All of lore.kernel.org
 help / color / mirror / Atom feed
* radeon framebuffer tiling
@ 2016-01-04 17:57 Jay Cornwall
  2016-01-04 18:14 ` Alex Deucher
  0 siblings, 1 reply; 3+ messages in thread
From: Jay Cornwall @ 2016-01-04 17:57 UTC (permalink / raw)
  To: dri-devel

Hi,

dce4_crtc_do_set_base is the only location in which 
EVERGREEN_GRPH_CONTROL is updated. This currently configures a non-tiled 
framebuffer on my SI system.

This design prevents drmModePageFlip from handling a framebuffer with a 
different tiling configuration.

Is this the intended architecture? i.e. All surfaces provided to 
drmModePageFlip should match the initial framebuffer layout, or should 
EVERGREEN_GRPH_CONTROL be updated on every page flip?

-- 
Jay Cornwall
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: radeon framebuffer tiling
  2016-01-04 17:57 radeon framebuffer tiling Jay Cornwall
@ 2016-01-04 18:14 ` Alex Deucher
  2016-01-04 20:03   ` Jay Cornwall
  0 siblings, 1 reply; 3+ messages in thread
From: Alex Deucher @ 2016-01-04 18:14 UTC (permalink / raw)
  To: Jay Cornwall; +Cc: Maling list - DRI developers

On Mon, Jan 4, 2016 at 12:57 PM, Jay Cornwall <jay@jcornwall.me> wrote:
> Hi,
>
> dce4_crtc_do_set_base is the only location in which EVERGREEN_GRPH_CONTROL
> is updated. This currently configures a non-tiled framebuffer on my SI
> system.
>
> This design prevents drmModePageFlip from handling a framebuffer with a
> different tiling configuration.
>
> Is this the intended architecture? i.e. All surfaces provided to
> drmModePageFlip should match the initial framebuffer layout, or should
> EVERGREEN_GRPH_CONTROL be updated on every page flip?

Pageflipping is only available between surfaces of the same format
(i.e., we only update the base address, not the other parameters).

Alex
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: radeon framebuffer tiling
  2016-01-04 18:14 ` Alex Deucher
@ 2016-01-04 20:03   ` Jay Cornwall
  0 siblings, 0 replies; 3+ messages in thread
From: Jay Cornwall @ 2016-01-04 20:03 UTC (permalink / raw)
  To: Alex Deucher; +Cc: Maling list - DRI developers

On 2016-01-04 12:14, Alex Deucher wrote:
> On Mon, Jan 4, 2016 at 12:57 PM, Jay Cornwall <jay@jcornwall.me> wrote:
>> Hi,
>> 
>> dce4_crtc_do_set_base is the only location in which 
>> EVERGREEN_GRPH_CONTROL
>> is updated. This currently configures a non-tiled framebuffer on my SI
>> system.
>> 
>> This design prevents drmModePageFlip from handling a framebuffer with 
>> a
>> different tiling configuration.
>> 
>> Is this the intended architecture? i.e. All surfaces provided to
>> drmModePageFlip should match the initial framebuffer layout, or should
>> EVERGREEN_GRPH_CONTROL be updated on every page flip?
> 
> Pageflipping is only available between surfaces of the same format
> (i.e., we only update the base address, not the other parameters).

Thanks, I got it now. KWin/Wayland is only calling drmModeSetCrtc when 
the buffer stride changes, but flipping once from a dumb buffer to a GBM 
buffer.

I've filed: https://bugs.kde.org/show_bug.cgi?id=357543

-- 
Jay Cornwall
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2016-01-04 20:04 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-01-04 17:57 radeon framebuffer tiling Jay Cornwall
2016-01-04 18:14 ` Alex Deucher
2016-01-04 20:03   ` Jay Cornwall

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.