[PATCH v2] drm/amdkfd: Ignored various return code
Martin, Andrew
Andrew.Martin at amd.com
Mon Dec 30 18:15:01 UTC 2024
[Public]
> -----Original Message-----
> From: Russell, Kent <Kent.Russell at amd.com>
> Sent: Monday, December 30, 2024 12:44 PM
> To: Martin, Andrew <Andrew.Martin at amd.com>; amd-
> gfx at lists.freedesktop.org
> Cc: Kuehling, Felix <Felix.Kuehling at amd.com>; Martin, Andrew
> <Andrew.Martin at amd.com>
> Subject: RE: [PATCH v2] drm/amdkfd: Ignored various return code
>
> [Public]
>
> Might be worth changing the commit headline to what the patch does,
> something like Don't ignore return codes Or Various return code cleanup
>
> Otherwise it sounds like your patch is intentionally ignoring return codes,
> which is the opposite of what you're doing. Also 2 comments below.
Thanks, will do!, Please see response to your 2 comments also!
>
> > -----Original Message-----
> > From: amd-gfx <amd-gfx-bounces at lists.freedesktop.org> On Behalf Of
> > Andrew Martin
> > Sent: Monday, December 30, 2024 12:02 PM
> > To: amd-gfx at lists.freedesktop.org
> > Cc: Kuehling, Felix <Felix.Kuehling at amd.com>; Russell, Kent
> > <Kent.Russell at amd.com>; Martin, Andrew <Andrew.Martin at amd.com>
> > Subject: [PATCH v2] drm/amdkfd: Ignored various return code
> >
> > This patch checks and handles the return code of the called function.
> >
> > Signed-off-by: Andrew Martin <Andrew.Martin at amd.com>
> > ---
> > drivers/gpu/drm/amd/amdkfd/kfd_chardev.c | 2 +-
> > drivers/gpu/drm/amd/amdkfd/kfd_crat.c | 2 ++
> > drivers/gpu/drm/amd/amdkfd/kfd_debug.c | 19 ++++++++++++++-----
> > .../drm/amd/amdkfd/kfd_device_queue_manager.c | 11 ++++++-----
> > 4 files changed, 23 insertions(+), 11 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
> > b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
> > index 065d87841459..52a11dc01422 100644
> > --- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
> > +++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
> > @@ -2864,7 +2864,7 @@ static int runtime_disable(struct kfd_process *p)
> >
> > pdd->dev->vm_info.last_vmid_kfd);
> >
> > if (!pdd->dev->kfd->shared_resources.enable_mes)
> > - debug_refresh_runlist(pdd->dev->dqm);
> > +
> > + (void)debug_refresh_runlist(pdd->dev->dqm);
> > else
> > kfd_dbg_set_mes_debug_mode(pdd,
> >
> > !kfd_dbg_has_cwsr_workaround(pdd->dev));
> > diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
> > b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
> > index 693469c18c60..f335ed9e0b3a 100644
> > --- a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
> > +++ b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
> > @@ -2351,6 +2351,8 @@ static int kfd_create_vcrat_image_gpu(void
> *pcrat_image,
> > if (kdev->kfd->hive_id) {
> > for (nid = 0; nid < proximity_domain; ++nid) {
> > peer_dev =
> > kfd_topology_device_by_proximity_domain_no_lock(nid);
> > + if (!peer_dev)
> > + return -ENODEV;
> > if (!peer_dev->gpu)
> > continue;
> > if (peer_dev->gpu->kfd->hive_id !=
> > kdev->kfd->hive_id) diff --git
> > a/drivers/gpu/drm/amd/amdkfd/kfd_debug.c
> > b/drivers/gpu/drm/amd/amdkfd/kfd_debug.c
> > index a8abc3091801..b4db409d90a1 100644
> > --- a/drivers/gpu/drm/amd/amdkfd/kfd_debug.c
> > +++ b/drivers/gpu/drm/amd/amdkfd/kfd_debug.c
> > @@ -518,6 +518,9 @@ int kfd_dbg_trap_set_flags(struct kfd_process
> > *target, uint32_t *flags)
> > for (i = 0; i < target->n_pdds; i++) {
> > struct kfd_topology_device *topo_dev =
> >
> > kfd_topology_device_by_id(target->pdds[i]->dev->id);
> > + if (!topo_dev)
> > + return -EINVAL;
> > +
> > uint32_t caps = topo_dev->node_props.capability;
> >
> > if (!(caps &
> > HSA_CAP_TRAP_DEBUG_PRECISE_MEMORY_OPERATIONS_SUPPORTED)
> > &&
> > @@ -565,9 +568,9 @@ int kfd_dbg_trap_set_flags(struct kfd_process
> > *target, uint32_t *flags)
> > continue;
> >
> > if (!pdd->dev->kfd->shared_resources.enable_mes)
> > - debug_refresh_runlist(pdd->dev->dqm);
> > +
> > + (void)debug_refresh_runlist(pdd->dev->dqm);
> > else
> > - kfd_dbg_set_mes_debug_mode(pdd, true);
> > + (void)kfd_dbg_set_mes_debug_mode(pdd,
> > + true);
> > }
> > }
> >
> > @@ -584,8 +587,8 @@ int kfd_dbg_trap_set_flags(struct kfd_process
> > *target, uint32_t *flags)
> > */
> > void kfd_dbg_trap_deactivate(struct kfd_process *target, bool unwind,
> > int
> > unwind_count)
> > {
> > - int i;
> >
> > + int i, r = 0;
> > if (!unwind) {
> > uint32_t flags = 0;
> > int resume_count = resume_queues(target, 0, NULL); @@
> > -627,9 +630,11 @@ void kfd_dbg_trap_deactivate(struct kfd_process
> > *target, bool unwind, int unwind
> > pr_err("Failed to release debug vmid on [%i]\n",
> > pdd->dev-
> > >id);
> >
> > if (!pdd->dev->kfd->shared_resources.enable_mes)
> > - debug_refresh_runlist(pdd->dev->dqm);
> > + r = debug_refresh_runlist(pdd->dev->dqm);
> > else
> > - kfd_dbg_set_mes_debug_mode(pdd,
> > !kfd_dbg_has_cwsr_workaround(pdd->dev));
> > + r = kfd_dbg_set_mes_debug_mode(pdd,
> > !kfd_dbg_has_cwsr_workaround(pdd->dev));
> > + if (r)
> > + break;
> > }
> >
> > kfd_dbg_set_workaround(target, false); @@ -1071,6 +1076,10 @@
> > int kfd_dbg_trap_device_snapshot(struct kfd_process *target,
> > for (i = 0; i < tmp_num_devices; i++) {
> > struct kfd_process_device *pdd = target->pdds[i];
> > struct kfd_topology_device *topo_dev =
> > kfd_topology_device_by_id(pdd->dev->id);
> > + if (!topo_dev) {
> > + r = EINVAL;
> > + break;
> > + }
> >
> > device_info.gpu_id = pdd->dev->id;
> > device_info.exception_status = pdd->exception_status;
> > 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 1405e8affd48..250aa43a39c8 100644
> > --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> > +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> > @@ -1907,16 +1907,18 @@ static int start_cpsch(struct
> > device_queue_manager
> > *dqm)
> >
> > static int stop_cpsch(struct device_queue_manager *dqm) {
> > + int ret = 0;
> > dqm_lock(dqm);
> > if (!dqm->sched_running) {
> > dqm_unlock(dqm);
> > - return 0;
> > + return ret;
I can agree with this one, since it minimizes the changes, but it does flow the same.
>
> I think it's cleaner to just leave the "return 0" . e.g. If the code gets refactored
> and new stuff gets added above, then this might flow differently.
>
>
> > }
> >
> > if (!dqm->dev->kfd->shared_resources.enable_mes)
> > - unmap_queues_cpsch(dqm,
> > KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0,
> USE_DEFAULT_GRACE_PERIOD,
> > false);
> > + ret = unmap_queues_cpsch(dqm,
> > KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES,
> > + 0, USE_DEFAULT_GRACE_PERIOD,
> > + false);
> > else
> > - remove_all_kfd_queues_mes(dqm);
> > + ret = remove_all_kfd_queues_mes(dqm);
> >
> > dqm->sched_running = false;
> >
> > @@ -1930,7 +1932,7 @@ static int stop_cpsch(struct
> > device_queue_manager
> > *dqm)
> > dqm->detect_hang_info = NULL;
> > dqm_unlock(dqm);
> >
> > - return 0;
> > + return ret;
>
> Same here
>
If I change this back to "return 0 ;" then I would have lost the return value from "unmap_queues_cpsch()" or "remove_all_kfd_queues_mes()" which was the whole point of touching this function. But since we in the "stop_cpch()" perhaps it is a more optimal solution is to ignore the return code with
...
(void)unmap_queues_cpsch()
else
(void)remove_all_kfd_queues_mes()
...
Which would be the only change here. Otherwise, I would need to introduce a 3rd return statement. So do let me know which path complements the design more.
> Kent
>
> > }
> >
> > static int create_kernel_queue_cpsch(struct device_queue_manager
> > *dqm, @@ -3439,7 +3441,6 @@ int suspend_queues(struct kfd_process *p,
> > else
> > per_device_suspended++;
> > } else if (err != -EBUSY) {
> > - r = err;
> > queue_ids[q_idx] |=
> > KFD_DBG_QUEUE_ERROR_MASK;
> > break;
> > }
> > --
> > 2.43.0
>
More information about the amd-gfx
mailing list