Ryzen 7 7700 iGPU support

Would depend on which release / stable branch you use.

On most recent latest (aka main) branch of ports tree, graphics/drm-612-kmod is added and made default for releng/15.1 (upcoming 15.1-RELEASE), stable/15 and main (aka CURRENT) of src (base).
(This is done by using OSVERSION >= 1500509 for 6.12, >= 1500031 for 6.6 and other supported branches/archs to be 6.1.)

So if you're using 14.4 or 14.3, graphics/drm-61-kmod would be chosen and this is known NOT supporting Ryzen 7 7700 iGPU.

If you're using 15.0, graphics/drm-66-kmod would be chosen and this states supports for Ryzen 7 7700 iGPU. But cannot find good reports on searching Internet.

If you're using stable/15, 15.1-{RELEASE|BETA*|RC*} or main, latest branch of ports would prefer newly added graphics/drm-612-kmod and 6.12 has "partially" good report.
But if you're using quarterly (currently 2026Q2) branch of ports/pkgs, newest version of the driver is at 6.9 on graphics/drm-latest-kmod, and cannot find good reports on searching Internet.

Anyway, at this moment [May 24, 2026 at 10:28 JST (+9)], I cannot find pkg for graphics/drm-612-kmod both at FreeBSD-ports repo and FreeBSD-ports-kmod repo for stable/15. So you'll need to build from ports locally if you're on 15.1-{BETA*|RC*}, stable/15 or main and latest branch of ports tree.
 
ArgentoSoma said:
But truth be told, amdgpu is well known on FreeBSD for its strange glitches—something we have been suffering from since FreeBSD 13.
Click to expand...
Hm, I'm not sure about amdgpu driver issues, as I'm using NVIDIA dGPUs only (and maintaining FreeBSD driver ports for NVIDIA GPUs).

For NVIDIA GPUs, Blackwell generation of architectures (RTX 5xxx series) still seem to have rough edges, but what I have in hand are RTX A400 (older, Ampere generation of architecture) and Quadro P1000 (notebook, Pascal generation of architecture supported by -580 variant) only.
 
ArgentoSoma said:
graphics/drm-latest-kmod (6.9) is available on FreeBSD15, this DRM-KMOD work with this Radeon card.

But truth be told, amdgpu is well known on FreeBSD for its strange glitches—something we have been suffering from since FreeBSD 13.
Click to expand...
Well, the crashes and drm soft panics also happen on Linux.
The main difference is that the firmware packages get updated more frequently and most distros are on 6.18+.
Dumbbell who is porting drm will sometimes backport fixes for amdgpu and intel, so the stability of the drm code should improve with the 612 release.
But there is no new PR that I know of for upgrading the firmware packages (january 2025) and the latest ported (august 2025) branch has issues with some new boards. We probably want a new firmware package from march 2026 or later.

Nvidia suffers less from firmware issues because they pack it with the drivers of I'm not mistaken.
 
fahrenheit said:
Nvidia suffers less from firmware issues because they pack it with the drivers of I'm not mistaken.
Click to expand...
True and false.

Codes for GSP (GPU System Processor), that is implemented in Turing generation of architectures and later, are implemented as separate firmwares (nvidia_gsp_tu10x_fw.ko for Turing and nvidia_gsp_ga10x_fw.ko for later architectures).

As most of the codes to manipulate GPUs are included in nvidia.ko (thus, so huge!), GSP firmware is "basically" optional that can run as "Graphics" Processing Unit without it.

But GPUs with Blackwell (RTX 5xxx) generation of architecture (and maybe later, too) are known to require it even on this "basic" use-cases.
 
fahrenheit said:
Well, the crashes and drm soft panics also happen on Linux.
The main difference is that the firmware packages get updated more frequently and most distros are on 6.18+.
Dumbbell who is porting drm will sometimes backport fixes for amdgpu and intel, so the stability of the drm code should improve with the 612 release.
But there is no new PR that I know of for upgrading the firmware packages (january 2025) and the latest ported (august 2025) branch has issues with some new boards. We probably want a new firmware package from march 2026 or later.

Nvidia suffers less from firmware issues because they pack it with the drivers of I'm not mistaken.
Click to expand...
Old user of RX 580 and RX 6600...amdgpu on FreeBSD 13 to 14 always panic on my PC in both graphics, the drm-kmod GitHub is full of issues (include mine).

This is not a trivial issue; it is a problem that affects a significant number of users and is related to how LinuxKPI operates on certain hardware configurations, specifically regarding ACPI and MSI interrupts.

Let's hope this is resolved in FreeBSD 16, provided the plan to merge DRM into the base system comes to fruition.

This is particularly important because it would resolve these kernel panics and enable Intel Arc dGPUs to function correctly (as things stand, my A380 triggers a panic the moment the i915kms module begins loading).
 
ArgentoSoma said:
graphics/drm-latest-kmod (6.9) is available on FreeBSD15, this DRM-KMOD work with this Radeon card.

But truth be told, amdgpu is well known on FreeBSD for its strange glitches—something we have been suffering from since FreeBSD 13.
Click to expand...
Truth be told. I have Ryzen 7 7700 iGPU and RX 9070, and I couldn't make neither work unfortunately. 7700 gives me the same panic with drm-kmod 6.6/6.9 and I tried installing 15.1 RC1, I even managed to figure out, download some missing firmware from the linux repo but unfortunately I get "panic : Unregistered use of FPU in kernel" and other errors.

(sorry for screenshots, I deleted the system since it seems RX 9070 should work when 15.1 is out and I'm wasting time, I would be glad to be proven wrong though

Screenshot 1 iGPU
IMG_4458.jpg

Screenshot 2 RX 9070
IMG_4468.jpg
 
I can confirm that physically removing the cables from my RX 9070 made my iGPU work, it can output 4K@120hz. Then after confirming that RX 9070 being detected is the problem, I managed to make it work without unplugging in with:

```
vmm_load="YES"
ppt_load="YES"
pptdevs="3/0/0 3/0/1"
```

Tested on FreeBSD 15.1 RC1 with drm-612-kmod
1780246061867.webp
 
DoctorRainer As others said before: The problem is the RX 9070. It is to new. Have a look at https://forums.freebsd.org/threads/amd-gpu-support-matrix.102822/.


As you find out, your iGPU works. I have the successor AMD 9700X running only with iGPU on FreeBSD 15.0 p9. So I would expect the AMD 7700 iGPU to work fine under FreeBSD 15.0.

My iGPU is:
sh: 
dvl@mybsd:~ % doas pciconf -lvc 2>&1| grep vga                       
Password:
vgapci0@pci0:13:0:0:    class=0x030000 rev=0xc5 hdr=0x00 vendor=0x1002 device=0x13c0 subvendor=0x1849 subdevice=0x364e
dvl@mybsd:~ % doas pciconf -lvc vgapci0@pci0:13:0:0 2>&1 | grep device
Password:
vgapci0@pci0:13:0:0:    class=0x030000 rev=0xc5 hdr=0x00 vendor=0x1002 device=0x13c0 subvendor=0x1849 subdevice=0x364e
    device     = 'Granite Ridge [Radeon Graphics]'

Maybe you could try the following.

Update your firmware and drivers:
sh: 
fwget -v

Check, if you have latest microcode installed. If not:
Code: 
pkg info cpu-microcode

And don't forget in /boot/loader.conf:
Code: 
# enable cpu microcode updates as early as possible; patching is done by cpucontrol(8) triggered in /etc/sysrc
# loads /usr/local/share/cpucontrol/amd-ucode.bin (contains all CPU families) early for boot-CPU
cpu_microcode_load="YES"
cpu_microcode_name="/boot/firmware/amd-ucode.bin"

I have a possibly redundant setting in /etc/rc.conf:
Code: 
# additional setting to /boot/loader.conf
# this one performs patch /usr/local/share/cpucontrol/microcode_amd_fam1ah.bin to all cores of CPU with cpucontrol(8)
microcode_update_enable="YES"

My system crashed from time to time. For a stable system I had to add some parameters in /boot/loader.conf:
Code: 
 ============  AMDGPU stability
amdgpu.dc="1" # enable display core - maybe redundand like hw.pci.enable_msix="1"
hw.amdgpu.ppfeaturemask="0xffff7fff" # disable SMU Power Features which triggers freezes on RDNA3/3.5
hw.amdgpu.gpu_recovery="1" # soft recovery instead of freezes
hw.amdgpu.lockup_timeout="10000" # higher timeout to prevent false positives on heavy load

# ============  PCI / interrupt stability (check by 'vmstat -i': irqXXX is INTx)
# PCI Interrupt generations
# INTx: legacy (bad)
# MSI: better
# MSI-X: modern (best)
hw.pci.enable_msix="1" # maybe redundand like amdgpu.dc="1"

# preemtive scheduler latency: lower means faster context switch. default = 40 # 0010.1000 PREEMPTION
kern.sched.preempt_thresh=224 # FULL_PREEMPTION = 1110.0000

I'm eagerly waiting for FreeBSD 15.1 before I buy a RX 9060 to continue my AI experiments with llama-cpp...
 
kirsche said:
DoctorRainer As others said before: The problem is the RX 9070. It is to new. Have a look at https://forums.freebsd.org/threads/amd-gpu-support-matrix.102822/.


As you find out, your iGPU works. I have the successor AMD 9700X running only with iGPU on FreeBSD 15.0 p9. So I would expect the AMD 7700 iGPU to work fine under FreeBSD 15.0.

My iGPU is:
sh: 
dvl@mybsd:~ % doas pciconf -lvc 2>&1| grep vga                      
Password:
vgapci0@pci0:13:0:0:    class=0x030000 rev=0xc5 hdr=0x00 vendor=0x1002 device=0x13c0 subvendor=0x1849 subdevice=0x364e
dvl@mybsd:~ % doas pciconf -lvc vgapci0@pci0:13:0:0 2>&1 | grep device
Password:
vgapci0@pci0:13:0:0:    class=0x030000 rev=0xc5 hdr=0x00 vendor=0x1002 device=0x13c0 subvendor=0x1849 subdevice=0x364e
    device     = 'Granite Ridge [Radeon Graphics]'

Maybe you could try the following.

Update your firmware and drivers:
sh: 
fwget -v

Check, if you have latest microcode installed. If not:
Code: 
pkg info cpu-microcode

And don't forget in /boot/loader.conf:
Code: 
# enable cpu microcode updates as early as possible; patching is done by cpucontrol(8) triggered in /etc/sysrc
# loads /usr/local/share/cpucontrol/amd-ucode.bin (contains all CPU families) early for boot-CPU
cpu_microcode_load="YES"
cpu_microcode_name="/boot/firmware/amd-ucode.bin"

I have a possibly redundant setting in /etc/rc.conf:
Code: 
# additional setting to /boot/loader.conf
# this one performs patch /usr/local/share/cpucontrol/microcode_amd_fam1ah.bin to all cores of CPU with cpucontrol(8)
microcode_update_enable="YES"

My system crashed from time to time. For a stable system I had to add some parameters in /boot/loader.conf:
Code: 
 ============  AMDGPU stability
amdgpu.dc="1" # enable display core - maybe redundand like hw.pci.enable_msix="1"
hw.amdgpu.ppfeaturemask="0xffff7fff" # disable SMU Power Features which triggers freezes on RDNA3/3.5
hw.amdgpu.gpu_recovery="1" # soft recovery instead of freezes
hw.amdgpu.lockup_timeout="10000" # higher timeout to prevent false positives on heavy load

# ============  PCI / interrupt stability (check by 'vmstat -i': irqXXX is INTx)
# PCI Interrupt generations
# INTx: legacy (bad)
# MSI: better
# MSI-X: modern (best)
hw.pci.enable_msix="1" # maybe redundand like amdgpu.dc="1"

# preemtive scheduler latency: lower means faster context switch. default = 40 # 0010.1000 PREEMPTION
kern.sched.preempt_thresh=224 # FULL_PREEMPTION = 1110.0000

I'm eagerly waiting for FreeBSD 15.1 before I buy a RX 9060 to continue my AI experiments with llama-cpp...
Click to expand...
Thanks :)

Everything seems to work stable without any additional adjustments you mentioned. Well, maybe because I'm on FreeBSD 15.1 RC2 with drm-612-kmod, I'm not sure.

It works pretty well and smooth, I use ly+plasma(wayland), sddm sadly doesn't work but ly actually looks neat, especially with `allscreens_flags="-f terminus-b32"` in /etc/rc.conf
 
DoctorRainer said:
I can confirm that physically removing the cables from my RX 9070 made my iGPU work, it can output 4K@120hz. Then after confirming that RX 9070 being detected is the problem, I managed to make it work without unplugging in with:

```
vmm_load="YES"
ppt_load="YES"
pptdevs="3/0/0 3/0/1"
```

Tested on FreeBSD 15.1 RC1 with drm-612-kmod
View attachment 26346
Click to expand...
Could you list the output of ls -la /dev/dri/card*?
I believe there is card0, which points to GPU 0, and card1, which points to GPU 1.
Try swapping them so that card0 → 1 and card1 → 0.I used this trick yesterday to get Wayland running on the iGPU without the dGPU.
However, some Rust applications require additional environment variables to prevent them from loading on the dGPU.

Remember that FreeBSD versions prior to 16 have issues with Resizable BAR, and some releases do not support it at all. Make sure to verify whether it is enabled or disabled during testing (host and/or bhyve).
 
I tried using my RX 6400 the other day and it kept having a kernel panic. It works on 14. I didn't know there was an issue. I am getting ready to start a new build based on the HP85 layout using the RX 6400. Hopefully this issue will be fixed soon. :/
 
akil said:
Could you list the output of ls -la /dev/dri/card*?
I believe there is card0, which points to GPU 0, and card1, which points to GPU 1.
Try swapping them so that card0 → 1 and card1 → 0.I used this trick yesterday to get Wayland running on the iGPU without the dGPU.
However, some Rust applications require additional environment variables to prevent them from loading on the dGPU.

Remember that FreeBSD versions prior to 16 have issues with Resizable BAR, and some releases do not support it at all. Make sure to verify whether it is enabled or disabled during testing (host and/or bhyve).
Click to expand...
Well, I mean if:
Code: 
vmm_load="YES"
ppt_load="YES"
pptdevs="3/0/0 3/0/1"

Works for me, I don't see a point in trying to fix what is already fixed.

> Remember that FreeBSD versions prior to 16 have issues with Resizable BAR, and some releases do not support it at all. Make sure to verify whether it is enabled or disabled during testing (host and/or bhyve).

I don't use bhyve actually, heh. But the main system works perfect like this, I had 0 problems so far, the system is SUPER stable, I have yet anything crashing on me EXCEPT when the system goes to sleep, so I disabled sleep......
 
LibreQuest Could be wrong drivers or firmware. Did you ever run fwget -v? Maybe this article of vermaden helps: https://vermaden.wordpress.com/2025/10/13/new-fwget-freebsd-firmware-tool/

My PC crashed because of the wrong firmware and drivers. With my tuning (see my last post) the crashes recover automatically:
Code: 
Jun  2 17:21:19 mybsd kernel: drmn0: [gfxhub] page fault (src_id:0 ring:24 vmid:4 pasid:32780)
Jun  2 17:21:19 mybsd kernel: drmn0:  in process  pid 120475 thread  pid 120475
Jun  2 17:21:19 mybsd kernel: drmn0:   in page starting at address 0x000003c03f7ff000 from client 0x1b (UTCL2)
Jun  2 17:21:19 mybsd kernel: drmn0: GCVM_L2_PROTECTION_FAULT_STATUS:0x00401430
Jun  2 17:21:19 mybsd kernel: drmn0:     Faulty UTCL2 client ID: SQC (data) (0xa)
Jun  2 17:21:19 mybsd kernel: drmn0:     MORE_FAULTS: 0x0
Jun  2 17:21:19 mybsd kernel: drmn0:     WALKER_ERROR: 0x0
Jun  2 17:21:19 mybsd kernel: drmn0:     PERMISSION_FAULTS: 0x3
Jun  2 17:21:19 mybsd kernel: drmn0:     MAPPING_ERROR: 0x0
Jun  2 17:21:19 mybsd kernel: drmn0:     RW: 0x0
Jun  2 17:21:29 mybsd kernel: [drm ERROR :amdgpu_job_timedout] ring gfx_0.0.0 timeout, signaled seq=1806143, emitted seq=1806145
Jun  2 17:21:29 mybsd kernel: [drm ERROR :amdgpu_job_timedout] Process information: process  pid 120475 thread  pid 120475
Jun  2 17:21:29 mybsd kernel: drmn0: GPU reset begin!
Jun  2 17:21:29 mybsd kernel: drmn0: [drm] REG_WAIT timeout 1us * 100000 tries - optc1_wait_for_state line:841
Jun  2 17:21:30 mybsd syslogd: last message repeated 2 times
Jun  2 17:21:30 mybsd kernel: drmn0: MODE2 reset
Jun  2 17:21:30 mybsd kernel: drmn0: GPU reset succeeded, trying to resume
Jun  2 17:21:30 mybsd kernel: [drm] PCIE GART of 1024M enabled (table at 0x000000F47FC00000).
Jun  2 17:21:30 mybsd kernel: drmn0: PSP is resuming...
Jun  2 17:21:30 mybsd kernel: drmn0: reserve 0xa00000 from 0xf47e000000 for PSP TMR
Jun  2 17:21:30 mybsd kernel: drmn0: RAS: optional ras ta ucode is not available
Jun  2 17:21:30 mybsd kernel: drmn0: RAP: optional rap ta ucode is not available
Jun  2 17:21:30 mybsd kernel: drmn0: SECUREDISPLAY: securedisplay ta ucode is not available
Jun  2 17:21:30 mybsd kernel: drmn0: SMU is resuming...
Jun  2 17:21:30 mybsd kernel: drmn0: SMU is resumed successfully!
Jun  2 17:21:30 mybsd kernel: [drm] DMUB hardware initialized: version=0x05001C00
Jun  2 17:21:30 mybsd kernel: [drm] kiq ring mec 2 pipe 1 q 0
Jun  2 17:21:30 mybsd kernel: [drm] VCN decode and encode initialized successfully(under DPG Mode).
Jun  2 17:21:30 mybsd kernel: [drm] JPEG decode initialized successfully.
Jun  2 17:21:30 mybsd kernel: drmn0: ring gfx_0.0.0 uses VM inv eng 0 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.0.0 uses VM inv eng 1 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.1.0 uses VM inv eng 4 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.2.0 uses VM inv eng 5 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.3.0 uses VM inv eng 6 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.0.1 uses VM inv eng 7 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.1.1 uses VM inv eng 8 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.2.1 uses VM inv eng 9 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.3.1 uses VM inv eng 10 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring kiq_0.2.1.0 uses VM inv eng 11 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring sdma0 uses VM inv eng 12 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring vcn_dec_0 uses VM inv eng 0 on hub 8
Jun  2 17:21:30 mybsd kernel: drmn0: ring vcn_enc_0.0 uses VM inv eng 1 on hub 8
Jun  2 17:21:30 mybsd kernel: drmn0: ring vcn_enc_0.1 uses VM inv eng 4 on hub 8
Jun  2 17:21:30 mybsd kernel: drmn0: ring jpeg_dec uses VM inv eng 5 on hub 8
Jun  2 17:21:30 mybsd kernel: drmn0: recover vram bo from shadow start
Jun  2 17:21:30 mybsd kernel: drmn0: recover vram bo from shadow done
Jun  2 17:21:30 mybsd kernel: drmn0: GPU reset(4) succeeded!
Jun  2 17:21:30 mybsd kernel: [drm] *ERROR*

Check whether there are old firmware / drivers installed. In my case I had to remove some to get to this state (all are 1500068 now):
Code: 
# Which gpu do I have and does it need firmware?
dvl@mybsd:~ % doas fwget -v 2>&1 | grep firmware   
Password:
Needed firmware packages: 'gpu-firmware-amd-kmod-kaveri'                                   <==== iGPU of Ryzen 9700X is incorrectly identified



# Which firmware is available? ==> "-amd-" is newer driver 'amdgpu' and "-radeon-" is older driver 'radeonkms'
dvl@mybsd:~ % pkg search kaveri
gpu-firmware-amd-kmod-kaveri-20250109.1500068 Firmware modules for kaveri AMD GPUs
gpu-firmware-radeon-kmod-kaveri-20220511.1500068 Firmware modules for kaveri Radeon GPUs

dvl@mybsd:~ % pkg info | grep gpu-firmware
gpu-firmware-amd-kmod-dcn-3-1-5-20250109.1500068 Firmware modules for dcn_3_1_5 AMD GPUs
gpu-firmware-amd-kmod-gc-10-3-6-20250109.1500068 Firmware modules for gc_10_3_6 AMD GPUs
gpu-firmware-amd-kmod-kaveri-20250109.1500068 Firmware modules for kaveri AMD GPUs
gpu-firmware-amd-kmod-psp-13-0-5-20250109.1500068 Firmware modules for psp_13_0_5 AMD GPUs
gpu-firmware-amd-kmod-sdma-5-2-6-20250109.1500068 Firmware modules for sdma_5_2_6 AMD GPUs
gpu-firmware-amd-kmod-vcn-3-1-2-20250109.1500068 Firmware modules for vcn_3_1_2 AMD GPUs

# Any DRM installed?
dvl@mybsd:~ % pkg info "drm-*"
drm-latest-kmod-6.9.1500068_1

# Any unusual drivers loaded? ==> either amdgpu OR radeon!!!
dvl@mybsd:~ % kldstat | egrep 'amdgpu|radeon'
39    1 0xffffffff83600000   7ba930 amdgpu.ko
46    1 0xffffffff83e67000     27e0 amdgpu_psp_13_0_5_toc_bin.ko
47    1 0xffffffff83e6a000    394e0 amdgpu_psp_13_0_5_ta_bin.ko
48    1 0xffffffff83ea4000    3cc80 amdgpu_dcn_3_1_5_dmcub_bin.ko
49    1 0xffffffff83ee1000    425e0 amdgpu_gc_10_3_6_pfp_bin.ko
50    1 0xffffffff83f24000    425e0 amdgpu_gc_10_3_6_me_bin.ko
51    1 0xffffffff83f67000    42560 amdgpu_gc_10_3_6_ce_bin.ko
52    1 0xffffffff83faa000    2d4b0 amdgpu_gc_10_3_6_rlc_bin.ko
53    1 0xffffffff83fd8000    43a90 amdgpu_gc_10_3_6_mec_bin.ko
54    1 0xffffffff8401c000    43a90 amdgpu_gc_10_3_6_mec2_bin.ko
55    1 0xffffffff84060000     a5e0 amdgpu_sdma_5_2_6_bin.ko
56    1 0xffffffff8406b000    8f810 amdgpu_vcn_3_1_2_bin.ko
As you can see there is no kaveri loaded. Maybe you have a similar issue.
 
Pobb
kirsche said:
LibreQuest Could be wrong drivers or firmware. Did you ever run fwget -v? Maybe this article of vermaden helps: https://vermaden.wordpress.com/2025/10/13/new-fwget-freebsd-firmware-tool/

My PC crashed because of the wrong firmware and drivers. With my tuning (see my last post) the crashes recover automatically:
Code: 
Jun  2 17:21:19 mybsd kernel: drmn0: [gfxhub] page fault (src_id:0 ring:24 vmid:4 pasid:32780)
Jun  2 17:21:19 mybsd kernel: drmn0:  in process  pid 120475 thread  pid 120475
Jun  2 17:21:19 mybsd kernel: drmn0:   in page starting at address 0x000003c03f7ff000 from client 0x1b (UTCL2)
Jun  2 17:21:19 mybsd kernel: drmn0: GCVM_L2_PROTECTION_FAULT_STATUS:0x00401430
Jun  2 17:21:19 mybsd kernel: drmn0:     Faulty UTCL2 client ID: SQC (data) (0xa)
Jun  2 17:21:19 mybsd kernel: drmn0:     MORE_FAULTS: 0x0
Jun  2 17:21:19 mybsd kernel: drmn0:     WALKER_ERROR: 0x0
Jun  2 17:21:19 mybsd kernel: drmn0:     PERMISSION_FAULTS: 0x3
Jun  2 17:21:19 mybsd kernel: drmn0:     MAPPING_ERROR: 0x0
Jun  2 17:21:19 mybsd kernel: drmn0:     RW: 0x0
Jun  2 17:21:29 mybsd kernel: [drm ERROR :amdgpu_job_timedout] ring gfx_0.0.0 timeout, signaled seq=1806143, emitted seq=1806145
Jun  2 17:21:29 mybsd kernel: [drm ERROR :amdgpu_job_timedout] Process information: process  pid 120475 thread  pid 120475
Jun  2 17:21:29 mybsd kernel: drmn0: GPU reset begin!
Jun  2 17:21:29 mybsd kernel: drmn0: [drm] REG_WAIT timeout 1us * 100000 tries - optc1_wait_for_state line:841
Jun  2 17:21:30 mybsd syslogd: last message repeated 2 times
Jun  2 17:21:30 mybsd kernel: drmn0: MODE2 reset
Jun  2 17:21:30 mybsd kernel: drmn0: GPU reset succeeded, trying to resume
Jun  2 17:21:30 mybsd kernel: [drm] PCIE GART of 1024M enabled (table at 0x000000F47FC00000).
Jun  2 17:21:30 mybsd kernel: drmn0: PSP is resuming...
Jun  2 17:21:30 mybsd kernel: drmn0: reserve 0xa00000 from 0xf47e000000 for PSP TMR
Jun  2 17:21:30 mybsd kernel: drmn0: RAS: optional ras ta ucode is not available
Jun  2 17:21:30 mybsd kernel: drmn0: RAP: optional rap ta ucode is not available
Jun  2 17:21:30 mybsd kernel: drmn0: SECUREDISPLAY: securedisplay ta ucode is not available
Jun  2 17:21:30 mybsd kernel: drmn0: SMU is resuming...
Jun  2 17:21:30 mybsd kernel: drmn0: SMU is resumed successfully!
Jun  2 17:21:30 mybsd kernel: [drm] DMUB hardware initialized: version=0x05001C00
Jun  2 17:21:30 mybsd kernel: [drm] kiq ring mec 2 pipe 1 q 0
Jun  2 17:21:30 mybsd kernel: [drm] VCN decode and encode initialized successfully(under DPG Mode).
Jun  2 17:21:30 mybsd kernel: [drm] JPEG decode initialized successfully.
Jun  2 17:21:30 mybsd kernel: drmn0: ring gfx_0.0.0 uses VM inv eng 0 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.0.0 uses VM inv eng 1 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.1.0 uses VM inv eng 4 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.2.0 uses VM inv eng 5 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.3.0 uses VM inv eng 6 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.0.1 uses VM inv eng 7 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.1.1 uses VM inv eng 8 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.2.1 uses VM inv eng 9 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.3.1 uses VM inv eng 10 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring kiq_0.2.1.0 uses VM inv eng 11 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring sdma0 uses VM inv eng 12 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring vcn_dec_0 uses VM inv eng 0 on hub 8
Jun  2 17:21:30 mybsd kernel: drmn0: ring vcn_enc_0.0 uses VM inv eng 1 on hub 8
Jun  2 17:21:30 mybsd kernel: drmn0: ring vcn_enc_0.1 uses VM inv eng 4 on hub 8
Jun  2 17:21:30 mybsd kernel: drmn0: ring jpeg_dec uses VM inv eng 5 on hub 8
Jun  2 17:21:30 mybsd kernel: drmn0: recover vram bo from shadow start
Jun  2 17:21:30 mybsd kernel: drmn0: recover vram bo from shadow done
Jun  2 17:21:30 mybsd kernel: drmn0: GPU reset(4) succeeded!
Jun  2 17:21:30 mybsd kernel: [drm] *ERROR*

Check whether there are old firmware / drivers installed. In my case I had to remove some to get to this state (all are 1500068 now):
Code: 
# Which gpu do I have and does it need firmware?
dvl@mybsd:~ % doas fwget -v 2>&1 | grep firmware  
Password:
Needed firmware packages: 'gpu-firmware-amd-kmod-kaveri'                                   <==== iGPU of Ryzen 9700X is incorrectly identified



# Which firmware is available? ==> "-amd-" is newer driver 'amdgpu' and "-radeon-" is older driver 'radeonkms'
dvl@mybsd:~ % pkg search kaveri
gpu-firmware-amd-kmod-kaveri-20250109.1500068 Firmware modules for kaveri AMD GPUs
gpu-firmware-radeon-kmod-kaveri-20220511.1500068 Firmware modules for kaveri Radeon GPUs

dvl@mybsd:~ % pkg info | grep gpu-firmware
gpu-firmware-amd-kmod-dcn-3-1-5-20250109.1500068 Firmware modules for dcn_3_1_5 AMD GPUs
gpu-firmware-amd-kmod-gc-10-3-6-20250109.1500068 Firmware modules for gc_10_3_6 AMD GPUs
gpu-firmware-amd-kmod-kaveri-20250109.1500068 Firmware modules for kaveri AMD GPUs
gpu-firmware-amd-kmod-psp-13-0-5-20250109.1500068 Firmware modules for psp_13_0_5 AMD GPUs
gpu-firmware-amd-kmod-sdma-5-2-6-20250109.1500068 Firmware modules for sdma_5_2_6 AMD GPUs
gpu-firmware-amd-kmod-vcn-3-1-2-20250109.1500068 Firmware modules for vcn_3_1_2 AMD GPUs

# Any DRM installed?
dvl@mybsd:~ % pkg info "drm-*"
drm-latest-kmod-6.9.1500068_1

# Any unusual drivers loaded? ==> either amdgpu OR radeon!!!
dvl@mybsd:~ % kldstat | egrep 'amdgpu|radeon'
39    1 0xffffffff83600000   7ba930 amdgpu.ko
46    1 0xffffffff83e67000     27e0 amdgpu_psp_13_0_5_toc_bin.ko
47    1 0xffffffff83e6a000    394e0 amdgpu_psp_13_0_5_ta_bin.ko
48    1 0xffffffff83ea4000    3cc80 amdgpu_dcn_3_1_5_dmcub_bin.ko
49    1 0xffffffff83ee1000    425e0 amdgpu_gc_10_3_6_pfp_bin.ko
50    1 0xffffffff83f24000    425e0 amdgpu_gc_10_3_6_me_bin.ko
51    1 0xffffffff83f67000    42560 amdgpu_gc_10_3_6_ce_bin.ko
52    1 0xffffffff83faa000    2d4b0 amdgpu_gc_10_3_6_rlc_bin.ko
53    1 0xffffffff83fd8000    43a90 amdgpu_gc_10_3_6_mec_bin.ko
54    1 0xffffffff8401c000    43a90 amdgpu_gc_10_3_6_mec2_bin.ko
55    1 0xffffffff84060000     a5e0 amdgpu_sdma_5_2_6_bin.ko
56    1 0xffffffff8406b000    8f810 amdgpu_vcn_3_1_2_bin.ko
As you can see there is no kaveri loaded. Maybe you have a similar issue.
Click to expand...
Possibly, I tried fwget but it didn't change anything. I even rebuilt all the packages. It's maybe a system setting somewhere. I'll be building a new system for the Radeon soon. I'll likely discover it's working! 😄
 
kirsche said:
LibreQuest Could be wrong drivers or firmware. Did you ever run fwget -v? Maybe this article of vermaden helps: https://vermaden.wordpress.com/2025/10/13/new-fwget-freebsd-firmware-tool/

My PC crashed because of the wrong firmware and drivers. With my tuning (see my last post) the crashes recover automatically:
Code: 
Jun  2 17:21:19 mybsd kernel: drmn0: [gfxhub] page fault (src_id:0 ring:24 vmid:4 pasid:32780)
Jun  2 17:21:19 mybsd kernel: drmn0:  in process  pid 120475 thread  pid 120475
Jun  2 17:21:19 mybsd kernel: drmn0:   in page starting at address 0x000003c03f7ff000 from client 0x1b (UTCL2)
Jun  2 17:21:19 mybsd kernel: drmn0: GCVM_L2_PROTECTION_FAULT_STATUS:0x00401430
Jun  2 17:21:19 mybsd kernel: drmn0:     Faulty UTCL2 client ID: SQC (data) (0xa)
Jun  2 17:21:19 mybsd kernel: drmn0:     MORE_FAULTS: 0x0
Jun  2 17:21:19 mybsd kernel: drmn0:     WALKER_ERROR: 0x0
Jun  2 17:21:19 mybsd kernel: drmn0:     PERMISSION_FAULTS: 0x3
Jun  2 17:21:19 mybsd kernel: drmn0:     MAPPING_ERROR: 0x0
Jun  2 17:21:19 mybsd kernel: drmn0:     RW: 0x0
Jun  2 17:21:29 mybsd kernel: [drm ERROR :amdgpu_job_timedout] ring gfx_0.0.0 timeout, signaled seq=1806143, emitted seq=1806145
Jun  2 17:21:29 mybsd kernel: [drm ERROR :amdgpu_job_timedout] Process information: process  pid 120475 thread  pid 120475
Jun  2 17:21:29 mybsd kernel: drmn0: GPU reset begin!
Jun  2 17:21:29 mybsd kernel: drmn0: [drm] REG_WAIT timeout 1us * 100000 tries - optc1_wait_for_state line:841
Jun  2 17:21:30 mybsd syslogd: last message repeated 2 times
Jun  2 17:21:30 mybsd kernel: drmn0: MODE2 reset
Jun  2 17:21:30 mybsd kernel: drmn0: GPU reset succeeded, trying to resume
Jun  2 17:21:30 mybsd kernel: [drm] PCIE GART of 1024M enabled (table at 0x000000F47FC00000).
Jun  2 17:21:30 mybsd kernel: drmn0: PSP is resuming...
Jun  2 17:21:30 mybsd kernel: drmn0: reserve 0xa00000 from 0xf47e000000 for PSP TMR
Jun  2 17:21:30 mybsd kernel: drmn0: RAS: optional ras ta ucode is not available
Jun  2 17:21:30 mybsd kernel: drmn0: RAP: optional rap ta ucode is not available
Jun  2 17:21:30 mybsd kernel: drmn0: SECUREDISPLAY: securedisplay ta ucode is not available
Jun  2 17:21:30 mybsd kernel: drmn0: SMU is resuming...
Jun  2 17:21:30 mybsd kernel: drmn0: SMU is resumed successfully!
Jun  2 17:21:30 mybsd kernel: [drm] DMUB hardware initialized: version=0x05001C00
Jun  2 17:21:30 mybsd kernel: [drm] kiq ring mec 2 pipe 1 q 0
Jun  2 17:21:30 mybsd kernel: [drm] VCN decode and encode initialized successfully(under DPG Mode).
Jun  2 17:21:30 mybsd kernel: [drm] JPEG decode initialized successfully.
Jun  2 17:21:30 mybsd kernel: drmn0: ring gfx_0.0.0 uses VM inv eng 0 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.0.0 uses VM inv eng 1 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.1.0 uses VM inv eng 4 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.2.0 uses VM inv eng 5 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.3.0 uses VM inv eng 6 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.0.1 uses VM inv eng 7 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.1.1 uses VM inv eng 8 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.2.1 uses VM inv eng 9 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring comp_1.3.1 uses VM inv eng 10 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring kiq_0.2.1.0 uses VM inv eng 11 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring sdma0 uses VM inv eng 12 on hub 0
Jun  2 17:21:30 mybsd kernel: drmn0: ring vcn_dec_0 uses VM inv eng 0 on hub 8
Jun  2 17:21:30 mybsd kernel: drmn0: ring vcn_enc_0.0 uses VM inv eng 1 on hub 8
Jun  2 17:21:30 mybsd kernel: drmn0: ring vcn_enc_0.1 uses VM inv eng 4 on hub 8
Jun  2 17:21:30 mybsd kernel: drmn0: ring jpeg_dec uses VM inv eng 5 on hub 8
Jun  2 17:21:30 mybsd kernel: drmn0: recover vram bo from shadow start
Jun  2 17:21:30 mybsd kernel: drmn0: recover vram bo from shadow done
Jun  2 17:21:30 mybsd kernel: drmn0: GPU reset(4) succeeded!
Jun  2 17:21:30 mybsd kernel: [drm] *ERROR*

Check whether there are old firmware / drivers installed. In my case I had to remove some to get to this state (all are 1500068 now):
Code: 
# Which gpu do I have and does it need firmware?
dvl@mybsd:~ % doas fwget -v 2>&1 | grep firmware  
Password:
Needed firmware packages: 'gpu-firmware-amd-kmod-kaveri'                                   <==== iGPU of Ryzen 9700X is incorrectly identified



# Which firmware is available? ==> "-amd-" is newer driver 'amdgpu' and "-radeon-" is older driver 'radeonkms'
dvl@mybsd:~ % pkg search kaveri
gpu-firmware-amd-kmod-kaveri-20250109.1500068 Firmware modules for kaveri AMD GPUs
gpu-firmware-radeon-kmod-kaveri-20220511.1500068 Firmware modules for kaveri Radeon GPUs

dvl@mybsd:~ % pkg info | grep gpu-firmware
gpu-firmware-amd-kmod-dcn-3-1-5-20250109.1500068 Firmware modules for dcn_3_1_5 AMD GPUs
gpu-firmware-amd-kmod-gc-10-3-6-20250109.1500068 Firmware modules for gc_10_3_6 AMD GPUs
gpu-firmware-amd-kmod-kaveri-20250109.1500068 Firmware modules for kaveri AMD GPUs
gpu-firmware-amd-kmod-psp-13-0-5-20250109.1500068 Firmware modules for psp_13_0_5 AMD GPUs
gpu-firmware-amd-kmod-sdma-5-2-6-20250109.1500068 Firmware modules for sdma_5_2_6 AMD GPUs
gpu-firmware-amd-kmod-vcn-3-1-2-20250109.1500068 Firmware modules for vcn_3_1_2 AMD GPUs

# Any DRM installed?
dvl@mybsd:~ % pkg info "drm-*"
drm-latest-kmod-6.9.1500068_1

# Any unusual drivers loaded? ==> either amdgpu OR radeon!!!
dvl@mybsd:~ % kldstat | egrep 'amdgpu|radeon'
39    1 0xffffffff83600000   7ba930 amdgpu.ko
46    1 0xffffffff83e67000     27e0 amdgpu_psp_13_0_5_toc_bin.ko
47    1 0xffffffff83e6a000    394e0 amdgpu_psp_13_0_5_ta_bin.ko
48    1 0xffffffff83ea4000    3cc80 amdgpu_dcn_3_1_5_dmcub_bin.ko
49    1 0xffffffff83ee1000    425e0 amdgpu_gc_10_3_6_pfp_bin.ko
50    1 0xffffffff83f24000    425e0 amdgpu_gc_10_3_6_me_bin.ko
51    1 0xffffffff83f67000    42560 amdgpu_gc_10_3_6_ce_bin.ko
52    1 0xffffffff83faa000    2d4b0 amdgpu_gc_10_3_6_rlc_bin.ko
53    1 0xffffffff83fd8000    43a90 amdgpu_gc_10_3_6_mec_bin.ko
54    1 0xffffffff8401c000    43a90 amdgpu_gc_10_3_6_mec2_bin.ko
55    1 0xffffffff84060000     a5e0 amdgpu_sdma_5_2_6_bin.ko
56    1 0xffffffff8406b000    8f810 amdgpu_vcn_3_1_2_bin.ko
As you can see there is no kaveri loaded. Maybe you have a similar issue.
Click to expand...
Those firmware files say that you have a Raphael (Ryzen 7xxx), Granite Ridge (Ryzen 9xxx) or Dragon Range (Ryzen 7x45) igpu.
If that's the case it's normal you don't see any kaveri firmware loaded.
If you do have a Kaveri Axx-7yyyy from 2014 your issue is broken detection. Open an issue on drm-kmod or on the linux drm big tracker if you get the same issue with upstream linux.
 
You must log in or register to reply here.
Back
Top