Hi Dan,
On Fri, May 26, 2023 at 02:59:31PM +0300, Dan Carpenter wrote:
> Return the error code if i915_gem_object_create_internal() fails,
> instead of returning success.
>
> Fixes: cf586021642d ("drm/i915/gt: Pipelined page migration")
> Signed-off-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
> ---
> drivers/gpu/drm/i915/gt/selftest_migrate.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/i915/gt/selftest_migrate.c b/drivers/gpu/drm/i915/gt/selftest_migrate.c
> index e677f2da093d..a26429fd5326 100644
> --- a/drivers/gpu/drm/i915/gt/selftest_migrate.c
> +++ b/drivers/gpu/drm/i915/gt/selftest_migrate.c
> @@ -55,8 +55,10 @@ static int copy(struct intel_migrate *migrate,
>
> sz = src->base.size;
> dst = i915_gem_object_create_internal(i915, sz);
> - if (IS_ERR(dst))
> + if (IS_ERR(dst)) {
> + err = PTR_ERR(dst);
> goto err_free_src;
> + }
I think it was intentional to return '0' when
i915_gem_object_create_internal() failed, as we are not testing
object creation here.
I don't really mind this patch, but, on the other hand, returning
an error value here might provide a biased information.
Thanks,
Andi
>
> for_i915_gem_ww(&ww, err, true) {
> err = i915_gem_object_lock(src, &ww);
> --
> 2.39.2
