Bugzilla – Bug 302016
openSUSE 10.3 beta1 installation/boot fails on SB700+RX780
Last modified: 2007-09-12 02:51:18 UTC
Steps: 1. Install openSUSE Alpha7 i386 on one SB600 board; 2. Upgrade the kernel to the Beta1/2 test version: kernel-default-2.6.22.3-20070816053233 downloaded from ftp://ftp.suse.com/pub/people/aj/10.3-Beta1-Kernel-Test2/ 3. Connect this HDD to one SB700 board(SB700+RX780) 4. Boot the openSUSE update to date kernel, but failed. here are the boot messages: pnp: 00:09: iomem range 0xfee00000-0xfee00fff has been reserved pnp: 00:0a: iomem range 0xffb80000-0xffbfffff could not be reserved pnp: 00:0a: iomem range 0xfff00000-0xffffffff could not be reserved pnp: 00:0b: iomem range 0xe0000000-0xefffffff has been reserved pnp: 00:0c: iomem range 0x0-0x9ffff could not be reserved pnp: 00:0c: iomem range 0x0-0x0 could not be reserved pnp: 00:0c: iomem range 0xe0000-0xfffff could not be reserved pnp: 00:0c: iomem range 0x100000-0x3fbfffff could not be reserved Time: hpet clocksource has been installed. Switched to NOHz mode on CPU #0 Switched to NOHz mode on CPU #1 PCI: Bridge: 0000:00:02.0 IO window: e000-efff MEM window: fea00000-feafffff PREFETCH window: d8000000-dfffffff PCI: Bridge: 0000:00:0a.0 IO window: disabled. MEM window: feb00000-febfffff PREFETCH window: disabled. PCI: Bridge: 0000:00:14.4 IO window: disabled. MEM window: disabled. PREFETCH window: disabled. NET: Registered protocol family 2 IP route cache hash table entries: 32768 (order: 5, 131072 bytes) TCP established hash table entries: 131072 (order: 8, 1572864 bytes) TCP bind hash table entries: 65536 (order: 7, 524288 bytes) TCP: Hash tables configured (established 131072 bind 65536) TCP reno registered Machine check exception polling timer started. audit: initializing netlink socket (disabled) audit:(1187691115.952:1): initialized highmem bounce pool size: 64 pages Total HugeTLB memory allocated, 0 VFS: Disk quotas dquot_6.5.1 Dquot-cache hash table entries: 1024 (order 0, 4096 bytes) io scheduler noop registered io scheduler anticipatory registered io scheduler deadline registered io scheduler cfq registered (default)
When I add "acpi=off pci=nomsi" in GRUB config, the kernel can boot up well. Are you able to disable RX780 msi before the release of openSUSE10.3 Beta2 ? So that we can test it. And the "acpi=off" is also necessary, otherwise kernel can NOT boot either. Thanks
[Cc'ing Thomas for ACPI.] Please post the result of 'lspci -nn'. What happens if you specify "acpi=off" without "pci=nomsi"?
Created attachment 158582 [details] output of lspci -nn as requested
Okay, patch committed to SL103 and forwarded upstream. Shane, can you please tell me all PCI IDs of ATI host bridges with this problem? We currently have RS400_200 (7833), RS480 (5950), RD580 (5952), RX790 (5957) and RS690 (7910) on the blacklist.
Maybe most of the ATI chipsets contain this problem except the MSI is disabled in system BIOS, but I'm not able to provide all those pci device IDs immediately here, and this guess also has not been proved by any guys.
Created attachment 158600 [details] boot messages when specify "acpi=off" without "pci=nomsi"
Boot messages when specify "acpi=off" without "pci=nomsi" are provided as requested.
Thanks. Reassigning to Thomas for ACPI.
You might want to increase ACPI debug level via boot param: acpi.debug_level=0x1011F or if output is too much maybe we already see something suspicious with: acpi.debug_level=0x1001F (if acpi.debug_level does not work, 2.6.22 uses still the old param e.g.: acpi_dbg_level=0x1001F) Increase the kernel/dmesg log buffer with boot param log_buf_len=VALUE_IN_BYTES, the output can be huge, you should try with 16,32 or even more MB of log_buf_len. Please attach results or if it's really more than some MB, you might want to zip this or point us to an internet address where the output can be found.
I'll be off for two weeks, Alexey can you have a look at this meanwhile...
Hi Tejun: I don't find MSI disablement to RX790|RX780 (5957) in openSUSE10.3 Beta2 linux kernel source code. Will it be added from Beta 3? As to the ACPI problem, we will debug and discuss it by ourselves at first, and post our questions here if we need your help. Thomas, thank you for your debug suggestion. Shane
Yeah, B3 should have it.
I have verified that openSUSE Beta3 can be installed on SB700+RX780 directly without any "pci=nomsi". thanks for your work. And it seems that the ACPI problem also have been resolved with our up to date BIOS.
fixed from openSUSE Beta3
So all is fine, right? Pls reopen if not -> closing.
yep, we will reopen it if need. Thanks.