Bug 475082

Summary:	sys-kernel/gentoo-sources-3.9.7: radeon module freezes the whole system
Product:	Gentoo Linux	Reporter:	Rafał Mużyło <galtgendo>
Component:	[OLD] Core system	Assignee:	Gentoo Kernel Bug Wranglers and Kernel Maintainers <kernel>
Status:	RESOLVED NEEDINFO
Severity:	normal	CC:	x11
Priority:	Normal
Version:	unspecified
Hardware:	AMD64
OS:	Linux
Whiteboard:
Package list:		Runtime testing required:	---
Attachments:	kernel config with the freeze

Description Rafał Mużyło 2013-06-28 09:46:30 UTC

Created attachment 352148 [details]
kernel config with the freeze

This will be a very bad bug report, as I don't have any experience with kernel bugs.

I've got Asrock h67m motherboard with i3-2120 CPU that additionally has Radeon HD 4850 card (RV770). Till awhile (but I don't recall the exact time/kernel) I could use the intel graphics, while having radeon module loaded (though no monitor was plugged into radeon). With kernels 3.8.13, 3.9.3, 3.9.7 loading radeon module causes the whole system to freeze.

I can't really tell more - not like a kernel freeze is something debuggable on my current skill level.

sys-devel/gcc-4.5.4 (if that could matter)

Odd part is that I was able to get past the freeze once or twice during boot under some unknown circumstances, but ever since it was consistently freezing.

Due to intel, radeon module is KMS-only, but I don't think it's the recent bug (one addressed by https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/commit/?id=e335ccae2ae4caef26a1e81711c44aec1cc83bd2), cause that one was supposed to be fixed in 3.9.6.

The radeon module is definitely the problem, as simply modprobing it triggers the freeze (kmod-13-r1, if it matters). For the time being, I'm blacklisting the module, otherwise the machine freezes on boot on "waiting for uevents to be processed".

Comment 1 Chí-Thanh Christopher Nguyễn gentoo-dev

2013-06-28 10:19:20 UTC

If you modprobe radeon after the system has booted will it freeze too? Does the system also stop responding to ping/ssh/magic sysrq?

Comment 2 Rafał Mużyło 2013-06-28 10:59:33 UTC

(In reply to Chí-Thanh Christopher Nguyễn from comment #1)
> If you modprobe radeon after the system has booted will it freeze too?
It definite does (I think I've wrote that, but perhaps I wasn't clear enough)

> Does the system also stop responding to ping/ssh/magic sysrq?
Well, as far as ssh goes, the response is "No route to host", so it does look like a total freeze.

Comment 3 Chí-Thanh Christopher Nguyễn gentoo-dev

2013-06-28 11:02:40 UTC

There are several possible causes to this. It could be a hardware problem (like PSU going bad and initializing the radeon card will stress it too much), or BIOS issue (ensure that you are running latest BIOS, boot with pci=nomsi, ...) or a kernel bug.

What happens if you disable the Intel graphics in your BIOS and connect the monitor to the radeon card on boot?

Comment 4 Rafał Mużyło 2013-06-28 12:04:42 UTC

Well, when I blacklisted i915 *and* radeon, it didn't help.

After boot, while still in console, I've rechecked with lsmod - neither of two was loaded. Upon 'modprobe radeon' - instant freeze.

'pci=nomsi' didn't make a change either.

Comment 5 Rafał Mużyło 2013-06-28 16:08:10 UTC

...I'm beginning to wonder if it's a kernel bug or a hardware failure: when I set in UEFI primary video to that PCIe card, I got a hard freeze at POST (at least I think if was at POST, as nothing was shown). The annoying part is the card was hardly used...

Comment 6 Chí-Thanh Christopher Nguyễn gentoo-dev

2013-06-28 16:20:59 UTC

From experience, I think it is more likely PSU failure than card failure.

Comment 7 Tom Wijsman (TomWij) (RETIRED) gentoo-dev

2013-07-01 16:38:53 UTC

(In reply to Rafał Mużyło from comment #5)
> I'm beginning to wonder if it's a kernel bug or a hardware failure

This is something that you will need to figure out first.

Some questions to ask yourself:

1. Does this happen on another operating system? On a LiveCD perhaps?
2. Does this happen on older kernels, for example on the LTS kernels? (3.0-3.4)
3. Does this happen on the latest release candidate kernel? (3.10-rc7)

If the answers to all three questions are yes then you would have confirmed that this is a hardware issue or an issue at least present in a big share of the kernel versions, that this is not a regression (but to be sure, you could check whether 2.4 or 2.6 works?) and that this has not been fixed.

If one of these is false, it is probably not a hardware failure and fixable.