Nouveau - May 2015 - [PATCH] nvc0: fix context destruction for partly implemented tesselation

Backtrace:
...
0x00007ffff57fc392 in __assert_fail () from /lib64/libc.so.6
0x00007ffff0cf5bec in nvc0_shader_stage (pipe=<optimized out>) at
./nvc0/nvc0_context.h:204
nvc0_set_constant_buffer (pipe=0x631080, shader=<optimized out>,
index=<optimized out>, cb=0x0)
  at nvc0/nvc0_state.c:771
0x00007ffff0a2cf63 in st_destroy_context (st=0x68a9f0) at
state_tracker/st_context.c:382
...

Signed-off-by: Tobias Klausmann <tobias.johannes.klausmann at mni.thm.de>
---
 src/gallium/drivers/nouveau/nvc0/nvc0_context.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_context.h
b/src/gallium/drivers/nouveau/nvc0/nvc0_context.h
index 09d08e4..f910541 100644
--- a/src/gallium/drivers/nouveau/nvc0/nvc0_context.h
+++ b/src/gallium/drivers/nouveau/nvc0/nvc0_context.h
@@ -195,8 +195,8 @@ nvc0_shader_stage(unsigned pipe)
 {
    switch (pipe) {
    case PIPE_SHADER_VERTEX: return 0;
-/* case PIPE_SHADER_TESSELLATION_CONTROL: return 1; */
-/* case PIPE_SHADER_TESSELLATION_EVALUATION: return 2; */
+   case PIPE_SHADER_TESS_CTRL: return 1;
+   case PIPE_SHADER_TESS_EVAL: return 2;
    case PIPE_SHADER_GEOMETRY: return 3;
    case PIPE_SHADER_FRAGMENT: return 4;
    case PIPE_SHADER_COMPUTE: return 5;
-- 
2.4.1

Hm, oops. But I guess my tess patchset will take care of this. I guess
it's a side-effect of adding to the gallium enums.

Marek, you ended up pushing those gallium patches out... I was going
to wait until the branchpoint was done. Should we just fix up the
drivers as necessary, or fix up st_destroy_context to not call it for
tess shaders when !tess? [But then we have to teach st/mesa about
tess]

On Sun, May 17, 2015 at 8:56 PM, Tobias Klausmann
<tobias.johannes.klausmann at mni.thm.de> wrote:
> Backtrace:
> ...
> 0x00007ffff57fc392 in __assert_fail () from /lib64/libc.so.6
> 0x00007ffff0cf5bec in nvc0_shader_stage (pipe=<optimized out>) at
./nvc0/nvc0_context.h:204
> nvc0_set_constant_buffer (pipe=0x631080, shader=<optimized out>,
index=<optimized out>, cb=0x0)
>   at nvc0/nvc0_state.c:771
> 0x00007ffff0a2cf63 in st_destroy_context (st=0x68a9f0) at
state_tracker/st_context.c:382
> ...
>
> Signed-off-by: Tobias Klausmann <tobias.johannes.klausmann at
mni.thm.de>
> ---
>  src/gallium/drivers/nouveau/nvc0/nvc0_context.h | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_context.h
b/src/gallium/drivers/nouveau/nvc0/nvc0_context.h
> index 09d08e4..f910541 100644
> --- a/src/gallium/drivers/nouveau/nvc0/nvc0_context.h
> +++ b/src/gallium/drivers/nouveau/nvc0/nvc0_context.h
> @@ -195,8 +195,8 @@ nvc0_shader_stage(unsigned pipe)
>  {
>     switch (pipe) {
>     case PIPE_SHADER_VERTEX: return 0;
> -/* case PIPE_SHADER_TESSELLATION_CONTROL: return 1; */
> -/* case PIPE_SHADER_TESSELLATION_EVALUATION: return 2; */
> +   case PIPE_SHADER_TESS_CTRL: return 1;
> +   case PIPE_SHADER_TESS_EVAL: return 2;
>     case PIPE_SHADER_GEOMETRY: return 3;
>     case PIPE_SHADER_FRAGMENT: return 4;
>     case PIPE_SHADER_COMPUTE: return 5;
> --
> 2.4.1
>
> _______________________________________________
> Nouveau mailing list
> Nouveau at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/nouveau

We should fix st/mesa and move the code to cso_destroy_context, where
it belongs.

Marek

On Mon, May 18, 2015 at 3:49 AM, Ilia Mirkin <imirkin at alum.mit.edu>
wrote:
> Hm, oops. But I guess my tess patchset will take care of this. I guess
> it's a side-effect of adding to the gallium enums.
>
> Marek, you ended up pushing those gallium patches out... I was going
> to wait until the branchpoint was done. Should we just fix up the
> drivers as necessary, or fix up st_destroy_context to not call it for
> tess shaders when !tess? [But then we have to teach st/mesa about
> tess]
>
> On Sun, May 17, 2015 at 8:56 PM, Tobias Klausmann
> <tobias.johannes.klausmann at mni.thm.de> wrote:
>> Backtrace:
>> ...
>> 0x00007ffff57fc392 in __assert_fail () from /lib64/libc.so.6
>> 0x00007ffff0cf5bec in nvc0_shader_stage (pipe=<optimized out>) at
./nvc0/nvc0_context.h:204
>> nvc0_set_constant_buffer (pipe=0x631080, shader=<optimized out>,
index=<optimized out>, cb=0x0)
>>   at nvc0/nvc0_state.c:771
>> 0x00007ffff0a2cf63 in st_destroy_context (st=0x68a9f0) at
state_tracker/st_context.c:382
>> ...
>>
>> Signed-off-by: Tobias Klausmann <tobias.johannes.klausmann at
mni.thm.de>
>> ---
>>  src/gallium/drivers/nouveau/nvc0/nvc0_context.h | 4 ++--
>>  1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_context.h
b/src/gallium/drivers/nouveau/nvc0/nvc0_context.h
>> index 09d08e4..f910541 100644
>> --- a/src/gallium/drivers/nouveau/nvc0/nvc0_context.h
>> +++ b/src/gallium/drivers/nouveau/nvc0/nvc0_context.h
>> @@ -195,8 +195,8 @@ nvc0_shader_stage(unsigned pipe)
>>  {
>>     switch (pipe) {
>>     case PIPE_SHADER_VERTEX: return 0;
>> -/* case PIPE_SHADER_TESSELLATION_CONTROL: return 1; */
>> -/* case PIPE_SHADER_TESSELLATION_EVALUATION: return 2; */
>> +   case PIPE_SHADER_TESS_CTRL: return 1;
>> +   case PIPE_SHADER_TESS_EVAL: return 2;
>>     case PIPE_SHADER_GEOMETRY: return 3;
>>     case PIPE_SHADER_FRAGMENT: return 4;
>>     case PIPE_SHADER_COMPUTE: return 5;
>> --
>> 2.4.1
>>
>> _______________________________________________
>> Nouveau mailing list
>> Nouveau at lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/nouveau