rpi5-eeprom support M.2 hat, coming soon…
I had some down time today and decided to build the -rc kernel packages for the pi5 and pushed them to the unstable branch when the mirrors sync. I have no way yet to test it so if some one feels adventurous enough to test please do and tell me how it turns out.
linux-rpi5-rc 6.7.rc2-1
linux-rpi5-rc-headers 6.7.rc2-1
Added 11-23-2023:
Test this kernel with today’s commit and see if you still see this. Following links here there are a lot of similarity’s with your DE and v4l drm_prime. Kernel packages pushed to the unstable branch when the mirrors sync.
drm: vc4: Block swiotlb bounce buffers being imported as dmabuf
linux-rpi4 6.1.63-2
linux-rpi4-headers 6.1.63-2
linux-rpi5 6.1.63-2
linux-rpi5-headers 6.1.63-2
Always the same message :
rpivid feb10000.codec: SPS changed
nov. 23 15:26:00 bloodmoon-pc kernel: rpivid feb10000.codec: PPS changed
nov. 23 15:26:00 bloodmoon-pc kernel: rpivid feb10000.codec: Missing DPB ent 0, timestamp=0
nov. 23 15:26:00 bloodmoon-pc kernel: rpivid feb10000.codec: Missing DPB ent 1, timestamp=0
nov. 23 15:26:00 bloodmoon-pc kernel: rpivid feb10000.codec: Missing DPB ent 2, timestamp=0
nov. 23 15:26:00 bloodmoon-pc kernel: rpivid feb10000.codec: Missing DPB ent 3, timestamp=0
nov. 23 15:26:03 bloodmoon-pc kernel: swiotlb_tbl_map_single: 63 callbacks suppressed
nov. 23 15:26:03 bloodmoon-pc kernel: vc4-drm gpu: swiotlb buffer is full (sz: 524288 bytes), total 32768 (slots), used 2349 (slots)
nov. 23 15:26:06 bloodmoon-pc kernel: rpivid feb10000.codec: PPS changed
nov. 23 15:26:06 bloodmoon-pc kernel: vc4-drm gpu: swiotlb buffer is full (sz: 294912 bytes), total 32768 (slots), used 3019 (slots)
nov. 23 15:26:18 bloodmoon-pc kernel: vc4-drm gpu: swiotlb buffer is full (sz: 524288 bytes), total 32768 (slots), used 1923 (slots)
God news : I have no more the first bug with
vc4-drm gpu: [drm] ERROR Failed to allocate DLIST entry: -28
And for freeze with no TTY (second bug)
gnome-shell[1146] :Page flip failed: drmModeAtomicCommit: No space available on the device
gnome-shell[1146] :Cursor update failed: drmModeAtomicCommit: No space available on the device
I have no problem with gnome xorg (tested one week) but only with wayland. Thus i m testing with the latest PR of mesa-git (26200.patch) .
Edit : crash with 26200 patch too…
And i have another type of crash. I think it’s more a mutter’s bug.
nov. 23 01:19:40 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (WW) modeset(0): Present-unflip: queue flip during flip on CRTC 3 failed: No space left on device
nov. 23 01:19:40 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (EE) modeset(0): failed to set mode: No space left on device
nov. 23 01:21:22 bloodmoon-pc firefox.desktop[42503]: Crash Annotation GraphicsCriticalError: |[C0][GFX1-]: Couldn't sanitize RENDERER device: V3D 4.2 (t=10328.2) |[C91][GFX1-]: Couldn't sa>
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "63"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "63"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (II) event4 - vc4-hdmi-0: device removed
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "54"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "54"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (II) event5 - vc4-hdmi-1: device removed
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "38"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "38"
nov. 23 01:16:46 bloodmoon-pc firefox.desktop[42503]: Crash Annotation GraphicsCriticalError: |[C0][GFX1-]: Couldn't sanitize RENDERER device: V3D 4.2 (t=10328.2) |[C91][GFX1-]: Couldn't sa>
nov. 23 01:19:40 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (WW) modeset(0): Present-unflip: queue flip during flip on CRTC 3 failed: No space left on device
nov. 23 01:19:40 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (EE) modeset(0): failed to set mode: No space left on device
nov. 23 01:21:22 bloodmoon-pc firefox.desktop[42503]: Crash Annotation GraphicsCriticalError: |[C0][GFX1-]: Couldn't sanitize RENDERER device: V3D 4.2 (t=10328.2) |[C91][GFX1-]: Couldn't sa>
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "63"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "63"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (II) event4 - vc4-hdmi-0: device removed
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "54"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "54"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (II) event5 - vc4-hdmi-1: device removed
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "38"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "38"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (II) event3 - Logitech Gaming Mouse G502 Keyboard: device removed
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "37"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (II) event0 - Logitech LogiG MKeyboard: device removed
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "36"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (II) event2 - Logitech Gaming Mouse G502: device removed
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "35"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (**) Option "fd" "35"
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (II) event1 - Logitech LogiG MKeyboard: device removed
nov. 23 01:21:25 bloodmoon-pc /usr/lib/gdm-x-session[1490]: (II) AIGLX: Suspending AIGLX clients for VT switch
That does not look right for the pi5 should be RENDERER device: V3D 7.1.
I have a Pi4. It seems that the only viable DE is xfce now. I don’t know is this is useful to compile gnome-shell or mutter with debug. The second bug is the more annoying.
I am keeping an eye on this new Issue. I am wondering if it is related.
Quote from within the thread:
I just confirmed that this is some sort of Wayland issue.
Well after 6 days of my pi5 lounging around in England it decided to move but in the opposite direction of the US. It is now “Chilling Out” in Germany for a while.
Pi5s make a stop and ate some turkey 
2 Likes
I have that with kernel 6.7rc, no sound.
pcm512x 1-004d: ASoC: error at snd_soc_pcm_component_pm_runtime_get on pcm512x.1-004d: -22
Is the dtoverlay for hifiberry sound card enabled ?
I’m testing to not use firefox/wayland and
MUTTER_DEBUG_DISABLE_HW_CURSORS=1
MUTTER_DEBUG_FORCE_KMS_MODE=simple
for this crash
gnome-shell[1146] :Page flip failed: drmModeAtomicCommit: No space available on the device
gnome-shell[1146] :Cursor update failed: drmModeAtomicCommit: No space available on the device
I have not see a report about vc4-drm gpu: swiotlb buffer is full when playing a video under gnome wayland ? That one is easy to trigger.
These latest kernel packages has been pushed to the unstable branch when the mirrors sync.
linux-rpi4-rc 6.7.rc3-1
linux-rpi4-rc-headers 6.7.rc3-1
linux-rpi5-rc 6.7.rc3-1
linux-rpi5-rc-headers 6.7.rc3-1
rpi-overlays 6.1.63-2
Update 11-30-23
These latest pi related kernel packages has been pushed to the unstable branch when the mirrors sync.
linux-rpi4 6.1.64-1
linux-rpi4-headers 6.1.64-1
linux-rpi4-mainline 6.6.3-1
linux-rpi4-mainline-headers 6.6.3-1
linux-rpi5 6.1.64-1
linux-rpi5-headers 6.1.64-1
6.6.3 scare to shutdown 
Generating module dependencies
==> Creating gzip-compressed initcpio image: ‘/boot/initramfs8’
==> WARNING: errors were encountered during the build. The image may not be complete.
The new mkinitcpio tries to add kms and vc4 twice. I have had no issues rebooting here.
there a commit comes, can you rebuild and remove scare warning.
I will pull it from the arch-arm server when they make the upgrade.
These kernel updates has been pushed to the unstable branch when the mirrors sync.
@tartanpion test and see if RPi’s revert they did for you fixes your sound issue.
linux-rpi4 6.1.64-2
linux-rpi4-headers 6.1.64-2
linux-rpi5 6.1.64-2
linux-rpi5-headers 6.1.64-2
Done. It’s OK now. When i have time i will post in https://github.com/raspberrypi/linux/issues/5674 because the problem is not closed in wayland.
Edit :
@Darksky When the next crash is going to be here, what can I try in a Terminal (with SSH in my phone) to help besides writing this ?
sudo cat /sys/kernel/debug/dri/1/hvs_dlist_allocs > crash.txt
Well I would say use the watch command to continuously write until it locks up but for some reason it gets garbage that is not seen in the terminal. Might use the watch command to monitor with out directing to a file until things lock up.
What I am seeing here writing to a file.
^[[?1049h^[[22;0;0t^[[1;24r^[(B^[[m^[[4l^[[?7h^[[H^[[2JAllocated nodes:
^[[2dnode [00000020 + 0000000b]
^[[3dnode [0000002b + 00000011]
^[[4dStale nodes:^[[24;80H^[[24;1H^[[?1049l^[[23;0;0t
^[[?1l^[>
I would say monitor with this until things lock up in a terminal. It will refresh every second. Even if you could write to a file when things lock up things cached may not get written to the file anyway if the system locks up.
Run the command as root:
watch -n 1 -t cat /sys/kernel/debug/dri/1/hvs_dlist_allocs
Allocated nodes:
node [00000020 + 0000000b]
node [00000034 + 00000011]
Stale nodes:
6.6, firefox crash randomly and sometime system halted.
back to 6.1 solved.
Can you test arch-arm’s new mkinitcpio and see if all is good there. I have disabled my plymouth stuff here. Reboot after installing to see if the spinning circle shows up.
sudo pacman -U http://fl.us.mirror.archlinuxarm.org/armv7h/core/mkinitcpio-37.1-1-any.pkg.tar.xz