Re: [Intel-gfx] [PATCH v3] drm/i915: Replace *_reference/unreference() or *_ref/unref with _get/put()

[Sean Paul]

On Mon, Oct 9, 2017 at 4:59 AM, Daniel Vetter <daniel@xxxxxxxx> wrote:
> On Sun, Oct 08, 2017 at 03:43:35PM +0100, Chris Wilson wrote:
>> Quoting Harsha Sharma (2017-10-08 15:04:07)
>> > @@ -624,7 +624,7 @@ static bool intel_fbdev_init_bios(struct drm_device *dev,
>> >         ifbdev->preferred_bpp = fb->base.format->cpp[0] * 8;
>> >         ifbdev->fb = fb;
>> >
>> > -       drm_framebuffer_reference(&ifbdev->fb->base);
>> > +       drm_framebuffer_put(&ifbdev->fb->base);
>>
>> Whoops.
>
> Hm yeah, how did this happen? Does cocci really do this, or is that an
> accident from manually fixing stuff up?


Running the spatch from the commit message gives me the correct substitution:
@@ -627,7 +627,7 @@ static bool intel_fbdev_init_bios(struct drm_device *dev,
        ifbdev->preferred_bpp = fb->base.format->cpp[0] * 8;
        ifbdev->fb = fb;

-       drm_framebuffer_reference(&ifbdev->fb->base);
+       drm_framebuffer_get(&ifbdev->fb->base);

        /* Final pass to check if any active pipes don't have fbs */
        for_each_crtc(dev, crtc) {


Probably just finger slip since this is the last chunk before the
omitted selftests changes.

Harsha: the "better" way to omit the selftests without hand tuning the
patch would be to run the cocci spatch on i915 as normal, and then run
"git checkout -- drivers/gpu/drm/i915/selftests/" before committing.
It's dangerous to edit patches by hand, or to misrepresent a patch as
being the result of a cocci spatch when it's not.

Sean



> -Daniel
> --
> Daniel Vetter
> Software Engineer, Intel Corporation
> http://blog.ffwll.ch