Bug 482518

Summary:	=sys-kernel/gentoo-sources-3.10.7 amd modesetting freezes the hardware
Product:	Gentoo Linux	Reporter:	Zoltán Halassy <zhalassy>
Component:	[OLD] Core system	Assignee:	Gentoo Kernel Bug Wranglers and Kernel Maintainers <kernel>
Status:	RESOLVED FIXED
Severity:	normal	CC:	galtgendo
Priority:	Normal
Version:	unspecified
Hardware:	AMD64
OS:	Linux
URL:	https://bugzilla.kernel.org/show_bug.cgi?id=60797
Whiteboard:
Package list:		Runtime testing required:	---
Attachments:	working config for 3.8 freezing config for 3.10 Working config with 3.11

Description Zoltán Halassy 2013-08-26 09:08:28 UTC

Have a Dell OptiPlex 380, it has the following hardware:

00:00.0 Host bridge [0600]: Intel Corporation 4 Series Chipset DRAM Controller [8086:2e30] (rev 03)
00:01.0 PCI bridge [0604]: Intel Corporation 4 Series Chipset PCI Express Root Port [8086:2e31] (rev 03)
00:1b.0 Audio device [0403]: Intel Corporation NM10/ICH7 Family High Definition Audio Controller [8086:27d8] (rev 01)
00:1c.0 PCI bridge [0604]: Intel Corporation NM10/ICH7 Family PCI Express Port 1 [8086:27d0] (rev 01)
00:1d.0 USB controller [0c03]: Intel Corporation NM10/ICH7 Family USB UHCI Controller #1 [8086:27c8] (rev 01)
00:1d.1 USB controller [0c03]: Intel Corporation NM10/ICH7 Family USB UHCI Controller #2 [8086:27c9] (rev 01)
00:1d.2 USB controller [0c03]: Intel Corporation NM10/ICH7 Family USB UHCI Controller #3 [8086:27ca] (rev 01)
00:1d.3 USB controller [0c03]: Intel Corporation NM10/ICH7 Family USB UHCI Controller #4 [8086:27cb] (rev 01)
00:1d.7 USB controller [0c03]: Intel Corporation NM10/ICH7 Family USB2 EHCI Controller [8086:27cc] (rev 01)
00:1e.0 PCI bridge [0604]: Intel Corporation 82801 PCI Bridge [8086:244e] (rev e1)
00:1f.0 ISA bridge [0601]: Intel Corporation 82801GB/GR (ICH7 Family) LPC Interface Bridge [8086:27b8] (rev 01)
00:1f.1 IDE interface [0101]: Intel Corporation 82801G (ICH7 Family) IDE Controller [8086:27df] (rev 01)
00:1f.2 IDE interface [0101]: Intel Corporation NM10/ICH7 Family SATA Controller [IDE mode] [8086:27c0] (rev 01)
00:1f.3 SMBus [0c05]: Intel Corporation NM10/ICH7 Family SMBus Controller [8086:27da] (rev 01)
01:00.0 VGA compatible controller [0300]: Advanced Micro Devices [AMD] nee ATI RV710 [Radeon HD 4350/4550] [1002:954f]
01:00.1 Audio device [0403]: Advanced Micro Devices [AMD] nee ATI RV710/730 HDMI Audio [Radeon HD 4000 series] [1002:aa38]
02:00.0 Ethernet controller [0200]: Broadcom Corporation NetLink BCM57780 Gigabit Ethernet PCIe [14e4:1692] (rev 01)

With sys-kernel/gentoo-sources-3.8.13 the kernel mode setting works perfectly, it changes into graphics mode during boot (FB console). Invoking startx works, then switching back and forth X and console also works perfectly.

With sys-kernel/gentoo-sources-3.10.7 during initial mode setting, the system hangs. The screen stays in text mode, and the last thing can be seen is "ATOM BIOS", when support compiled in. The system hangs during module load when it is compiled into module. No kernel panic. Tried with almost-same settings, then tried to compile in DRM and FB stuff, the latter only caused to hang the system sooner.

The strange thing is: if 3.8 is booted, invoking kexec on the fb console boots up 3.10. It does not get back into text mode, but at the mode setting step it continues to work. kexec-ing again from 3.10 into the same 3.10 hangs the system.

Attaching both configs, the working 3.8 one, and the freezing 3.10 one.

Reproducible: Always

Comment 1 Zoltán Halassy 2013-08-26 09:10:04 UTC

Created attachment 357058 [details]
working config for 3.8

Comment 2 Zoltán Halassy 2013-08-26 09:10:23 UTC

Created attachment 357060 [details]
freezing config for 3.10

Comment 3 Zoltán Halassy 2013-08-26 19:44:36 UTC

Seem to be vanilla kernel regression. Upstream bugreport:

https://bugzilla.kernel.org/show_bug.cgi?id=60797

Comment 4 Tom Wijsman (TomWij) (RETIRED) gentoo-dev

2013-09-03 18:15:05 UTC

From a quick look, there happened about 122 commits in drivers/gpu/drm/radeon between 3.9.11 and 3.10.7; so, if you keep dividing that number by two you can conclude that you can find the offending commit in about 7 or 8 compilations.

So, if you don't mind compiling your kernel that many times; could you please do https://wiki.gentoo.org/wiki/Kernel_git-bisect between v3.9.11 and v3.10.7 but instead of the "start bisect" step there use this instead:

    git bisect start 3.10.7 3.9.11 -- drivers/gpu/drm/radeon

So, follow the instructions further. When your system works once you reboot into a new compiled kernel, mark it as a good bisect; when it breaks due to the artifacts or blank screen, mark it as bad.

There is a small change the commit happened outside of the directory, but given that this isn't the first time I hear of this problem with Radeon and I don't see problems with others; I think it is worth doing it by directory. Worst case we end up with a small enough range between two radeon commits we can skim through or in worst case further test.

Thank you in advance.

Comment 5 Zoltán Halassy 2013-09-12 17:50:00 UTC

Added results to upstream bugreport. Thank you for the tip. One thing is strange for me though, the version bisec gave me was 3.9.0-rc4 . Why does my box work with vanilla 3.9.11?

Comment 6 Zoltán Halassy 2013-09-12 19:15:23 UTC

So, new kernels needs different firmware. The problem in 3.10.7 still exists (freezing hardware), but it's fixed in 3.11 when the correct firmware is loaded.

Is this mean this bug will closed as wontfix?

Comment 7 Zoltán Halassy 2013-09-12 19:47:41 UTC

=sys-kernel/gentoo-sources-3.11.0 works with the attached config, only needs R700_rlc.bin, RV710_smc.bin and RV710_uvd.bin compiled in, if radeon driver is compiled in aswell.

Comment 8 Zoltán Halassy 2013-09-12 19:48:25 UTC

Created attachment 358556 [details]
Working config with 3.11