Created attachment 1523018 [details] dmesg excerpt showing amdgpu error Description of problem: Fedora 29 with kernel 4.20.3-200.fc29 does not boot with Polaris 10 Version-Release number of selected component (if applicable): Name : linux-firmware Version : 20181219 Release : 89.git0f22c852.fc29 How reproducible: Always, assuming a Polaris 10 card (e.g. RX 580) and an up-to-date Fedora 29 Steps to Reproduce: 1. Ensure a Polaris 10 card is installed 2. Update to kernel 4.20.3-200.fc29 3. Reboot machine Actual results: EFI splash screen stays on screen Expected results: Plymouth shows the Fedora boot logo and then GDM asks you to log in Additional info: dmesg shows that it is unable to load polaris10_mc.bin (error -2), which is provided by linux-firmware. See attached excerpt from the dmesg output.
Debian had a similar problem, they now have a newer linux-firmware package that apparently fixed it on their end: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=916816 (added as external bug tracker as well)
Installed update from updates-testing, problem persists. New version information from DNF is: Name : linux-firmware Version : 20190118 Release : 91.gita8b75cac.fc29
After you install the new linux-firmware package you need to make sure the new firmware gets in the initramfs for you current kernel. # dracut -f ...
That doesn't give much of an improvement unfortunately, it still can't find the firmware. See the new dmesg-after-dracut.txt attachment.
Created attachment 1523119 [details] dmesg excerpt after running dracut -f
Created attachment 1524793 [details] dracut -fv output
This is still an issue after installing kernel 4.20.4-200.fc29.x86_64. It still fails to load amdgpu/polaris10_mc.bin: [ 4.764344] [drm] amdgpu kernel modesetting enabled. [ 4.782472] fb0: switching to amdgpudrmfb from EFI VGA [ 4.782843] amdgpu 0000:05:00.0: enabling device (0106 -> 0107) [ 4.783296] amdgpu 0000:05:00.0: Direct firmware load for amdgpu/polaris10_mc.bin failed with error -2 [ 4.783298] mc: Failed to load firmware "amdgpu/polaris10_mc.bin" [ 4.783404] [drm:gmc_v8_0_sw_init [amdgpu]] *ERROR* Failed to load mc firmware! [ 4.783475] [drm:amdgpu_device_init.cold.31 [amdgpu]] *ERROR* sw_init of IP block <gmc_v8_0> failed -2 [ 4.783477] amdgpu 0000:05:00.0: amdgpu_device_ip_init failed [ 4.783479] amdgpu 0000:05:00.0: Fatal error during GPU init [ 4.783480] [drm] amdgpu: finishing device. [ 4.783865] amdgpu: probe of 0000:05:00.0 failed with error -2 The file clearly exists though: $ ls -l /usr/lib/firmware/amdgpu/polaris10_mc.bin -rw-r--r--. 1 root root 32732 Jan 22 08:50 /usr/lib/firmware/amdgpu/polaris10_mc.bin Please see the attachment for the output of `dracut -fv`. Is there anything I can try to get this working?
linux-firmware-20190213-93.git710963fe.fc29 has been submitted as an update to Fedora 29. https://bodhi.fedoraproject.org/updates/FEDORA-2019-42cbff0fd6
linux-firmware-20190213-93.git710963fe.fc28 has been submitted as an update to Fedora 28. https://bodhi.fedoraproject.org/updates/FEDORA-2019-f27089e66c
linux-firmware-20190213-93.git710963fe.fc28 has been pushed to the Fedora 28 testing repository. If problems still persist, please make note of it in this bug report. See https://fedoraproject.org/wiki/QA:Updates_Testing for instructions on how to install test updates. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2019-f27089e66c
linux-firmware-20190213-93.git710963fe.fc29 has been pushed to the Fedora 29 testing repository. If problems still persist, please make note of it in this bug report. See https://fedoraproject.org/wiki/QA:Updates_Testing for instructions on how to install test updates. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2019-42cbff0fd6
Not a fix on the latest kernel: [ 4.740392] fb0: switching to amdgpudrmfb from EFI VGA [ 4.740785] amdgpu 0000:05:00.0: enabling device (0106 -> 0107) [ 4.741033] [drm] initializing kernel modesetting (POLARIS10 0x1002:0x67DF 0x1DA2:0xE366 0xE7). [ 4.741044] [drm] register mmio base: 0x8C500000 [ 4.741045] [drm] register mmio size: 262144 [ 4.741058] [drm] add ip block number 0 <vi_common> [ 4.741059] [drm] add ip block number 1 <gmc_v8_0> [ 4.741060] [drm] add ip block number 2 <tonga_ih> [ 4.741061] [drm] add ip block number 3 <gfx_v8_0> [ 4.741062] [drm] add ip block number 4 <sdma_v3_0> [ 4.741072] [drm] add ip block number 5 <powerplay> [ 4.741073] [drm] add ip block number 6 <dm> [ 4.741075] [drm] add ip block number 7 <uvd_v6_0> [ 4.741076] [drm] add ip block number 8 <vce_v3_0> [ 4.741082] kfd kfd: skipped device 1002:67df, PCI rejects atomics [ 4.741094] [drm] UVD is enabled in VM mode [ 4.741095] [drm] UVD ENC is enabled in VM mode [ 4.741097] [drm] VCE enabled in VM mode [ 4.741180] ATOM BIOS: 113-1E366CU-S5S [ 4.741231] [drm] vm size is 512 GB, 2 levels, block size is 10-bit, fragment size is 9-bit [ 4.741269] amdgpu 0000:05:00.0: Direct firmware load for amdgpu/polaris10_mc.bin failed with error -2 [ 4.741271] mc: Failed to load firmware "amdgpu/polaris10_mc.bin" [ 4.741395] [drm:gmc_v8_0_sw_init [amdgpu]] *ERROR* Failed to load mc firmware! [ 4.741466] [drm:amdgpu_device_init.cold.31 [amdgpu]] *ERROR* sw_init of IP block <gmc_v8_0> failed -2 [ 4.741468] amdgpu 0000:05:00.0: amdgpu_device_ip_init failed [ 4.741470] amdgpu 0000:05:00.0: Fatal error during GPU init [ 4.741473] [drm] amdgpu: finishing device. [ 4.741843] amdgpu: probe of 0000:05:00.0 failed with error -2 Also... Why is it trying to load a Polaris 10 firmware when it should be loading Polaris 20? We are talking about a RX580 here according to lspci. Should this maybe be reclassified as a kernel bug?
linux-firmware-20190213-93.git710963fe.fc29 has been pushed to the Fedora 29 stable repository. If problems still persist, please make note of it in this bug report.
linux-firmware-20190213-93.git710963fe.fc28 has been pushed to the Fedora 28 stable repository. If problems still persist, please make note of it in this bug report.