[Intel-gfx] [PATCH 3/4 v3] drm/i915: always mark the target of pwrite() as dirty

Chris Wilson chris at chris-wilson.co.uk
Thu Dec 10 13:09:51 PST 2015

On Thu, Dec 10, 2015 at 06:51:25PM +0000, Dave Gordon wrote:
> Currently, the target object being written *may* be marked dirty, either
> in i915_gem_gtt_pwrite_fast() (as a side-effect of setting its domain to
> GTT!), or in i915_gem_shmem_pwrite() (if it's a shmfs-backed object).
> While these two are the common cases, it's not obvious that they cover
> every possible path through the pwrite code, for every possible type
> of object (e.g. phys, stolen, etc). So here we move setting-the-mark
> to the top level so that it is obvious that it applies no matter which
> subsequent path is followed.
> Signed-off-by: Dave Gordon <david.s.gordon at intel.com>
> Cc: Chris Wilson <chris at chris-wilson.co.uk>

I don't like this patch - I feel like it divorces the information that
we are dirtying the pages from the actual copy. Especially as some paths
don't actually dirty the object's backing storage (for extra confusion).

Chris Wilson, Intel Open Source Technology Centre

More information about the Intel-gfx mailing list