-Wmissing-braces in drivers/gpu/drm/amd/amdgpu/gmc_v*

[nickdesaulniers]

  CC      drivers/gpu/drm/amd/amdgpu/gmc_v8_0.o
drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c:1447:41: warning: suggest braces around
      initialization of subobject [-Wmissing-braces]
                struct amdgpu_task_info task_info = { 0 };
                                                      ^
                                                      {}
1 warning generated.
  CC      drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.o
  CC      drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.o
  CC      drivers/gpu/drm/amd/amdgpu/gmc_v9_0.o
drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c:262:41: warning: suggest braces around
      initialization of subobject [-Wmissing-braces]
                struct amdgpu_task_info task_info = { 0 };
                                                      ^
                                                      {}
1 warning generated.

IIRC, { 0 } does not initialize subobjects in C? Should be an easy fix. CONFIG_DRM_AMDGPU=y

[nathanchance]

Patch posted: https://lore.kernel.org/lkml/20180912002559.1853-1-natechancellor@gmail.com/

[nathanchance]

v2 posted: https://lore.kernel.org/lkml/20180912200558.7899-1-natechancellor@gmail.com/

[nathanchance]

Was there ever a consensus on whether { { 0 } } or {} was preferred (or correct)?

Looking at Nick's example, Clang and GCC emit the same assembly for them so it'd probably be better to take the smaller change in this case (unless I'm missing something).

Also looks like someone else found the warnings and sent patches as well:

https://lore.kernel.org/lkml/1536919552-116245-1-git-send-email-peng.hao2@zte.com.cn/

https://lore.kernel.org/lkml/1537026417-3798-1-git-send-email-peng.hao2@zte.com.cn/

[nathanchance]

There was a lengthy back and forth between Guenter Roeck and Christian Koenig in the first link in the comment above.

TL;DR: The maintainers are not accepting either { } or { { 0 } } as a solution for this warning, they'd like to move to memset to account for potential padding issues.

I'll look to see what a patch like this looks like, probably something nasty like my v2 above...

Interestingly enough, Clang actually already calls memset (it looks like at least) for both of those initializers, as seen by extending Nick's example (all three of these functions are identical in assembly): https://godbolt.org/z/AsxFLu

Some other reading I found:

https://stackoverflow.com/questions/10788861/is-memsetmystruct-0-sizeof-mystruct-same-as-mystruct-0

http://www.ex-parrot.com/~chris/random/initialise.html

[nathanchance]

This has been fixed in one of the AMD WIP branches: https://cgit.freedesktop.org/~agd5f/linux/commit/?id=3de3a056e011939dfe27b5b16de2b08d0ed97169

I'll keep an eye on it.

[nathanchance]

Merged into mainline: https://git.kernel.org/torvalds/c/05794eff1aa6060248bfca34ee936c613f94a942