[PATCH] drm/xe: Fix warning on unreachable statement

[Nathan Chancellor]

Hi Lucas,

On Fri, Jul 19, 2024 at 12:15:34PM -0700, Lucas De Marchi wrote:
> eu_type_to_str() relies on -Wswitch to warn (and -Werror) to make sure
> it handles all enum values. However it's perfectly legal to pass an int
> to that function so in the end that function may happen to return
> nothing. A smart compiler could notice eu_type is never assigned to
> anything other than those values.

Well that assignment happens in load_eu_mask(), which is obviously
called before xe_gt_topology_dump() in xe_gt_topology_init(), but I
don't see how the compiler could assume that eu_type is always one of
those two values when xe_gt_topology_dump() is not static and could be
called from anywhere (i.e., when eu_type is potentially some other
value)? I might be missing something though.

> Trying to reproduce this issue, none of gcc-9, gcc-10 and gcc-13
> triggered for me, but this was reported in a different system with
> gcc-10:
> 
> 	drivers/gpu/drm/xe/xe.o: warning: objtool: xe_gt_topology_dump() falls through to next function xe_gt_topology_init()

I have been seeing similar warnings with clang as well, such as:

  drivers/gpu/drm/xe/xe.o: warning: objtool: xe_gt_topology_dump+0x77: sibling call from callable instruction with modified stack frame

  drivers/gpu/drm/xe/xe.o: warning: objtool: xe_gt_topology_dump() falls through to next function xe_dss_mask_group_ffs()

  drivers/gpu/drm/xe/xe.o: warning: objtool: xe_gt_topology_dump+0x77: can't find jump dest instruction at .text.xe_gt_topology_dump+0xc0

The final warning happens when LTO and CFI are enabled and I see a CFI
failure that appears to be a result of that.

[  +0.008116] Missing ENDBR: __cfi_init_module+0x0/0x10 [xe]
[  +0.000226] ------------[ cut here ]------------
[  +0.000001] kernel BUG at arch/x86/kernel/cet.c:102!
[  +0.000014] Oops: invalid opcode: 0000 [#1] PREEMPT SMP NOPTI
[  +0.000004] CPU: 3 PID: 335 Comm: (udev-worker) Not tainted 6.10.0-rc3-debug-01858-g7108b4a589cd #1 6163aa9290b9c012af9bc3cd405d04758721ac11
[  +0.000005] Hardware name: AZW MINI S/MINI S, BIOS ADLNV105 12/12/2023
...
[  +0.000002]  ? __cfi_init_module+0x10/0x10 [xe 257ff34a29c86b7d55c1c838833ff13eec5c3393]
[  +0.000167]  ? xe_hw_fence_module_init+0x40/0x40 [xe 257ff34a29c86b7d55c1c838833ff13eec5c3393]
[  +0.000171]  ? __cfi_init_module+0x10/0x10 [xe 257ff34a29c86b7d55c1c838833ff13eec5c3393]
[  +0.000174]  ? do_one_initcall+0x147/0x350

> Since that is not really possible, just take the simple approach and
> return NULL.

Indeed, bare unreachable() is generally considered harmful and can introduce
undefined behavior. Commits

  d652d5f1eeeb ("drm/edid: fix objtool warning in drm_cvt_modes()")

and

  3764647b255a ("bcachefs: Remove undefined behavior in bch2_dev_buckets_reserved()")

come to mind as instances like this.

> Cc: Michal Wajdeczko <michal.wajdeczko at intel.com>
> Signed-off-by: Lucas De Marchi <lucas.demarchi at intel.com>

This change resolves all the issues I have noted above, so thank you!

Reviewed-by: Nathan Chancellor <nathan at kernel.org>
Tested-by: Nathan Chancellor <nathan at kernel.org>

> ---
>  drivers/gpu/drm/xe/xe_gt_topology.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/xe/xe_gt_topology.c b/drivers/gpu/drm/xe/xe_gt_topology.c
> index 5a1559edf3e9..0662f71c6ede 100644
> --- a/drivers/gpu/drm/xe/xe_gt_topology.c
> +++ b/drivers/gpu/drm/xe/xe_gt_topology.c
> @@ -233,7 +233,7 @@ static const char *eu_type_to_str(enum xe_gt_eu_type eu_type)
>  		return "simd8";
>  	}
>  
> -	unreachable();
> +	return NULL;
>  }
>  
>  void
> -- 
> 2.43.0
>