OSS 4.2 fails to load - Intel ICH9, 82801I

[Wintershade]

Hello everyone.

I'm having problems with OSS 4.2 on my Acer Aspire 7730 laptop. The soundcard in it is Intel ICH9 hdaudio, and the OSS 4.1 worked fine with it.
However, ever since OSS 4.2 came out, I'm having trouble with it. When I try to load the oss daemon, it gives me a bunch of error messages, and fails to load.

I've put my /var/log/soundon.log in the pastebin, as well as the ossdetect -vvv output.
As for my OS, I'm running Arch Linux (rolling release), with kernel 2.6.33.
If anyone could help me with this issue, I'd appreciate it a lot.


Thanks in advance!

[igorzwx]

Did you remove /usr/lib/oss before installing v.4.2 ?

[cesium]

"BUG: unable to handle kernel paging request at ffffc900100ed000" hmm...

Can you try latest version from mercurial? Erase /usr/lib/oss entirely, get the source from here and build according to this. Does it work?

[Wintershade]

Yes, I've removed the /usr/lib/oss directory.

I've also compiled the package from mercurial using the AUR (took the time to adopt and update the abandoned package as well).
However, the issue still remains.
Any other ideas? Is this a bug in the driver, or in the kernel perhaps?

[cesium]

Hmm.. Do you remember which version of 4.1 worked? We'll have to bisect the source to find out where the regression was...
I suggest starting with revisions 583 and 586 of mercurial - Perhaps this causes the regression? If so, the former will work while the latter will not. ('hg update -r (REV)' can be used to do this. Also 'hg bisect' might turn out to be useful).

[igorzwx]

Hmm.. Do you remember which version of 4.1 worked? We'll have to bisect the source to find out where the regression was...
I suggest starting with revisions 583 and 586 of mercurial - Perhaps this causes the regression? If so, the former will work while the latter will not. ('hg update -r (REV)' can be used to do this. Also 'hg bisect' might turn out to be useful).

I would rather suggest kernel. Does v.4.1 work with the new kernel?

There is also rt-kernel. Why not try it too?

[Wintershade]

Hi guys, thanks everyone for all your engagement, I hope you'll be able (and willing) to help me some more.
(and sorry for the late reply, I've been in a rush for the past few days)

Anyway. I'd be more than happy to bisect the source. However, I'm not really skillful with using mercurial. Can someone give me a few pointers (and perhaps a link to a "howto", or some documentation) on doing this with hg?
I've done this once with git (when I was trying to locate a bug in WINE), so I hope I'll figure it out with little trouble.

Also, I've been told on AUR that the mercurial repository for OSS is now on sourceforge, rather than the one mentioned in the wiki. Is this true? I mean, is the latter really outdated?

Many thanks again.

[cesium]

Yeah, that's true. I changed the wiki to point to the correct repo...

[MadCatX]

I believe I have a similar issue, though not exactly the same. I'm also running Arch Linux and 2.6.34 kernel and when I try to start oss I get a lot of error messages in dmesg (see attached file) and oss fails to load. I tried it with 4.2_2002 and 4.2_2003, both fail in the same way. 4.1 seems to load with no problems, but I don't get any sound with it.

My sound card is "Intel Corporation 82801I (ICH9 Family) HD Audio Controller (rev 03)", ALSA detects it as ALC272 codec. The machine is Toshiba Qosmio X300 laptop.

[cesium]

Err.. no file is attached. Can you try again?

[MadCatX]

Great, this happens me all the time. It's not that long, so I'll paste it right in here. I also tried it with the latest mercurial revision 897 and it behaves in the same way.

Code: Select all

oss_hdaudio 0000:00:1b.0: PCI INT A -> GSI 22 (level, low) -> IRQ 22
oss_hdaudio: HDA codec 0x10ec0272 not known yet
oss_hdaudio: HDA codec 0x10ec0272 not known yet
oss_hdaudio: Too many output endpoints
general protection fault: 0000 [#1] PREEMPT SMP
last sysfs file: /sys/module/osscore/initstate
CPU 1
Modules linked in: oss_hdaudio(+) osscore rfcomm sco bridge stp llc bnep l2cap btusb bluetooth ext2 cpufreq_powersave cpufreq_ondemand uvcvideo iwlagn iwlcore videodev v4l1_compat mac80211 joydev cfg80211 v4l2_compat_ioctl32 toshiba_bluetooth acpi_cpufreq rfkill uhci_hcd battery nvidia(P) video output jmb38x_ms memstick ac ehci_hcd usbcore psmouse button firewire_ohci firewire_core thermal crc_itu_t sdhci_pci sdhci mmc_core led_class r8169 mii i2c_i801 i2c_core iTCO_wdt iTCO_vendor_support intel_agp pcspkr serio_raw evdev freq_table processor thermal_sys coretemp rtc_cmos rtc_core rtc_lib

Pid: 2131, comm: modprobe Tainted: P           2.6.34-ck1-ARCHMOD #1 KSRAA/Qosmio X300
RIP: 0010:[<ffffffffa0df7ef3>]  [<ffffffffa0df7ef3>] hdaudio_mix_init+0x193/0x1a0 [oss_hdaudio]
RSP: 0018:ffff88013d6b5988  EFLAGS: 00010206
RAX: 1010000400000000 RBX: 0000000000000002 RCX: ffffc9001008e010
RDX: ffffc90010084028 RSI: 0000000000000002 RDI: 0000000000000000
RBP: ffffc90000051020 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 000000000000000c R12: ffffc90000051010
R13: 0000000000000000 R14: 0000000000000001 R15: 0000000000000000
FS:  00007f0000f53700(0000) GS:ffff880001900000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 00007f0000f1f000 CR3: 000000013f3b0000 CR4: 00000000000406a0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process modprobe (pid: 2131, threadinfo ffff88013d6b4000, task ffff88013d79f1a0)
Stack:
ffffc90000000018 ffffffffa0d8213b 0000000000000000 ffffc900100f7120
<0> ffffc900100f7108 0000000000000000 ffff88013d6b5a18 0000000000000000
<0> 000000000000001c 0000000000000000 0000000000000070 0000000000000000
Call Trace:
[<ffffffffa0d8213b>] ? mixer_ext_create_control+0x18b/0x380 [osscore]
[<ffffffffa0d80f7b>] ? touch_mixer+0x30b/0x480 [osscore]
[<ffffffff811a6e33>] ? sprintf+0x43/0x50
[<ffffffffa0df9149>] ? hdaudio_mixer_create+0x1039/0x2800 [oss_hdaudio]
[<ffffffff8107d691>] ? __setup_irq+0x1e1/0x340
[<ffffffff8100370e>] ? apic_timer_interrupt+0xe/0x20
[<ffffffffa0d5dc59>] ? oss_reserve_pages+0x29/0x60 [osscore]
[<ffffffffa0e06136>] ? oss_hdaudio_attach+0xb16/0x14f0 [oss_hdaudio]
[<ffffffff811e013b>] ? acpi_pci_irq_enable+0x212/0x23c
[<ffffffff811b87f3>] ? pci_enable_resources+0x23/0x160
[<ffffffffa0e06db1>] ? osspci_probe+0xd3/0x116 [oss_hdaudio]
[<ffffffff811b5b30>] ? pci_device_probe+0x80/0xb0
[<ffffffff8123443a>] ? driver_sysfs_add+0x5a/0x90
[<ffffffff812346e3>] ? driver_probe_device+0x93/0x190
[<ffffffff81234873>] ? __driver_attach+0x93/0xa0
[<ffffffff812347e0>] ? __driver_attach+0x0/0xa0
[<ffffffff8123380e>] ? bus_for_each_dev+0x4e/0x80
[<ffffffff81233f5a>] ? bus_add_driver+0xaa/0x260
[<ffffffff81234acc>] ? driver_register+0x6c/0x130
[<ffffffff811b5dcc>] ? __pci_register_driver+0x4c/0xc0
[<ffffffffa0e1a000>] ? pcidrv_init+0x0/0x35 [oss_hdaudio]
[<ffffffffa0e1a01e>] ? pcidrv_init+0x1e/0x35 [oss_hdaudio]
[<ffffffff810002e4>] ? do_one_initcall+0x34/0x1a0
[<ffffffff8106535c>] ? sys_init_module+0xdc/0x250
[<ffffffff81002dab>] ? system_call_fastpath+0x16/0x1b
Code: 48 c7 82 28 88 00 00 10 04 e0 a0 48 c7 c3 10 04 e0 a0 e9 e6 fe ff ff 90 49 8b 84 24 d8 0e 00 00 45 31 ed 48 85 c0 74 c0 44 89 ff <ff> d0 eb b9 66 0f 1f 84 00 00 00 00 00 41 57 41 56 41 55 49 89
RIP  [<ffffffffa0df7ef3>] hdaudio_mix_init+0x193/0x1a0 [oss_hdaudio]
RSP <ffff88013d6b5988>
---[ end trace 4ee734f861a2bd88 ]---
usbcore: registered new interface driver oss_usb


This is all the info regarding oss I get from dmesg. The oss_usb driver also appears to be "stuck", "rmmod oss_usb" always runs for eternity doing nothing.

[cesium]

Can you try to rebuild on mercurial? It seems you already know how to do that. You can now revert oss_hdaudio to an earlier version and see which rev. introduces the problem... (Note that to test you don't need "soundon". You can just modprobe osscore/oss_hdaudio manually and check dmesg afterwards). Once we get it to load properly, we'll deal with sound playback etc.

[MadCatX]

After some painstaking work I managed to get as low as to revision 806, which still crashes in the same manner. The crash occurs when I try to load "oss_hdaudio" module. I tried aforementioned revs 583 and 586, but they both fail to compile with

Code: Select all

make[5]: Leaving directory `/home/madcat/abs/oss-hg/src/oss/build/noregparm/build/lib/libOSSlib'
make[4]: Leaving directory `/home/madcat/abs/oss-hg/src/oss/build/noregparm/build/lib/libOSSlib'
make[3]: Leaving directory `/home/madcat/abs/oss-hg/src/oss/build/noregparm/build/lib'
noregparm
/bin/sh: line 0: cd: noregparm: No such file or directory
make[2]: *** [subdirs] Error 1
make[2]: Leaving directory `/home/madcat/abs/oss-hg/src/oss/build/noregparm/build'
make[1]: *** [subdirs] Error 1
make[1]: Leaving directory `/home/madcat/abs/oss-hg/src/oss/build/noregparm'
make: *** [subdirs] Error 1


I had to manually patch /setup/dirsetup.c and /setup/srcconf.c to make their compilation even start. If anyone has any suggestion how to get 58* revs working, I'm certainly willing to give them a try.

[cesium]

Weird, I don't get that build error on 586 (this failed during build not install, right?)... Try running configure with "--regparm --only-drv=oss_hdaudio --enable-libsalsa=NO" to minimize compiling and dependencies.

[MadCatX]

I've been fooling around these configure parameters before, the problem is that when I use "--regparm" and run "/etc/rc.d/ossv4 start" I get build errors too when the script is trying to build osscore. Without "--regparm" the build fails with the errors I posted before. Other configure parameteres doesn't seem to affect anything. Shall I try this with 2.6.33-ARCH kernel instead of 2.6.34 with CK patches?