amd-gfx.lists.freedesktop.org archive mirror
 help / color / mirror / Atom feed
From: Alex Deucher <alexdeucher@gmail.com>
To: "Christian König" <christian.koenig@amd.com>
Cc: "open list:EFIFB FRAMEBUFFER DRIVER"
	<linux-fbdev@vger.kernel.org>,
	"Christian König" <ckoenig.leichtzumerken@gmail.com>,
	"amd-gfx list" <amd-gfx@lists.freedesktop.org>,
	"Deucher, Alexander" <alexander.deucher@amd.com>,
	"Benjamin Herrenschmidt" <benh@kernel.crashing.org>,
	"Alberto Salvia Novella" <es20490446e@gmail.com>
Subject: Re: Interlaced resolutions hang the desktop
Date: Wed, 31 Mar 2021 14:01:46 -0400	[thread overview]
Message-ID: <CADnq5_PAYJs1GtnpaUuZZDho0Y04wDcCQ__JBOoDquOq6RQ4SA@mail.gmail.com> (raw)
In-Reply-To: <e9dd4112-c0d6-8081-8f4a-e857b3097d58@amd.com>

Does disabling pageflipping via the xorg.conf help?

Alex

On Wed, Mar 31, 2021 at 1:40 PM Christian König
<christian.koenig@amd.com> wrote:
>
> Yeah, agree that must be some kind of bug in the upper layer of the stack.
>
> Most likely some userspace component is not handling the specialties of interlacing correctly (different vblank timing every other frame).
>
> It probably only works on Intel and after restarting the display manager by coincident.
>
> Sorry, but as I said this is a use case which basically nobody is using any more and because of this the different parts of the stack are not tested well enough for this.
>
> Christian.
>
> Am 31.03.21 um 17:47 schrieb Alberto Salvia Novella:
>
> Restarting the display manager service works and, more importantly, makes the bug no longer reproducible.
>
> Restarting the window manager doesn't work.
>
> Changing display manager makes the bug still reproducible.
>
> Maybe this is due to xorg-server, isn't it?
>
> On Wed, 31 Mar 2021 at 16:55, Christian König <ckoenig.leichtzumerken@gmail.com> wrote:
>>
>> Well the hardware is working fine as far as I can see.
>>
>> Can you try to kill the X server over SSH and see if you then get some screen update?
>>
>> Regards,
>> Christian.
>>
>> Am 31.03.21 um 16:52 schrieb Alberto Salvia Novella:
>>
>> Output.
>>
>> On Wed, 31 Mar 2021 at 16:36, Christian König <christian.koenig@amd.com> wrote:
>>>
>>> Mhm strange.
>>>
>>> Can you get me the output of "sudo cat /sys/kernel/debug/dri/0/radeon_fence_info" when the problem happens?
>>>
>>> Thanks,
>>> Christian.
>>>
>>> Am 31.03.21 um 16:33 schrieb Alberto Salvia Novella:
>>>
>>> - The computer still replies to ping.
>>> - The journal shows no errors, but a few warnings.
>>> - The mouse doesn't freeze.
>>>
>>> On Wed, 31 Mar 2021 at 10:09, Christian König <christian.koenig@amd.com> wrote:
>>>>
>>>> Can you access the system over the network and see if there is anything in the system log?
>>>>
>>>> It sounds like the display stack has crashed, but when the sound keeps playing the system is most likely still responsive over network.
>>>>
>>>> Thanks,
>>>> Christian.
>>>>
>>>> Am 31.03.21 um 10:05 schrieb Alberto Salvia Novella:
>>>>
>>>> What happens is this simple: after a few minutes, about 6 or so, the entire content of the screen stays still. In some minor situations only the applications panel of KDE Plasma.
>>>>
>>>> If music is playing it continues playing, so only graphics are hung. Yet in most cases the power button won't shut down the computer, as it usually does.
>>>>
>>>> At least this is the case using kwin on x11, and not on wayland. It only happens on "radeon" and not on Intel or "radeonhd".
>>>>
>>>> On Wed, 31 Mar 2021 at 09:48, Christian König <christian.koenig@amd.com> wrote:
>>>>>
>>>>> Correct, but a TV is intended for videos only. That's why it implements only the lower HDMI standard.
>>>>>
>>>>> Interlaced transmits only halve the lines with each frame, so a 60Hz mode effectively either becomes a 30Hz mode, halving the vertical resolution or adaptive motion compensated which the know visual artifacts. Depending on what the deinterlacing setting on your TV is.
>>>>>
>>>>> You could just add a progressive 1920x540@60 or 1920x1080@30 mode manually and would have the same effect with probably better quality. See https://de.wikipedia.org/wiki/Deinterlacing for reference.
>>>>>
>>>>> If you can give us some more information what is happening when the system freeze we could try to narrow this down, but we can't spend much time on a very specific use case in a driver which is in maintenance mode.
>>>>>
>>>>> Regards,
>>>>> Christian.
>>>>>
>>>>> Am 31.03.21 um 09:21 schrieb Alberto Salvia Novella:
>>>>>
>>>>> 24fps is intended for video only. Anything interactive at 24fps, as just moving the mouse around, is extremely choppy.
>>>>>
>>>>> No way anyone would prefer that over an interlaced resolution or a lower resolution. That is, by far, the worst option.
>>>>>
>>>>> Just try it on your screen, set it to 24Hz or alike, and tell me your experience. You can't even tell where the mouse is going to go.
>>>>>
>>>>> On Wed, 31 Mar 2021 at 08:44, Christian König <christian.koenig@amd.com> wrote:
>>>>>>
>>>>>> Hi Alberto,
>>>>>>
>>>>>> well a frame rate of 24Hz is perfectly reasonable for a TV and desktop usage.
>>>>>>
>>>>>> This is probably caused by the TVs limited HDMI bandwidth and a refresh rate of 30/25 Hz for the interlaced mode isn't much better either.
>>>>>>
>>>>>> Regards,
>>>>>> Christian.
>>>>>>
>>>>>> Am 30.03.21 um 22:59 schrieb Alberto Salvia Novella:
>>>>>>
>>>>>> The frame-rate at 24Hz is extremely poor for normal desktop usage.
>>>>>>
>>>>>> If the highest resolution, aka 1080p, uses that refresh rate then the desktop will default to that frame-rate.
>>>>>>
>>>>>> Other progressive modes don't exhibit any issue.
>>>>>>
>>>>>> On Tue, 30 Mar 2021 at 18:26, Christian König <christian.koenig@amd.com> wrote:
>>>>>>>
>>>>>>> Hi Alberto,
>>>>>>>
>>>>>>> I think the driver should only support resolutions that are progressive, but also at least of 50Hz.
>>>>>>>
>>>>>>>
>>>>>>> Why do you think so?, the 24Hz resolution seems to be the native one of the display.
>>>>>>>
>>>>>>> Regards,
>>>>>>> Christian.
>>>>>>>
>>>>>>> Am 30.03.21 um 17:37 schrieb Alberto Salvia Novella:
>>>>>>>
>>>>>>> This is why I'm using interlaced:
>>>>>>>
>>>>>>> $ xrandr
>>>>>>> Screen 0: minimum 320 x 200, current 1920 x 1080, maximum 8192 x 8192
>>>>>>> DisplayPort-0 disconnected (normal left inverted right x axis y axis)
>>>>>>> HDMI-0 connected primary 1920x1080+0+0 (normal left inverted right x axis y axis) 16mm x 9mm
>>>>>>>    1920x1080i    60.00*+  50.00    59.94
>>>>>>>    1920x1080     24.00    23.98
>>>>>>>    1280x720      60.00    50.00    59.94
>>>>>>>    1024x768      75.03    70.07    60.00
>>>>>>>    832x624       74.55
>>>>>>>    800x600       72.19    75.00    60.32    56.25
>>>>>>>    720x576       50.00
>>>>>>>    720x576i      50.00
>>>>>>>    720x480       60.00    59.94
>>>>>>>    720x480i      60.00    59.94
>>>>>>>    640x480       75.00    72.81    66.67    60.00    59.94
>>>>>>>    720x400       70.08
>>>>>>> DVI-0 disconnected (normal left inverted right x axis y axis)
>>>>>>>
>>>>>>> I think the driver should only support resolutions that are progressive, but also at least of 50Hz.
>>>>>>>
>>>>>>> On Tue, 30 Mar 2021 at 15:41, Christian König <ckoenig.leichtzumerken@gmail.com> wrote:
>>>>>>>>
>>>>>>>> Mhm, no idea why an interlaced resolution would cause a crash. Maybe some miscalculation in the display code.
>>>>>>>>
>>>>>>>> But apart from that if you just connected your PC to a TV I also wouldn't recommend using an interlaced resolution in the first place.
>>>>>>>>
>>>>>>>> See those resolutions only exists for backward compatibility with analog hardware.
>>>>>>>>
>>>>>>>> I think we would just disable those modes instead of searching for the bug.
>>>>>>>>
>>>>>>>> Regards,
>>>>>>>> Christian.
>>>>>>>>
>>>>>>>> Am 30.03.21 um 11:07 schrieb Alberto Salvia Novella:
>>>>>>>>
>>>>>>>> I guessed so.
>>>>>>>>
>>>>>>>> The GPU is a Radeon HD5870, and the screen is an old Telefunken TV (TLFK22LEDPVR1).
>>>>>>>>
>>>>>>>> Since my real display got into repair I used this TV meanwhile, and to my surprise it froze the system.
>>>>>>>>
>>>>>>>> On Tue, 30 Mar 2021 at 10:15, Christian König <christian.koenig@amd.com> wrote:
>>>>>>>>>
>>>>>>>>> Hi Alberto,
>>>>>>>>>
>>>>>>>>> well what hardware do you have?
>>>>>>>>>
>>>>>>>>> Interlaced resolutions are not used any more on modern hardware, so they
>>>>>>>>> are not well tested.
>>>>>>>>>
>>>>>>>>> Regards,
>>>>>>>>> Christian.
>>>>>>>>>
>>>>>>>>> Am 30.03.21 um 10:04 schrieb Alberto Salvia Novella:
>>>>>>>>> > The entire desktop hangs after some minutes when using the module
>>>>>>>>> > "radeon" with an interlaced resolution.
>>>>>>>>> >
>>>>>>>>> > Easier to trigger by playing a video on Firefox, at least on kwin_x11.
>>>>>>>>> > Wayland didn't exhibit the problem.
>>>>>>>>> >
>>>>>>>>> > Other display drivers, from different computers I have tried, didn't
>>>>>>>>> > allow those interlaced resolutions all together. It seems they know
>>>>>>>>> > there will be problems.
>>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> amd-gfx mailing list
>>>>>>>> amd-gfx@lists.freedesktop.org
>>>>>>>> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

  reply	other threads:[~2021-03-31 18:01 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-03-30  8:04 Interlaced resolutions hang the desktop Alberto Salvia Novella
2021-03-30  8:15 ` Christian König
2021-03-30  9:07   ` Alberto Salvia Novella
2021-03-30 13:41     ` Christian König
2021-03-30 15:37       ` Alberto Salvia Novella
2021-03-30 16:08         ` Alex Deucher
2021-03-30 16:26         ` Christian König
2021-03-30 20:59           ` Alberto Salvia Novella
2021-03-31  6:44             ` Christian König
2021-03-31  7:21               ` Alberto Salvia Novella
2021-03-31  7:48                 ` Christian König
2021-03-31  8:05                   ` Alberto Salvia Novella
2021-03-31  8:09                     ` Christian König
2021-03-31 14:33                       ` Alberto Salvia Novella
2021-03-31 14:36                         ` Christian König
2021-03-31 14:52                           ` Alberto Salvia Novella
2021-03-31 14:55                             ` Christian König
2021-03-31 15:47                               ` Alberto Salvia Novella
2021-03-31 17:39                                 ` Christian König
2021-03-31 18:01                                   ` Alex Deucher [this message]
2021-04-01 15:13                                     ` Alberto Salvia Novella

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=CADnq5_PAYJs1GtnpaUuZZDho0Y04wDcCQ__JBOoDquOq6RQ4SA@mail.gmail.com \
    --to=alexdeucher@gmail.com \
    --cc=alexander.deucher@amd.com \
    --cc=amd-gfx@lists.freedesktop.org \
    --cc=benh@kernel.crashing.org \
    --cc=christian.koenig@amd.com \
    --cc=ckoenig.leichtzumerken@gmail.com \
    --cc=es20490446e@gmail.com \
    --cc=linux-fbdev@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).