diff mbox series

[PATH,i-g-t,2/2] core: Show backtrace from igt_skip_on_simulation

Message ID 20180912093306.23537-2-tvrtko.ursulin@linux.intel.com (mailing list archive)
State New, archived
Headers show
Series [PATH,i-g-t,1/2] intel: Be consistent with test results on simulation | expand

Commit Message

Tvrtko Ursulin Sept. 12, 2018, 9:33 a.m. UTC
From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

igt_skip_on_simulation is called both directly from tests but also from
library helpers. In the latter case especially the logged caller name is
useless since it is always the helper itself. What we instead want to know
is who is the caller.

Trivial approach would be to move the helper to a header as static inline,
but due the longjmp in it it can never be inlined. Alternative option is
to print a backtrace from it.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Radoslaw Szwichtenberg <radoslaw.szwichtenberg@intel.com>
---
 lib/igt_core.c | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)

Comments

Daniel Vetter Sept. 14, 2018, 9:12 a.m. UTC | #1
On Wed, Sep 12, 2018 at 10:33:06AM +0100, Tvrtko Ursulin wrote:
> From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> 
> igt_skip_on_simulation is called both directly from tests but also from
> library helpers. In the latter case especially the logged caller name is
> useless since it is always the helper itself. What we instead want to know
> is who is the caller.
> 
> Trivial approach would be to move the helper to a header as static inline,
> but due the longjmp in it it can never be inlined. Alternative option is
> to print a backtrace from it.
> 
> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> Cc: Radoslaw Szwichtenberg <radoslaw.szwichtenberg@intel.com>
> ---
>  lib/igt_core.c | 20 ++++++++++++++++----
>  1 file changed, 16 insertions(+), 4 deletions(-)
> 
> diff --git a/lib/igt_core.c b/lib/igt_core.c
> index 23bb858fd886..990abc5a36b3 100644
> --- a/lib/igt_core.c
> +++ b/lib/igt_core.c
> @@ -2065,14 +2065,26 @@ bool igt_run_in_simulation(void)
>   */
>  void igt_skip_on_simulation(void)
>  {
> +	bool in_simulation;
> +
>  	if (igt_only_list_subtests())
>  		return;
>  
> +	in_simulation = igt_run_in_simulation();
> +
>  	if (!igt_can_fail()) {
> -		igt_fixture
> -			igt_require(!igt_run_in_simulation());
> -	} else
> -		igt_require(!igt_run_in_simulation());
> +		igt_fixture {
> +			if (in_simulation) {
> +				print_backtrace();
> +				igt_require(!in_simulation);
> +			}
> +		}
> +	} else {
> +		if (in_simulation) {
> +			print_backtrace();
> +			igt_require(!in_simulation);

Hm, why don't we go right ahead and push this into igt_skip()? There's a
pile of other igt_require, and we tend to push a lot of these into the
library. So they have all the same problem.
-Daniel

> +		}
> +	}
>  }
>  
>  /* structured logging */
> -- 
> 2.17.1
> 
> _______________________________________________
> igt-dev mailing list
> igt-dev@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/igt-dev
Tvrtko Ursulin Sept. 14, 2018, 9:19 a.m. UTC | #2
On 14/09/2018 10:12, Daniel Vetter wrote:
> On Wed, Sep 12, 2018 at 10:33:06AM +0100, Tvrtko Ursulin wrote:
>> From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
>>
>> igt_skip_on_simulation is called both directly from tests but also from
>> library helpers. In the latter case especially the logged caller name is
>> useless since it is always the helper itself. What we instead want to know
>> is who is the caller.
>>
>> Trivial approach would be to move the helper to a header as static inline,
>> but due the longjmp in it it can never be inlined. Alternative option is
>> to print a backtrace from it.
>>
>> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
>> Cc: Radoslaw Szwichtenberg <radoslaw.szwichtenberg@intel.com>
>> ---
>>   lib/igt_core.c | 20 ++++++++++++++++----
>>   1 file changed, 16 insertions(+), 4 deletions(-)
>>
>> diff --git a/lib/igt_core.c b/lib/igt_core.c
>> index 23bb858fd886..990abc5a36b3 100644
>> --- a/lib/igt_core.c
>> +++ b/lib/igt_core.c
>> @@ -2065,14 +2065,26 @@ bool igt_run_in_simulation(void)
>>    */
>>   void igt_skip_on_simulation(void)
>>   {
>> +	bool in_simulation;
>> +
>>   	if (igt_only_list_subtests())
>>   		return;
>>   
>> +	in_simulation = igt_run_in_simulation();
>> +
>>   	if (!igt_can_fail()) {
>> -		igt_fixture
>> -			igt_require(!igt_run_in_simulation());
>> -	} else
>> -		igt_require(!igt_run_in_simulation());
>> +		igt_fixture {
>> +			if (in_simulation) {
>> +				print_backtrace();
>> +				igt_require(!in_simulation);
>> +			}
>> +		}
>> +	} else {
>> +		if (in_simulation) {
>> +			print_backtrace();
>> +			igt_require(!in_simulation);
> 
> Hm, why don't we go right ahead and push this into igt_skip()? There's a
> pile of other igt_require, and we tend to push a lot of these into the
> library. So they have all the same problem.

Maybe.. I wasn't 100% this was a good idea to start with, or in another 
words, that other people would consider it a problem. Since the downside 
is test output gets more verbose on skips, or some could say more noisy.

So I basically floated the patch to see if it will provoke some 
responses. :)

Regards,

Tvrtko
Daniel Vetter Sept. 14, 2018, 9:46 a.m. UTC | #3
On Fri, Sep 14, 2018 at 10:19:29AM +0100, Tvrtko Ursulin wrote:
> 
> On 14/09/2018 10:12, Daniel Vetter wrote:
> > On Wed, Sep 12, 2018 at 10:33:06AM +0100, Tvrtko Ursulin wrote:
> > > From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> > > 
> > > igt_skip_on_simulation is called both directly from tests but also from
> > > library helpers. In the latter case especially the logged caller name is
> > > useless since it is always the helper itself. What we instead want to know
> > > is who is the caller.
> > > 
> > > Trivial approach would be to move the helper to a header as static inline,
> > > but due the longjmp in it it can never be inlined. Alternative option is
> > > to print a backtrace from it.
> > > 
> > > Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> > > Cc: Radoslaw Szwichtenberg <radoslaw.szwichtenberg@intel.com>
> > > ---
> > >   lib/igt_core.c | 20 ++++++++++++++++----
> > >   1 file changed, 16 insertions(+), 4 deletions(-)
> > > 
> > > diff --git a/lib/igt_core.c b/lib/igt_core.c
> > > index 23bb858fd886..990abc5a36b3 100644
> > > --- a/lib/igt_core.c
> > > +++ b/lib/igt_core.c
> > > @@ -2065,14 +2065,26 @@ bool igt_run_in_simulation(void)
> > >    */
> > >   void igt_skip_on_simulation(void)
> > >   {
> > > +	bool in_simulation;
> > > +
> > >   	if (igt_only_list_subtests())
> > >   		return;
> > > +	in_simulation = igt_run_in_simulation();
> > > +
> > >   	if (!igt_can_fail()) {
> > > -		igt_fixture
> > > -			igt_require(!igt_run_in_simulation());
> > > -	} else
> > > -		igt_require(!igt_run_in_simulation());
> > > +		igt_fixture {
> > > +			if (in_simulation) {
> > > +				print_backtrace();
> > > +				igt_require(!in_simulation);
> > > +			}
> > > +		}
> > > +	} else {
> > > +		if (in_simulation) {
> > > +			print_backtrace();
> > > +			igt_require(!in_simulation);
> > 
> > Hm, why don't we go right ahead and push this into igt_skip()? There's a
> > pile of other igt_require, and we tend to push a lot of these into the
> > library. So they have all the same problem.
> 
> Maybe.. I wasn't 100% this was a good idea to start with, or in another
> words, that other people would consider it a problem. Since the downside is
> test output gets more verbose on skips, or some could say more noisy.
> 
> So I basically floated the patch to see if it will provoke some responses.
> :)

We have the backtrace already in igt_fail, makes total sense to add it to
igt_skip too. Has my ack at least.

Aside: I kinda wonder whether we need an __igt_require, which embeds the
if (igt_can_fail) igt_require() else igt_fixture igt_require() thing. But
that's just an aside.

Cheers, Daniel
Chris Wilson Sept. 14, 2018, 9:49 a.m. UTC | #4
Quoting Daniel Vetter (2018-09-14 10:46:25)
> On Fri, Sep 14, 2018 at 10:19:29AM +0100, Tvrtko Ursulin wrote:
> > 
> > On 14/09/2018 10:12, Daniel Vetter wrote:
> > > On Wed, Sep 12, 2018 at 10:33:06AM +0100, Tvrtko Ursulin wrote:
> > > > From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> > > > 
> > > > igt_skip_on_simulation is called both directly from tests but also from
> > > > library helpers. In the latter case especially the logged caller name is
> > > > useless since it is always the helper itself. What we instead want to know
> > > > is who is the caller.
> > > > 
> > > > Trivial approach would be to move the helper to a header as static inline,
> > > > but due the longjmp in it it can never be inlined. Alternative option is
> > > > to print a backtrace from it.
> > > > 
> > > > Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> > > > Cc: Radoslaw Szwichtenberg <radoslaw.szwichtenberg@intel.com>
> > > > ---
> > > >   lib/igt_core.c | 20 ++++++++++++++++----
> > > >   1 file changed, 16 insertions(+), 4 deletions(-)
> > > > 
> > > > diff --git a/lib/igt_core.c b/lib/igt_core.c
> > > > index 23bb858fd886..990abc5a36b3 100644
> > > > --- a/lib/igt_core.c
> > > > +++ b/lib/igt_core.c
> > > > @@ -2065,14 +2065,26 @@ bool igt_run_in_simulation(void)
> > > >    */
> > > >   void igt_skip_on_simulation(void)
> > > >   {
> > > > + bool in_simulation;
> > > > +
> > > >           if (igt_only_list_subtests())
> > > >                   return;
> > > > + in_simulation = igt_run_in_simulation();
> > > > +
> > > >           if (!igt_can_fail()) {
> > > > -         igt_fixture
> > > > -                 igt_require(!igt_run_in_simulation());
> > > > - } else
> > > > -         igt_require(!igt_run_in_simulation());
> > > > +         igt_fixture {
> > > > +                 if (in_simulation) {
> > > > +                         print_backtrace();
> > > > +                         igt_require(!in_simulation);
> > > > +                 }
> > > > +         }
> > > > + } else {
> > > > +         if (in_simulation) {
> > > > +                 print_backtrace();
> > > > +                 igt_require(!in_simulation);
> > > 
> > > Hm, why don't we go right ahead and push this into igt_skip()? There's a
> > > pile of other igt_require, and we tend to push a lot of these into the
> > > library. So they have all the same problem.
> > 
> > Maybe.. I wasn't 100% this was a good idea to start with, or in another
> > words, that other people would consider it a problem. Since the downside is
> > test output gets more verbose on skips, or some could say more noisy.
> > 
> > So I basically floated the patch to see if it will provoke some responses.
> > :)
> 
> We have the backtrace already in igt_fail, makes total sense to add it to
> igt_skip too. Has my ack at least.

Skip is rather more frequent than fail, and more often than not,
expected. So I am not entirely enjoying the prospect of a lot more noise,
the requirement message was supposed to be sufficient to understand why
we skipped. Maybe enforce that we have no igt_skip without a message?
-Chris
Daniel Vetter Sept. 14, 2018, 3:27 p.m. UTC | #5
On Fri, Sep 14, 2018 at 10:49:47AM +0100, Chris Wilson wrote:
> Quoting Daniel Vetter (2018-09-14 10:46:25)
> > On Fri, Sep 14, 2018 at 10:19:29AM +0100, Tvrtko Ursulin wrote:
> > > 
> > > On 14/09/2018 10:12, Daniel Vetter wrote:
> > > > On Wed, Sep 12, 2018 at 10:33:06AM +0100, Tvrtko Ursulin wrote:
> > > > > From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> > > > > 
> > > > > igt_skip_on_simulation is called both directly from tests but also from
> > > > > library helpers. In the latter case especially the logged caller name is
> > > > > useless since it is always the helper itself. What we instead want to know
> > > > > is who is the caller.
> > > > > 
> > > > > Trivial approach would be to move the helper to a header as static inline,
> > > > > but due the longjmp in it it can never be inlined. Alternative option is
> > > > > to print a backtrace from it.
> > > > > 
> > > > > Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> > > > > Cc: Radoslaw Szwichtenberg <radoslaw.szwichtenberg@intel.com>
> > > > > ---
> > > > >   lib/igt_core.c | 20 ++++++++++++++++----
> > > > >   1 file changed, 16 insertions(+), 4 deletions(-)
> > > > > 
> > > > > diff --git a/lib/igt_core.c b/lib/igt_core.c
> > > > > index 23bb858fd886..990abc5a36b3 100644
> > > > > --- a/lib/igt_core.c
> > > > > +++ b/lib/igt_core.c
> > > > > @@ -2065,14 +2065,26 @@ bool igt_run_in_simulation(void)
> > > > >    */
> > > > >   void igt_skip_on_simulation(void)
> > > > >   {
> > > > > + bool in_simulation;
> > > > > +
> > > > >           if (igt_only_list_subtests())
> > > > >                   return;
> > > > > + in_simulation = igt_run_in_simulation();
> > > > > +
> > > > >           if (!igt_can_fail()) {
> > > > > -         igt_fixture
> > > > > -                 igt_require(!igt_run_in_simulation());
> > > > > - } else
> > > > > -         igt_require(!igt_run_in_simulation());
> > > > > +         igt_fixture {
> > > > > +                 if (in_simulation) {
> > > > > +                         print_backtrace();
> > > > > +                         igt_require(!in_simulation);
> > > > > +                 }
> > > > > +         }
> > > > > + } else {
> > > > > +         if (in_simulation) {
> > > > > +                 print_backtrace();
> > > > > +                 igt_require(!in_simulation);
> > > > 
> > > > Hm, why don't we go right ahead and push this into igt_skip()? There's a
> > > > pile of other igt_require, and we tend to push a lot of these into the
> > > > library. So they have all the same problem.
> > > 
> > > Maybe.. I wasn't 100% this was a good idea to start with, or in another
> > > words, that other people would consider it a problem. Since the downside is
> > > test output gets more verbose on skips, or some could say more noisy.
> > > 
> > > So I basically floated the patch to see if it will provoke some responses.
> > > :)
> > 
> > We have the backtrace already in igt_fail, makes total sense to add it to
> > igt_skip too. Has my ack at least.
> 
> Skip is rather more frequent than fail, and more often than not,
> expected. So I am not entirely enjoying the prospect of a lot more noise,
> the requirement message was supposed to be sufficient to understand why
> we skipped. Maybe enforce that we have no igt_skip without a message?

Hm yeah, adding a const char * to igt_skip_on_simulation that explains why
we're skipping could be the pretty solution here. Would also serve as some
self-documentation.
-Daniel
diff mbox series

Patch

diff --git a/lib/igt_core.c b/lib/igt_core.c
index 23bb858fd886..990abc5a36b3 100644
--- a/lib/igt_core.c
+++ b/lib/igt_core.c
@@ -2065,14 +2065,26 @@  bool igt_run_in_simulation(void)
  */
 void igt_skip_on_simulation(void)
 {
+	bool in_simulation;
+
 	if (igt_only_list_subtests())
 		return;
 
+	in_simulation = igt_run_in_simulation();
+
 	if (!igt_can_fail()) {
-		igt_fixture
-			igt_require(!igt_run_in_simulation());
-	} else
-		igt_require(!igt_run_in_simulation());
+		igt_fixture {
+			if (in_simulation) {
+				print_backtrace();
+				igt_require(!in_simulation);
+			}
+		}
+	} else {
+		if (in_simulation) {
+			print_backtrace();
+			igt_require(!in_simulation);
+		}
+	}
 }
 
 /* structured logging */