http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #2 from Alan Horsfield --- Created attachment 614965 --> http://bugzilla.opensuse.org/attachment.cgi?id=614965&action=edit Requested from 13.1 bootup sound works fine with 13.1 kernel -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #4 from Alan Horsfield --- Created attachment 614969 --> http://bugzilla.opensuse.org/attachment.cgi?id=614969&action=edit Rebooted back with kernel 13.2 again sound not working. attached dmesg.txt obtained with 13.2. will now attach hwinfo. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #6 from Takashi Iwai --- Thanks. Could you try a KMP in OBS home:tiwai:bnc907092/hda repo? Install the kmp corresponding to your running kernel, and reboot with 13.2 kernel. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #8 from Takashi Iwai --- Give the kernel message again. Also what shows /sbin/modinfo snd-hda-intel | grep filename and /sbin/modinfo snd-hda-controller | grep filename ? -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #10 from Takashi Iwai --- (In reply to Alan Horsfield from comment #9)

(In reply to Takashi Iwai from comment #8)

...

Give the kernel message again. Also what shows /sbin/modinfo snd-hda-intel | grep filename and /sbin/modinfo snd-hda-controller | grep filename ?

Is the kernel message the file generated by hwinfo? Is this the filename to be used in the above two commands?

No and no. I'd need the output of dmesg command, and "filename" is really the word "filename". -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #12 from Alan Horsfield --- Created attachment 615150 --> http://bugzilla.opensuse.org/attachment.cgi?id=615150&action=edit modinfo output controller -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #14 from Alan Horsfield --- Created attachment 615152 --> http://bugzilla.opensuse.org/attachment.cgi?id=615152&action=edit output from dmesg -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #16 from Takashi Iwai --- Thanks. So, it's not the CORB_RIRB_SELF_CLEAR thing that was recently introduced. Hmm, I need to check a wider range, then... For a better hint, could you try the following? - Uninstall hda-kmp-desktop package. rpm -e hda-kmp-desktop - Install the old kernels to narrow down the regression range. Try to install 3.13 at first. The package is available at OBS home:tiwai:kernel:3.13, http://download.opensuse.org/repositories/home:/tiwai:/kernel:/3.13/openSUSE... Take kernel-desktop.rpm, and install it via rpm -ivh kernel-desktop.rpm --oldpackage --nodeps Boot with this kernel, and check whether the sound device appears. - If 3.13 works, try 3.14 and 3.15 kernels, available at OBS home:tiwai:kernel:3.14 and home:tiwai:kernel:3.15 repos, respectively. - If 3.13 doesn't work, try 3.12 kernel from OBS Kernel:SLE12 repo: http://download.opensuse.org/repositories/Kernel:/SLE12/standard/ In that way, we can narrow down the range. After checking that the sound, you can uninstall these test kernels at any time. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #18 from Takashi Iwai --- Check the output of "rpm -q kernel-desktop". There is a systemd purge-kernel service that automatically cleans up old kernels, and some might have been already cleaned up. You can manage how many kernels are kept via /etc/zypp/zypp.conf; look for "multiversion" there. Back to the regression: so, from your experiment, the regression was introduced between 3.13 and 3.14. Hm, I didn't expect that... With 3.14 kernel, you see the similar kernel errors "no AFG or MFG node found" a few times and "no codecs initialized", right? Also, what happens if you stop graphics environment once, and reload snd-hda-intel module? That is, - Go to VT1 via ctl-alt-F1, login as root - Run "init 3", to stop X11 - Run "modprobe -r snd-hda-intel" If it's busy, try again later - Run "modprobe snd-hda-intel" again. Check the kernel message. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #20 from Alan Horsfield --- Created attachment 615231 --> http://bugzilla.opensuse.org/attachment.cgi?id=615231&action=edit dmesg report for 3.16-default -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 Mobeen Azhar changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |moby@pcsn.net --- Comment #22 from Mobeen Azhar --- I am in the same boat. Toshiba laptop, installed OSS 13.2, lost sound. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #26 from Takashi Iwai --- Currently I have no clue what broke. The best would be to bisect the kernel git commits to find out the culprit. But it can't be done without a real hardware here. Could you perform git bisect? -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #31 from Alan Horsfield --- After installing gcc it went smoothly till I tried to boot the the new kernel, 3.14.0testing. I got: Kernel panic-not syncing : VFS : unable to mount root fs on unknown -block(0,0). "make install" I believe said I may need to "make a raqmdisk now". I just hit return but what should I have done? -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #33 from Alan Horsfield --- Created attachment 623310 --> http://bugzilla.opensuse.org/attachment.cgi?id=623310&action=edit Output from running lsinitrd Thank you for your help and patience. I reran the whole process for 3.14 and it still would not boot. I booted to 3.16 and ran lsinitrd and there was no reference to ext2, ext3 or ext4. I have attached the output and an extract from linux/.boot. I am not sure where to gom from here. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #35 from Takashi Iwai --- (In reply to Alan Horsfield from comment #33)

Created attachment 623310 [details] Output from running lsinitrd

Thank you for your help and patience. I reran the whole process for 3.14 and it still would not boot.

OK, in your case, ext4 is built-in, so there is no module indeed. Did you try to build 3.16 kernel manually from localmodconfig and succeed? Changing between kernel versions may bring the config differences and break sometimes if you don't reset the config carefully. If you already managed to make the manual 3.16 kernel working, try similarly for 3.14; i.e. boot the packaged 3.14 kernel, rerun make localmodconfig from it, then build 3.14 manually. BTW, it's better to save .config file if the built kernel is properly running. Then you can reuse it when git bisect switch to that kernel again. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #41 from Takashi Iwai --- (In reply to Alan Horsfield from comment #40)

The sequence of the good (G) and bad (B) bisects was: BGBBGBBBBBBBB. Has something gone wrong as I expected another G near the end?

Don't worry, it's not always so. If it went really wrong, git bisect would complain you. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #43 from Takashi Iwai --- (In reply to Alan Horsfield from comment #42)

Thanks for quick reply. What do I do now?

Continue git bisect until it finishes. Git will tell you. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #45 from Takashi Iwai --- (In reply to Alan Horsfield from comment #44)

Thought I had finished but here is the output from the last git bisect:

linux-wda0:~/linux # git bisect bad 40c368c1ef6379db08d2ebd36f5a45efb93cff73 is the first bad commit commit 40c368c1ef6379db08d2ebd36f5a45efb93cff73 Author: Yijing Wang Date: Thu Dec 5 19:52:53 2013 +0800

PCI: Use dev_is_pci() to identify PCI devices

Use dev_is_pci() instead of checking bus type directly.

Signed-off-by: Yijing Wang Signed-off-by: Bjorn Helgaas

:040000 040000 c38465519330f18f0e4a9c9c262f884cee377add 3977ec42757beb69619de06f834ca1f9765cb1fb M drivers

Hmm, this looks very unlikely as a right spot, unfortunately. Maybe some git bisect went wrong then pointed to this end. It's often sensitive... You can see the whole git bisect history via "git bisect log" command. Could you put this to bugzilla? Then, retry git bisection again, but at this time, assuming that the regression has happend in the sound driver code. For doing this, start git bisection like % git bisect start sound include/sound This makes git bisect concentrating only on changes in sound/* and include/sound/* files. This should decrease the number of iterations, too. (You might need to reset the previous git bisect via "git bisect reset" once before restarting like the above.) -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #47 from Takashi Iwai --- (In reply to Alan Horsfield from comment #46)

Created attachment 623455 [details] git bisect log

Attached is the log file. I suspect that on one occasion I omitted "make modules-install" and the corresponding commit may well have been good. However I will do as you suggest and start again with the extra parameter.

Thanks. Another often seen failure is to boot a wrong kernel. The kernel version during bisection might change up and down occasionally depending on the development branch. That is, the later commit doesn't have to be a greater version, but sometimes the version goes back. So, be careful which kernel to boot during bisection. Always choose the right kernel manually from GRUB2 menu. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 Alan Horsfield changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |tiwai@suse.com Flags| |needinfo?(tiwai@suse.com) --- Comment #49 from Alan Horsfield --- Has this bug been abandoned or has there been any action since the bit bisect? At present I am unable to playback my CD's so I may have to consider moving from SuSE. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #51 from Alan Horsfield --- i installed kernel-obs-build-3.16.7-59.1.gfde41e8.x86-64.rpm but it was not recognised by my bootloader (YAST2). I have since installed and am now running with kernel-desktop-4.0.2-1.1.ga425d38.x86-64.rpm. It gave no sound so went to the YAST2-sound and it threw up the following: "An error occurred during installation of NM10/ICH7 Family H.D.Audio Controller. The kernel module snd-hda-intel for sound support could not be loaded. Can be caused by incorrect module parameters, including invalid IO or IRQ parameters." Your help would be much appreciated and hopefully with this new kernel it will resolve the problem. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 --- Comment #54 from Takashi Iwai --- Alan, could you try to boot with "pci=use_crs" option? This should be effective for all recent broken kernels. The very similar bug report on bugzilla.kernel.org was worked around with this. https://bugzilla.kernel.org/show_bug.cgi?id=99221 -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 http://bugzilla.opensuse.org/show_bug.cgi?id=907092#c56 Bjorn Helgaas changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |bhelgaas@google.com --- Comment #56 from Bjorn Helgaas --- Created attachment 638098 --> http://bugzilla.opensuse.org/attachment.cgi?id=638098&action=edit quirk to auto-enable pci=use_crs Could you please try this patch? With this patch, you should not need to boot with "pci=use_crs". Please attach the complete dmesg log if you test this. Also, I'd like to credit you as reporting the bug and testing the fix, so please let me know whether it's OK to publish your name and email address. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 http://bugzilla.opensuse.org/show_bug.cgi?id=907092#c58 Alan Horsfield changed: What |Removed |Added ---------------------------------------------------------------------------- Flags|needinfo?(alan@hazelgarth.c | |o.uk) | --- Comment #58 from Alan Horsfield --- (In reply to Bjorn Helgaas from comment #56)

Created attachment 638098 [details] quirk to auto-enable pci=use_crs

Could you please try this patch? With this patch, you should not need to boot with "pci=use_crs". Please attach the complete dmesg log if you test this.

Also, I'd like to credit you as reporting the bug and testing the fix, so please let me know whether it's OK to publish your name and email address.

Bjorn, I removed the pci=use_crs boot option via yast2, rebooted and clicked "link" on the dif page of attachment 638098, and rebooted. There was no sound. Is this what I should have done or am I supposed to edit a file myself. I am attaching the dmesg output file. No problem using my name and email. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 http://bugzilla.opensuse.org/show_bug.cgi?id=907092#c60 --- Comment #60 from Takashi Iwai --- Alan, try the kernel package mentioned in comment 57 to be sure. You didn't seem to test the thing correctly. (And please the correct dmesg output. Obviously a wrong thing was uploaded :) -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 http://bugzilla.opensuse.org/show_bug.cgi?id=907092#c62 --- Comment #62 from Alan Horsfield --- (In reply to Takashi Iwai from comment #60)

Alan, try the kernel package mentioned in comment 57 to be sure. You didn't seem to test the thing correctly.

(And please the correct dmesg output. Obviously a wrong thing was uploaded :)

Takashi, Thanks for comment 57. I have looked at the packages and there are 10 to choose from. Would you kindly tell me which one I should rpm? -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 http://bugzilla.opensuse.org/show_bug.cgi?id=907092#c64 --- Comment #64 from Takashi Iwai --- (In reply to Alan Horsfield from comment #62)

Takashi, Thanks for comment 57. I have looked at the packages and there are 10 to choose from. Would you kindly tell me which one I should rpm?

Try standard/i686/kernel-desktop.rpm (In reply to Alan Horsfield from comment #63)

Can you please tell me how I should have tested "the thing"?

Check the running kernel version. We can check from the dmesg output. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 http://bugzilla.opensuse.org/show_bug.cgi?id=907092#c66 --- Comment #66 from Takashi Iwai --- (In reply to Takashi Iwai from comment #64)

...

(In reply to Alan Horsfield from comment #62)

...

Takashi, Thanks for comment 57. I have looked at the packages and there are 10 to choose from. Would you kindly tell me which one I should rpm?

Try standard/i686/kernel-desktop.rpm

(In reply to Alan Horsfield from comment #63)

...

Can you please tell me how I should have tested "the thing"?

Check the running kernel version. We can check from the dmesg output.

Thanks, As I did not want both i686 and x86_64 on my laptop, I chose kernel-desktop-4.0.5-1.1.g4f0b419.x86_64.rpm.

I somehow wrongly remembered as if you were using 32bit kernel, but it was 64bit. Then x86-64 kernel-desktop.rpm is the right choice, don't worry.

You will be pleased to know it recovered sound.

Great. Would you mind if Bjorn put your name to the upstream commit as a tester?

Do I now stay with this until the next update or perhaps 13.3?

Keep it for a while. Once when the patch gets upstreamed, they will be backported to stable kernels, and I'll backport to openSUSE 13.1 and 13.2 updates, too.

Do you want the current dmesg output?

Yes. -- You are receiving this mail because: You are on the CC list for the bug.

http://bugzilla.opensuse.org/show_bug.cgi?id=907092 http://bugzilla.opensuse.org/show_bug.cgi?id=907092#c68 --- Comment #68 from Alan Horsfield --- (In reply to Takashi Iwai from comment #66)

(In reply to Takashi Iwai from comment #64)

...

...

(In reply to Alan Horsfield from comment #62)

...

Takashi, Thanks for comment 57. I have looked at the packages and there are 10 to choose from. Would you kindly tell me which one I should rpm?

Try standard/i686/kernel-desktop.rpm

(In reply to Alan Horsfield from comment #63)

...

Can you please tell me how I should have tested "the thing"?

Check the running kernel version. We can check from the dmesg output.

Thanks, As I did not want both i686 and x86_64 on my laptop, I chose kernel-desktop-4.0.5-1.1.g4f0b419.x86_64.rpm.

I somehow wrongly remembered as if you were using 32bit kernel, but it was 64bit. Then x86-64 kernel-desktop.rpm is the right choice, don't worry.

...

You will be pleased to know it recovered sound.

Great. Would you mind if Bjorn put your name to the upstream commit as a tester?

...

Do I now stay with this until the next update or perhaps 13.3?

Keep it for a while. Once when the patch gets upstreamed, they will be backported to stable kernels, and I'll backport to openSUSE 13.1 and 13.2 updates, too.

...

Do you want the current dmesg output?

Yes.

Quite happy for Bjorn to use my name, not that I have done anything. -- You are receiving this mail because: You are on the CC list for the bug.