diff mbox

dma-fence: Make dma_fence_add_callback() fail if signaled with error

Message ID 20180509201449.27452-1-ezequiel@collabora.com (mailing list archive)
State New, archived
Headers show

Commit Message

Ezequiel Garcia May 9, 2018, 8:14 p.m. UTC
Change how dma_fence_add_callback() behaves, when the fence
has error-signaled by the time it is being add. After this commit,
dma_fence_add_callback() returns the fence error, if it
has error-signaled before dma_fence_add_callback() is called.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
---
 drivers/dma-buf/dma-fence.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

Comments

Ezequiel Garcia May 10, 2018, 12:51 p.m. UTC | #1
On Wed, 2018-05-09 at 19:42 -0300, Gustavo Padovan wrote:
> Hi Ezequiel,
> 
> On Wed, 2018-05-09 at 17:14 -0300, Ezequiel Garcia wrote:
> > Change how dma_fence_add_callback() behaves, when the fence
> > has error-signaled by the time it is being add. After this commit,
> > dma_fence_add_callback() returns the fence error, if it
> > has error-signaled before dma_fence_add_callback() is called.
> > 
> > Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
> > ---
> >  drivers/dma-buf/dma-fence.c | 10 ++++++----
> >  1 file changed, 6 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/dma-buf/dma-fence.c b/drivers/dma-buf/dma-
> > fence.c
> > index 4edb9fd3cf47..298b440c5b68 100644
> > --- a/drivers/dma-buf/dma-fence.c
> > +++ b/drivers/dma-buf/dma-fence.c
> > @@ -226,7 +226,8 @@ EXPORT_SYMBOL(dma_fence_enable_sw_signaling);
> >   *
> >   * Note that the callback can be called from an atomic context.  If
> >   * fence is already signaled, this function will return -ENOENT (and
> > - * *not* call the callback)
> > + * *not* call the callback). If the fence is error-signaled, this
> > + * function returns the fence error.
> >   *
> >   * Add a software callback to the fence. Same restrictions apply to
> >   * refcount as it does to dma_fence_wait, however the caller doesn't
> > need to
> > @@ -235,8 +236,8 @@ EXPORT_SYMBOL(dma_fence_enable_sw_signaling);
> >   * after it signals with dma_fence_signal. The callback itself can
> > be called
> >   * from irq context.
> >   *
> > - * Returns 0 in case of success, -ENOENT if the fence is already
> > signaled
> > - * and -EINVAL in case of error.
> > + * Returns 0 in case of success, -ENOENT (or the error value) if the
> > fence is
> > + * already signaled and -EINVAL in case of error.
> >   */
> >  int dma_fence_add_callback(struct dma_fence *fence, struct
> > dma_fence_cb *cb,
> >  			   dma_fence_func_t func)
> > @@ -250,7 +251,8 @@ int dma_fence_add_callback(struct dma_fence
> > *fence, struct dma_fence_cb *cb,
> >  
> >  	if (test_bit(DMA_FENCE_FLAG_SIGNALED_BIT, &fence->flags)) {
> >  		INIT_LIST_HEAD(&cb->node);
> > -		return -ENOENT;
> > +		ret = (fence->error < 0) ? fence->error : -ENOENT;
> > +		return ret;
> >  	}
> 
> It looks good to me, but I'd first go check all place we call it in the
> kernel because I have some memory of callers relying on the -ENOENT
> return code for some decision. I might be wrong though.
> 
> 

I checked all users before submitting this patch.

git grep " = dma_fence_add_callback"
drivers/gpu/drm/i915/i915_sw_fence.c:   ret = dma_fence_add_callback(dma, &cb->base, func);
drivers/gpu/drm/scheduler/gpu_scheduler.c:                      r = dma_fence_add_callback(fence, &s_fence->cb,
drivers/gpu/drm/scheduler/gpu_scheduler.c:                      r = dma_fence_add_callback(fence, &s_fence->cb,

And from what I could see, all of them handle the error
properly.

Thanks,
Eze
Chris Wilson May 11, 2018, 7:23 a.m. UTC | #2
Quoting Ezequiel Garcia (2018-05-10 13:51:56)
> On Wed, 2018-05-09 at 19:42 -0300, Gustavo Padovan wrote:
> > Hi Ezequiel,
> > 
> > On Wed, 2018-05-09 at 17:14 -0300, Ezequiel Garcia wrote:
> > > Change how dma_fence_add_callback() behaves, when the fence
> > > has error-signaled by the time it is being add. After this commit,
> > > dma_fence_add_callback() returns the fence error, if it
> > > has error-signaled before dma_fence_add_callback() is called.
> > > 
> > > Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
> > > ---
> > >  drivers/dma-buf/dma-fence.c | 10 ++++++----
> > >  1 file changed, 6 insertions(+), 4 deletions(-)
> > > 
> > > diff --git a/drivers/dma-buf/dma-fence.c b/drivers/dma-buf/dma-
> > > fence.c
> > > index 4edb9fd3cf47..298b440c5b68 100644
> > > --- a/drivers/dma-buf/dma-fence.c
> > > +++ b/drivers/dma-buf/dma-fence.c
> > > @@ -226,7 +226,8 @@ EXPORT_SYMBOL(dma_fence_enable_sw_signaling);
> > >   *
> > >   * Note that the callback can be called from an atomic context.  If
> > >   * fence is already signaled, this function will return -ENOENT (and
> > > - * *not* call the callback)
> > > + * *not* call the callback). If the fence is error-signaled, this
> > > + * function returns the fence error.
> > >   *
> > >   * Add a software callback to the fence. Same restrictions apply to
> > >   * refcount as it does to dma_fence_wait, however the caller doesn't
> > > need to
> > > @@ -235,8 +236,8 @@ EXPORT_SYMBOL(dma_fence_enable_sw_signaling);
> > >   * after it signals with dma_fence_signal. The callback itself can
> > > be called
> > >   * from irq context.
> > >   *
> > > - * Returns 0 in case of success, -ENOENT if the fence is already
> > > signaled
> > > - * and -EINVAL in case of error.
> > > + * Returns 0 in case of success, -ENOENT (or the error value) if the
> > > fence is
> > > + * already signaled and -EINVAL in case of error.
> > >   */
> > >  int dma_fence_add_callback(struct dma_fence *fence, struct
> > > dma_fence_cb *cb,
> > >                        dma_fence_func_t func)
> > > @@ -250,7 +251,8 @@ int dma_fence_add_callback(struct dma_fence
> > > *fence, struct dma_fence_cb *cb,
> > >  
> > >     if (test_bit(DMA_FENCE_FLAG_SIGNALED_BIT, &fence->flags)) {
> > >             INIT_LIST_HEAD(&cb->node);
> > > -           return -ENOENT;
> > > +           ret = (fence->error < 0) ? fence->error : -ENOENT;
> > > +           return ret;
> > >     }
> > 
> > It looks good to me, but I'd first go check all place we call it in the
> > kernel because I have some memory of callers relying on the -ENOENT
> > return code for some decision. I might be wrong though.
> > 
> > 
> 
> I checked all users before submitting this patch.
> 
> git grep " = dma_fence_add_callback"
> drivers/gpu/drm/i915/i915_sw_fence.c:   ret = dma_fence_add_callback(dma, &cb->base, func);
> 
> And from what I could see, all of them handle the error
> properly.

Err, no. That error then is propagated back to userspace, and that is
not part of our ABI...
-Chris
Chris Wilson May 11, 2018, 7:27 a.m. UTC | #3
Quoting Ezequiel Garcia (2018-05-09 21:14:49)
> Change how dma_fence_add_callback() behaves, when the fence
> has error-signaled by the time it is being add. After this commit,
> dma_fence_add_callback() returns the fence error, if it
> has error-signaled before dma_fence_add_callback() is called.

Why? What problem are you trying to solve? fence->error does not imply
that the fence has yet been signaled, and the caller wants a callback
when it is signaled.
-Chris
Daniel Vetter May 14, 2018, 4:48 p.m. UTC | #4
On Fri, May 11, 2018 at 08:27:41AM +0100, Chris Wilson wrote:
> Quoting Ezequiel Garcia (2018-05-09 21:14:49)
> > Change how dma_fence_add_callback() behaves, when the fence
> > has error-signaled by the time it is being add. After this commit,
> > dma_fence_add_callback() returns the fence error, if it
> > has error-signaled before dma_fence_add_callback() is called.
> 
> Why? What problem are you trying to solve? fence->error does not imply
> that the fence has yet been signaled, and the caller wants a callback
> when it is signaled.

On top this is incosistent, e.g. we don't do the same for any of the other
dma_fence interfaces. Plus there's the issue that you might alias errno
values with fence errno values.

I think keeping the error codes from the functions you're calling distinct
from the error code of the fence itself makes a lot of sense. The first
tells you whether your request worked out (or why not), the second tells
you whether the asynchronous dma operation (gpu rendering, page flip,
whatever) that the dma_fence represents worked out (or why not). That's 2
distinct things imo.

Might be good to show us the driver code that needs this behaviour so we
can discuss how to best handle your use-case.

Cheers, Daniel
Ezequiel Garcia May 14, 2018, 9:28 p.m. UTC | #5
On Mon, 2018-05-14 at 18:48 +0200, Daniel Vetter wrote:
> On Fri, May 11, 2018 at 08:27:41AM +0100, Chris Wilson wrote:
> > Quoting Ezequiel Garcia (2018-05-09 21:14:49)
> > > Change how dma_fence_add_callback() behaves, when the fence
> > > has error-signaled by the time it is being add. After this commit,
> > > dma_fence_add_callback() returns the fence error, if it
> > > has error-signaled before dma_fence_add_callback() is called.
> > 
> > Why? What problem are you trying to solve? fence->error does not imply
> > that the fence has yet been signaled, and the caller wants a callback
> > when it is signaled.
> 
> On top this is incosistent, e.g. we don't do the same for any of the other
> dma_fence interfaces. Plus there's the issue that you might alias errno
> values with fence errno values.
> 

Right.

> I think keeping the error codes from the functions you're calling distinct
> from the error code of the fence itself makes a lot of sense. The first
> tells you whether your request worked out (or why not), the second tells
> you whether the asynchronous dma operation (gpu rendering, page flip,
> whatever) that the dma_fence represents worked out (or why not). That's 2
> distinct things imo.
> 
> Might be good to show us the driver code that needs this behaviour so we
> can discuss how to best handle your use-case.
> 

This change arose while discussing the in-fences support for video4linux.
Here's the patch that calls dma_fence_add_callback https://lkml.org/lkml/2018/5/4/766.

The code snippet currently looks something like this:

	if (vb->in_fence) {
		ret = dma_fence_add_callback(vb->in_fence, &vb->fence_cb,
				
	     vb2_qbuf_fence_cb);
		/* is the fence signaled? */
		if (ret == -ENOENT) {
	
		dma_fence_put(vb->in_fence);
			vb->in_fence = NULL;
		} else if (ret)
{
			goto unlock;
		}
	}

In this use case, if the callback is added successfully,
the video4linux core defers the activation of the buffer
until the fence signals.

If the fence is signaled (currently disregarding of errors)
then the buffer is assumed to be ready to be activated,
and so it gets queued for hardware usage.

Giving some more thought to this, I'm not so sure what is
the right action if a fence signaled with error. In this case,
it appears to me that we shouldn't be using this buffer
if its in-fence is in error, but perhaps I'm missing
something.

Thanks!
Eze
Chris Wilson May 15, 2018, 12:16 p.m. UTC | #6
Quoting Ezequiel Garcia (2018-05-14 22:28:31)
> On Mon, 2018-05-14 at 18:48 +0200, Daniel Vetter wrote:
> > On Fri, May 11, 2018 at 08:27:41AM +0100, Chris Wilson wrote:
> > > Quoting Ezequiel Garcia (2018-05-09 21:14:49)
> > > > Change how dma_fence_add_callback() behaves, when the fence
> > > > has error-signaled by the time it is being add. After this commit,
> > > > dma_fence_add_callback() returns the fence error, if it
> > > > has error-signaled before dma_fence_add_callback() is called.
> > > 
> > > Why? What problem are you trying to solve? fence->error does not imply
> > > that the fence has yet been signaled, and the caller wants a callback
> > > when it is signaled.
> > 
> > On top this is incosistent, e.g. we don't do the same for any of the other
> > dma_fence interfaces. Plus there's the issue that you might alias errno
> > values with fence errno values.
> > 
> 
> Right.
> 
> > I think keeping the error codes from the functions you're calling distinct
> > from the error code of the fence itself makes a lot of sense. The first
> > tells you whether your request worked out (or why not), the second tells
> > you whether the asynchronous dma operation (gpu rendering, page flip,
> > whatever) that the dma_fence represents worked out (or why not). That's 2
> > distinct things imo.
> > 
> > Might be good to show us the driver code that needs this behaviour so we
> > can discuss how to best handle your use-case.
> > 
> 
> This change arose while discussing the in-fences support for video4linux.
> Here's the patch that calls dma_fence_add_callback https://lkml.org/lkml/2018/5/4/766.
> 
> The code snippet currently looks something like this:
> 
>         if (vb->in_fence) {
>                 ret = dma_fence_add_callback(vb->in_fence, &vb->fence_cb,
>                                 
>              vb2_qbuf_fence_cb);
>                 /* is the fence signaled? */
>                 if (ret == -ENOENT) {
>         
>                 dma_fence_put(vb->in_fence);
>                         vb->in_fence = NULL;
>                 } else if (ret)
> {
>                         goto unlock;
>                 }
>         }
> 
> In this use case, if the callback is added successfully,
> the video4linux core defers the activation of the buffer
> until the fence signals.
> 
> If the fence is signaled (currently disregarding of errors)
> then the buffer is assumed to be ready to be activated,
> and so it gets queued for hardware usage.
> 
> Giving some more thought to this, I'm not so sure what is
> the right action if a fence signaled with error. In this case,
> it appears to me that we shouldn't be using this buffer
> if its in-fence is in error, but perhaps I'm missing
> something.

What I have in mind for async errors is to skip the operation and
propagate the error onto the next fence. Mostly because those async
errors may include fatal errors such as unable to pin the backing
storage for the operation, but even "trivial" errors such as an early
operation failing means that this request is then subject to garbage-in,
garbage-out. However, for trivial errors I would just propagate the
error status (so the caller knows something went wrong if they care, but
in all likelihood no one will notice) and continue on with the glitchy
operation.
-Chris
Daniel Vetter May 16, 2018, 9:42 a.m. UTC | #7
On Tue, May 15, 2018 at 01:16:30PM +0100, Chris Wilson wrote:
> Quoting Ezequiel Garcia (2018-05-14 22:28:31)
> > On Mon, 2018-05-14 at 18:48 +0200, Daniel Vetter wrote:
> > > On Fri, May 11, 2018 at 08:27:41AM +0100, Chris Wilson wrote:
> > > > Quoting Ezequiel Garcia (2018-05-09 21:14:49)
> > > > > Change how dma_fence_add_callback() behaves, when the fence
> > > > > has error-signaled by the time it is being add. After this commit,
> > > > > dma_fence_add_callback() returns the fence error, if it
> > > > > has error-signaled before dma_fence_add_callback() is called.
> > > > 
> > > > Why? What problem are you trying to solve? fence->error does not imply
> > > > that the fence has yet been signaled, and the caller wants a callback
> > > > when it is signaled.
> > > 
> > > On top this is incosistent, e.g. we don't do the same for any of the other
> > > dma_fence interfaces. Plus there's the issue that you might alias errno
> > > values with fence errno values.
> > > 
> > 
> > Right.
> > 
> > > I think keeping the error codes from the functions you're calling distinct
> > > from the error code of the fence itself makes a lot of sense. The first
> > > tells you whether your request worked out (or why not), the second tells
> > > you whether the asynchronous dma operation (gpu rendering, page flip,
> > > whatever) that the dma_fence represents worked out (or why not). That's 2
> > > distinct things imo.
> > > 
> > > Might be good to show us the driver code that needs this behaviour so we
> > > can discuss how to best handle your use-case.
> > > 
> > 
> > This change arose while discussing the in-fences support for video4linux.
> > Here's the patch that calls dma_fence_add_callback https://lkml.org/lkml/2018/5/4/766.
> > 
> > The code snippet currently looks something like this:
> > 
> >         if (vb->in_fence) {
> >                 ret = dma_fence_add_callback(vb->in_fence, &vb->fence_cb,
> >                                 
> >              vb2_qbuf_fence_cb);
> >                 /* is the fence signaled? */
> >                 if (ret == -ENOENT) {
> >         
> >                 dma_fence_put(vb->in_fence);
> >                         vb->in_fence = NULL;
> >                 } else if (ret)
> > {
> >                         goto unlock;
> >                 }
> >         }
> > 
> > In this use case, if the callback is added successfully,
> > the video4linux core defers the activation of the buffer
> > until the fence signals.
> > 
> > If the fence is signaled (currently disregarding of errors)
> > then the buffer is assumed to be ready to be activated,
> > and so it gets queued for hardware usage.
> > 
> > Giving some more thought to this, I'm not so sure what is
> > the right action if a fence signaled with error. In this case,
> > it appears to me that we shouldn't be using this buffer
> > if its in-fence is in error, but perhaps I'm missing
> > something.
> 
> What I have in mind for async errors is to skip the operation and
> propagate the error onto the next fence. Mostly because those async
> errors may include fatal errors such as unable to pin the backing
> storage for the operation, but even "trivial" errors such as an early
> operation failing means that this request is then subject to garbage-in,
> garbage-out. However, for trivial errors I would just propagate the
> error status (so the caller knows something went wrong if they care, but
> in all likelihood no one will notice) and continue on with the glitchy
> operation.

In general, there's not really any hard rule about propagating fence
errors across devices. It's mostly just used by drivers internally to keep
track of failed stuff (gpu hangs or anything else async like Chris
describes here).

For v4l I'm not sure you want to care much about this, since right now the
main use of fence errors is gpu hang recovery (whether it's the driver or
hw that's hung doesn't matter here).
-Daniel
Lucas Stach May 16, 2018, 10:26 a.m. UTC | #8
Am Mittwoch, den 16.05.2018, 11:42 +0200 schrieb Daniel Vetter:
> On Tue, May 15, 2018 at 01:16:30PM +0100, Chris Wilson wrote:
> > Quoting Ezequiel Garcia (2018-05-14 22:28:31)
> > > On Mon, 2018-05-14 at 18:48 +0200, Daniel Vetter wrote:
> > > > On Fri, May 11, 2018 at 08:27:41AM +0100, Chris Wilson wrote:
> > > > > Quoting Ezequiel Garcia (2018-05-09 21:14:49)
> > > > > > Change how dma_fence_add_callback() behaves, when the fence
> > > > > > has error-signaled by the time it is being add. After this commit,
> > > > > > dma_fence_add_callback() returns the fence error, if it
> > > > > > has error-signaled before dma_fence_add_callback() is called.
> > > > > 
> > > > > Why? What problem are you trying to solve? fence->error does not imply
> > > > > that the fence has yet been signaled, and the caller wants a callback
> > > > > when it is signaled.
> > > > 
> > > > On top this is incosistent, e.g. we don't do the same for any of the other
> > > > dma_fence interfaces. Plus there's the issue that you might alias errno
> > > > values with fence errno values.
> > > > 
> > > 
> > > Right.
> > > 
> > > > I think keeping the error codes from the functions you're calling distinct
> > > > from the error code of the fence itself makes a lot of sense. The first
> > > > tells you whether your request worked out (or why not), the second tells
> > > > you whether the asynchronous dma operation (gpu rendering, page flip,
> > > > whatever) that the dma_fence represents worked out (or why not). That's 2
> > > > distinct things imo.
> > > > 
> > > > Might be good to show us the driver code that needs this behaviour so we
> > > > can discuss how to best handle your use-case.
> > > > 
> > > 
> > > This change arose while discussing the in-fences support for video4linux.
> > > Here's the patch that calls dma_fence_add_callback https://lkml.org/lkml/2018/5/4/766.
> > > 
> > > The code snippet currently looks something like this:
> > > 
> > >         if (vb->in_fence) {
> > >                 ret = dma_fence_add_callback(vb->in_fence, &vb->fence_cb,
> > >                                 
> > >              vb2_qbuf_fence_cb);
> > >                 /* is the fence signaled? */
> > >                 if (ret == -ENOENT) {
> > >         
> > >                 dma_fence_put(vb->in_fence);
> > >                         vb->in_fence = NULL;
> > >                 } else if (ret)
> > > {
> > >                         goto unlock;
> > >                 }
> > >         }
> > > 
> > > In this use case, if the callback is added successfully,
> > > the video4linux core defers the activation of the buffer
> > > until the fence signals.
> > > 
> > > If the fence is signaled (currently disregarding of errors)
> > > then the buffer is assumed to be ready to be activated,
> > > and so it gets queued for hardware usage.
> > > 
> > > Giving some more thought to this, I'm not so sure what is
> > > the right action if a fence signaled with error. In this case,
> > > it appears to me that we shouldn't be using this buffer
> > > if its in-fence is in error, but perhaps I'm missing
> > > something.
> > 
> > What I have in mind for async errors is to skip the operation and
> > propagate the error onto the next fence. Mostly because those async
> > errors may include fatal errors such as unable to pin the backing
> > storage for the operation, but even "trivial" errors such as an early
> > operation failing means that this request is then subject to garbage-in,
> > garbage-out. However, for trivial errors I would just propagate the
> > error status (so the caller knows something went wrong if they care, but
> > in all likelihood no one will notice) and continue on with the glitchy
> > operation.
> 
> In general, there's not really any hard rule about propagating fence
> errors across devices. It's mostly just used by drivers internally to keep
> track of failed stuff (gpu hangs or anything else async like Chris
> describes here).
> 
> For v4l I'm not sure you want to care much about this, since right now the
> main use of fence errors is gpu hang recovery (whether it's the driver or
> hw that's hung doesn't matter here).

Yes, my understanding is that fence signal and errors are two distinct
things that should not be conflated like it is done in this patch.

In my understanding signaling a fence means the HW or SW component
which added the fence is done with the buffer and will not touch it
anymore. In case of an unrecoverable error the fence will be signaled
with error status set, so we correctly reflect the buffer status as
being free to be used by whoever is waiting for it, but may contain
garbage.

If a fence waiter cares about the buffer content and may wish to skip
its operation if the fence is signaled with an error it should do it by
explicitly checking the fence error status, instead of making this
implicit behavior.

Regards,
Lucas
Ezequiel Garcia May 16, 2018, 6:27 p.m. UTC | #9
On Wed, 2018-05-16 at 12:26 +0200, Lucas Stach wrote:
> Am Mittwoch, den 16.05.2018, 11:42 +0200 schrieb Daniel Vetter:
> > On Tue, May 15, 2018 at 01:16:30PM +0100, Chris Wilson wrote:
> > > Quoting Ezequiel Garcia (2018-05-14 22:28:31)
> > > > On Mon, 2018-05-14 at 18:48 +0200, Daniel Vetter wrote:
> > > > > On Fri, May 11, 2018 at 08:27:41AM +0100, Chris Wilson wrote:
> > > > > > Quoting Ezequiel Garcia (2018-05-09 21:14:49)
> > > > > > > Change how dma_fence_add_callback() behaves, when the fence
> > > > > > > has error-signaled by the time it is being add. After this commit,
> > > > > > > dma_fence_add_callback() returns the fence error, if it
> > > > > > > has error-signaled before dma_fence_add_callback() is called.
> > > > > > 
> > > > > > Why? What problem are you trying to solve? fence->error does not imply
> > > > > > that the fence has yet been signaled, and the caller wants a callback
> > > > > > when it is signaled.
> > > > > 
> > > > > On top this is incosistent, e.g. we don't do the same for any of the other
> > > > > dma_fence interfaces. Plus there's the issue that you might alias errno
> > > > > values with fence errno values.
> > > > > 
> > > > 
> > > > Right.
> > > > 
> > > > > I think keeping the error codes from the functions you're calling distinct
> > > > > from the error code of the fence itself makes a lot of sense. The first
> > > > > tells you whether your request worked out (or why not), the second tells
> > > > > you whether the asynchronous dma operation (gpu rendering, page flip,
> > > > > whatever) that the dma_fence represents worked out (or why not). That's 2
> > > > > distinct things imo.
> > > > > 
> > > > > Might be good to show us the driver code that needs this behaviour so we
> > > > > can discuss how to best handle your use-case.
> > > > > 
> > > > 
> > > > This change arose while discussing the in-fences support for video4linux.
> > > > Here's the patch that calls dma_fence_add_callback https://lkml.org/lkml/2018/5/4/766.
> > > > 
> > > > The code snippet currently looks something like this:
> > > > 
> > > >         if (vb->in_fence) {
> > > >                 ret = dma_fence_add_callback(vb->in_fence, &vb->fence_cb,
> > > >                                 
> > > >              vb2_qbuf_fence_cb);
> > > >                 /* is the fence signaled? */
> > > >                 if (ret == -ENOENT) {
> > > >         
> > > >                 dma_fence_put(vb->in_fence);
> > > >                         vb->in_fence = NULL;
> > > >                 } else if (ret)
> > > > {
> > > >                         goto unlock;
> > > >                 }
> > > >         }
> > > > 
> > > > In this use case, if the callback is added successfully,
> > > > the video4linux core defers the activation of the buffer
> > > > until the fence signals.
> > > > 
> > > > If the fence is signaled (currently disregarding of errors)
> > > > then the buffer is assumed to be ready to be activated,
> > > > and so it gets queued for hardware usage.
> > > > 
> > > > Giving some more thought to this, I'm not so sure what is
> > > > the right action if a fence signaled with error. In this case,
> > > > it appears to me that we shouldn't be using this buffer
> > > > if its in-fence is in error, but perhaps I'm missing
> > > > something.
> > > 
> > > What I have in mind for async errors is to skip the operation and
> > > propagate the error onto the next fence. Mostly because those async
> > > errors may include fatal errors such as unable to pin the backing
> > > storage for the operation, but even "trivial" errors such as an early
> > > operation failing means that this request is then subject to garbage-in,
> > > garbage-out. However, for trivial errors I would just propagate the
> > > error status (so the caller knows something went wrong if they care, but
> > > in all likelihood no one will notice) and continue on with the glitchy
> > > operation.
> > 
> > In general, there's not really any hard rule about propagating fence
> > errors across devices. It's mostly just used by drivers internally to keep
> > track of failed stuff (gpu hangs or anything else async like Chris
> > describes here).
> > 
> > For v4l I'm not sure you want to care much about this, since right now the
> > main use of fence errors is gpu hang recovery (whether it's the driver or
> > hw that's hung doesn't matter here).
> 
> Yes, my understanding is that fence signal and errors are two distinct
> things that should not be conflated like it is done in this patch.
> 
> In my understanding signaling a fence means the HW or SW component
> which added the fence is done with the buffer and will not touch it
> anymore. In case of an unrecoverable error the fence will be signaled
> with error status set, so we correctly reflect the buffer status as
> being free to be used by whoever is waiting for it, but may contain
> garbage.
> 
> If a fence waiter cares about the buffer content and may wish to skip
> its operation if the fence is signaled with an error it should do it by
> explicitly checking the fence error status, instead of making this
> implicit behavior.
> 

Yes, that sounds right.

Thanks for the help,
Eze
diff mbox

Patch

diff --git a/drivers/dma-buf/dma-fence.c b/drivers/dma-buf/dma-fence.c
index 4edb9fd3cf47..298b440c5b68 100644
--- a/drivers/dma-buf/dma-fence.c
+++ b/drivers/dma-buf/dma-fence.c
@@ -226,7 +226,8 @@  EXPORT_SYMBOL(dma_fence_enable_sw_signaling);
  *
  * Note that the callback can be called from an atomic context.  If
  * fence is already signaled, this function will return -ENOENT (and
- * *not* call the callback)
+ * *not* call the callback). If the fence is error-signaled, this
+ * function returns the fence error.
  *
  * Add a software callback to the fence. Same restrictions apply to
  * refcount as it does to dma_fence_wait, however the caller doesn't need to
@@ -235,8 +236,8 @@  EXPORT_SYMBOL(dma_fence_enable_sw_signaling);
  * after it signals with dma_fence_signal. The callback itself can be called
  * from irq context.
  *
- * Returns 0 in case of success, -ENOENT if the fence is already signaled
- * and -EINVAL in case of error.
+ * Returns 0 in case of success, -ENOENT (or the error value) if the fence is
+ * already signaled and -EINVAL in case of error.
  */
 int dma_fence_add_callback(struct dma_fence *fence, struct dma_fence_cb *cb,
 			   dma_fence_func_t func)
@@ -250,7 +251,8 @@  int dma_fence_add_callback(struct dma_fence *fence, struct dma_fence_cb *cb,
 
 	if (test_bit(DMA_FENCE_FLAG_SIGNALED_BIT, &fence->flags)) {
 		INIT_LIST_HEAD(&cb->node);
-		return -ENOENT;
+		ret = (fence->error < 0) ? fence->error : -ENOENT;
+		return ret;
 	}
 
 	spin_lock_irqsave(fence->lock, flags);