Bugzilla – Bug 140238
Install hung at "Hardware Configuration" when probing snd_hda_intel
Last modified: 2008-06-25 09:48:30 UTC
During OSS 10.1 Aplha4 install, after SW install during Hardware Configuration, the detection of the sound device hangs with busy cursor in graphical install (f7) console. switching to F10 cosole reveals continuous repetition of the following messages: Dec 19 20:52:10 Linux Kernel: printk: 12 messages suppressed Dec 19 20:52:10 Linux Kernel: ALSA sound/pci/hda/hda_intel.c:520: azx_get_response timeout ps reveals modprobe of snd_hda_intel Attempts to rmmod -f or kill the process have no result on the modprobe or the install shell. The only apparent way to continue is force a reboot, which has unknown impact due to incomplete HW configuration step of install. Install kernel has acpi=off since with acpi on, install kernel fails to load.
Created attachment 61435 [details] /var/log/YaST2 files
Created attachment 61436 [details] hwinfo --sound results
Created attachment 61437 [details] dmesg results
Changing to Major from Blocker since it is possible to continue by performing a hard reset. This bypasses the HW config step of the install, putting the resulting system in an unconfigured state, but it will boot and I can perform manual configuration as needed. This is still major since average user will not know this and install should not hang like this. After reboot, any attempt to load the sound drivers (snd_hda_intel) has the same results as install, modprobe is stuck and neverending messages at console about "azx_get_response timeout"
Takashi, HDA driver cannot be loaded in this case.
It looks like a problem with interrupts. Try add 'lapic' or 'pci=biosirq'. Anyway, the main problem is that ACPI must be enabled for your hardware, otherwise not all devices are properly initialized (see unitialized USB controllers). So it looks like an ACPI not sound bug.
Well, I was never able to install or boot with ACPI enabled (see #140649). I have now upgraded to 2.6.15-rc6 from SL-OSS-Factory and can boot with ACPI, and have no problems with sound now. Alpha4 kernel (2.6.15-rc5) seems to have been a corrupt or generally broken version for many people with ACPI dependant HW. Resolve this assuming Beta1 will verify Intel HDA sound driver functions and does not hang on load.
I also bumped into this problem while installing SUSE10... so I did the network installation which allows for skiping of sound card detection. I have kernel 2.6.15-rc6 from SL-OSS-factory installation via network (DISTVERSION 10.0.42-Alpha2). I have no problem with ACPI, but when it comes to sound I get the originally mentioned problem of repeated "azx_get_response timeout" messages. Therefore, I highly doubt ACPI or kernel being the problem but the sound and most likely ALSA driver seem to be the issue.
Shane Helms: Could you attach 'dmesg' output immediately after boot of problematic kernel?
Created attachment 63245 [details] dmesg-kernel-2.6.13-default
Created attachment 63246 [details] var-log-messages
I had too many problems with SL-OSS-factory version, so I've reinstalled the SUSE10 release version. I've attached my default kernel dmesg log, as well as the /var/log/messages at the point of loading the snd drivers and hence the whole hanging issue. I'm going to start upgrading my kernel and ALSA drivers on this release. I've also found this link which i might follow on if all other means fail. http://downloadfinder.intel.com/scripts-df-external/Detail_Desc.aspx?agr=N&DwnldID=7760
We don't debug 10.0 kernel any more for these problems. So, try the latest kernel (preferably KOTD). If it still doesn't work, too, try "single_cmd=1" module option for snd-hda-intel driver.
Created attachment 65018 [details] dmesg of kernel of the day (2.6.16-rc1-git3-20060124182340-default)
Created attachment 65019 [details] dmesg (2.6.16-rc1-git3-20060124182340-default) with "single_cmd=1" module option
First attachment is dmesg from kernel of the day following loading of the ALSA driver, once again timeouts and hanging issues. The second is the log following ALSA loading with the mentioned "single_cmd=1" option. The driver loads without hanging but is buggy. The playback is intermittent and noisy. Don't know if this is of any interest to you or not, I could bringup the alsamixer, and noticed the main volume is not associated with the PCM or Front channels, but the Headphone channel.
Looks like the interrupt handling is still pretty broken. Thomas, do you have new kernel packages with the latest ACPI patchset for external testers?
pci=noacpi and/or noapic could be also worth trying to boot with. The newest ACPI patches are included here(could need some hours until they pop up): ftp.suse.com/pub/people/trenn/2.6.16_rc1_git3-ACPI_test/*.rpm Chances are low that it helps but it's worth trying, ACPI guys are currently thinking about rewriting the pci/driver/interrupt_assigning parts of ACPI, but this will take some months until anything is popping up mainline (if ever).
Try the latest kernel from KOTD.
No reaction since more than 2 month on needinfo. Shane, please response till Sunday midnight, I will close it invalid on Monday, if we geht no feeback.
Which Shane, there are two of us providing info on this report. I have no issue anymore as noted in comment #7.
I'm using Suse 10 on a Travelmate 8103 and have the same problem as listed on https://bugzilla.novell.com/show_bug.cgi?id=167029 Basicaly I found a way to turn on the sound, but have to turn it on every time I turn on the laptop, which I described on the last comment of that bug post but would like to try "single_cmd=1" module option for snd-hda-intel driver, but am not sure of where to go to do that.
Hi Harald Mueller-Ney, Frankly speaking, I couldn't live without the sound on my laptop, and various attempts and tests of the kernel-of-the-day revisions lead me to no good. So, I took a friends advice on trying the gentoo distro (despite its ultra-painful configuration process). I tried 4 different kernel sources from gentoo distro: - vanilla-sources-2.6.16 (Full sources for the Linux kernel) - mm-sources-2.6.16 (Andrew Morton's kernel) - gentoo-sources-2.6.14 - suspend-sources-2.6.14 & 16 (Software Suspend 2 + Gentoo patchset sources) and every single one of them worked with my sound card! I don't know if this info can be of any help to you, or if I can provide you with any useful configurations from my current operating system, but I no longer have SUSE on my laptop. I hope others don't experience this problem, and as Shane Bryan's comment suggests that the bug is progressively resolved in newer verions of SUSE distro. Thanks, Shane.
Releasing the needinfo to get it back on Takashis radar.
As I mentioned, it's not the sound driver itself but the interrupt routing.
This looks suspicious: PCI: Bus #07 (-#0a) may be hidden behind transparent bridge #06 (-#07) (try 'pci=assign-busses') Please try pci=assign-busses boot param.
This bug is still in NEEDINFO. What is the status here?
This does not look good: usb usb5: configuration #1 chosen from 1 choice hub 5-0:1.0: USB hub found hub 5-0:1.0: 8 ports detected irq 7: nobody cared (try booting with the "irqpoll" option) [<c0134d47>] __report_bad_irq+0x2b/0x69 [<c0134f04>] note_interrupt+0x17f/0x1ab [<c013486a>] handle_IRQ_event+0x23/0x4c [<c013492d>] __do_IRQ+0x9a/0xcd [<c0104f1c>] do_IRQ+0x46/0x56 [<c0103afa>] common_interrupt+0x1a/0x20 [<c028007b>] inet6_lookup+0x1c0/0x1d4 [<c011c1b9>] __do_softirq+0x2c/0x7f [<c011c22e>] do_softirq+0x22/0x26 [<c0104f21>] do_IRQ+0x4b/0x56 [<c0103afa>] common_interrupt+0x1a/0x20 [<c014ba78>] check_poison_obj+0x3c/0x149 [<c014bbc9>] cache_alloc_debugcheck_after+0x1f/0xea [<c014cc43>] kmem_cache_alloc+0x70/0x79 [<c015b772>] getname+0x1a/0x98 [<c015b772>] getname+0x1a/0x98 [<c010188a>] sys_execve+0xc/0x69 [<c0102a3b>] sysenter_past_esp+0x54/0x79 handlers: [<e0c45548>] (usb_hcd_irq+0x0/0x4f [usbcore]) Disabling IRQ #7 IIRC similar problems were popping up on other machines (mobile team). I remember that some machines needed irqpoll and that something got fixed (by Andi or Greg?) to avoid that. If Shane could confirm whether final Code10 does (not-)work we could close this one or give it another try (best with latest dmesg of Code10 kernel).
I'm having the same problem here, System is a fresh suse 10.1 install. Machine in question is an acer aspire 3102WLMi laptop, amd turion CPU, 64bit installation, 512mb ram. After quite a lot of those "azx_get_response timeout" messages, the kernel crashes. And since my only way to connect that laptop to the internet is a madwifi card, I wasn't even able to install any updates during installation, and now the machine won't boot anymore. "safe mode" boot doesn't work either. How do I get rid of that soundcard configuration, so that I at least can boot to install updates?
... did I write turion cpu? its an amd sempron 3200+
ok, i booted a rescue system, moved the hda-intel modules out of the way, and was able to upgrade. with the latest you kernel, the machine continues to boot, but sound still crashes the kernel module. here's dmesg: ALSA sound/pci/hda/hda_codec.c:204: hda_codec: invalid dep_range_val 0:7fff ALSA sound/pci/hda/hda_codec.c:204: hda_codec: invalid dep_range_val 0:7fff ALSA sound/pci/hda/hda_intel.c:581: hda-intel: get_response timeout: IRS=0x0 ALSA sound/pci/hda/hda_codec.c:204: hda_codec: invalid dep_range_val 0:7fff ALSA sound/pci/hda/hda_codec.c:204: hda_codec: invalid dep_range_val 0:7fff ALSA sound/pci/hda/hda_intel.c:581: hda-intel: get_response timeout: IRS=0x0 ALSA sound/pci/hda/hda_codec.c:204: hda_codec: invalid dep_range_val 0:7fff ALSA sound/pci/hda/hda_codec.c:204: hda_codec: invalid dep_range_val 0:7fff ALSA sound/pci/hda/hda_intel.c:581: hda-intel: get_response timeout: IRS=0x0 ALSA sound/pci/hda/hda_codec.c:204: hda_codec: invalid dep_range_val 0:7fff ALSA sound/pci/hda/hda_intel.c:581: hda-intel: get_response timeout: IRS=0x0 ALSA sound/pci/hda/hda_intel.c:581: hda-intel: get_response timeout: IRS=0x0 ALSA sound/pci/hda/hda_generic.c:621: hda_generic: no proper input path found ALSA sound/pci/hda/hda_generic.c:398: hda_generic: no proper output path found ALSA sound/pci/hda/hda_generic.c:864: hda_generic: no PCM found Unable to handle kernel NULL pointer dereference at 0000000000000014 RIP: <ffffffff881fccb0>{:snd_hda_codec:create_mixer+85} PGD 19661067 PUD 19625067 PMD 0 Oops: 0000 [1] last sysfs file: /block/hda/removable CPU 0 Modules linked in: wlan_scan_sta snd_hda_intel ath_pci snd_hda_codec ath_rate_sample ide_cd pcmcia firmware_class shpchp pci_hotplug 8139cp wlan snd_pcm snd_timer snd soundcore snd_page_alloc cdrom 8139too ath_hal mii ohci_hcd ehci_hcd yenta_socket rsrc_nonstatic pcmcia_core usbcore xfs exportfs fan thermal processor atiixp ide_disk ide_core Pid: 1428, comm: modprobe Tainted: P U 2.6.16.21-0.13-default #1 RIP: 0010:[<ffffffff881fccb0>] <ffffffff881fccb0>{:snd_hda_codec:create_mixer+85} RSP: 0018:ffff810018ecd9c8 EFLAGS: 00010296 RAX: 0000000000000013 RBX: ffff810019fc1400 RCX: 0000000000000008 RDX: 0000000000000008 RSI: fffffffffffffff6 RDI: ffffffff882062dc RBP: 0000000000000000 R08: 00000000ffffffff R09: 0000000000000000 R10: 0000000000000011 R11: 0000000000000000 R12: 0000000000000000 R13: ffff810019fbc800 R14: ffffffff882062dc R15: ffffffff882062ed FS: 00002b0323a346d0(0000) GS:ffffffff803f7000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 0000000000000014 CR3: 0000000019677000 CR4: 00000000000006e0 Process modprobe (pid: 1428, threadinfo ffff810018ecc000, task ffff81001b76d7d0) Stack: 0000000000000002 00000000ffffffff 0000000000000400 ffffffff80387b20 ffffffff881654fd 0000000000000400 ffffffff80318460 0000000000000256 0000004419fbc810 000200d000000000 Call Trace: <ffffffff8014ad84>{__alloc_pages+101} <ffffffff8014ad84>{__alloc_pages+101} <ffffffff8015f89e>{cache_alloc_refill+657} <ffffffff8014ef97>{kzalloc+9} <ffffffff88161652>{:snd:snd_ctl_new+177} <ffffffff881fd3ba>{:snd_hda_codec:build_generic_controls+153} <ffffffff881fa583>{:snd_hda_codec:snd_hda_codec_read+96} <ffffffff881fb7c3>{:snd_hda_codec:snd_hda_create_spdif_out_ctls+104} <ffffffff881fc740>{:snd_hda_codec:snd_hda_build_controls+29} <ffffffff8823f4a6>{:snd_hda_intel:azx_probe+1846} <ffffffff8823e5d7>{:snd_hda_intel:azx_send_cmd+0} <ffffffff8823e4fb>{:snd_hda_intel:azx_get_response+0} <ffffffff801d46d5>{pci_device_probe+76} <ffffffff80230d88>{driver_probe_device+82} <ffffffff80230ead>{__driver_attach+107} <ffffffff80230e42>{__driver_attach+0} <ffffffff802307af>{bus_for_each_dev+67} <ffffffff8023041f>{bus_add_driver+126} <ffffffff801d4870>{__pci_register_driver+110} <ffffffff8013c000>{sys_init_module+4819} <ffffffff8010a3b6>{system_call+126} Code: 8b 45 14 a8 02 0f 84 96 00 00 00 83 7d 28 00 0f 89 8c 00 00 RIP <ffffffff881fccb0>{:snd_hda_codec:create_mixer+85} RSP <ffff810018ecd9c8> CR2: 0000000000000014 is there any chance that this bug will be fixed? or will this stay on "NEEDINFO" forever, and i should open a new one?
(In reply to comment #29) I'm also reporting the same problem on a SUSE 10.1 installation. Computer: laptop Acer Aspire 5102WLMi, with Turion 64x2. Linux 64-bit version, updated with the latest patches. I tried to boot the Live/Installation DVD of Ubuntu 6.06 LTS, and the sound worked perfectly! So, to me, it looks a SUSE's problem... :-( Actually, I didn't check if this working Ubuntu was 32- or 64-bit. > After quite a lot of those "azx_get_response timeout" messages, the kernel > crashes. I think i get some hundreds of those message before the kernel crashes. Mathias, if you'll open a new bug, please inform me so I can report my experience there too! Even though, I'm not sure if another bug should be opened, because in my opinion the problem is already well described here...
but this bug seems to be frozen in "NEEDINFO" since the person who needs to supply the info doesn't supply... so, if suse needs more info about this, they should ask someone else, or tell us to open a new bug.
after putting "options snd-hda-intel single_cmd=1" in /etc/modprobe.conf.local, the error messages change to a different message, something about an invalid range 0:7fff or somesuch, and the machine hangs on the kernel crash.
Has anyone already played with some apic boot options as suggested in comment #6 and #18? lapic noapic pci=noacpi there is also: apic=verbose (some apic output) apic=debug (some more apic output) If none above boot params help, can someone boot with apic=debug and attach whole dmesg output, please.
(In reply to comment #35) > Has anyone already played with some apic boot options as suggested in comment > #6 and #18? > lapic > noapic > pci=noacpi Nothing works with any of these, including 'pci=biosirq' (I tried them in turn). > If none above boot params help, can someone > boot with apic=debug and attach whole dmesg output, please. It wasn't the easiest job, because the dmesg's internal buffer gets instantly saturated by thousands of: ALSA sound/pci/hda/hda_intel.c:581: hda-intel: get_response timeout: IRS=0x0 ALSA sound/pci/hda/hda_codec.c:204: hda_codec: invalid dep_range_val 0:7fff ALSA sound/pci/hda/hda_codec.c:204: hda_codec: invalid dep_range_val 0:7fff I'm attaching the dmesg some instants before the flooding; you can see that the last line is a message by "our" module.
Created attachment 97794 [details] dmesg just before module insertion. SUSE 10.1 on AMD Turion 64x2 with "apic=debug" boot parameter File referred by (comment #36)
from dmesg above: > pcie_portdrv_probe->Dev[5a37:1002] has invalid IRQ. Check vendor BIOS > assign_interrupt_mode Found MSI capability -> maybe boot param nomsi helps? Jaroslav also stated in comment #6 to try: pci=biosirq boot param, has this already be tested? > PCI: Cannot allocate resource region 7 of bridge 0000:00:04.0 > PCI: Cannot allocate resource region 8 of bridge 0000:00:04.0 > PCI: Cannot allocate resource region 9 of bridge 0000:00:04.0 > PCI: Cannot allocate resource region 7 of bridge 0000:00:05.0 > PCI: Cannot allocate resource region 8 of bridge 0000:00:05.0 This also looks suspicious. But this should be part of acpi's irq/resource processsing and should get disabled with pci=noacpi. When this does also not work I think it's something else. Hmm, I don't have the sources right now. I am back next week and will check. If you could also attach acpidump, I can try to check the resources that should get used for that device (but it's likely not this if pci=noacpi does not help). Can you also paste /proc/interrupts and /proc/iomem, maybe there is something obvious strange... Could you also describe this machine(s) a bit. CPU, how old, vendor, ...
brand new acer aspire 3102 WLMi laptop, amd mobile sempron 3200+ cpu pci=noacpi didnt help, still getting those hda_codec: invalid dep_range_val 0:7fff lines, followed by a crashing module. boot param "nomsi" didnt help either, same crash in snd_hda_codec:create_mixer+85 boot param "noapic" didnt help either, same as above.
even in "safe mode" it crashes. the only way right now to get a partially usable machine is boot the rescue system, and move the snd-hda-intel and snd-hda-codec modules out of the way.
the other boot parameters that have been suggested didnt help either. lapic, noapic, pci=biosirq...
Created attachment 97829 [details] acpidump from acer aspire 3102 WLMi
Created attachment 97830 [details] /proc/interrupts and /proc/iomem from acer aspire 3102 WLMi
(In reply to comment #38) > -> maybe boot param nomsi helps? No, it doesn't help. :-( > Jaroslav also stated in comment #6 to try: > pci=biosirq boot param, has this already be tested? Yes, I did, as I wrote in comment #36. It doesn't help. > This also looks suspicious. But this should be part of acpi's irq/resource > processsing and should get disabled with pci=noacpi. When this does also not > work I think it's something else. Hmm, I don't have the sources right now. I am > back next week and will check. Thank you. > Could you also describe this machine(s) a bit. CPU, how old, vendor, ... Machine is an Acer laptop. Model: Aspire 5102WLMi. Turion 64x2 1.6GHz 1 GB RAM ATI Radeon Xpress 1100 with ATI's official driver. Wireless LAN by Atheros, drived by ndiswrapper Realtek ethernet card ENE Technology multimedia card reader (unused) Some chipset by ATI The laptop was bought around one week ago.
Created attachment 97836 [details] acpidump on Acer Aspire 5102WLMi
Created attachment 97837 [details] /proc/interrupts and /proc/iomem on Acer Aspire 5102WLMi
just for information: after booting into livecd mode from the kubuntu 6.06 dvd, sound works with single_cmd=1 so i guess this really is a suse problem. btw, except for the cpu the 3102 is the same as the 5102. and a comment for #44: http://madwifi.org/suse/ to get rid of ndiswrapper ;)
Can you try to build externally ALSA modules from alsa-driver-1.0.13rc1 release on www.alsa-project.org? This includes fixes for Acer laptops, and some workaround for broken BIOS. Also, the latest kernel from openSUSE 10.2 is worth to try. It's based on 2.6.18 kernel.
where do i get that kernel? and if that fixes the problem, will that fix make it back into official 10.1 patches?
I'm going to install kernel and alsa from factory this evening...
I have same problem with Acer 5102WLMi, 10.1 SuSE, i tried to install alsa-driver-1.0.12rc4 with a latest kernel (?2.16.21), and not resolved the problem. I find a missing link on kernel source tree, when tried to add sound card, from kernel source ..../kernel/sound/pci/... the pci directory is mising. Maybe this is the source of prolem, but not sure. In the same time my touchpad are gone away - it is possibly a indication of PCI recognition problem, written above.
It's alsa-driver-1.0.13rc1 that includes some fixes for Acer laptops. 1.0.12 doesn't work.
I tried from realtek homepage: realtek-linux-audiopack-4.05b.tar.bz2 driver (2006.sep.01) same results.
ok, kernel and alsa from factory didnt help. same problem as before.
10.2 kernel is pretty close to vanilla 2.6.18, which doesn't include recent fixes for Acer laptops. Try to build ALSA modules externally from alsa-driver 1.0.13rc1 pacakge.
latest "you" kernel (2.6.16.21_0.8) + alsa-driver 1.0.13rc1 didnt help that much; still the same messages, but with single_cmd=1 the soundcard partially works (left channel only, no decent mixer settings, KDE arts doesnt recognize it as a valid sound device).
(In reply to comment #53) > I tried from realtek homepage: > realtek-linux-audiopack-4.05b.tar.bz2 > driver (2006.sep.01) same results. > > where did you find that?
In reply to comment #57) > where did you find that? http://www.realtek.com.tw/downloads/dlhd-2.aspx?lineid=2004052&famid=2004052&series=2004061&Software=True&title=HD%20Audio%20CODECs
update: there's a bios update on www.acer.de for the 3100/5100 series. version number goes up from (on my laptop) 1.30 to 2.00. didn't help, though.
partial success by now: alsa-driver 1.0.13rc1, kernel-default-2.6.16.21_0.13, options snd-hda-intel model=acer position_fix=1 in /etc/modules.conf.local, sound works on my acer aspire 3102 wlmi, but alsa seems not to have a mixer device? pippin:/proc/asound # alsamixer alsamixer: function snd_mixer_load failed: Invalid argument
(In reply to comment #60) things are starting to work here too. But with the same problems you encountered. That means: I get sound but I don't get alsamixer, and my log file is flooded. System information: alsa-driver 1.0.13rc1, kernel 2.6.16.21-0.13-smp by SUSE Laptop's BIOS updated (thanks for the news! :-) The mixer actually worked the very moment I installed the driver with alsaconf. Now I rebooted and it stopped working, as you report. > options snd-hda-intel model=acer position_fix=1 I set them too, hoping they could stop the log-flooding. But they didn't. For this boot, I've got 2537 lines in my /var/log/messages: Sep 7 21:56:59 linux kernel: ALSA /home/carlo/temp/alsa-driver-1.0.13rc1/pci/hda/hda_codec.c:213: hda_codec: invalid dep_range_val 0:7fff ...and a little reply to comment #47: if madwifi worked in Ad-Hoc mode I'd use that! But it doesn't... :-(
(In reply to comment #61) > ...and a little reply to comment #47: if madwifi worked in Ad-Hoc mode I'd use > that! But it doesn't... :-( > it doesn't? didn't know that. but then i'm running in managed mode... why would you need ad-hoc mode, anyways?
alsa-driver 1.0.13rc2, kernel-default-2.6.16.21_0.21, after installation I could set the driver in yast. I get system sounds, amarok player failed and crashed, and i Kmixer showed the chanels I can change volume, but not worked. After rebooting crash, with messeages .../hda_codec.c:213: hda_codec: invalid dep_range_val 0:7fff After system start with safe settings is half OK, I could log into, and sound settings changed, mixer is disabled. :-(
alsa-driver 1.0.14rc1, kernel-default-2.6.16.27-0.6 : same problem, partial sound, no alsa mixer. is there ANYTHING happening here at all?
Yes, the fix is in the upstream for 2.6.21 kernel. Resolved to LATER for now until openSUSE kernel gets updated.
just for the record: i've moved on to 10.2 by now, and 10.2 has the same problem, see bug 230740 it seems NOT to be a driver or kernel problem; at least, the fact that the alsa mixer DOES work for a while (or rather, until your first reboot after the completed installation) makes it look like that. even more so if you consider the fact that a) realtek themselves supply a really old alsa version (1.0.9) as their driver for this chipset, so they either know that it should work, or they don't give a damn. b) other linux versions work just fine with it (also see bug 230740)
mass reopening all SuSE Linux bugs that are set to REMIND+LATER to change the resolution to WONTFIX (adapting to new policy)
Already fixed in the recent release.