On Tue, Oct 19, 2021 at 06:51:14PM +1300, Avon Robertson wrote: > On Mon, Oct 18, 2021 at 03:52:34PM -0700, Chris Cappuccio wrote: > > Here's a thread where people are seeing similar hangs on similar hardware > > under Linux: > > > > https://bugs.freedesktop.org/show_bug.cgi?id=108900 > > > > "VERIFIED WONTFIX" because the kernel driver is probably closer to the > > issue, not Mesa > > > > here's another: https://bugzilla.kernel.org/show_bug.cgi?id=201957 > > > > The "fix" in the second one seems to be downgrading to an earlier firmware. > > > > Perhaps you can try to do that, the older radeondrm firmwares should be > > available from > > http://firmware.openbsd.org/firmware/6.8/radeondrm-firmware-20181218.tgz > > > > You can just unpack various versions and place the appropriate files under > > /etc/firmware for quick testing... > > > > I have no idea if the 20181218 is the right version to test, it may not be. > > It's just the first earlier version available on firmware.openbsd.org. You > > may need to grab a different version from a Linux repository. > > > > Chris > > > > Avon Robertson [avo...@xtra.co.nz] wrote: > > > > > > drm:pid70131:gmc_v8_0_process_interrupt *ERROR* GPU fault detected: 147 > > > 0x0582c802 for process pid 0 thread Xorg pid 9818 > > > drm:pid70131:gmc_v8_0_process_interrupt *ERROR* > > > VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x000086B0 > > > drm:pid70131:gmc_v8_0_process_interrupt *ERROR* > > > VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x070C8002 > > > drm:pid70131:gmc_v8_0_vm_decode_fault *ERROR* VM fault (0x02, vmid 3, > > > pasid 32769) at page 34480, write from 'TC2' (0x54433200) (200) > > > [drm] *ERROR* Illegal register access in command stream > > > [drm] *ERROR* ring gfx timeout, signaled seq=36730, emitted seq=36732 > > > [drm] *ERROR* Process information: process pid 0 thread Xorg pid 9818 > > > amdgpu_device_suspend_display_audio: stub > > > amdgpu: cp is busy, skip halt cp > > > amdgpu: rlc is busy, skip halt rlc > > > [drm] *ERROR* Failed to initialize parser -88! > > > <Removed 58 exact same lines as above and below here.> > > > [drm] *ERROR* Failed to initialize parser -88! > > > [drm] *ERROR* ring gfx timeout, signaled seq=36733, emitted seq=36733 > > > [drm] *ERROR* Process information: process pid 0 thread Xorg pid 9818 > > > amdgpu_device_suspend_display_audio: stub > > > [drm] *ERROR* Failed to initialize parser -88! > > > <Removed 1294 exact same lines as above and below here.> > > > [drm] *ERROR* Failed to initialize parser -88! > > > usb_insert_transfer: xfer=0xfffffd901e91d000 not free > > > ucomstart: err=INVAL > > > ucom1: read start failed > > > [drm] *ERROR* Failed to initialize parser -88! > > > <Removed 22 exact same lines as above and below here.> > > > [drm] *ERROR* Failed to initialize parser -88! > > > > > > > -- > > Thank you for the above reference information URLs Chris. I will see if > any fix they suggest fixes the freezes that I have been experiencing. > > -- aer
Greetings Chris and misc@, As suggested above by Chris? 1. Downloaded radeondrm-firmware-20181218.tgz to ~/download/. 2. # rm -fr /etc/firmware/amdgpu/* 3. # tar -C /etc -xzvf ~/download/radeondrm-firmware-20181218.tgz The machine has been running without freezing for almost 24 hours, so the radeon firmware may be a permanent fix. If the freezes occur again I will send another email to this thread. Is there some way to prevent the amdgpu firmware being reinstalled and or updated, during the reboot following a snapshot upgrade procedure?