Re: [PATCH 1/3] drm/panthor: Don't invalidate OTHER caches

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] doc: comment contradicts code on RO cache invalidation
--

commit 97f084533582df463bbfd6d17bb8670c85068160
Author: Boris Brezillon <boris.brezillon@collabora.com>

drm/panthor: Don't invalidate OTHER caches

This commit stops invalidating OTHER (L1 read-only) caches during MMU
operations to prevent issues with flush-elimination logic on v13 hardware.
The invalidation is deemed unnecessary as TLB invalidation prevents context
data leaks and the cache is read-only.

> diff --git a/drivers/gpu/drm/panthor/panthor_mmu.c b/drivers/gpu/drm/panthor/panthor_mmu.c
> index 9f63a048df61c..7e98084b9a1ed 100644
> --- a/drivers/gpu/drm/panthor/panthor_mmu.c
> +++ b/drivers/gpu/drm/panthor/panthor_mmu.c

[ ... ]

> @@ -1841,8 +1841,7 @@ static void panthor_vm_unlock_region(struct panthor_vm *vm)
>  		 * range is narrow enough and the HW supports it.
>  		 */
>  		ret = panthor_gpu_flush_caches(ptdev, CACHE_CLEAN | CACHE_INV,
> -					       CACHE_CLEAN | CACHE_INV,
> -					       CACHE_INV);
> +					       CACHE_CLEAN | CACHE_INV, 0);

[Severity: Low]
This isn't a bug, but should the comment just above this call in
panthor_vm_unlock_region() be updated?

The comment still explicitly says "and invalidate RO ones", which now
contradicts the implementation since 0 is being passed for the OTHER (RO)
caches.

>  
>  		/* Unlock the region if the flush is effective. */
>  		if (!ret)

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260924-panthor-mmu-fixes-v1-0-e25f9b106466@collabora.com?part=1