drm/i915/execlists: Assert tasklet is locked for process_csb()
diff mbox series

Message ID 20191013193115.16844-1-chris@chris-wilson.co.uk
State New
Headers show
Series
  • drm/i915/execlists: Assert tasklet is locked for process_csb()
Related show

Commit Message

Chris Wilson Oct. 13, 2019, 7:31 p.m. UTC
We rely on only the tasklet being allowed to call into process_csb(), so
assert that is locked when we do. As the tasklet uses a simple bitlock,
there is no strong lockdep checking so we must make do with a plain
assertion that the tasklet is running and assume that we are the
tasklet!

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
---
 drivers/gpu/drm/i915/gt/intel_lrc.c | 1 +
 drivers/gpu/drm/i915/i915_gem.h     | 5 +++++
 2 files changed, 6 insertions(+)

Comments

Tvrtko Ursulin Oct. 14, 2019, 9:25 a.m. UTC | #1
On 13/10/2019 20:31, Chris Wilson wrote:
> We rely on only the tasklet being allowed to call into process_csb(), so
> assert that is locked when we do. As the tasklet uses a simple bitlock,
> there is no strong lockdep checking so we must make do with a plain
> assertion that the tasklet is running and assume that we are the
> tasklet!
> 
> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> ---
>   drivers/gpu/drm/i915/gt/intel_lrc.c | 1 +
>   drivers/gpu/drm/i915/i915_gem.h     | 5 +++++
>   2 files changed, 6 insertions(+)
> 
> diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c
> index 8be9e69d5718..ab20433182d1 100644
> --- a/drivers/gpu/drm/i915/gt/intel_lrc.c
> +++ b/drivers/gpu/drm/i915/gt/intel_lrc.c
> @@ -1984,6 +1984,7 @@ static void process_csb(struct intel_engine_cs *engine)
>   	u8 head, tail;
>   
>   	GEM_BUG_ON(USES_GUC_SUBMISSION(engine->i915));
> +	GEM_BUG_ON(!tasklet_is_locked(&execlists->tasklet));

I see some reset paths calling process_csb which looks like a problem 
for this assert.

Regards,

Tvrtko

>   
>   	/*
>   	 * Note that csb_write, csb_status may be either in HWSP or mmio.
> diff --git a/drivers/gpu/drm/i915/i915_gem.h b/drivers/gpu/drm/i915/i915_gem.h
> index db20d2b0842b..f6f9675848b8 100644
> --- a/drivers/gpu/drm/i915/i915_gem.h
> +++ b/drivers/gpu/drm/i915/i915_gem.h
> @@ -86,6 +86,11 @@ static inline void tasklet_lock(struct tasklet_struct *t)
>   		cpu_relax();
>   }
>   
> +static inline bool tasklet_is_locked(const struct tasklet_struct *t)
> +{
> +	return test_bit(TASKLET_STATE_RUN, &t->state);
> +}
> +
>   static inline void __tasklet_disable_sync_once(struct tasklet_struct *t)
>   {
>   	if (!atomic_fetch_inc(&t->count))
>
Chris Wilson Oct. 14, 2019, 9:27 a.m. UTC | #2
Quoting Tvrtko Ursulin (2019-10-14 10:25:04)
> 
> On 13/10/2019 20:31, Chris Wilson wrote:
> > We rely on only the tasklet being allowed to call into process_csb(), so
> > assert that is locked when we do. As the tasklet uses a simple bitlock,
> > there is no strong lockdep checking so we must make do with a plain
> > assertion that the tasklet is running and assume that we are the
> > tasklet!
> > 
> > Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
> > Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> > ---
> >   drivers/gpu/drm/i915/gt/intel_lrc.c | 1 +
> >   drivers/gpu/drm/i915/i915_gem.h     | 5 +++++
> >   2 files changed, 6 insertions(+)
> > 
> > diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c
> > index 8be9e69d5718..ab20433182d1 100644
> > --- a/drivers/gpu/drm/i915/gt/intel_lrc.c
> > +++ b/drivers/gpu/drm/i915/gt/intel_lrc.c
> > @@ -1984,6 +1984,7 @@ static void process_csb(struct intel_engine_cs *engine)
> >       u8 head, tail;
> >   
> >       GEM_BUG_ON(USES_GUC_SUBMISSION(engine->i915));
> > +     GEM_BUG_ON(!tasklet_is_locked(&execlists->tasklet));
> 
> I see some reset paths calling process_csb which looks like a problem 
> for this assert.

Oh, no it's not :)

execlists_reset() is surrounded by reset_prepare and reset_finish who
are responsible for disabling the tasklet and serialising with direct
submission around the reset.
-Chris
Chris Wilson Oct. 14, 2019, 9:28 a.m. UTC | #3
Quoting Chris Wilson (2019-10-14 10:27:59)
> Quoting Tvrtko Ursulin (2019-10-14 10:25:04)
> > 
> > On 13/10/2019 20:31, Chris Wilson wrote:
> > > We rely on only the tasklet being allowed to call into process_csb(), so
> > > assert that is locked when we do. As the tasklet uses a simple bitlock,
> > > there is no strong lockdep checking so we must make do with a plain
> > > assertion that the tasklet is running and assume that we are the
> > > tasklet!
> > > 
> > > Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
> > > Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> > > ---
> > >   drivers/gpu/drm/i915/gt/intel_lrc.c | 1 +
> > >   drivers/gpu/drm/i915/i915_gem.h     | 5 +++++
> > >   2 files changed, 6 insertions(+)
> > > 
> > > diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c
> > > index 8be9e69d5718..ab20433182d1 100644
> > > --- a/drivers/gpu/drm/i915/gt/intel_lrc.c
> > > +++ b/drivers/gpu/drm/i915/gt/intel_lrc.c
> > > @@ -1984,6 +1984,7 @@ static void process_csb(struct intel_engine_cs *engine)
> > >       u8 head, tail;
> > >   
> > >       GEM_BUG_ON(USES_GUC_SUBMISSION(engine->i915));
> > > +     GEM_BUG_ON(!tasklet_is_locked(&execlists->tasklet));
> > 
> > I see some reset paths calling process_csb which looks like a problem 
> > for this assert.
> 
> Oh, no it's not :)
> 
> execlists_reset() is surrounded by reset_prepare and reset_finish who
> are responsible for disabling the tasklet and serialising with direct
> submission around the reset.

Same being true for cancel_requests, on wedging we have to disable the
tasklet before taking control of the state.
-Chris
Chris Wilson Oct. 14, 2019, 9:31 a.m. UTC | #4
Quoting Chris Wilson (2019-10-14 10:28:53)
> Quoting Chris Wilson (2019-10-14 10:27:59)
> > Quoting Tvrtko Ursulin (2019-10-14 10:25:04)
> > > 
> > > On 13/10/2019 20:31, Chris Wilson wrote:
> > > > We rely on only the tasklet being allowed to call into process_csb(), so
> > > > assert that is locked when we do. As the tasklet uses a simple bitlock,
> > > > there is no strong lockdep checking so we must make do with a plain
> > > > assertion that the tasklet is running and assume that we are the
> > > > tasklet!
> > > > 
> > > > Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
> > > > Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> > > > ---
> > > >   drivers/gpu/drm/i915/gt/intel_lrc.c | 1 +
> > > >   drivers/gpu/drm/i915/i915_gem.h     | 5 +++++
> > > >   2 files changed, 6 insertions(+)
> > > > 
> > > > diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c
> > > > index 8be9e69d5718..ab20433182d1 100644
> > > > --- a/drivers/gpu/drm/i915/gt/intel_lrc.c
> > > > +++ b/drivers/gpu/drm/i915/gt/intel_lrc.c
> > > > @@ -1984,6 +1984,7 @@ static void process_csb(struct intel_engine_cs *engine)
> > > >       u8 head, tail;
> > > >   
> > > >       GEM_BUG_ON(USES_GUC_SUBMISSION(engine->i915));
> > > > +     GEM_BUG_ON(!tasklet_is_locked(&execlists->tasklet));
> > > 
> > > I see some reset paths calling process_csb which looks like a problem 
> > > for this assert.
> > 
> > Oh, no it's not :)
> > 
> > execlists_reset() is surrounded by reset_prepare and reset_finish who
> > are responsible for disabling the tasklet and serialising with direct
> > submission around the reset.
> 
> Same being true for cancel_requests, on wedging we have to disable the
> tasklet before taking control of the state.

Hmm. That being said we do have that sanitize path which cuts a few
corners.
-Chris

Patch
diff mbox series

diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c
index 8be9e69d5718..ab20433182d1 100644
--- a/drivers/gpu/drm/i915/gt/intel_lrc.c
+++ b/drivers/gpu/drm/i915/gt/intel_lrc.c
@@ -1984,6 +1984,7 @@  static void process_csb(struct intel_engine_cs *engine)
 	u8 head, tail;
 
 	GEM_BUG_ON(USES_GUC_SUBMISSION(engine->i915));
+	GEM_BUG_ON(!tasklet_is_locked(&execlists->tasklet));
 
 	/*
 	 * Note that csb_write, csb_status may be either in HWSP or mmio.
diff --git a/drivers/gpu/drm/i915/i915_gem.h b/drivers/gpu/drm/i915/i915_gem.h
index db20d2b0842b..f6f9675848b8 100644
--- a/drivers/gpu/drm/i915/i915_gem.h
+++ b/drivers/gpu/drm/i915/i915_gem.h
@@ -86,6 +86,11 @@  static inline void tasklet_lock(struct tasklet_struct *t)
 		cpu_relax();
 }
 
+static inline bool tasklet_is_locked(const struct tasklet_struct *t)
+{
+	return test_bit(TASKLET_STATE_RUN, &t->state);
+}
+
 static inline void __tasklet_disable_sync_once(struct tasklet_struct *t)
 {
 	if (!atomic_fetch_inc(&t->count))