Bug 14438 - [PATCH] IDE Probe fails to find CD drive
Summary: [PATCH] IDE Probe fails to find CD drive
Status: CLOSED CODE_FIX
Alias: None
Product: IO/Storage
Classification: Unclassified
Component: IDE (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: io_ide@kernel-bugs.osdl.org
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2009-10-18 19:53 UTC by Matt
Modified: 2012-06-13 20:27 UTC (History)
5 users (show)

See Also:
Kernel Version: 2.6.31
Subsystem:
Regression: No
Bisected commit-id:


Attachments
System information for 2.6.30 and 2.6.31 (53.02 KB, application/x-gtar)
2009-10-18 19:53 UTC, Matt
Details
System info as above, but for 2.6.30, 2.6.31 and 2.6.32 (78.04 KB, application/x-gtar)
2009-10-18 20:48 UTC, Matt
Details
Fix for vmware IDE cdrom detection (1.16 KB, patch)
2010-03-03 12:08 UTC, Thomas Jarosch
Details | Diff

Description Matt 2009-10-18 19:53:35 UTC
Created attachment 23463 [details]
System information for 2.6.30 and 2.6.31

Since updating to 2.6.31 my CD drive is no longer detected.  The drive is still detected in version 2.6.30.

The offending part of dmesg appears to be:

  Probing IDE interface ide1...
  ide1: no devices on the port

Compared with 2.6.30's: 

  Probing IDE interface ide1...
  hdc: PLEXTOR CD-R PX-W8432T, ATAPI CD/DVD-ROM drive 
  hdc: host max PIO5 wanted PIO255(auto-tune) selected PIO4
  hdc: MWDMA2 mode selected

I have checked to make sure that the CD drive is set to master (it is), and have also tried kernel 2.6.32-rc5 with identical results.

I cannot find any obvious differences in the kernel configurations that may be causing the problem.

I have attached various logs / proc / config files that i think are appropriate.  However, i cannot find much information that distinguishes them.
Comment 1 Matt 2009-10-18 20:48:49 UTC
Created attachment 23464 [details]
System info as above, but for 2.6.30, 2.6.31 and 2.6.32

Resubmitted with information for 2.6.32-rc4
Comment 2 Jean Delvare 2009-10-18 20:55:21 UTC
Please don't create bugs in bugzilla when you're drunk.
Comment 3 kernel 2009-10-18 21:10:25 UTC
(In reply to comment #2)
> Please don't create bugs in bugzilla when you're drunk.

He clearly isn't drunk, and you clearly are not as funny or clever as you think you are.

This kind of arrogance and unprovoked rudeness should not be tolerated by any open source project, you drag the project your working for down as much as you do people's opinions of yourself.
Comment 4 Borislav Petkov 2009-10-20 06:58:38 UTC
Hi,

does libata also fail to detect your drive? I.e., turn off CONFIG_IDE and enable CONFIG_PATA_VIA.
Comment 5 Matt 2009-10-20 08:07:21 UTC
libata seems to find the drive fine -- abcde and mplayer are happy with it, vlc won't play it, but i suspect that's vlc's problem.

I play around a bit tonight to see if the change is fine -- my drives all have different names now...

Thanks.
Comment 6 Borislav Petkov 2009-10-20 11:45:15 UTC
(In reply to comment #5)
> I play around a bit tonight to see if the change is fine -- my drives all
> have
> different names now...

You can always have stable device names with udev:
http://reactivated.net/writing_udev_rules.html
Comment 7 andrea 2010-01-11 13:49:29 UTC
I had a problem which sounds very similar last week.

I started a forum thread here (before I discovered this DB):
http://www.linuxquestions.org/questions/linux-virtualization-90/2.6.31-kernel-guest-os-on-vmware-esx-3.5.0-780042/ 

The solution (for us) was to regress out this kernel change:
http://kerneltrap.org/mailarchive/git-commits-head/2009/6/12/5948973

A more resilient fix might be to leave ide_probe_port() as-is but to fix 
ide_wait_not_busy() to return the last successful rc code (only returning -ENODEV if strictly no device is detected on the IDE port). I do not know what other repercussions such a fix might have however.
Comment 8 Thomas Jarosch 2010-03-03 12:07:18 UTC
Same issue here on vmware server 1.08: The IDE cdrom is not detected anymore on kernel 2.6.32.9. I've reapplied the patch to 2.6.32.9 and will attach it here for easier access.
Comment 9 Thomas Jarosch 2010-03-03 12:08:01 UTC
Created attachment 25337 [details]
Fix for vmware IDE cdrom detection
Comment 10 Alan 2010-03-03 12:15:54 UTC
See Documentation/SubmittingPatches for how to get patches accepted. The person handling IDE now is Dave Miller.
Comment 11 Thomas Jarosch 2010-03-03 12:22:52 UTC
Thanks for the tip, Alan. The patch was meant as a hint or PoC, there must be a reason why the IDE controller is returning "no devices attached". Just added it for the IDE maintainer to get an idea what might be wrong.

I also can't sign the patch as I didn't write it in the first place, can I?

Reverting the commit mentioned above should do the same trick:
http://kerneltrap.org/mailarchive/git-commits-head/2009/6/12/5948973

btw: I guess it's a good idea to add the original author as CC:.
Comment 12 Bartlomiej Zolnierkiewicz 2010-03-27 10:30:37 UTC
Please always bring your upstream kernel problem with current upstream maintainer
(I'm really sorry but I have absolutely no time to spend on investigating current IDE problems).
Comment 13 Thomas Jarosch 2010-05-07 09:29:13 UTC
The 2.6.33 code looked fixed to me and I just verified it works again with 2.6.33.2 and VMware server 1.0.8.

This issue is resolved in 2.6.33, please close it.

Note You need to log in before you can comment on or make changes to this bug.