *kfd) if (!kfd->init_complete) return; - /* For first KFD device suspend all the KFD processes */ - if (atomic_inc_return(&kfd_locked) == 1) - kfd_suspend_all_processes(); - - kfd->dqm->ops.stop(kfd->dqm); - - kfd_iommu_suspend(kfd); + kfd_suspend(kfd, false); } int kgd2kfd_resume(struct kfd_dev *kfd)@@ -795,6 +790,17 @@ int kgd2kfd_resume(struct kfd_dev *kfd) return ret; } +static void kfd_suspend (struct kfd_dev *kfd,bool pre_reset) +{ + /* For first KFD device suspend all the KFD processes */ + if (atomic_inc_return(&kfd_locked) == 1) + kfd_suspend_all_processes(pre_reset); + + kfd->dqm->ops.stop(kfd->dqm, pre_reset); + + kfd_iommu_suspend(kfd); +} + static int kfd_resume(struct kfd_dev *kfd) { int err = 0;@@ -877,7 +883,7 @@ int kgd2kfd_quiesce_mm(struct mm_struct *mm) if (!p) return -ESRCH; - r = kfd_process_evict_queues(p); + r = kfd_process_evict_queues(p, false); kfd_unref_process(p); return r;(quoted diff) +0
-0
b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.cdrivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c static int update_queue(struct
Quoted 4 lines
index
f7f6df40875e..3a49944164da100644drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +0
-0@@ -592,7 +592,8 @@ static int update_queue(struct
device_queue_manager*dqm, struct queue *q)Quoted 3 lines
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +0
-0}static int evict_process_queues_nocpsch(struct
device_queue_manager*dqm,drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c static int evict_process_queues_nocpsch(struct
Quoted 7 lines
- struct
qcm_process_device*qpd) + structqcm_process_device*qpd, + boolpre_reset)
{
struct queue *q;
structmqd_manager*mqd_mgr;drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +0
-0@@ -622,6 +623,8 @@ static int evict_process_queues_nocpsch(struct
device_queue_manager*dqm,Quoted 2 lines
if (WARN_ONCE(!dqm->sched_running, "Evict when
stopped\n"))
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c static int evict_process_queues_nocpsch(struct
Quoted 7 lines
continue;
+ if (pre_reset)
+ continue;retval = mqd_mgr->destroy_mqd(mqd_mgr,
q->mqd,KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN,drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +0
-0@@ -639,7 +642,8 @@ static int evict_process_queues_nocpsch(struct
device_queue_manager*dqm,Quoted 3 lines
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +0
-0}static int evict_process_queues_cpsch(struct
device_queue_manager*dqm,
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c static int evict_process_queues_cpsch(struct
Quoted 7 lines
- struct
qcm_process_device*qpd) + structqcm_process_device*qpd, + boolpre_reset)
{
struct queue *q;
structkfd_process_device*pdd;drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +0
-0@@ -664,6 +668,10 @@ static int evict_process_queues_cpsch(struct
device_queue_manager*dqm,drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c static int start_nocpsch(struct
Quoted 11 lines
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +4
-0q->properties.is_active = false; dqm->queue_count--; } + + if (pre_reset) + goto out; + retval = execute_queues_cpsch(dqm,
qpd->is_debug?KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES:drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +0
-0@@ -944,10 +952,10 @@ static int start_nocpsch(struct
device_queue_manager*dqm)drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c static int start_nocpsch(struct device_queue_manager *dqm)
Quoted 5 lines
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +1
-1return 0; } -static int stop_nocpsch (struct device_queue_manager *dqm) +static int stop_nocpsch(struct device_queue_manager *dqm, bool
pre_reset)drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c static int start_cpsch(struct
Quoted 8 lines
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +1
-1{ if (dqm->dev->device_info->asic_family == CHIP_HAWAII) - pm_uninit(&dqm->packets); + pm_uninit(&dqm->packets, pre_reset); dqm->sched_running = false; return 0;@@ -1107,20 +1115,21 @@ static int start_cpsch(struct
device_queue_manager*dqm)Quoted 11 lines
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +0
-0return 0;
fail_allocate_vidmem:
fail_set_sched_resources:
-pm_uninit(&dqm->packets);
+pm_uninit(&dqm->packets, false);
fail_packet_manager_init:
return retval;
}drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +1
-1-static int stop_cpsch (struct device_queue_manager *dqm) +static int stop_cpsch (struct device_queue_manager *dqm,bool pre_reset)Quoted 5 lines
{
dqm_lock(dqm);
-unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0); + if (!pre_reset)
+ unmap_queues_cpsch(dqm,
KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0);drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c int kfd_process_vm_fault(struct
Quoted 10 lines
dqm->sched_running= false;
dqm_unlock(dqm);
kfd_gtt_sa_free(dqm->dev, dqm->fence_mem); -pm_uninit(&dqm->packets);
+pm_uninit(&dqm->packets, pre_reset);return 0;
}drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +0
-0@@ -1891,7 +1900,7 @@ int kfd_process_vm_fault(struct
device_queue_manager*dqm,drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c int kfd_process_vm_fault(struct device_queue_manager *dqm,
Quoted 5 lines
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +1
-1return -EINVAL; pdd = kfd_get_process_device_data(dqm->dev, p); if (pdd) - ret = dqm->ops.evict_process_queues(dqm, &pdd->qpd); + ret = dqm->ops.evict_process_queues(dqm, &pdd->qpd,false);
Review
Quoted 4 lines
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +0
-0kfd_unref_process(p); return ret;(quoted diff) +0
-0
b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.hReview
Quoted 23 lines
index
a8c37e6da027..9f82f95f6a92100644drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h +3
-2@@ -103,7 +103,7 @@ struct device_queue_manager_ops { int (*initialize)(struct device_queue_manager *dqm); int (*start)(struct device_queue_manager *dqm); - int (*stop)(struct device_queue_manager *dqm); + int (*stop)(struct device_queue_manager *dqm, bool pre_reset); void (*uninitialize)(struct device_queue_manager *dqm); int (*create_kernel_queue)(struct device_queue_manager *dqm, struct kernel_queue *kq,@@ -129,7 +129,8 @@ struct device_queue_manager_ops { struct qcm_process_device *qpd); int (*evict_process_queues)(struct device_queue_manager *dqm, - struct qcm_process_device *qpd); + struct qcm_process_device *qpd, + bool pre_reset); int (*restore_process_queues)(struct device_queue_manager *dqm, struct qcm_process_device *qpd);(quoted diff) +0
-0
b/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.cdrivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c static bool kq_initialize(struct kernel_queue
Quoted 4 lines
index
2d56dc534459..dbd2e8e9ae69100644drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c +0
-0@@ -195,9 +195,9 @@ static bool kq_initialize(struct kernel_queue*kq, struct
kfd_dev*dev,drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c static bool kq_initialize(struct kernel_queue *kq, struct kfd_dev *dev,
Quoted 8 lines
drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c +2
-2} /* Uninitialize a kernel queue and free all its memory usages. */ -static void kq_uninitialize (struct kernel_queue *kq) +static void kq_uninitialize (struct kernel_queue *kq,bool pre_reset) { - if (kq->queue->properties.type == KFD_QUEUE_TYPE_HIQ) + if (kq->queue->properties.type == KFD_QUEUE_TYPE_HIQ &&!pre_reset)
drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c struct kernel_queue *kernel_queue_init(struct
Quoted 4 lines
drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c +0
-0kq->mqd_mgr->destroy_mqd(kq->mqd_mgr, kq->queue->mqd, KFD_PREEMPT_TYPE_WAVEFRONT_RESET,@@ -337,9 +337,9 @@ struct kernel_queue *kernel_queue_init(struct
kfd_dev*dev,Review
Quoted 12 lines
drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c +2
-2return NULL; } -void kernel_queue_uninit (struct kernel_queue *kq) +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset) { - kq_uninitialize(kq); + kq_uninitialize(kq, pre_reset); kfree(kq); }(quoted diff) +0
-0
b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.cdrivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c int pm_init(struct packet_manager *pm, struct
Quoted 4 lines
index
6cabed06ef5d..7732a3bbebd6100644drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c +0
-0@@ -264,10 +264,10 @@ int pm_init(struct packet_manager *pm, struct
device_queue_manager*dqm)Review
Quoted 13 lines
drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c +2
-2return 0; } -void pm_uninit (struct packet_manager *pm) +void pm_uninit (struct packet_manager *pm,bool pre_reset) { mutex_destroy(&pm->lock); - kernel_queue_uninit(pm->priv_queue); + kernel_queue_uninit(pm->priv_queue, pre_reset); } int pm_send_set_resources(struct packet_manager *pm,(quoted diff) +0
-0
b/drivers/gpu/drm/amd/amdkfd/kfd_priv.hdrivers/gpu/drm/amd/amdkfd/kfd_priv.h struct kfd_process *kfd_get_process(const struct
Quoted 4 lines
index
087e96838997..9bc4ced4acba100644drivers/gpu/drm/amd/amdkfd/kfd_priv.h +0
-0@@ -762,9 +762,9 @@ struct kfd_process *kfd_get_process(const struct
task_struct*);Quoted 2 lines
drivers/gpu/drm/amd/amdkfd/kfd_priv.h +0
-0struct kfd_process *kfd_lookup_process_by_pasid(unsigned int pasid);struct
kfd_process*kfd_lookup_process_by_mm(const struct
mm_struct*mm);drivers/gpu/drm/amd/amdkfd/kfd_priv.h struct device_queue_manager
Quoted 10 lines
void
kfd_unref_process(struct kfd_process *p);drivers/gpu/drm/amd/amdkfd/kfd_priv.h +1
-1-int kfd_process_evict_queues (struct kfd_process *p); +int kfd_process_evict_queues (struct kfd_process *p,bool pre_reset);int
kfd_process_restore_queues(struct kfd_process *p);drivers/gpu/drm/amd/amdkfd/kfd_priv.h +1
-1-void kfd_suspend_all_processes (void ); +void kfd_suspend_all_processes (bool pre_reset);int
kfd_resume_all_processes(void);int kfd_process_device_init_vm(struct
kfd_process_device*pdd,drivers/gpu/drm/amd/amdkfd/kfd_priv.h +0
-0@@ -883,7 +883,7 @@ struct device_queue_manager*device_queue_manager_init(struct
kfd_dev*dev);drivers/gpu/drm/amd/amdkfd/kfd_priv.h struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev);
Quoted 6 lines
drivers/gpu/drm/amd/amdkfd/kfd_priv.h +1
-1void device_queue_manager_uninit(struct device_queue_manager *dqm); struct kernel_queue *kernel_queue_init(struct kfd_dev *dev, enum kfd_queue_type type); -void kernel_queue_uninit (struct kernel_queue *kq); +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset); int kfd_process_vm_fault(struct device_queue_manager *dqm,unsigned int pasid);
drivers/gpu/drm/amd/amdkfd/kfd_priv.h extern const struct packet_manager_funcs
Quoted 3 lines
drivers/gpu/drm/amd/amdkfd/kfd_priv.h +0
-0/* Process Queue Manager */@@ -974,7 +974,7 @@ extern const struct packet_manager_funcs
kfd_vi_pm_funcs;Quoted 3 lines
drivers/gpu/drm/amd/amdkfd/kfd_priv.h +0
-0extern const struct packet_manager_funcs kfd_v9_pm_funcs;int pm_init(struct
packet_manager*pm, structdevice_queue_manager*dqm);
Quoted 5 lines
drivers/gpu/drm/amd/amdkfd/kfd_priv.h +1
-1-void pm_uninit (struct packet_manager *pm); +void pm_uninit (struct packet_manager *pm,bool pre_reset);int
pm_send_set_resources(struct packet_manager *pm, struct scheduling_resources *res); int pm_send_runlist(structpacket_manager*pm, structlist_head
*dqm_queues);Review
Quoted 1 line
(quoted diff) +0
-0
b/drivers/gpu/drm/amd/amdkfd/kfd_process.cdrivers/gpu/drm/amd/amdkfd/kfd_process.c struct kfd_process
Quoted 4 lines
index
536a153ac9a4..7bcadd3a1e3b100644drivers/gpu/drm/amd/amdkfd/kfd_process.c +0
-0@@ -948,7 +948,7 @@ struct kfd_process*kfd_lookup_process_by_mm(const struct
mm_struct*mm)drivers/gpu/drm/amd/amdkfd/kfd_process.c struct kfd_process · line 948
Quoted 1 line
drivers/gpu/drm/amd/amdkfd/kfd_process.c +0
-0* Eviction is reference-counted per process-device. This meansmultiple
Quoted 11 lines
drivers/gpu/drm/amd/amdkfd/kfd_process.c +1
-1* evictions from different sources can be nested safely. */ -int kfd_process_evict_queues (struct kfd_process *p) +int kfd_process_evict_queues (struct kfd_process *p,bool pre_reset) { struct kfd_process_device *pdd; int r = 0;@@ -956,7 +956,8 @@ int kfd_process_evict_queues(struct kfd_process *p) list_for_each_entry(pdd, &p->per_device_data, per_device_list) {r =
pdd->dev->dqm->ops.evict_process_queues(pdd->dev->dqm,
drivers/gpu/drm/amd/amdkfd/kfd_process.c static void evict_process_worker(struct
Quoted 7 lines
-
&pdd->qpd);
+&pdd->qpd,
+pre_reset);
if (r) {
pr_err("Failed to evict process queues\n"); goto fail;drivers/gpu/drm/amd/amdkfd/kfd_process.c +0
-0@@ -1026,7 +1027,7 @@ static void evict_process_worker(struct
work_struct*work)drivers/gpu/drm/amd/amdkfd/kfd_process.c static void restore_process_worker(struct
Quoted 9 lines
drivers/gpu/drm/amd/amdkfd/kfd_process.c +1
-1flush_delayed_work(&p->restore_work); pr_debug("Started evicting pasid 0x%x\n", p->pasid); - ret = kfd_process_evict_queues(p); + ret = kfd_process_evict_queues(p, false); if (!ret) { dma_fence_signal(p->ef); dma_fence_put(p->ef);@@ -1082,7 +1083,7 @@ static void restore_process_worker(struct
work_struct*work)drivers/gpu/drm/amd/amdkfd/kfd_process.c static void restore_process_worker(struct · line 1083
Quoted 1 line
drivers/gpu/drm/amd/amdkfd/kfd_process.c +0
-0pr_err("Failed to restore queues of pasid 0x%x\n",
p->pasid);Quoted 8 lines
drivers/gpu/drm/amd/amdkfd/kfd_process.c +1
-1} -void kfd_suspend_all_processes (void ) +void kfd_suspend_all_processes (bool pre_reset) { struct kfd_process *p; unsigned int temp;drivers/gpu/drm/amd/amdkfd/kfd_process.c void kfd_suspend_all_processes(void) · line 1097
Quoted 6 lines
drivers/gpu/drm/amd/amdkfd/kfd_process.c +1
-1@@ -1092,7 +1093,7 @@ void kfd_suspend_all_processes(void)cancel_delayed_work_sync (&p->eviction_work);cancel_delayed_work_sync (&p->restore_work); - if (kfd_process_evict_queues(p)) + if (kfd_process_evict_queues(p, pre_reset)) pr_err("Failed to suspend process 0x%x\n",
p->pasid);Review
Quoted 3 lines
drivers/gpu/drm/amd/amdkfd/kfd_process.c +0
-0dma_fence_signal(p->ef); dma_fence_put(p->ef);(quoted diff) +0
-0
b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
Quoted 13 lines
drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c int pqm_destroy_queue(struct
Quoted 4 lines
index
d3eacf72e8db..8fa856e6a03f100644drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +0
-0@@ -374,7 +374,7 @@ int pqm_destroy_queue(struct
process_queue_manager*pqm, unsigned int qid)Quoted 8 lines
drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +1
-1/* destroy kernel queue (DIQ) */ dqm = pqn->kq->dev->dqm; dqm->ops.destroy_kernel_queue(dqm, pqn->kq, &pdd->qpd); - kernel_queue_uninit(pqn->kq); + kernel_queue_uninit(pqn->kq, false); } if (pqn->q) {
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Hi Felix
Do you know how I can get a "xxx.patch" file from the email from you ??
_____________________________________
Monk Liu|GPU Virtualization Team |AMD
-----Original Message-----
From: Kuehling, Felix <[email protected]>
Sent: Friday, December 20, 2019 10:09 AM
To: [email protected]
Cc: Liu, Shaoyun <[email protected]>; Liu, Monk <[email protected]>
Subject: [PATCH 1/1] drm/amdkfd: Don't touch the hardware in pre_reset callback
The intention of the pre_reset callback is to update the driver state to reflect that all user mode queues are preempted and the HIQ is destroyed. However we should not actually preempt any queues or otherwise touch the hardware because it is presumably hanging. The impending reset will take care of actually stopping all queues.
This should prevent KIQ register access hanging on SRIOV function level reset (FLR). It should also speed up the reset by avoiding unnecessary timeouts from a potentially hanging GPU scheduler.
--- drivers/gpu/drm/amd/amdkfd/kfd_device.c | 24 ++++++++++------- .../drm/amd/amdkfd/kfd_device_queue_manager.c | 27 ++++++++++++------- .../drm/amd/amdkfd/kfd_device_queue_manager.h | 5 ++-- drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c | 8 +++--- .../gpu/drm/amd/amdkfd/kfd_packet_manager.c | 4 +-- drivers/gpu/drm/amd/amdkfd/kfd_priv.h | 8 +++--- drivers/gpu/drm/amd/amdkfd/kfd_process.c | 11 ++++---- .../amd/amdkfd/kfd_process_queue_manager.c | 2 +- 8 files changed, 53 insertions(+), 36 deletions(-)
drivers/gpu/drm/amd/amdkfd/kfd_device.c +1 -0
@@ -486,6 +486,7 @@ static int kfd_gtt_sa_init(struct kfd_dev *kfd, unsigned int buf_size, unsigned int chunk_size);static void kfd_gtt_sa_fini (struct kfd_dev *kfd); +static void kfd_suspend (struct kfd_dev *kfd,bool pre_reset);static int kfd_resume (struct kfd_dev *kfd); struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd, @@ -728,7 +729,7 @@ int kgd2kfd_pre_reset(struct kfd_dev *kfd) {
if (!kfd->init_complete)
return 0;
- kgd2kfd_suspend(kfd);
+ kfd_suspend(kfd, true);
kfd_signal_reset_event(kfd);
return 0;
@@ -767,13 +768,7 @@ void kgd2kfd_suspend(struct kfd_dev *kfd)
if (!kfd->init_complete)
return;
- /* For first KFD device suspend all the KFD processes */
- if (atomic_inc_return(&kfd_locked) == 1)
- kfd_suspend_all_processes();
-
- kfd->dqm->ops.stop(kfd->dqm);
-
- kfd_iommu_suspend(kfd);
+ kfd_suspend(kfd, false);
}
int kgd2kfd_resume(struct kfd_dev *kfd) @@ -795,6 +790,17 @@ int kgd2kfd_resume(struct kfd_dev *kfd)
return ret;
}
+static void kfd_suspend(struct kfd_dev *kfd, bool pre_reset) {
+ /* For first KFD device suspend all the KFD processes */
+ if (atomic_inc_return(&kfd_locked) == 1)
+ kfd_suspend_all_processes(pre_reset);
+
+ kfd->dqm->ops.stop(kfd->dqm, pre_reset);
+
+ kfd_iommu_suspend(kfd);
+}
+
static int kfd_resume(struct kfd_dev *kfd) {
int err = 0;
@@ -877,7 +883,7 @@ int kgd2kfd_quiesce_mm(struct mm_struct *mm)
if (!p)
return -ESRCH;
- r = kfd_process_evict_queues(p);
+ r = kfd_process_evict_queues(p, false);
kfd_unref_process(p);
return r;
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +13 -4
@@ -592,7 +592,8 @@ static int update_queue(struct device_queue_manager *dqm, struct queue *q) }static int evict_process_queues_nocpsch (struct device_queue_manager *dqm, -struct qcm_process_device *qpd) +struct qcm_process_device *qpd, +bool pre_reset) {struct queue *q;struct mqd_manager *mqd_mgr;
@@ -622,6 +623,8 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm,if (WARN_ONCE (!dqm->sched_running,"Evict when stopped\n" ))continue ; +if (pre_reset) +continue ; retval = mqd_mgr->destroy_mqd(mqd_mgr, q->mqd, KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN,
@@ -639,7 +642,8 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm, }static int evict_process_queues_cpsch (struct device_queue_manager *dqm, -struct qcm_process_device *qpd) +struct qcm_process_device *qpd, +bool pre_reset) {struct queue *q;struct kfd_process_device *pdd;
@@ -664,6 +668,10 @@ static int evict_process_queues_cpsch(struct device_queue_manager *dqm, q->properties.is_active = false; dqm->queue_count--; } + +if (pre_reset) + goto out; + retval = execute_queues_cpsch(dqm, qpd->is_debug ? KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES :
@@ -944,10 +952,10 @@ static int start_nocpsch(struct device_queue_manager *dqm)return 0 ; } -static int stop_nocpsch (struct device_queue_manager *dqm) +static int stop_nocpsch (struct device_queue_manager *dqm,bool +pre_reset) {if (dqm->dev->device_info->asic_family ==CHIP_HAWAII ) -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, pre_reset); dqm->sched_running = false;
return 0;
@@ -1107,20 +1115,21 @@ static int start_cpsch(struct device_queue_manager *dqm)
return 0;
fail_allocate_vidmem:
fail_set_sched_resources:
- pm_uninit(&dqm->packets);
+ pm_uninit(&dqm->packets, false);
fail_packet_manager_init:
return retval;
}
-static int stop_cpsch(struct device_queue_manager *dqm) +static int stop_cpsch(struct device_queue_manager *dqm, bool pre_reset)
{
dqm_lock(dqm);
- unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0);
+ if (!pre_reset)
+ unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0);
dqm->sched_running = false;
dqm_unlock(dqm);
kfd_gtt_sa_free(dqm->dev, dqm->fence_mem);
- pm_uninit(&dqm->packets);
+ pm_uninit(&dqm->packets, pre_reset);
return 0;
}
@@ -1891,7 +1900,7 @@ int kfd_process_vm_fault(struct device_queue_manager *dqm,
return -EINVAL;
pdd = kfd_get_process_device_data(dqm->dev, p);
if (pdd)
- ret = dqm->ops.evict_process_queues(dqm, &pdd->qpd);
+ ret = dqm->ops.evict_process_queues(dqm, &pdd->qpd, false);
kfd_unref_process(p);
return ret;
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h +3 -2
@@ -103,7 +103,7 @@ struct device_queue_manager_ops {int (*initialize)(struct device_queue_manager *dqm);int (*start)(struct device_queue_manager *dqm); -int (*stop)(struct device_queue_manager *dqm); +int (*stop)(struct device_queue_manager *dqm,bool pre_reset);void (*uninitialize)(struct device_queue_manager *dqm); int (*create_kernel_queue)(struct device_queue_manager *dqm, struct kernel_queue *kq,
@@ -129,7 +129,8 @@ struct device_queue_manager_ops { struct qcm_process_device *qpd);int (*evict_process_queues)(struct device_queue_manager *dqm, -struct qcm_process_device *qpd); +struct qcm_process_device *qpd, +bool pre_reset);int (*restore_process_queues)(struct device_queue_manager *dqm,struct qcm_process_device *qpd);
drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c +4 -3
@@ -195,9 +195,9 @@ static bool kq_initialize(struct kernel_queue *kq, struct kfd_dev *dev, } /* Uninitialize a kernel queue and free all its memory usages. */ -static void kq_uninitialize(struct kernel_queue *kq) +static void kq_uninitialize(struct kernel_queue *kq, bool pre_reset) { -if (kq->queue->properties.type ==KFD_QUEUE_TYPE_HIQ ) +if (kq->queue->properties.type ==KFD_QUEUE_TYPE_HIQ && !pre_reset) kq->mqd_mgr->destroy_mqd(kq->mqd_mgr, kq->queue->mqd, KFD_PREEMPT_TYPE_WAVEFRONT_RESET,
@@ -337,9 +337,9 @@ struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,return NULL ; } -void kernel_queue_uninit (struct kernel_queue *kq) +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset) { -kq_uninitialize (kq); +kq_uninitialize (kq, pre_reset);kfree (kq); }
drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c +2 -2
@@ -264,10 +264,10 @@ int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm)return 0 ; } -void pm_uninit (struct packet_manager *pm) +void pm_uninit (struct packet_manager *pm,bool pre_reset) {mutex_destroy (&pm->lock); -kernel_queue_uninit (pm->priv_queue); +kernel_queue_uninit (pm->priv_queue, pre_reset); } int pm_send_set_resources(struct packet_manager *pm, diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index 087e96838997..9bc4ced4acba 100644
drivers/gpu/drm/amd/amdkfd/kfd_priv.h +3 -1
@@ -762,9 +762,9 @@ struct kfd_process *kfd_get_process(const struct task_struct *); struct kfd_process *kfd_lookup_process_by_pasid(unsigned int pasid); struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm); void kfd_unref_process(struct kfd_process *p); -int kfd_process_evict_queues(struct kfd_process *p); +int kfd_process_evict_queues (struct kfd_process *p,bool pre_reset); int kfd_process_restore_queues(struct kfd_process *p); -void kfd_suspend_all_processes(void); +void kfd_suspend_all_processes (bool pre_reset);int kfd_resume_all_processes (void ); int kfd_process_device_init_vm(struct kfd_process_device *pdd, @@ -883,7 +883,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev); void device_queue_manager_uninit(struct device_queue_manager *dqm); struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,enum kfd_queue_type type); -void kernel_queue_uninit (struct kernel_queue *kq); +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset);int kfd_process_vm_fault (struct device_queue_manager *dqm,unsigned int pasid);
/* Process Queue Manager */
@@ -974,7 +974,7 @@ extern const struct packet_manager_funcs kfd_vi_pm_funcs; extern const struct packet_manager_funcs kfd_v9_pm_funcs;
int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm); -void pm_uninit(struct packet_manager *pm);
+void pm_uninit(struct packet_manager *pm, bool pre_reset);
int pm_send_set_resources(struct packet_manager *pm,
struct scheduling_resources *res);
int pm_send_runlist(struct packet_manager *pm, struct list_head *dqm_queues); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
index 536a153ac9a4..7bcadd3a1e3b 100644
drivers/gpu/drm/amd/amdkfd/kfd_process.c +6 -5
@@ -948,7 +948,7 @@ struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm)* Eviction is reference-counted per process-device. This means multiple * evictions from different sources can be nested safely. */ -int kfd_process_evict_queues (struct kfd_process *p) +int kfd_process_evict_queues (struct kfd_process *p,bool pre_reset) {struct kfd_process_device *pdd;int r =0 ;
@@ -956,7 +956,8 @@ int kfd_process_evict_queues(struct kfd_process *p)list_for_each_entry (pdd, &p->per_device_data, per_device_list) { r = pdd->dev->dqm->ops.evict_process_queues (pdd->dev->dqm, - &pdd->qpd); + &pdd->qpd, + pre_reset);if (r) {pr_err ("Failed to evict process queues\n" ); goto fail;
@@ -1026,7 +1027,7 @@ static void evict_process_worker(struct work_struct *work)flush_delayed_work (&p->restore_work);pr_debug ("Started evicting pasid 0x%x\n" , p->pasid); - ret =kfd_process_evict_queues (p); + ret =kfd_process_evict_queues (p, false);if (!ret) {dma_fence_signal (p->ef);dma_fence_put (p->ef);
@@ -1082,7 +1083,7 @@ static void restore_process_worker(struct work_struct *work)pr_err ("Failed to restore queues of pasid 0x%x\n" , p->pasid); } -void kfd_suspend_all_processes (void ) +void kfd_suspend_all_processes (bool pre_reset) {struct kfd_process *p;unsigned int temp;
@@ -1092,7 +1093,7 @@ void kfd_suspend_all_processes(void)cancel_delayed_work_sync (&p->eviction_work);cancel_delayed_work_sync (&p->restore_work); -if (kfd_process_evict_queues (p)) +if (kfd_process_evict_queues (p, pre_reset))pr_err ("Failed to suspend process 0x%x\n" , p->pasid);dma_fence_signal (p->ef);dma_fence_put (p->ef);
drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +1 -1
@@ -374,7 +374,7 @@ int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid)/* destroy kernel queue (DIQ) */ dqm = pqn->kq->dev->dqm; dqm->ops.destroy_kernel_queue (dqm, pqn->kq, &pdd->qpd); -kernel_queue_uninit (pqn->kq); +kernel_queue_uninit (pqn->kq, false); }if (pqn->q) {
--
2.24.1
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Felix Kuehling
maintainerYou should be able to save the plain text email and pass that to "git am". It's trivial with Thunderbird on a Linux system. If you're using outlook, I'm not sure.
Anyway, I'm already reworking the patch based on Shaoyun's suggestion and some ideas it gave me.
Regards,
Felix
Quoted 332 lines · Monk
Hi Felix
Do you know how I can get a "xxx.patch" file from the email from you ??
_____________________________________
Monk Liu|GPU Virtualization Team |AMD-----Original Message-----
From: Kuehling, Felix <[email protected]>
Sent: Friday, December 20, 2019 10:09 AM
To: [email protected]
Cc: Liu, Shaoyun <[email protected]>; Liu, Monk <[email protected]>
Subject: [PATCH 1/1] drm/amdkfd: Don't touch the hardware inpre_resetcallbackThe intention of the
pre_resetcallback is to update the driver state to reflect that all user mode queues are preempted and the HIQ is destroyed. However we should not actually preempt any queues or otherwise touch the hardware because it is presumably hanging. The impending reset will take care of actually stopping all queues.This should prevent KIQ register access hanging on SRIOV function level reset (FLR). It should also speed up the reset by avoiding unnecessary timeouts from a potentially hanging GPU scheduler.
CC shaoyunl <[email protected]>CC Liu Monk <[email protected]>Signed-off-by Felix Kuehling <[email protected]>--- drivers/gpu/drm/amd/amdkfd/kfd_device.c | 24 ++++++++++------- .../drm/amd/amdkfd/kfd_device_queue_manager.c | 27 ++++++++++++------- .../drm/amd/amdkfd/kfd_device_queue_manager.h | 5 ++-- drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c | 8 +++--- .../gpu/drm/amd/amdkfd/kfd_packet_manager.c | 4 +-- drivers/gpu/drm/amd/amdkfd/kfd_priv.h | 8 +++--- drivers/gpu/drm/amd/amdkfd/kfd_process.c | 11 ++++---- .../amd/amdkfd/kfd_process_queue_manager.c | 2 +- 8 files changed, 53 insertions(+), 36 deletions(-)drivers/gpu/drm/amd/amdkfd/kfd_device.c +1
-0@@ -486,6 +486,7 @@ static int kfd_gtt_sa_init(struct kfd_dev *kfd, unsigned int buf_size, unsigned int chunk_size);static void kfd_gtt_sa_fini (struct kfd_dev *kfd); +static void kfd_suspend (struct kfd_dev *kfd,bool pre_reset);static int kfd_resume (struct kfd_dev *kfd); struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd, @@ -728,7 +729,7 @@ int kgd2kfd_pre_reset(struct kfd_dev *kfd) {if (!kfd->init_complete)
return 0;drivers/gpu/drm/amd/amdkfd/kfd_device.c +2
-8-kgd2kfd_suspend (kfd); +kfd_suspend (kfd, true);kfd_signal_reset_event (kfd);return 0 ;@@ -767,13 +768,7 @@ void kgd2kfd_suspend(struct kfd_dev *kfd)if (!kfd->init_complete)return ; -/* For first KFD device suspend all the KFD processes */ -if (atomic_inc_return (&kfd_locked) ==1 ) -kfd_suspend_all_processes (); - - kfd->dqm->ops.stop (kfd->dqm); - -kfd_iommu_suspend (kfd); +kfd_suspend (kfd, false); } int kgd2kfd_resume(struct kfd_dev *kfd) @@ -795,6 +790,17 @@ int kgd2kfd_resume(struct kfd_dev *kfd)return ret; }drivers/gpu/drm/amd/amdkfd/kfd_device.c +11
-1+static void kfd_suspend (struct kfd_dev *kfd,bool pre_reset) { +/* For first KFD device suspend all the KFD processes */ +if (atomic_inc_return (&kfd_locked) ==1 ) +kfd_suspend_all_processes (pre_reset); + + kfd->dqm->ops.stop (kfd->dqm, pre_reset); + +kfd_iommu_suspend (kfd); +} +static int kfd_resume (struct kfd_dev *kfd) {int err =0 ;@@ -877,7 +883,7 @@ int kgd2kfd_quiesce_mm(struct mm_struct *mm)if (!p)return -ESRCH ; - r =kfd_process_evict_queues (p); + r =kfd_process_evict_queues (p, false);kfd_unref_process (p);return r;drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +19
-9@@ -592,7 +592,8 @@ static int update_queue(struct device_queue_manager *dqm, struct queue *q) }static int evict_process_queues_nocpsch (struct device_queue_manager *dqm, -struct qcm_process_device *qpd) +struct qcm_process_device *qpd, +bool pre_reset) {struct queue *q;struct mqd_manager *mqd_mgr;@@ -622,6 +623,8 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm,if (WARN_ONCE (!dqm->sched_running,"Evict when stopped\n" ))continue ; +if (pre_reset) +continue ; retval = mqd_mgr->destroy_mqd(mqd_mgr, q->mqd, KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN,@@ -639,7 +642,8 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm, }static int evict_process_queues_cpsch (struct device_queue_manager *dqm, -struct qcm_process_device *qpd) +struct qcm_process_device *qpd, +bool pre_reset) {struct queue *q;struct kfd_process_device *pdd;@@ -664,6 +668,10 @@ static int evict_process_queues_cpsch(struct device_queue_manager *dqm, q->properties.is_active = false; dqm->queue_count--; } + +if (pre_reset) + goto out; + retval = execute_queues_cpsch(dqm, qpd->is_debug ? KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES :@@ -944,10 +952,10 @@ static int start_nocpsch(struct device_queue_manager *dqm)return 0 ; } -static int stop_nocpsch (struct device_queue_manager *dqm) +static int stop_nocpsch (struct device_queue_manager *dqm,bool +pre_reset) {if (dqm->dev->device_info->asic_family ==CHIP_HAWAII ) -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, pre_reset); dqm->sched_running = false;return 0 ;@@ -1107,20 +1115,21 @@ static int start_cpsch(struct device_queue_manager *dqm)return 0 ; fail_allocate_vidmem: fail_set_sched_resources: -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, false); fail_packet_manager_init:return retval; } -static int stop_cpsch (struct device_queue_manager *dqm) +static int stop_cpsch (struct device_queue_manager *dqm,bool pre_reset) {dqm_lock (dqm); -unmap_queues_cpsch (dqm,KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES ,0 ); +if (!pre_reset) +unmap_queues_cpsch (dqm,KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES ,0 ); dqm->sched_running = false;dqm_unlock (dqm);kfd_gtt_sa_free (dqm->dev, dqm->fence_mem); -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, pre_reset);return 0 ; }@@ -1891,7 +1900,7 @@ int kfd_process_vm_fault(struct device_queue_manager *dqm,return -EINVAL ; pdd =kfd_get_process_device_data (dqm->dev, p);if (pdd) - ret = dqm->ops.evict_process_queues (dqm, &pdd->qpd); + ret = dqm->ops.evict_process_queues (dqm, &pdd->qpd, false);kfd_unref_process (p);return ret;drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h +3
-2@@ -103,7 +103,7 @@ struct device_queue_manager_ops {int (*initialize)(struct device_queue_manager *dqm);int (*start)(struct device_queue_manager *dqm); -int (*stop)(struct device_queue_manager *dqm); +int (*stop)(struct device_queue_manager *dqm,bool pre_reset);void (*uninitialize)(struct device_queue_manager *dqm); int (*create_kernel_queue)(struct device_queue_manager *dqm, struct kernel_queue *kq,@@ -129,7 +129,8 @@ struct device_queue_manager_ops { struct qcm_process_device *qpd);int (*evict_process_queues)(struct device_queue_manager *dqm, -struct qcm_process_device *qpd); +struct qcm_process_device *qpd, +bool pre_reset);int (*restore_process_queues)(struct device_queue_manager *dqm,struct qcm_process_device *qpd);drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c +4
-3@@ -195,9 +195,9 @@ static bool kq_initialize(struct kernel_queue *kq, struct kfd_dev *dev, } /* Uninitialize a kernel queue and free all its memory usages. */ -static void kq_uninitialize(struct kernel_queue *kq) +static void kq_uninitialize(struct kernel_queue *kq, bool pre_reset) { -if (kq->queue->properties.type ==KFD_QUEUE_TYPE_HIQ ) +if (kq->queue->properties.type ==KFD_QUEUE_TYPE_HIQ && !pre_reset) kq->mqd_mgr->destroy_mqd(kq->mqd_mgr, kq->queue->mqd, KFD_PREEMPT_TYPE_WAVEFRONT_RESET,@@ -337,9 +337,9 @@ struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,return NULL ; } -void kernel_queue_uninit (struct kernel_queue *kq) +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset) { -kq_uninitialize (kq); +kq_uninitialize (kq, pre_reset);kfree (kq); }drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c +2
-2@@ -264,10 +264,10 @@ int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm)return 0 ; } -void pm_uninit (struct packet_manager *pm) +void pm_uninit (struct packet_manager *pm,bool pre_reset) {mutex_destroy (&pm->lock); -kernel_queue_uninit (pm->priv_queue); +kernel_queue_uninit (pm->priv_queue, pre_reset); } int pm_send_set_resources(struct packet_manager *pm, diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.hindex
087e96838997..9bc4ced4acba100644drivers/gpu/drm/amd/amdkfd/kfd_priv.h +4
-1@@ -762,9 +762,9 @@ struct kfd_process *kfd_get_process(const struct task_struct *); struct kfd_process *kfd_lookup_process_by_pasid(unsigned int pasid); struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm); void kfd_unref_process(struct kfd_process *p); -int kfd_process_evict_queues(struct kfd_process *p); +int kfd_process_evict_queues (struct kfd_process *p,bool pre_reset); int kfd_process_restore_queues(struct kfd_process *p); -void kfd_suspend_all_processes(void); +void kfd_suspend_all_processes (bool pre_reset);int kfd_resume_all_processes (void ); int kfd_process_device_init_vm(struct kfd_process_device *pdd, @@ -883,7 +883,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev); void device_queue_manager_uninit(struct device_queue_manager *dqm); struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,enum kfd_queue_type type); -void kernel_queue_uninit (struct kernel_queue *kq); +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset);int kfd_process_vm_fault (struct device_queue_manager *dqm,unsigned int pasid);/* Process Queue Manager */ @@ -974,7 +974,7 @@ extern const struct packet_manager_funcs kfd_vi_pm_funcs; extern const struct packet_manager_funcs kfd_v9_pm_funcs; int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm); -void pm_uninit(struct packet_manager *pm); +void pm_uninit (struct packet_manager *pm,bool pre_reset);int pm_send_set_resources (struct packet_manager *pm,struct scheduling_resources *res); int pm_send_runlist(struct packet_manager *pm, struct list_head *dqm_queues); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.cindex
536a153ac9a4..7bcadd3a1e3b100644drivers/gpu/drm/amd/amdkfd/kfd_process.c +6
-5@@ -948,7 +948,7 @@ struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm)* Eviction is reference-counted per process-device. This means multiple * evictions from different sources can be nested safely. */ -int kfd_process_evict_queues (struct kfd_process *p) +int kfd_process_evict_queues (struct kfd_process *p,bool pre_reset) {struct kfd_process_device *pdd;int r =0 ;@@ -956,7 +956,8 @@ int kfd_process_evict_queues(struct kfd_process *p)list_for_each_entry (pdd, &p->per_device_data, per_device_list) { r = pdd->dev->dqm->ops.evict_process_queues (pdd->dev->dqm, - &pdd->qpd); + &pdd->qpd, + pre_reset);if (r) {pr_err ("Failed to evict process queues\n" ); goto fail;@@ -1026,7 +1027,7 @@ static void evict_process_worker(struct work_struct *work)flush_delayed_work (&p->restore_work);pr_debug ("Started evicting pasid 0x%x\n" , p->pasid); - ret =kfd_process_evict_queues (p); + ret =kfd_process_evict_queues (p, false);if (!ret) {dma_fence_signal (p->ef);dma_fence_put (p->ef);@@ -1082,7 +1083,7 @@ static void restore_process_worker(struct work_struct *work)pr_err ("Failed to restore queues of pasid 0x%x\n" , p->pasid); } -void kfd_suspend_all_processes (void ) +void kfd_suspend_all_processes (bool pre_reset) {struct kfd_process *p;unsigned int temp;@@ -1092,7 +1093,7 @@ void kfd_suspend_all_processes(void)cancel_delayed_work_sync (&p->eviction_work);cancel_delayed_work_sync (&p->restore_work); -if (kfd_process_evict_queues (p)) +if (kfd_process_evict_queues (p, pre_reset))pr_err ("Failed to suspend process 0x%x\n" , p->pasid);dma_fence_signal (p->ef);dma_fence_put (p->ef);drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +1
-2@@ -374,7 +374,7 @@ int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid)/* destroy kernel queue (DIQ) */ dqm = pqn->kq->dev->dqm; dqm->ops.destroy_kernel_queue (dqm, pqn->kq, &pdd->qpd); -kernel_queue_uninit (pqn->kq); +kernel_queue_uninit (pqn->kq, false); }if (pqn->q) { --
Quoted 2 lines
2.24.1
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Okay
Can you send me the patch (in attachment) once you finished it, I need to verify it on SRIOV
Thanks
_____________________________________
Monk Liu|GPU Virtualization Team |AMD
-----Original Message-----
From: Kuehling, Felix <[email protected]>
Sent: Friday, December 20, 2019 3:56 PM
To: Liu, Monk <[email protected]>; [email protected]
Cc: Liu, Shaoyun <[email protected]>
Subject: Re: [PATCH 1/1] drm/amdkfd: Don't touch the hardware in pre_reset callback
You should be able to save the plain text email and pass that to "git am". It's trivial with Thunderbird on a Linux system. If you're using outlook, I'm not sure.
Anyway, I'm already reworking the patch based on Shaoyun's suggestion and some ideas it gave me.
Regards,
Felix
Quoted 359 lines · Monk
Hi Felix
Do you know how I can get a "xxx.patch" file from the email from you ??
_____________________________________
Monk Liu|GPU Virtualization Team |AMD-----Original Message-----
From: Kuehling, Felix <[email protected]>
Sent: Friday, December 20, 2019 10:09 AM
To: [email protected]
Cc: Liu, Shaoyun <[email protected]>; Liu, Monk <[email protected]>
Subject: [PATCH 1/1] drm/amdkfd: Don't touch the hardware inpre_resetcallbackThe intention of the
pre_resetcallback is to update the driver state to reflect that all user mode queues are preempted and the HIQ is destroyed. However we should not actually preempt any queues or otherwise touch the hardware because it is presumably hanging. The impending reset will take care of actually stopping all queues.This should prevent KIQ register access hanging on SRIOV function level reset (FLR). It should also speed up the reset by avoiding unnecessary timeouts from a potentially hanging GPU scheduler.
CC shaoyunl <[email protected]>CC Liu Monk <[email protected]>Signed-off-by Felix Kuehling <[email protected]>--- drivers/gpu/drm/amd/amdkfd/kfd_device.c | 24 ++++++++++------- .../drm/amd/amdkfd/kfd_device_queue_manager.c | 27 ++++++++++++------- .../drm/amd/amdkfd/kfd_device_queue_manager.h | 5 ++-- drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c | 8 +++--- .../gpu/drm/amd/amdkfd/kfd_packet_manager.c | 4 +-- drivers/gpu/drm/amd/amdkfd/kfd_priv.h | 8 +++--- drivers/gpu/drm/amd/amdkfd/kfd_process.c | 11 ++++---- .../amd/amdkfd/kfd_process_queue_manager.c | 2 +- 8 files changed, 53 insertions(+), 36 deletions(-)drivers/gpu/drm/amd/amdkfd/kfd_device.c +1
-0@@ -486,6 +486,7 @@ static int kfd_gtt_sa_init(struct kfd_dev *kfd, unsigned int buf_size, unsigned int chunk_size);static void kfd_gtt_sa_fini (struct kfd_dev *kfd); +static void kfd_suspend (struct kfd_dev *kfd,bool pre_reset);static int kfd_resume (struct kfd_dev *kfd); struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd, @@ -728,7 +729,7 @@ int kgd2kfd_pre_reset(struct kfd_dev *kfd) {if (!kfd->init_complete)
return 0;drivers/gpu/drm/amd/amdkfd/kfd_device.c +2
-8-kgd2kfd_suspend (kfd); +kfd_suspend (kfd, true);kfd_signal_reset_event (kfd);return 0 ;@@ -767,13 +768,7 @@ void kgd2kfd_suspend(struct kfd_dev *kfd)if (!kfd->init_complete)return ; -/* For first KFD device suspend all the KFD processes */ -if (atomic_inc_return (&kfd_locked) ==1 ) -kfd_suspend_all_processes (); - - kfd->dqm->ops.stop (kfd->dqm); - -kfd_iommu_suspend (kfd); +kfd_suspend (kfd, false); } int kgd2kfd_resume(struct kfd_dev *kfd) @@ -795,6 +790,17 @@ int kgd2kfd_resume(struct kfd_dev *kfd)return ret; }drivers/gpu/drm/amd/amdkfd/kfd_device.c +11
-1+static void kfd_suspend (struct kfd_dev *kfd,bool pre_reset) { +/* For first KFD device suspend all the KFD processes */ +if (atomic_inc_return (&kfd_locked) ==1 ) +kfd_suspend_all_processes (pre_reset); + + kfd->dqm->ops.stop (kfd->dqm, pre_reset); + +kfd_iommu_suspend (kfd); +} +static int kfd_resume (struct kfd_dev *kfd) {int err =0 ;@@ -877,7 +883,7 @@ int kgd2kfd_quiesce_mm(struct mm_struct *mm)if (!p)return -ESRCH ; - r =kfd_process_evict_queues (p); + r =kfd_process_evict_queues (p, false);kfd_unref_process (p);return r;drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +4
-1@@ -592,7 +592,8 @@ static int update_queue(struct device_queue_manager *dqm, struct queue *q) }static int evict_process_queues_nocpsch (struct device_queue_manager *dqm, -struct qcm_process_device *qpd) +struct qcm_process_device *qpd, +bool pre_reset) {struct queue *q;struct mqd_manager *mqd_mgr;@@ -622,6 +623,8 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm, if (WARN_ONCE(!dqm->sched_running, "Evict when stopped\n")) continue; +if (pre_reset) +continue ; retval = mqd_mgr->destroy_mqd(mqd_mgr, q->mqd,
KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN, @@ -639,7 +642,8 @@ static int
evict_process_queues_nocpsch(structdevice_queue_manager*dqm, }static int evict_process_queues_cpsch(struct
device_queue_manager*dqm,drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +16
-8- struct qcm_process_device *qpd) + struct qcm_process_device *qpd, + bool pre_reset) {struct queue *q;struct kfd_process_device *pdd;@@ -664,6 +668,10 @@ static int evict_process_queues_cpsch(struct device_queue_manager *dqm, q->properties.is_active = false; dqm->queue_count--; } + +if (pre_reset) + goto out; + retval = execute_queues_cpsch(dqm, qpd->is_debug ? KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES :@@ -944,10 +952,10 @@ static int start_nocpsch(struct device_queue_manager *dqm)return 0 ; } -static int stop_nocpsch (struct device_queue_manager *dqm) +static int stop_nocpsch (struct device_queue_manager *dqm,bool +pre_reset) {if (dqm->dev->device_info->asic_family ==CHIP_HAWAII ) -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, pre_reset); dqm->sched_running = false;return 0 ;@@ -1107,20 +1115,21 @@ static int start_cpsch(struct device_queue_manager *dqm)return 0 ; fail_allocate_vidmem: fail_set_sched_resources: -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, false); fail_packet_manager_init:return retval; } -static int stop_cpsch (struct device_queue_manager *dqm) +static int stop_cpsch (struct device_queue_manager *dqm,bool +pre_reset) {dqm_lock (dqm); -unmap_queues_cpsch (dqm,KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES ,0 ); +if (!pre_reset) +unmap_queues_cpsch (dqm,KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES ,0 ); dqm->sched_running = false;dqm_unlock (dqm);kfd_gtt_sa_free (dqm->dev, dqm->fence_mem); -pm_uninit (&dqm->packets); +pm_uninit (&dqm->packets, pre_reset);return 0 ; }@@ -1891,7 +1900,7 @@ int kfd_process_vm_fault(struct device_queue_manager *dqm,return -EINVAL ; pdd =kfd_get_process_device_data (dqm->dev, p);if (pdd) - ret = dqm->ops.evict_process_queues (dqm, &pdd->qpd); + ret = dqm->ops.evict_process_queues (dqm, &pdd->qpd, false);kfd_unref_process (p);return ret;drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h +3
-2@@ -103,7 +103,7 @@ struct device_queue_manager_ops {int (*initialize)(struct device_queue_manager *dqm);int (*start)(struct device_queue_manager *dqm); -int (*stop)(struct device_queue_manager *dqm); +int (*stop)(struct device_queue_manager *dqm,bool pre_reset);void (*uninitialize)(struct device_queue_manager *dqm); int (*create_kernel_queue)(struct device_queue_manager *dqm, struct kernel_queue *kq,@@ -129,7 +129,8 @@ struct device_queue_manager_ops { struct qcm_process_device *qpd);int (*evict_process_queues)(struct device_queue_manager *dqm, -struct qcm_process_device *qpd); +struct qcm_process_device *qpd, +bool pre_reset);int (*restore_process_queues)(struct device_queue_manager *dqm,struct qcm_process_device *qpd);drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c +2
-2@@ -195,9 +195,9 @@ static bool kq_initialize(struct kernel_queue *kq, struct kfd_dev *dev, }/* Uninitialize a kernel queue and free all its memory usages. */ -static void kq_uninitialize (struct kernel_queue *kq) +static void kq_uninitialize (struct kernel_queue *kq,bool pre_reset) { -if (kq->queue->properties.type ==KFD_QUEUE_TYPE_HIQ ) +if (kq->queue->properties.type ==KFD_QUEUE_TYPE_HIQ && !pre_reset) kq->mqd_mgr->destroy_mqd(kq->mqd_mgr, kq->queue->mqd, KFD_PREEMPT_TYPE_WAVEFRONT_RESET, @@ -337,9 +337,9 @@ struct
kernel_queue*kernel_queue_init(structkfd_dev*dev,
return NULL;
}drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c +2
-2-void kernel_queue_uninit (struct kernel_queue *kq) +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset) { -kq_uninitialize (kq); +kq_uninitialize (kq, pre_reset);kfree (kq); }drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c +2
-2@@ -264,10 +264,10 @@ int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm)return 0 ; } -void pm_uninit (struct packet_manager *pm) +void pm_uninit (struct packet_manager *pm,bool pre_reset) {mutex_destroy (&pm->lock); -kernel_queue_uninit (pm->priv_queue); +kernel_queue_uninit (pm->priv_queue, pre_reset); } int pm_send_set_resources(struct packet_manager *pm, diff --git
a/drivers/gpu/drm/amd/amdkfd/kfd_priv.hb/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index087e96838997..9bc4ced4acba100644drivers/gpu/drm/amd/amdkfd/kfd_priv.h +0
-0@@ -762,9 +762,9 @@ struct kfd_process *kfd_get_process(const struct
task_struct*); structkfd_process*kfd_lookup_process_by_pasid(unsigned int pasid); structkfd_process*kfd_lookup_process_by_mm(const structmm_struct*mm); voidkfd_unref_process(struct kfd_process *p); -intkfd_process_evict_queues(struct kfd_process *p);
+intkfd_process_evict_queues(struct kfd_process *p, bool pre_reset);
intkfd_process_restore_queues(struct kfd_process *p); -voidkfd_suspend_all_processes(void);drivers/gpu/drm/amd/amdkfd/kfd_priv.h +2
-1+void kfd_suspend_all_processes (bool pre_reset);int kfd_resume_all_processes (void ); int kfd_process_device_init_vm(struct kfd_process_device *pdd, @@ -883,7 +883,7 @@ struct device_queue_manager *device_queue_manager_init(struct kfd_dev *dev); void device_queue_manager_uninit(struct device_queue_manager *dqm); struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,enum kfd_queue_type type); -void kernel_queue_uninit (struct kernel_queue *kq); +void kernel_queue_uninit (struct kernel_queue *kq,bool pre_reset); int kfd_process_vm_fault(struct device_queue_manager *dqm, unsignedint pasid);
/* Process Queue Manager */
drivers/gpu/drm/amd/amdkfd/kfd_priv.h +0
-0@@ -974,7 +974,7 @@ extern const struct packet_manager_funcs
kfd_vi_pm_funcs; extern const structpacket_manager_funcskfd_v9_pm_funcs;int pm_init(struct
packet_manager*pm, structdevice_queue_manager
*dqm); -voidpm_uninit(struct packet_manager *pm);
+voidpm_uninit(struct packet_manager *pm, bool pre_reset);
int pm_send_set_resources(structpacket_manager*pm,
structscheduling_resources*res);
int pm_send_runlist(structpacket_manager*pm, structlist_head*dqm_queues); diff --gita/drivers/gpu/drm/amd/amdkfd/kfd_process.cb/drivers/gpu/drm/amd/amdkfd/kfd_process.c
index536a153ac9a4..7bcadd3a1e3b100644drivers/gpu/drm/amd/amdkfd/kfd_process.c +3
-2@@ -948,7 +948,7 @@ struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm)* Eviction is reference-counted per process-device. This means multiple * evictions from different sources can be nested safely. */ -int kfd_process_evict_queues (struct kfd_process *p) +int kfd_process_evict_queues (struct kfd_process *p,bool pre_reset) {struct kfd_process_device *pdd;int r =0 ;@@ -956,7 +956,8 @@ int kfd_process_evict_queues(struct kfd_process *p)list_for_each_entry (pdd, &p->per_device_data, per_device_list) { r = pdd->dev->dqm->ops.evict_process_queues (pdd->dev->dqm, - &pdd->qpd); + &pdd->qpd, + pre_reset);if (r) {pr_err ("Failed to evict process queues\n" );goto fail;
drivers/gpu/drm/amd/amdkfd/kfd_process.c +3
-3@@ -1026,7 +1027,7 @@ static void evict_process_worker(struct work_struct *work)flush_delayed_work (&p->restore_work);pr_debug ("Started evicting pasid 0x%x\n" , p->pasid); - ret =kfd_process_evict_queues (p); + ret =kfd_process_evict_queues (p, false);if (!ret) {dma_fence_signal (p->ef);dma_fence_put (p->ef);@@ -1082,7 +1083,7 @@ static void restore_process_worker(struct work_struct *work)pr_err ("Failed to restore queues of pasid 0x%x\n" , p->pasid); } -void kfd_suspend_all_processes (void ) +void kfd_suspend_all_processes (bool pre_reset) {struct kfd_process *p;unsigned int temp;@@ -1092,7 +1093,7 @@ void kfd_suspend_all_processes(void)cancel_delayed_work_sync (&p->eviction_work);cancel_delayed_work_sync (&p->restore_work); -if (kfd_process_evict_queues (p)) +if (kfd_process_evict_queues (p, pre_reset))pr_err ("Failed to suspend process 0x%x\n" , p->pasid);dma_fence_signal (p->ef);dma_fence_put (p->ef);drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +1
-2@@ -374,7 +374,7 @@ int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid)/* destroy kernel queue (DIQ) */ dqm = pqn->kq->dev->dqm; dqm->ops.destroy_kernel_queue (dqm, pqn->kq, &pdd->qpd); -kernel_queue_uninit (pqn->kq); +kernel_queue_uninit (pqn->kq, false); }if (pqn->q) { --
Quoted 2 lines
2.24.1
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx