nic handler?
>
We can reserve a vmap space for ATOMIC or NOWAIT allocations. As for PTE
part, we can also populate reserved space, because after that operation
those are never get released.
The question is how many users need this. As for this particular case i
am in line with Jocelyn Falempe. Allocate for DRM and write on panic.
--
Uladzislau Rezki
return NULL;
> +
> + addr = (unsigned long)area->addr;
> + if (vmap_pages_range(addr, addr + size, pgprot_nx(prot),
> + pages, PAGE_SHIFT) < 0) {
> + return NULL;
> + }
> +
> + return area->addr;
> +}
> +
> /**
> * vmap - map an array of pages into virtually contiguous space
> * @pages: array of page pointers
> --
> 2.48.1
>
It is copy-paste code, so it is odd. The proposal is not a way forward
to me. Unfortunately vmalloc is not compatible with GFP_ATOMIC, there
is at least one place it is a page-table allocation entries where it is
hard-coded to the GFP_KERNEL.
Doing this without locks and synchronizations is not possible.
--
Uladzislau Rezki
Cc: dri-devel@lists.freedesktop.org
Signed-off-by: Uladzislau Rezki (Sony)
---
drivers/gpu/drm/i915/Kconfig.debug | 1 -
drivers/gpu/drm/i915/i915_utils.h | 3 +--
2 files changed, 1 insertion(+), 3 deletions(-)
diff --git a/drivers/gpu/drm/i915/Kconfig.debug
b/drivers/gpu/drm/i915
