Re: My ongoing Radeon HD-5450 saga

[Andrew Randrianasulu <randrianasulu%gmail.com@localhost>]

On Sat, Feb 28, 2026 at 11:50 PM Rhialto <rhialto%falu.nl@localhost> wrote:
>
> Just reporting on the status of my ongoing Radeon HD-5450 saga with 10.1.
>
> After yet another graphics crash, I switched from Driver "modesetting"
> to Driver "radeon" (in xorg.conf).
>
> I had forgotten why I stopped using the radeon driver, but within 48
> hours I was reminded: this crashes even faster than the modesetting
> driver. The logged messages are very similar though:
>
> This is from the modesetting driver:
>
> Feb 25 22:15:46 murthe /netbsd: [ 1474856.8544041] radeon0: autoconfiguration error: error: ring 0 stalled for more than 10480msec
> Feb 25 22:15:46 murthe /netbsd: [ 1474856.8544041] radeon0: warn: GPU lockup (current fence id 0x0000000001424637 last fence id 0x0000000001424638 on ring 0)
> Feb 25 22:15:46 murthe /netbsd: [ 1474857.3543763] radeon0: autoconfiguration error: error: ring 0 stalled for more than 10980msec
> Feb 25 22:15:46 murthe /netbsd: [ 1474857.3543763] radeon0: warn: GPU lockup (current fence id 0x0000000001424637 last fence id 0x0000000001424638 on ring 0)
> Feb 25 22:15:47 murthe /netbsd: [ 1474857.8543477] radeon0: autoconfiguration error: error: ring 0 stalled for more than 11480msec
>
> ... etc etc ...
>
> and this from the radeon driver:
>
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5659760] radeon0: autoconfiguration error: error: ring 0 stalled for more than 10300msec
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5659760] radeon0: warn: GPU lockup (current fence id 0x000000000013585c last fence id 0x000000000013588a on ring 0)
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0: Saved 1458 dwords of commands on ring 0.
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0: GPU softreset: 0x00000019
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0: GPU softreset: 0x0000       = 0xA2733CA0
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   GRBM_STATUS_SE0           = 0x1C000007
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   GRBM_STATUS_SE1           = 0x00000007
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   SRBM_STATUS               = 0x200000C0
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   SRBM_STATUS2              = 0x00000000
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   R_008674_CP_STALLED_STAT1 = 0x01000000
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   SRBM8678_CP_STALLED_STAT2 = 0x000D1000
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   R_00867C_CP_BUSY_STAT     = 0x00068406
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   SRB08680_CP_STAT          = 0x80878647
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   R_00D034_DMA_STATUS_REG   = 0x44C83D57
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0: GRBM_SOFT_RESET=0x00007F6B
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0: SRBM_SOFT_RESET=0x00000100
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   GRBM_STATUS               = 0x00003828
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   GRBM_STATUS_SE0           = 0x00000007
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   GRBM_STATUS_SE0           = 0x00000007
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   SRBM_STATUS               = 0x200000C0
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   SRBM_STATUS2              = 0x00000000
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   R_008674_CP_STALLED_STAT1 = 0x00000000
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   R_008678_CP_STALLED_STAT2 = 0x00000000
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   R_00867C_CP_BUSY_STAT     = 0x00000000
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   R_008680_CP_STAT          = 0x00000000
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0:   R_00D034_DMA_STATUS_REG   = 0x44C83D57
> Feb 27 21:43:20 murthe /netbsd: [ 170573.5859730] radeon0: GPU reset succeeded, trying to resume
> Feb 27 21:43:20 murthe /netbsd: [ 170573.6259709] [drm] PCIE GART of 1024M enabled (table at 0x000000000014C000).
> Feb 27 21:43:20 murthe /netbsd: [ 170573.6259709] radeon0: WB enabled
> Feb 27 21:43:20 murthe /netbsd: [ 170573.6259709] [drm] PCIE GART of 1024M enabled (table addr 0x0000000040000c00 and cpu addr 0x0xffffc1a52e0f3c00
> Feb 27 21:43:20 murthe /netbsd: [ 170573.6259709] radeon0: fence driver on ring 3 use gpu addr 0x0000000040000c0c and cpu addr 0x0xffffc1a52e0f3c0c
> Feb 27 21:43:20 murthe /netbsd: [ 170573.6359703] radeon0: fence driver on ring 5 use gpu addr 0x000000000005c418 and cpu addr 0x0xffff9f926e3ec418
> Feb 27 21:43:32 murthe /netbsd: [ 170573.6459697] [drm] ring test on 0 succeeded in 1 usecs
> Feb 27 21:43:32 murthe /netbsd: [ 170573.6459697] [drm] ring test on 3 succeeded in 2 usecs
> Feb 27 21:43:32 murthe /netbsd: [ 170573.8359593] [drm] ring test on 5 succeeded in 1 usecs
> Feb 27 21:43:32 murthe /netbsd: [ 170573.8359593] [drm] UVD initialized successfully.
> Feb 27 21:43:32 murthe /netbsd: [ 170573.8659592] [drm] ib test on ring 0 succeeded in 0 usecs
> Feb 27 21:43:32 murthe /netbsd: [ 170573.8659592] [drm] ib test on ring 3 succeeded in 0 usecs
> Feb 27 21:43:32 murthe /netbsd: [ 170575.0258944] [drm] ib test on ring 5 succeeded
> Feb 27 21:43:32 murthe /netbsd: [ 170585.5553215] radeon0: autoconfiguration error: error: ring 0 stalled for more than 10430msec
> Feb 27 21:43:32 murthe /netbsd: [ 170573.6459697] [drm] ring test on 0 succeeded in 1 usecs[ 170586.0552945] radeon0: autoconfiguration error: error: ring 0 stalled for more than 10930msec
> Feb 27 21:43:32 murthe /netbsd: [ 170586.0552945] radeon0: warn: GPU lockup (current fence id 0x0000000000135a54 last fence id 0x0000000000135a5a on ring 0)
> Feb 27 21:43:33 murthe /netbsd: [ 170586.5552669] radeon0: autoconfiguration error: error: ring 0 stalled for more than 11430msec
> Feb 27 21:43:33 murthe /netbsd: [ 170586.5552669] radeon0: autoconfiguration error: error: ring 0 stalled for more than 1ce id 0x0000000000135a5a on ring 0)
> Feb 27 21:43:33 murthe /netbsd: [ 170587.0552396] radeon0: autoconfiguration error: error: ring 0 stalled for more than 11930msec
> Feb 27 21:43:33 murthe /netbsd: [ 170587.0552396] radeon0: autoconfiguration error: error: ring 0 stalled f35a54 last fence id 0x0000000000135a5a on ring 0)
> Feb 27 21:43:34 murthe /netbsd: [ 170587.5552125] radeon0: autoconfiguration error: error: ring 0 stalled for more than 12430msec
> Feb 27 21:43:34 murthe /netbsd: [ 170587.5552125] radeon0: autoconfiguration error: error: ring 0 stalled for more thafence id 0x0000000000135a5a on ring 0)
> Feb 27 21:43:34 murthe /netbsd: [ 170588.0551851] radeon0: autoconfiguration error: error: ring 0 stalled for more than 12930msec
> Feb 27 21:43:34 murthe /netbsd: [ 170588.0551851] radeon0: autoconfiguration error: error: ring 0 stalled 135a54 last fence id 0x0000000000135a5a on ring 0)
>
> I wonder why those very different drivers seem to run into the same
> problem. Could it be a hint to find the root cause?

well, hw 2d acceleration in modesetting case goes via 3D engine and
Glamor (X.org's acceleration architecture)) and EXA also uses 3D
engine...

May be something like Option "AccelMethod" "None" will stop
acceleration from happening?

On linux "man radeon" says:

 Option "AccelMethod" "string"
              Chooses between available acceleration architectures.
Valid values are EXA (for pre-TAHITI GPUs)
              and  glamor  (for R300 or higher). The default is glamor
with R600 or newer (with Xorg >= 1.18.3,
              otherwise with TAHITI or newer), otherwise EXA.

 Option "Accel" "boolean"
              Enables or disables all hardware acceleration.
              The default is on.




>
> So I plugged my monitor into the motherboard DVI port to use the onboard
> Intel graphics. This involved checking a mysterious BIOS setting (in a
> strange place like Advanced/System Agent (SA) Configuration/Graphics
> Configuration) but that's fine.
>
> Both the "intel" and the "modesetting" driver seemed to work in a quick
> test.
>
> After a slightly longer time, I noticed the following quirks:
>
> - In Firefox (not in other gtk3 programs that I tried) the menu
>   highlighting (and other kinds of pop-ups) looks like there is a cache
>   push missing (this is a known problem). (xcompmgr seems to work
>   around this, but causes other issues)
>
> - x11/redshift doesn't work. It uses xrandr to adjust screen brightness
>   and gamma. "xrandr --verbose" shows that it *thinks* it works, but it
>   has no actual effect on the screen. I don't know if this is a known
>   problem.
>
> (this is all still with 10.1, intel hardware, modesetting driver)
>
> -Olaf.
> --
> ___ Olaf 'Rhialto' Seibert                            <rhialto/at/falu.nl>
> \X/ There is no AI. There is just someone else's work.           --I. Rose