drm/radeon: Set depth on low mem to 16 bpp instead of 8 bpp

Commit Message

Takashi Iwai July 18, 2017, 3:20 p.m. UTC

From: Egbert Eich <[e-mail]>

The radeon driver reduces the framebuffer resolution to 8bpp if a
device with less than 32MB VRAM is found.  This causes the framebuffer
to run in 8 bit paletted mode.  For a text console this is not an
issue as 256 different colors is more than one gets on a VGA text
console.  However this leads to a poor 8bit pseudo-color visual when
running X on fbdev, too, which is quite ugly.

In this patch, we try to give some moderate compromise: limit the
framebuffer bpp to 8 only when VRAM is 8MB or less, and use 16 bpp
otherwise for 32MB or less VRAM.

Signed-off-by: Egbert Eich <[e-mail]>
Signed-off-by: Takashi Iwai <[e-mail]>
---

This has been included in SUSE kernel for quite some time, and I'm
trying to upstream this.  The 16bpp looks more reasonable nowadays,
but I know this is a matter of taste, too.  It'd be great if you guys
can comment on this, at least, whether it's uttly non-sense or not.

thanks,

Takashi

 drivers/gpu/drm/radeon/radeon_fb.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

Comments

Michel Dänzer July 18, 2017, 8:53 p.m. UTC | #1

On 18/07/17 11:20 AM, Takashi Iwai wrote:
> From: Egbert Eich <[e-mail]>
> 
> The radeon driver reduces the framebuffer resolution to 8bpp if a
> device with less than 32MB VRAM is found.  This causes the framebuffer
> to run in 8 bit paletted mode.  For a text console this is not an
> issue as 256 different colors is more than one gets on a VGA text
> console.  However this leads to a poor 8bit pseudo-color visual when
> running X on fbdev, too, which is quite ugly.
> 
> In this patch, we try to give some moderate compromise: limit the
> framebuffer bpp to 8 only when VRAM is 8MB or less, and use 16 bpp
> otherwise for 32MB or less VRAM.
> 
> Signed-off-by: Egbert Eich <[e-mail]>
> Signed-off-by: Takashi Iwai <[e-mail]>
> ---
> 
> This has been included in SUSE kernel for quite some time, and I'm
> trying to upstream this.  The 16bpp looks more reasonable nowadays,
> but I know this is a matter of taste, too.  It'd be great if you guys
> can comment on this, at least, whether it's uttly non-sense or not.

[...]

> diff --git a/drivers/gpu/drm/radeon/radeon_fb.c b/drivers/gpu/drm/radeon/radeon_fb.c
> index 356ad90a5238..b5f2642f124b 100644
> --- a/drivers/gpu/drm/radeon/radeon_fb.c
> +++ b/drivers/gpu/drm/radeon/radeon_fb.c
> @@ -347,9 +347,12 @@ int radeon_fbdev_init(struct radeon_device *rdev)
>   	if (list_empty(&rdev->ddev->mode_config.connector_list))
>   		return 0;
>   
> -	/* select 8 bpp console on RN50 or 16MB cards */
> -	if (ASIC_IS_RN50(rdev) || rdev->mc.real_vram_size <= (32*1024*1024))
> +	/* select 8 bpp console on 8MB cards, or 16 bpp on RN50 or 32MB */
> +	if (rdev->mc.real_vram_size <= (8*1024*1024))
>   		bpp_sel = 8;
> +	else if (ASIC_IS_RN50(rdev) ||
> +		 rdev->mc.real_vram_size <= (32*1024*1024))
> +		bpp_sel = 16;
>   
>   	rfbdev = kzalloc(sizeof(struct radeon_fbdev), GFP_KERNEL);
>   	if (!rfbdev)
> 

Looks reasonable to me — if nothing else, at least the comment and code 
will match again! :)

My only potential doubt is whether there could be another reason than 
VRAM size why RN50 should use 8 bpp. Assuming not,

Reviewed-by: Michel Dänzer <[e-mail]>

Takashi Iwai July 19, 2017, 8:45 a.m. UTC | #2

On Tue, 18 Jul 2017 22:53:48 +0200,
Michel D4nzer wrote:
> 
> On 18/07/17 11:20 AM, Takashi Iwai wrote:
> > From: Egbert Eich <[e-mail]>
> >
> > The radeon driver reduces the framebuffer resolution to 8bpp if a
> > device with less than 32MB VRAM is found.  This causes the framebuffer
> > to run in 8 bit paletted mode.  For a text console this is not an
> > issue as 256 different colors is more than one gets on a VGA text
> > console.  However this leads to a poor 8bit pseudo-color visual when
> > running X on fbdev, too, which is quite ugly.
> >
> > In this patch, we try to give some moderate compromise: limit the
> > framebuffer bpp to 8 only when VRAM is 8MB or less, and use 16 bpp
> > otherwise for 32MB or less VRAM.
> >
> > Signed-off-by: Egbert Eich <[e-mail]>
> > Signed-off-by: Takashi Iwai <[e-mail]>
> > ---
> >
> > This has been included in SUSE kernel for quite some time, and I'm
> > trying to upstream this.  The 16bpp looks more reasonable nowadays,
> > but I know this is a matter of taste, too.  It'd be great if you guys
> > can comment on this, at least, whether it's uttly non-sense or not.
> 
> [...]
> 
> > diff --git a/drivers/gpu/drm/radeon/radeon_fb.c b/drivers/gpu/drm/radeon/radeon_fb.c
> > index 356ad90a5238..b5f2642f124b 100644
> > --- a/drivers/gpu/drm/radeon/radeon_fb.c
> > +++ b/drivers/gpu/drm/radeon/radeon_fb.c
> > @@ -347,9 +347,12 @@ int radeon_fbdev_init(struct radeon_device *rdev)
> >   	if (list_empty(&rdev->ddev->mode_config.connector_list))
> >   		return 0;
> >   -	/* select 8 bpp console on RN50 or 16MB cards */
> > -	if (ASIC_IS_RN50(rdev) || rdev->mc.real_vram_size <= (32*1024*1024))
> > +	/* select 8 bpp console on 8MB cards, or 16 bpp on RN50 or 32MB */
> > +	if (rdev->mc.real_vram_size <= (8*1024*1024))
> >   		bpp_sel = 8;
> > +	else if (ASIC_IS_RN50(rdev) ||
> > +		 rdev->mc.real_vram_size <= (32*1024*1024))
> > +		bpp_sel = 16;
> >     	rfbdev = kzalloc(sizeof(struct radeon_fbdev), GFP_KERNEL);
> >   	if (!rfbdev)
> >
> 
> Looks reasonable to me — if nothing else, at least the comment and
> code will match again! :)
> 
> My only potential doubt is whether there could be another reason than
> VRAM size why RN50 should use 8 bpp. Assuming not,

We also hope there no such popular board :)
FWIW, at least we didn't get a relevant bug report, so far.

> Reviewed-by: Michel Dänzer <[e-mail]>

Thanks!


Takashi

Alex Deucher July 24, 2017, 8:46 p.m. UTC | #3

On Wed, Jul 19, 2017 at 4:45 AM, Takashi Iwai <[e-mail]> wrote:
> On Tue, 18 Jul 2017 22:53:48 +0200,
> Michel D4nzer wrote:
>>
>> On 18/07/17 11:20 AM, Takashi Iwai wrote:
>> > From: Egbert Eich <[e-mail]>
>> >
>> > The radeon driver reduces the framebuffer resolution to 8bpp if a
>> > device with less than 32MB VRAM is found.  This causes the framebuffer
>> > to run in 8 bit paletted mode.  For a text console this is not an
>> > issue as 256 different colors is more than one gets on a VGA text
>> > console.  However this leads to a poor 8bit pseudo-color visual when
>> > running X on fbdev, too, which is quite ugly.
>> >
>> > In this patch, we try to give some moderate compromise: limit the
>> > framebuffer bpp to 8 only when VRAM is 8MB or less, and use 16 bpp
>> > otherwise for 32MB or less VRAM.
>> >
>> > Signed-off-by: Egbert Eich <[e-mail]>
>> > Signed-off-by: Takashi Iwai <[e-mail]>
>> > ---
>> >
>> > This has been included in SUSE kernel for quite some time, and I'm
>> > trying to upstream this.  The 16bpp looks more reasonable nowadays,
>> > but I know this is a matter of taste, too.  It'd be great if you guys
>> > can comment on this, at least, whether it's uttly non-sense or not.
>>
>> [...]
>>
>> > diff --git a/drivers/gpu/drm/radeon/radeon_fb.c b/drivers/gpu/drm/radeon/radeon_fb.c
>> > index 356ad90a5238..b5f2642f124b 100644
>> > --- a/drivers/gpu/drm/radeon/radeon_fb.c
>> > +++ b/drivers/gpu/drm/radeon/radeon_fb.c
>> > @@ -347,9 +347,12 @@ int radeon_fbdev_init(struct radeon_device *rdev)
>> >     if (list_empty(&rdev->ddev->mode_config.connector_list))
>> >             return 0;
>> >   - /* select 8 bpp console on RN50 or 16MB cards */
>> > -   if (ASIC_IS_RN50(rdev) || rdev->mc.real_vram_size <= (32*1024*1024))
>> > +   /* select 8 bpp console on 8MB cards, or 16 bpp on RN50 or 32MB */
>> > +   if (rdev->mc.real_vram_size <= (8*1024*1024))
>> >             bpp_sel = 8;
>> > +   else if (ASIC_IS_RN50(rdev) ||
>> > +            rdev->mc.real_vram_size <= (32*1024*1024))
>> > +           bpp_sel = 16;
>> >             rfbdev = kzalloc(sizeof(struct radeon_fbdev), GFP_KERNEL);
>> >     if (!rfbdev)
>> >
>>
>> Looks reasonable to me — if nothing else, at least the comment and
>> code will match again! :)
>>
>> My only potential doubt is whether there could be another reason than
>> VRAM size why RN50 should use 8 bpp. Assuming not,
>
> We also hope there no such popular board :)
> FWIW, at least we didn't get a relevant bug report, so far.
>
>> Reviewed-by: Michel Dänzer <[e-mail]>
>
> Thanks!

Applied.  thanks!

Alex


>
>
> Takashi
> _______________________________________________
> amd-gfx mailing list
> [e-mail]
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Patch

diff --git a/drivers/gpu/drm/radeon/radeon_fb.c b/drivers/gpu/drm/radeon/radeon_fb.c
index 356ad90a5238..b5f2642f124b 100644
--- a/drivers/gpu/drm/radeon/radeon_fb.c
+++ b/drivers/gpu/drm/radeon/radeon_fb.c
@@ -347,9 +347,12 @@  int radeon_fbdev_init(struct radeon_device *rdev)
 	if (list_empty(&rdev->ddev->mode_config.connector_list))
 		return 0;
 
-	/* select 8 bpp console on RN50 or 16MB cards */
-	if (ASIC_IS_RN50(rdev) || rdev->mc.real_vram_size <= (32*1024*1024))
+	/* select 8 bpp console on 8MB cards, or 16 bpp on RN50 or 32MB */
+	if (rdev->mc.real_vram_size <= (8*1024*1024))
 		bpp_sel = 8;
+	else if (ASIC_IS_RN50(rdev) ||
+		 rdev->mc.real_vram_size <= (32*1024*1024))
+		bpp_sel = 16;
 
 	rfbdev = kzalloc(sizeof(struct radeon_fbdev), GFP_KERNEL);
 	if (!rfbdev)