|Summary:||(SI) latest stable & cant boot|
|Component:||DRM/AMDgpu||Assignee:||Default DRI bug account <dri-devel>|
|Status:||RESOLVED FIXED||QA Contact:|
|i915 platform:||i915 features:|
Description qmastery16 2017-06-16 22:22:51 UTC
Created attachment 132011 [details] cut_here KERNEL PANIC Trying to enable AMD R5-M230 Southern Islands (SI) laptop discrete GPU - by using the latest stable 4.11.5 Linux Kernel and the latest Mesa 17.1.2 . Already blacklisted radeon kernel module. However I can not boot with AMDGPU after it is giving me these errors: [drm] amdgpu kernel modesetting enabled. [drm] initializing kernel modesetting (HAINAN 0x1002:0x6665 0x0000:0x0000 0x00). [drm] register mmio base: 0xF0300000 [drm] register mmio size: 262144 amdgpu 0000:04:00.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0x0000 [drm] vbios mem is null or mem size is wrong amdgpu 0000:04:00.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0x0000 [drm] vbios mem is null or mem size is wrong [drm:amdgpu_get_bios [amdgpu]] *ERROR* Unable to locate a BIOS ROM amdgpu 0000:04:00.0: Fatal error during GPU init [drm] amdgpu: finishing device. <=== KERNEL PANIC, "cut here" is inside the attachments This is AMDGPU blocker issue, because I could boot only if I add "nomodeset" to kernel parameters
Comment 1 Alex Deucher 2017-10-31 20:38:46 UTC
Please attach your full dmesg output with amdgpu. Is this still an issue with a newer kernel?
Comment 2 denisgolovan 2017-11-16 13:59:54 UTC
Same error in dmesg for me. Except that I can boot in graphic mode. My hardware is XFX Radeon RX580. desktop64 denis # lspci | grep VG 06:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Device 67df (rev e7)
Comment 3 denisgolovan 2017-11-16 14:00:46 UTC
Created attachment 135516 [details] XFX Radeon RX580 dmesg
Comment 4 denisgolovan 2017-11-16 14:03:50 UTC
Tested under Linux 4.13.9 + ck1 patch, Mesa 17.2.5, xf64-video-amdgpu (git 875339c1064f666a2391b4a5a495eddda9407ab6), libdrm-2.4.88.
Comment 5 Alex Deucher 2017-11-16 14:40:48 UTC
The message: Invalid PCI ROM header structure: expecting 0xaa55, got 0x0000 is harmless and can be ignored.
Comment 6 denisgolovan 2017-11-16 14:50:31 UTC
(In reply to Alex Deucher from comment #5) > The message: > Invalid PCI ROM header structure: expecting 0xaa55, got 0x0000 > is harmless and can be ignored. Ok. Thanks
Comment 7 qmastery16 2019-04-18 15:26:21 UTC
Sorry, but this message "Invalid PCI ROM header structure: expecting 0xaa55, got 0x0000" is not always harmless, since it usually leads to "Fatal error during GPU init" : [ 5.732623] [drm:radeon_get_bios [radeon]] *ERROR* ACPI VFCT image header truncated [ 5.732646] radeon 0000:01:00.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0x0000 [ 5.732829] radeon 0000:01:00.0: Invalid PCI ROM header signature: expecting 0xaa55, got 0x0000 [ 5.732866] [drm:radeon_get_bios [radeon]] *ERROR* Unable to locate a BIOS ROM [ 5.732869] radeon 0000:01:00.0: Fatal error during GPU init [ 5.732872] [drm] radeon: finishing device.
Comment 8 qmastery16 2019-05-10 06:43:41 UTC
Our problem was caused by the opensource coreboot BIOS incorrectly initializing the discrete GPU. These unofficial patches are solving our problem - https://review.coreboot.org/c/coreboot/+/31929/ . And here is our latest board-status report, containing a new kernel log and other useful info - https://coreboot.org/status/board-status.html#lenovo/g505s . So this issue could be closed - and if someone has a similar problem he could reopen, although maybe it's better to do a fresh start by creating a new issue.