Kernel 2.6.39-2-686-pae panics on unmount of external hd

[grizzler]

Just helped my cousin with a new LMDE installation. Clicked 'Safely remove...' on one of the harddisc icons on the desktop to unmount an external harddisc (Samsung USB portable 640GB with 9 partitions) and got a kernel panic.
I figured it probably had something to do with his pc (a Medion) and tried the same on my own pc (an unbranded machine with ECS motherboard) after I got home. Same thing happened.

So it would seem to be unrelated to the computer hardware as such (could be something about this particular external harddisc, of course). Anyone else experience anything similar?

Edit: just for the record, these are 32 bit systems.

Final edit: no more panic here with kernel 3.2.0-1-686-pae.

[rich_roast]

Could you post the

Code: Select all

sudo fdisk -l

along with the drive you were unmounting, please?

[AdamS]

I encountered several issues with 2.6.39.2 kernel.

broadcom drivers b43 & sta broke

broke i7 and i5 turbo support

broke acpi on several intel based builds

failure to mount external media

went back to 2.6.39.1 or 2.6.38-10 and all was well.

That kernel needs work yet lol.

[grizzler]

Could you post the

Code: Select all

sudo fdisk -l

along with the drive you were unmounting, please?

Code: Select all

Disk /dev/sdb: 640.1 GB, 640135028736 bytes
255 heads, 63 sectors/track, 77825 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x6780db81

   Device Boot      Start         End      Blocks   Id  System
/dev/sdb1   *           1       13055   104857600+   7  HPFS/NTFS
/dev/sdb2           13055       15726    21461462+  83  Linux
/dev/sdb3           15727       21070    42925680   83  Linux
/dev/sdb4           21071       77825   455884537+   5  Extended
/dev/sdb5           21071       26414    42925648+  83  Linux
/dev/sdb6           26415       31758    42925648+  83  Linux
/dev/sdb7           31759       37102    42925648+  83  Linux
/dev/sdb8           37103       42446    42925648+  83  Linux
/dev/sdb9           42447       47790    42925648+  83  Linux
/dev/sdb10          47791       77564   239159623+  83  Linux
/dev/sdb11          77565       77825     2096451   82  Linux swap / Solaris

The portable (which now has 11 partitions, by the way) shows up like this in the lsusb listing:
Bus 001 Device 002: ID 04e8:1f06 Samsung Electronics Co., Ltd

It's a Samsung S2 portable, type HX-MU064DA.

[Linux n00b]

Same problem here. I'm still searching for fix. I have 2.6.39-2 kernel, and from the search result, this is either kernel or one of the patch (some reported stock kernel doesn't have this problem) to kernel by distro.

There are reported fix (have to recompile kernel with couple of simple changes) with different distro. I also searched debian bug report, it seems debian hasn't got the hold of the problem yet.

It is also reported going back to 2.6.38-5 may "fix" the problem.

Appears kernel panic can also happen with DVD if it is not fixed. This issue really sucks.

[jeffreyC]

The Liquorix kernel does not have that problem, try it

http://liquorix.net/

[Linux n00b]

I'm here to report that the workaround posted on the ubuntu and debian forum works (the problem is wide spread). Only problem is you have to manually umount.

I guess sodu umount is not so hard.

[hopimet]

I do not have this kernel panic on my system.

Maybe it could be related to some hardware configurations ?

[Linux n00b]

I do not have this kernel panic on my system.

Maybe it could be related to some hardware configurations ?

Are you using 64bit or 32bit kernel (I use 2.6.39-2 AMD64)?

The kernel panic issue is widely spread after 2.6.85 kernel. It is not only Debian. It seems also not tied to particular hardware at this moment.

Three theories:

1. Somewhere in the kernel no one knows yet, since it happens only to some, but not all, and there is particular cause at the moment.
2. Some reported compile stock 2.6.39-2 kernel fixed it, and suspect the problem is related to udev.
3. Some guy says he nailed the problem with kernel, and it require recompile with simple "patch".

[hopimet]

I do not have this kernel panic on my system.

Maybe it could be related to some hardware configurations ?

Are you using 64bit or 32bit kernel (I use 2.6.39-2 AMD64)?

32.

It could be an explanation.

[lmdeman]

Maybe it's a different issue, but yesterday I experienced a system freeze right after a "Safely remove drive" before unplugging an USB hdd holding two partitions, one FAT32 and one NTFS. The funny (not that much, actually) thing is that it happened with a 2.6.32 64-bit kernel (LMDE tied to squeeze repos). I'm using such system since 6 months and it did never happen before. Since most external hdd are NTFS, I'm wondering if the kernel issues are actually related to FUSE. Did anyone experience this problem with non-NTFS USB drives?

[Linux n00b]

Debian has a minor kernel update. I updated to the latest kernel, and the problem is not fixed yet.

Here I attached the screen (I took the photo from my phone) if anyone interested, and hopefully has a way to fix it easily.

[grizzler]

There are a few posts on the Debian User forum that suggest this problem is still present in kernel 3.0.0-rc6. It may take a while before this is fixed...

[Linux n00b]

Latest report. I was suspecting the problem was to do with NTFS formated drive. So it might be NTFS-3G, FUSE , or udev. There was update to NTFS-3G, I was hoping it may fix the problem. But, nope, it didn't.

[grizzler]

No, it hasn't been fixed yet. What's worse, someone who was using Ubuntu 10.04.2 with kernel 2.6.32-32 without any problems reports that after some updates it is also present in 2.6.32-33. So whatever the problem is, it seems to have been 'backported' to earlier kernels. Great job, guys! (NOT!)

Still, for some people the recent updates appear to have helped: http://forums.linuxmint.com/viewtopic.p ... 40#p451161

[mccfrank]

I don't use Mint...but run Debian Sid and I have never been able to get the PAE kernel to boot on my machine.
The developers seem to think it's a hardware problem - I don't think so as I NEVER before had any problems
getting any kernels to boot.

[Linux n00b]

Whatever the problem is, it is not fixed. i just tried after both FUSE and udev update.

Now, rather than waiting the "official" fix, there should be a easy solution.

Since unmount from terminal works fine, just the GUI "safely remove" would cause the kernel panic, it is clear the fault is directly related with how the script/code linked to that particular GUI option. I remember back to Mint 6, there was only unmount option from GUI which always works. So why can't we change it back? Anyone knows how to do this easily for less experienced users, like myself?

[wilberfan]

I've been getting a kernel panic with this kernel at shutdown: http://forums.linuxmint.com/viewtopic.php?f=141&t=78154. I've used Liquorix kernels in the past; might be time to try it again!

[The Diamond Geezer]

I, for one, don't get a kernel panic when I click on "Safely remove drive" to unmount an external 2TB HDD, but a system freeze. All I can do is a hard reset. Thankfully this incurs no data loss on the external drive, which leads me to think that it is indeed successfully unmounted before the freeze. My kernel is 2.6.39-2. This has happened for me since 2.6.32.

The interesting and frustrating thing is that I used to be able to right-click and "Unmount" the drive from Nautilus without a problem before Update Pack 2. Now this option has been replaced by the "Eject" black upward arrow next to the drive name in the "Places" pane. Clicking on the "Eject" symbol simply reloads the external volume without unmounting it.

I suspect some Gnome Shell package is to be blamed for this *regression*. Pardon my lack of objectivity but I had a very poor user experience with Gnome Shell in Fedora 15 and now dread it to no end..

Thankfully, the proposed workaround (umount in CLI) should provide some relief to this painful glitch...for now.

[Linux n00b]

If Gnome shell is found to be the source of the problem, I'm sure I will switch to KDE. First there was retarded Unity, then stupid Gnome3. This madness must stop.