* [PATCH] drm/etnaviv: fix ref count leak via pm_runtime_get_sync
@ 2020-06-14 6:46 Navid Emamdoost
2020-06-14 9:48 ` Andy Shevchenko
0 siblings, 1 reply; 6+ messages in thread
From: Navid Emamdoost @ 2020-06-14 6:46 UTC (permalink / raw)
To: Lucas Stach, Russell King, Christian Gmeiner, David Airlie,
Daniel Vetter, etnaviv, dri-devel, linux-kernel
Cc: emamd001, wu000273, kjlu, smccaman, Navid Emamdoost
in etnaviv_gpu_submit, etnaviv_gpu_recover_hang, etnaviv_gpu_debugfs,
and etnaviv_gpu_init the call to pm_runtime_get_sync increments the
counter even in case of failure, leading to incorrect ref count.
In case of failure, decrement the ref count before returning.
Signed-off-by: Navid Emamdoost <navid.emamdoost@gmail.com>
---
drivers/gpu/drm/etnaviv/etnaviv_gpu.c | 14 ++++++++++----
1 file changed, 10 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gpu.c b/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
index a31eeff2b297..16f5bc65771a 100644
--- a/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
+++ b/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
@@ -722,7 +722,7 @@ int etnaviv_gpu_init(struct etnaviv_gpu *gpu)
ret = pm_runtime_get_sync(gpu->dev);
if (ret < 0) {
dev_err(gpu->dev, "Failed to enable GPU power domain\n");
- return ret;
+ goto pm_put;
}
etnaviv_hw_identify(gpu);
@@ -819,6 +819,7 @@ int etnaviv_gpu_init(struct etnaviv_gpu *gpu)
fail:
pm_runtime_mark_last_busy(gpu->dev);
+pm_put:
pm_runtime_put_autosuspend(gpu->dev);
return ret;
@@ -859,7 +860,7 @@ int etnaviv_gpu_debugfs(struct etnaviv_gpu *gpu, struct seq_file *m)
ret = pm_runtime_get_sync(gpu->dev);
if (ret < 0)
- return ret;
+ goto pm_put;
dma_lo = gpu_read(gpu, VIVS_FE_DMA_LOW);
dma_hi = gpu_read(gpu, VIVS_FE_DMA_HIGH);
@@ -1003,6 +1004,7 @@ int etnaviv_gpu_debugfs(struct etnaviv_gpu *gpu, struct seq_file *m)
ret = 0;
pm_runtime_mark_last_busy(gpu->dev);
+pm_put:
pm_runtime_put_autosuspend(gpu->dev);
return ret;
@@ -1016,7 +1018,7 @@ void etnaviv_gpu_recover_hang(struct etnaviv_gpu *gpu)
dev_err(gpu->dev, "recover hung GPU!\n");
if (pm_runtime_get_sync(gpu->dev) < 0)
- return;
+ goto pm_put;
mutex_lock(&gpu->lock);
@@ -1035,6 +1037,7 @@ void etnaviv_gpu_recover_hang(struct etnaviv_gpu *gpu)
mutex_unlock(&gpu->lock);
pm_runtime_mark_last_busy(gpu->dev);
+pm_put:
pm_runtime_put_autosuspend(gpu->dev);
}
@@ -1308,8 +1311,10 @@ struct dma_fence *etnaviv_gpu_submit(struct etnaviv_gem_submit *submit)
if (!submit->runtime_resumed) {
ret = pm_runtime_get_sync(gpu->dev);
- if (ret < 0)
+ if (ret < 0) {
+ pm_runtime_put(gpu->dev);
return NULL;
+ }
submit->runtime_resumed = true;
}
@@ -1326,6 +1331,7 @@ struct dma_fence *etnaviv_gpu_submit(struct etnaviv_gem_submit *submit)
ret = event_alloc(gpu, nr_events, event);
if (ret) {
DRM_ERROR("no free events\n");
+ pm_runtime_put(gpu->dev);
return NULL;
}
--
2.17.1
^ permalink raw reply related [flat|nested] 6+ messages in thread* Re: [PATCH] drm/etnaviv: fix ref count leak via pm_runtime_get_sync
2020-06-14 6:46 [PATCH] drm/etnaviv: fix ref count leak via pm_runtime_get_sync Navid Emamdoost
@ 2020-06-14 9:48 ` Andy Shevchenko
2020-06-15 6:12 ` [PATCH v2] " Navid Emamdoost
2020-06-15 6:13 ` [PATCH] " Navid Emamdoost
0 siblings, 2 replies; 6+ messages in thread
From: Andy Shevchenko @ 2020-06-14 9:48 UTC (permalink / raw)
To: Navid Emamdoost
Cc: Lucas Stach, Russell King, Christian Gmeiner, David Airlie,
Daniel Vetter, etnaviv, dri-devel, Linux Kernel Mailing List,
Navid Emamdoost, wu000273, Kangjie Lu, Stephen McCamant
On Sun, Jun 14, 2020 at 9:48 AM Navid Emamdoost
<navid.emamdoost@gmail.com> wrote:
...
> + if (ret < 0) {
> + pm_runtime_put(gpu->dev);
Please, in all your patches fix this to be _put_noidle(). We wouldn't
bear the flag day of fixing these parts again.
Yes, I know that *now* behaviour is the same, but calling put here is
slightly inconsistent.
...
> + pm_runtime_put(gpu->dev);
--
With Best Regards,
Andy Shevchenko
^ permalink raw reply [flat|nested] 6+ messages in thread* [PATCH v2] drm/etnaviv: fix ref count leak via pm_runtime_get_sync
2020-06-14 9:48 ` Andy Shevchenko
@ 2020-06-15 6:12 ` Navid Emamdoost
2020-06-17 9:53 ` Lucas Stach
2020-06-15 6:13 ` [PATCH] " Navid Emamdoost
1 sibling, 1 reply; 6+ messages in thread
From: Navid Emamdoost @ 2020-06-15 6:12 UTC (permalink / raw)
To: Lucas Stach, Russell King, Christian Gmeiner, David Airlie,
Daniel Vetter, etnaviv, dri-devel, linux-kernel
Cc: emamd001, wu000273, kjlu, mccamant, andy.shevchenko, Navid Emamdoost
in etnaviv_gpu_submit, etnaviv_gpu_recover_hang, etnaviv_gpu_debugfs,
and etnaviv_gpu_init the call to pm_runtime_get_sync increments the
counter even in case of failure, leading to incorrect ref count.
In case of failure, decrement the ref count before returning.
Signed-off-by: Navid Emamdoost <navid.emamdoost@gmail.com>
---
Changes in v2:
- replace pm_runtime_put with pm_runtime_put_noidle
---
drivers/gpu/drm/etnaviv/etnaviv_gpu.c | 14 ++++++++++----
1 file changed, 10 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gpu.c b/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
index a31eeff2b297..7c9f3f9ba123 100644
--- a/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
+++ b/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
@@ -722,7 +722,7 @@ int etnaviv_gpu_init(struct etnaviv_gpu *gpu)
ret = pm_runtime_get_sync(gpu->dev);
if (ret < 0) {
dev_err(gpu->dev, "Failed to enable GPU power domain\n");
- return ret;
+ goto pm_put;
}
etnaviv_hw_identify(gpu);
@@ -819,6 +819,7 @@ int etnaviv_gpu_init(struct etnaviv_gpu *gpu)
fail:
pm_runtime_mark_last_busy(gpu->dev);
+pm_put:
pm_runtime_put_autosuspend(gpu->dev);
return ret;
@@ -859,7 +860,7 @@ int etnaviv_gpu_debugfs(struct etnaviv_gpu *gpu, struct seq_file *m)
ret = pm_runtime_get_sync(gpu->dev);
if (ret < 0)
- return ret;
+ goto pm_put;
dma_lo = gpu_read(gpu, VIVS_FE_DMA_LOW);
dma_hi = gpu_read(gpu, VIVS_FE_DMA_HIGH);
@@ -1003,6 +1004,7 @@ int etnaviv_gpu_debugfs(struct etnaviv_gpu *gpu, struct seq_file *m)
ret = 0;
pm_runtime_mark_last_busy(gpu->dev);
+pm_put:
pm_runtime_put_autosuspend(gpu->dev);
return ret;
@@ -1016,7 +1018,7 @@ void etnaviv_gpu_recover_hang(struct etnaviv_gpu *gpu)
dev_err(gpu->dev, "recover hung GPU!\n");
if (pm_runtime_get_sync(gpu->dev) < 0)
- return;
+ goto pm_put;
mutex_lock(&gpu->lock);
@@ -1035,6 +1037,7 @@ void etnaviv_gpu_recover_hang(struct etnaviv_gpu *gpu)
mutex_unlock(&gpu->lock);
pm_runtime_mark_last_busy(gpu->dev);
+pm_put:
pm_runtime_put_autosuspend(gpu->dev);
}
@@ -1308,8 +1311,10 @@ struct dma_fence *etnaviv_gpu_submit(struct etnaviv_gem_submit *submit)
if (!submit->runtime_resumed) {
ret = pm_runtime_get_sync(gpu->dev);
- if (ret < 0)
+ if (ret < 0) {
+ pm_runtime_put_noidle(gpu->dev);
return NULL;
+ }
submit->runtime_resumed = true;
}
@@ -1326,6 +1331,7 @@ struct dma_fence *etnaviv_gpu_submit(struct etnaviv_gem_submit *submit)
ret = event_alloc(gpu, nr_events, event);
if (ret) {
DRM_ERROR("no free events\n");
+ pm_runtime_put_noidle(gpu->dev);
return NULL;
}
--
2.17.1
^ permalink raw reply related [flat|nested] 6+ messages in thread* Re: [PATCH v2] drm/etnaviv: fix ref count leak via pm_runtime_get_sync
2020-06-15 6:12 ` [PATCH v2] " Navid Emamdoost
@ 2020-06-17 9:53 ` Lucas Stach
2020-06-17 17:57 ` Navid Emamdoost
0 siblings, 1 reply; 6+ messages in thread
From: Lucas Stach @ 2020-06-17 9:53 UTC (permalink / raw)
To: Navid Emamdoost, Russell King, Christian Gmeiner, David Airlie,
Daniel Vetter, etnaviv, dri-devel, linux-kernel
Cc: kjlu, mccamant, andy.shevchenko, emamd001, wu000273
Hi Navid,
Am Montag, den 15.06.2020, 01:12 -0500 schrieb Navid Emamdoost:
> in etnaviv_gpu_submit, etnaviv_gpu_recover_hang, etnaviv_gpu_debugfs,
> and etnaviv_gpu_init the call to pm_runtime_get_sync increments the
> counter even in case of failure, leading to incorrect ref count.
> In case of failure, decrement the ref count before returning.
While that change is correct with the current API, may I ask the
question why the way this API works is considered reasonable? A API
call that fails, but still changes internal state and expects the
caller to clean up the mess it not really what I would consider fool-
proof API design. Is there a specific reason why it is done this way
and not handled internally?
Regards,
Lucas
> Signed-off-by: Navid Emamdoost <navid.emamdoost@gmail.com>
> ---
> Changes in v2:
> - replace pm_runtime_put with pm_runtime_put_noidle
> ---
> drivers/gpu/drm/etnaviv/etnaviv_gpu.c | 14 ++++++++++----
> 1 file changed, 10 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gpu.c b/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
> index a31eeff2b297..7c9f3f9ba123 100644
> --- a/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
> +++ b/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
> @@ -722,7 +722,7 @@ int etnaviv_gpu_init(struct etnaviv_gpu *gpu)
> ret = pm_runtime_get_sync(gpu->dev);
> if (ret < 0) {
> dev_err(gpu->dev, "Failed to enable GPU power domain\n");
> - return ret;
> + goto pm_put;
> }
>
> etnaviv_hw_identify(gpu);
> @@ -819,6 +819,7 @@ int etnaviv_gpu_init(struct etnaviv_gpu *gpu)
>
> fail:
> pm_runtime_mark_last_busy(gpu->dev);
> +pm_put:
> pm_runtime_put_autosuspend(gpu->dev);
>
> return ret;
> @@ -859,7 +860,7 @@ int etnaviv_gpu_debugfs(struct etnaviv_gpu *gpu, struct seq_file *m)
>
> ret = pm_runtime_get_sync(gpu->dev);
> if (ret < 0)
> - return ret;
> + goto pm_put;
>
> dma_lo = gpu_read(gpu, VIVS_FE_DMA_LOW);
> dma_hi = gpu_read(gpu, VIVS_FE_DMA_HIGH);
> @@ -1003,6 +1004,7 @@ int etnaviv_gpu_debugfs(struct etnaviv_gpu *gpu, struct seq_file *m)
> ret = 0;
>
> pm_runtime_mark_last_busy(gpu->dev);
> +pm_put:
> pm_runtime_put_autosuspend(gpu->dev);
>
> return ret;
> @@ -1016,7 +1018,7 @@ void etnaviv_gpu_recover_hang(struct etnaviv_gpu *gpu)
> dev_err(gpu->dev, "recover hung GPU!\n");
>
> if (pm_runtime_get_sync(gpu->dev) < 0)
> - return;
> + goto pm_put;
>
> mutex_lock(&gpu->lock);
>
> @@ -1035,6 +1037,7 @@ void etnaviv_gpu_recover_hang(struct etnaviv_gpu *gpu)
>
> mutex_unlock(&gpu->lock);
> pm_runtime_mark_last_busy(gpu->dev);
> +pm_put:
> pm_runtime_put_autosuspend(gpu->dev);
> }
>
> @@ -1308,8 +1311,10 @@ struct dma_fence *etnaviv_gpu_submit(struct etnaviv_gem_submit *submit)
>
> if (!submit->runtime_resumed) {
> ret = pm_runtime_get_sync(gpu->dev);
> - if (ret < 0)
> + if (ret < 0) {
> + pm_runtime_put_noidle(gpu->dev);
> return NULL;
> + }
> submit->runtime_resumed = true;
> }
>
> @@ -1326,6 +1331,7 @@ struct dma_fence *etnaviv_gpu_submit(struct etnaviv_gem_submit *submit)
> ret = event_alloc(gpu, nr_events, event);
> if (ret) {
> DRM_ERROR("no free events\n");
> + pm_runtime_put_noidle(gpu->dev);
> return NULL;
> }
>
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [PATCH v2] drm/etnaviv: fix ref count leak via pm_runtime_get_sync
2020-06-17 9:53 ` Lucas Stach
@ 2020-06-17 17:57 ` Navid Emamdoost
0 siblings, 0 replies; 6+ messages in thread
From: Navid Emamdoost @ 2020-06-17 17:57 UTC (permalink / raw)
To: Lucas Stach
Cc: Russell King, Christian Gmeiner, David Airlie, Daniel Vetter,
etnaviv, dri-devel, LKML, Kangjie Lu, mccamant, Andy Shevchenko,
Navid Emamdoost, Qiushi Wu
Hi Lucas,
On Wed, Jun 17, 2020 at 4:53 AM Lucas Stach <l.stach@pengutronix.de> wrote:
>
> Hi Navid,
>
> Am Montag, den 15.06.2020, 01:12 -0500 schrieb Navid Emamdoost:
> > in etnaviv_gpu_submit, etnaviv_gpu_recover_hang, etnaviv_gpu_debugfs,
> > and etnaviv_gpu_init the call to pm_runtime_get_sync increments the
> > counter even in case of failure, leading to incorrect ref count.
> > In case of failure, decrement the ref count before returning.
>
> While that change is correct with the current API, may I ask the
> question why the way this API works is considered reasonable? A API
> call that fails, but still changes internal state and expects the
> caller to clean up the mess it not really what I would consider fool-
> proof API design. Is there a specific reason why it is done this way
> and not handled internally?
I share the same concern with you on the way this API is working now.
To the best of my knowledge, there are ongoing discussions on this
issue:
https://lkml.org/lkml/2020/6/14/76
https://patchwork.ozlabs.org/project/linux-tegra/patch/20200520095148.10995-1-dinghao.liu@zju.edu.cn/
>
> Regards,
> Lucas
>
> > Signed-off-by: Navid Emamdoost <navid.emamdoost@gmail.com>
> > ---
> > Changes in v2:
> > - replace pm_runtime_put with pm_runtime_put_noidle
> > ---
> > drivers/gpu/drm/etnaviv/etnaviv_gpu.c | 14 ++++++++++----
> > 1 file changed, 10 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gpu.c b/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
> > index a31eeff2b297..7c9f3f9ba123 100644
> > --- a/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
> > +++ b/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
> > @@ -722,7 +722,7 @@ int etnaviv_gpu_init(struct etnaviv_gpu *gpu)
> > ret = pm_runtime_get_sync(gpu->dev);
> > if (ret < 0) {
> > dev_err(gpu->dev, "Failed to enable GPU power domain\n");
> > - return ret;
> > + goto pm_put;
> > }
> >
> > etnaviv_hw_identify(gpu);
> > @@ -819,6 +819,7 @@ int etnaviv_gpu_init(struct etnaviv_gpu *gpu)
> >
> > fail:
> > pm_runtime_mark_last_busy(gpu->dev);
> > +pm_put:
> > pm_runtime_put_autosuspend(gpu->dev);
> >
> > return ret;
> > @@ -859,7 +860,7 @@ int etnaviv_gpu_debugfs(struct etnaviv_gpu *gpu, struct seq_file *m)
> >
> > ret = pm_runtime_get_sync(gpu->dev);
> > if (ret < 0)
> > - return ret;
> > + goto pm_put;
> >
> > dma_lo = gpu_read(gpu, VIVS_FE_DMA_LOW);
> > dma_hi = gpu_read(gpu, VIVS_FE_DMA_HIGH);
> > @@ -1003,6 +1004,7 @@ int etnaviv_gpu_debugfs(struct etnaviv_gpu *gpu, struct seq_file *m)
> > ret = 0;
> >
> > pm_runtime_mark_last_busy(gpu->dev);
> > +pm_put:
> > pm_runtime_put_autosuspend(gpu->dev);
> >
> > return ret;
> > @@ -1016,7 +1018,7 @@ void etnaviv_gpu_recover_hang(struct etnaviv_gpu *gpu)
> > dev_err(gpu->dev, "recover hung GPU!\n");
> >
> > if (pm_runtime_get_sync(gpu->dev) < 0)
> > - return;
> > + goto pm_put;
> >
> > mutex_lock(&gpu->lock);
> >
> > @@ -1035,6 +1037,7 @@ void etnaviv_gpu_recover_hang(struct etnaviv_gpu *gpu)
> >
> > mutex_unlock(&gpu->lock);
> > pm_runtime_mark_last_busy(gpu->dev);
> > +pm_put:
> > pm_runtime_put_autosuspend(gpu->dev);
> > }
> >
> > @@ -1308,8 +1311,10 @@ struct dma_fence *etnaviv_gpu_submit(struct etnaviv_gem_submit *submit)
> >
> > if (!submit->runtime_resumed) {
> > ret = pm_runtime_get_sync(gpu->dev);
> > - if (ret < 0)
> > + if (ret < 0) {
> > + pm_runtime_put_noidle(gpu->dev);
> > return NULL;
> > + }
> > submit->runtime_resumed = true;
> > }
> >
> > @@ -1326,6 +1331,7 @@ struct dma_fence *etnaviv_gpu_submit(struct etnaviv_gem_submit *submit)
> > ret = event_alloc(gpu, nr_events, event);
> > if (ret) {
> > DRM_ERROR("no free events\n");
> > + pm_runtime_put_noidle(gpu->dev);
> > return NULL;
> > }
> >
>
--
Navid.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] drm/etnaviv: fix ref count leak via pm_runtime_get_sync
2020-06-14 9:48 ` Andy Shevchenko
2020-06-15 6:12 ` [PATCH v2] " Navid Emamdoost
@ 2020-06-15 6:13 ` Navid Emamdoost
1 sibling, 0 replies; 6+ messages in thread
From: Navid Emamdoost @ 2020-06-15 6:13 UTC (permalink / raw)
To: Andy Shevchenko
Cc: Lucas Stach, Russell King, Christian Gmeiner, David Airlie,
Daniel Vetter, etnaviv, dri-devel, Linux Kernel Mailing List,
Navid Emamdoost, Qiushi Wu, Kangjie Lu, Stephen McCamant
On Sun, Jun 14, 2020 at 4:48 AM Andy Shevchenko
<andy.shevchenko@gmail.com> wrote:
>
> On Sun, Jun 14, 2020 at 9:48 AM Navid Emamdoost
> <navid.emamdoost@gmail.com> wrote:
>
> ...
>
> > + if (ret < 0) {
>
> > + pm_runtime_put(gpu->dev);
>
> Please, in all your patches fix this to be _put_noidle(). We wouldn't
> bear the flag day of fixing these parts again.
> Yes, I know that *now* behaviour is the same, but calling put here is
> slightly inconsistent.
v2 is sent.
>
> ...
>
> > + pm_runtime_put(gpu->dev);
>
> --
> With Best Regards,
> Andy Shevchenko
--
Navid.
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2020-06-17 17:57 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-06-14 6:46 [PATCH] drm/etnaviv: fix ref count leak via pm_runtime_get_sync Navid Emamdoost
2020-06-14 9:48 ` Andy Shevchenko
2020-06-15 6:12 ` [PATCH v2] " Navid Emamdoost
2020-06-17 9:53 ` Lucas Stach
2020-06-17 17:57 ` Navid Emamdoost
2020-06-15 6:13 ` [PATCH] " Navid Emamdoost
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).