| Message ID | 1481624214-13595-1-git-send-email-vidya.srinivas@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, 13 Dec 2016, Vidya Srinivas <vidya.srinivas@intel.com> wrote: > Currently in dual display connected boot scenarios, minimum of the resolutions > is taken for fb width and height as reference. Based on this resolution, other > modes are pruned. > > Example Scenario: If DSI mode is 2560x1440 and HDMI is 1920x1080, during the probing > the fb width and height is set to max 1920x1080 and the DSI mode gets pruned as it is > more than the reference. As a result, there is no DSI display. > Patch fixes this issue by taking the max of the resolutions and creating the fb > based on the same. > > Signed-off-by: Vidya Srinivas <vidya.srinivas@intel.com> > Signed-off-by: Uma Shankar <uma.shankar@intel.com> > --- > drivers/gpu/drm/drm_fb_helper.c | 8 ++++---- > 1 file changed, 4 insertions(+), 4 deletions(-) > > diff --git a/drivers/gpu/drm/drm_fb_helper.c b/drivers/gpu/drm/drm_fb_helper.c > index e934b54..6afc06f 100644 > --- a/drivers/gpu/drm/drm_fb_helper.c > +++ b/drivers/gpu/drm/drm_fb_helper.c > @@ -1482,8 +1482,8 @@ static int drm_fb_helper_single_fb_probe(struct drm_fb_helper *fb_helper, > memset(&sizes, 0, sizeof(struct drm_fb_helper_surface_size)); > sizes.surface_depth = 24; > sizes.surface_bpp = 32; > - sizes.fb_width = (unsigned)-1; > - sizes.fb_height = (unsigned)-1; > + sizes.fb_width = 0; > + sizes.fb_height = 0; > > /* if driver picks 8 or 16 by default use that > for both depth/bpp */ > @@ -1560,9 +1560,9 @@ static int drm_fb_helper_single_fb_probe(struct drm_fb_helper *fb_helper, > } > > if (lasth) > - sizes.fb_width = min_t(u32, desired_mode->hdisplay + x, sizes.fb_width); > + sizes.fb_width = max_t(u32, desired_mode->hdisplay + x, sizes.fb_width); > if (lastv) > - sizes.fb_height = min_t(u32, desired_mode->vdisplay + y, sizes.fb_height); > + sizes.fb_height = max_t(u32, desired_mode->vdisplay + y, sizes.fb_height); > } > > if (crtc_count == 0 || sizes.fb_width == -1 || sizes.fb_height == -1) { Not commenting on the main change itself, but just as a reminder, these checks will need updating too. BR, Jani.
On Tue, Dec 13, 2016 at 03:46:54PM +0530, Vidya Srinivas wrote: > Currently in dual display connected boot scenarios, minimum of the resolutions > is taken for fb width and height as reference. Based on this resolution, other > modes are pruned. > > Example Scenario: If DSI mode is 2560x1440 and HDMI is 1920x1080, during the probing > the fb width and height is set to max 1920x1080 and the DSI mode gets pruned as it is > more than the reference. As a result, there is no DSI display. > Patch fixes this issue by taking the max of the resolutions and creating the fb > based on the same. On the other hand, the viewable content is defined by the smaller mode. If that is the only output visible at panic time, you don't want that information lost due to it being invisible. This is only used for fbcon, which has to be the lowest common denominator. Any actual application can set their own modes and fb. -Chris
On Tue, 13 Dec 2016, Chris Wilson <chris@chris-wilson.co.uk> wrote: > On Tue, Dec 13, 2016 at 03:46:54PM +0530, Vidya Srinivas wrote: >> Currently in dual display connected boot scenarios, minimum of the resolutions >> is taken for fb width and height as reference. Based on this resolution, other >> modes are pruned. >> >> Example Scenario: If DSI mode is 2560x1440 and HDMI is 1920x1080, during the probing >> the fb width and height is set to max 1920x1080 and the DSI mode gets pruned as it is >> more than the reference. As a result, there is no DSI display. >> Patch fixes this issue by taking the max of the resolutions and creating the fb >> based on the same. > > On the other hand, the viewable content is defined by the smaller mode. > If that is the only output visible at panic time, you don't want that > information lost due to it being invisible. > > This is only used for fbcon, which has to be the lowest common > denominator. Any actual application can set their own modes and fb. Do we fail to take over non-native modes with scaling on DSI then? Ville? BR, Jani.
On Tue, Dec 13, 2016 at 04:52:01PM +0200, Jani Nikula wrote: > On Tue, 13 Dec 2016, Chris Wilson <chris@chris-wilson.co.uk> wrote: > > On Tue, Dec 13, 2016 at 03:46:54PM +0530, Vidya Srinivas wrote: > >> Currently in dual display connected boot scenarios, minimum of the resolutions > >> is taken for fb width and height as reference. Based on this resolution, other > >> modes are pruned. > >> > >> Example Scenario: If DSI mode is 2560x1440 and HDMI is 1920x1080, during the probing > >> the fb width and height is set to max 1920x1080 and the DSI mode gets pruned as it is > >> more than the reference. As a result, there is no DSI display. > >> Patch fixes this issue by taking the max of the resolutions and creating the fb > >> based on the same. > > > > On the other hand, the viewable content is defined by the smaller mode. > > If that is the only output visible at panic time, you don't want that > > information lost due to it being invisible. > > > > This is only used for fbcon, which has to be the lowest common > > denominator. Any actual application can set their own modes and fb. > > Do we fail to take over non-native modes with scaling on DSI then? > Ville? I believe we are taking the panel fitter into account when we reconstruct the BIOS fb. So in case the BIOS is upscaling we'll come up with some less than screen size framebuffer. The fb helper will want to use the native mode of the panel however, so at some point we'll drop the BIOS fb (as it's too small) and allocate a new one.
> -----Original Message----- > From: Ville Syrjälä [mailto:ville.syrjala@linux.intel.com] > Sent: Tuesday, December 13, 2016 9:17 PM > To: Jani Nikula <jani.nikula@linux.intel.com> > Cc: Chris Wilson <chris@chris-wilson.co.uk>; Srinivas, Vidya > <vidya.srinivas@intel.com>; intel-gfx@lists.freedesktop.org; Syrjala, Ville > <ville.syrjala@intel.com>; dri-devel@lists.freedesktop.org > Subject: Re: [Intel-gfx] [PATCH] drm: Fix for invalid pruning of modes in dual > display cases > > On Tue, Dec 13, 2016 at 04:52:01PM +0200, Jani Nikula wrote: > > On Tue, 13 Dec 2016, Chris Wilson <chris@chris-wilson.co.uk> wrote: > > > On Tue, Dec 13, 2016 at 03:46:54PM +0530, Vidya Srinivas wrote: > > >> Currently in dual display connected boot scenarios, minimum of the > > >> resolutions is taken for fb width and height as reference. Based on > > >> this resolution, other modes are pruned. > > >> > > >> Example Scenario: If DSI mode is 2560x1440 and HDMI is 1920x1080, > > >> during the probing the fb width and height is set to max 1920x1080 > > >> and the DSI mode gets pruned as it is more than the reference. As a > result, there is no DSI display. > > >> Patch fixes this issue by taking the max of the resolutions and > > >> creating the fb based on the same. > > > > > > On the other hand, the viewable content is defined by the smaller mode. > > > If that is the only output visible at panic time, you don't want > > > that information lost due to it being invisible. > > > > > > This is only used for fbcon, which has to be the lowest common > > > denominator. Any actual application can set their own modes and fb. > > > > Do we fail to take over non-native modes with scaling on DSI then? > > Ville? > > I believe we are taking the panel fitter into account when we reconstruct the > BIOS fb. So in case the BIOS is upscaling we'll come up with some less than > screen size framebuffer. The fb helper will want to use the native mode of > the panel however, so at some point we'll drop the BIOS fb (as it's too small) > and allocate a new one. We are not sure how to take this further. Can you please provide some inputs for next steps on this? We will try to make the changes and test as per your inputs/feedback. > > -- > Ville Syrjälä > Intel OTC
diff --git a/drivers/gpu/drm/drm_fb_helper.c b/drivers/gpu/drm/drm_fb_helper.c index e934b54..6afc06f 100644 --- a/drivers/gpu/drm/drm_fb_helper.c +++ b/drivers/gpu/drm/drm_fb_helper.c @@ -1482,8 +1482,8 @@ static int drm_fb_helper_single_fb_probe(struct drm_fb_helper *fb_helper, memset(&sizes, 0, sizeof(struct drm_fb_helper_surface_size)); sizes.surface_depth = 24; sizes.surface_bpp = 32; - sizes.fb_width = (unsigned)-1; - sizes.fb_height = (unsigned)-1; + sizes.fb_width = 0; + sizes.fb_height = 0; /* if driver picks 8 or 16 by default use that for both depth/bpp */ @@ -1560,9 +1560,9 @@ static int drm_fb_helper_single_fb_probe(struct drm_fb_helper *fb_helper, } if (lasth) - sizes.fb_width = min_t(u32, desired_mode->hdisplay + x, sizes.fb_width); + sizes.fb_width = max_t(u32, desired_mode->hdisplay + x, sizes.fb_width); if (lastv) - sizes.fb_height = min_t(u32, desired_mode->vdisplay + y, sizes.fb_height); + sizes.fb_height = max_t(u32, desired_mode->vdisplay + y, sizes.fb_height); } if (crtc_count == 0 || sizes.fb_width == -1 || sizes.fb_height == -1) {