linux-kernel - Re: [PATCH] drm/amdgpu: use dirty framebuffer helper

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CADnq5_MBF2j1Zwe2gFxN-bwe62Yo-RbOepUBwzYDR0APc=T0Zw@mail.gmail.com>
Date:   Mon, 19 Sep 2022 12:26:01 -0400
From:   Alex Deucher <alexdeucher@...il.com>
To:     Thomas Zimmermann <tzimmermann@...e.de>
Cc:     Hamza Mahfooz <hamza.mahfooz@....com>,
        linux-kernel@...r.kernel.org, Guchun Chen <guchun.chen@....com>,
        David Airlie <airlied@...ux.ie>,
        "Pan, Xinhui" <Xinhui.Pan@....com>, amd-gfx@...ts.freedesktop.org,
        Aurabindo Pillai <aurabindo.pillai@....com>,
        Sean Paul <seanpaul@...omium.org>,
        dri-devel@...ts.freedesktop.org,
        Alex Deucher <alexander.deucher@....com>,
        Christian König <christian.koenig@....com>,
        Fernando Ramos <greenfoo@....eu>
Subject: Re: [PATCH] drm/amdgpu: use dirty framebuffer helper

On Mon, Sep 19, 2022 at 2:44 AM Thomas Zimmermann <tzimmermann@...e.de> wrote:
>
> Hi
>
> Am 06.09.22 um 21:57 schrieb Hamza Mahfooz:
> > Currently, we aren't handling DRM_IOCTL_MODE_DIRTYFB. So, use
> > drm_atomic_helper_dirtyfb() as the dirty callback in the amdgpu_fb_funcs
> > struct.
>
> drm_atomic_helper_dirtyfb() creates a new atomic commit for the
> frambuffer's planes. Drivers can then updates these planes' output
> (e.g., writeback to video memory). I thought that amdgpu simply scans
> out from the framebuffer's memory regions in VRAM. So I'm curious why
> this patch is necessary.

I think in this particular case, the problem is that there are still
some asic which default to non-atomic code which is what is causing
the problem here.  Something like this would fix that:
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
index 5b09c8f4fe95..f5e9dd454c54 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
@@ -497,6 +497,11 @@ bool amdgpu_display_ddc_probe(struct
amdgpu_connector *amdgpu_connector,
 static const struct drm_framebuffer_funcs amdgpu_fb_funcs = {
        .destroy = drm_gem_fb_destroy,
        .create_handle = drm_gem_fb_create_handle,
+};
+
+static const struct drm_framebuffer_funcs amdgpu_fb_funcs_atomic = {
+       .destroy = drm_gem_fb_destroy,
+       .create_handle = drm_gem_fb_create_handle,
        .dirty = drm_atomic_helper_dirtyfb,
 };

@@ -1102,7 +1107,10 @@ static int
amdgpu_display_gem_fb_verify_and_init(struct drm_device *dev,
        if (ret)
                goto err;

-       ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs);
+       if (drm_drv_uses_atomic_modeset(adev_to_drm(adev)))
+               ret = drm_framebuffer_init(dev, &rfb->base,
&amdgpu_fb_funcs_atomic);
+       else
+               ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs);
        if (ret)
                goto err;

As for why we need the dirty callback, I think it's used for PSR.

Alex

>
> Best regards
> Thomas
>
> >
> > Signed-off-by: Hamza Mahfooz <hamza.mahfooz@....com>
> > ---
> >   drivers/gpu/drm/amd/amdgpu/amdgpu_display.c | 2 ++
> >   1 file changed, 2 insertions(+)
> >
> > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
> > index c20922a5af9f..5b09c8f4fe95 100644
> > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
> > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
> > @@ -38,6 +38,7 @@
> >   #include <linux/pci.h>
> >   #include <linux/pm_runtime.h>
> >   #include <drm/drm_crtc_helper.h>
> > +#include <drm/drm_damage_helper.h>
> >   #include <drm/drm_edid.h>
> >   #include <drm/drm_gem_framebuffer_helper.h>
> >   #include <drm/drm_fb_helper.h>
> > @@ -496,6 +497,7 @@ bool amdgpu_display_ddc_probe(struct amdgpu_connector *amdgpu_connector,
> >   static const struct drm_framebuffer_funcs amdgpu_fb_funcs = {
> >       .destroy = drm_gem_fb_destroy,
> >       .create_handle = drm_gem_fb_create_handle,
> > +     .dirty = drm_atomic_helper_dirtyfb,
> >   };
> >
> >   uint32_t amdgpu_display_supported_domains(struct amdgpu_device *adev,
>
> --
> Thomas Zimmermann
> Graphics Driver Developer
> SUSE Software Solutions Germany GmbH
> Maxfeldstr. 5, 90409 Nürnberg, Germany
> (HRB 36809, AG Nürnberg)
> Geschäftsführer: Ivo Totev