Skip to contentSkip to conversation

kernel BUG at drivers/ide/ide.c:1384!

DISCUSSION Ralf Hildebrandt 3 participants 6 messages Jan 9, 2006

lore
Share & tools
lore
#0 lore raw

I invoked "hdparm -w /dev/hda"

# uname -a
Linux hummus.charite.de 2.6.15 #1 Tue Jan 3 09:30:04 CET 2006 i686 GNU/Linux

Before you flame away at me for using the nvidia kernel module: I will reproduce this WITHOUT the nvidia kernel module. At least I'll try.

Thanks.

Jan  9 09:32:01 hummus kernel: end_request: I/O error, dev hda, sector 22604679
Jan  9 09:32:01 hummus kernel: ------------[ cut here ]------------
Jan  9 09:32:01 hummus kernel: kernel BUG at drivers/ide/ide.c:1384!
Jan  9 09:32:01 hummus kernel: invalid operand: 0000 [#1]
Jan  9 09:32:01 hummus kernel: Modules linked in: nvidia agpgart binfmt_misc lp thermal fan button processor ac battery pcmcia af_packet 8250_pnp rtc eth1394 ipw2200 ieee80211 ieee80211_crypt firmware_class ohci1394 ieee1394 yenta_socket rsrc_nonstatic pcmcia_core tg3 snd_intel8x0m 8250_pci 8250 serial_core snd_intel8x0 snd_ac97_codec snd_ac97_bus snd_pcm_oss snd_mixer_oss snd_pcm snd_timer snd soundcore snd_page_alloc usbhid ehci_hcd hci_usb bluetooth uhci_hcd usbcore deflate zlib_deflate zlib_inflate twofish serpent aes blowfish des sha256 sha1 md5 crypto_null af_key dm_mod
Jan  9 09:32:01 hummus kernel: CPU:    0
Jan  9 09:32:01 hummus kernel: EIP:    0060:[<c02064a1>]    Tainted: P      VLI
Jan  9 09:32:01 hummus kernel: EFLAGS: 00210086   (2.6.15)
Jan  9 09:32:01 hummus kernel: EIP is at generic_ide_ioctl+0x4e6/0x5b1
Jan  9 09:32:01 hummus kernel: eax: c1b8e0c0   ebx: 00200206   ecx: 00000002   edx: 00000001
Jan  9 09:32:01 hummus kernel: esi: c0379614   edi: c020e1ee   ebp: f61f0080   esp: c8f99ec0
Jan  9 09:32:01 hummus kernel: ds: 007b   es: 007b   ss: 0068
Jan  9 09:32:01 hummus kernel: Process hdparm (pid: 16818, threadinfo=c8f99000 task=ce026070)
Jan  9 09:32:01 hummus kernel: Stack: c02c9ee0 00000abc 000000c8 ebb1164c c1709ec0 c0132f0d 00000001 c01bacee
Jan  9 09:32:01 hummus kernel: c8f99f3c dd5d4c24 f693d440 f693d484 ebb115ac 00000000 00000002 c02c9ee0
Jan  9 09:32:01 hummus kernel: 00000abc c1709ec0 00000abc 00000000 0000031c c18e3b24 f61f0080 c0211e11
Jan  9 09:32:01 hummus kernel: Call Trace:
Jan  9 09:32:01 hummus kernel: [<c0132f0d>] filemap_nopage+0x2b3/0x363
Jan  9 09:32:01 hummus kernel: [<c01bacee>] bit_cursor+0x0/0x5b4
Jan  9 09:32:01 hummus kernel: [<c0211e11>] idedisk_ioctl+0x26/0x2c
Jan  9 09:32:01 hummus kernel: [<c0211deb>] idedisk_ioctl+0x0/0x2c
Jan  9 09:32:01 hummus kernel: [<c019e051>] blkdev_driver_ioctl+0x42/0x44
Jan  9 09:32:01 hummus kernel: [<c019e15f>] blkdev_ioctl+0x10c/0x143
Jan  9 09:32:01 hummus kernel: [<c0152ef5>] block_ioctl+0x18/0x1d
Jan  9 09:32:01 hummus kernel: [<c0152edd>] block_ioctl+0x0/0x1d
Jan  9 09:32:01 hummus kernel: [<c015aef9>] do_ioctl+0x19/0x55
Jan  9 09:32:01 hummus kernel: [<c01137e8>] do_page_fault+0x24f/0x5b3
Jan  9 09:32:01 hummus kernel: [<c015b02f>] vfs_ioctl+0x50/0x19e
Jan  9 09:32:01 hummus kernel: [<c015b1b1>] sys_ioctl+0x34/0x53
Jan  9 09:32:01 hummus kernel: [<c0102bcb>] sysenter_past_esp+0x54/0x75
Jan  9 09:32:01 hummus kernel: Code: 00 00 20 0f 84 fd fc ff ff 81 48 0c 00 01 00 00 9c 5b fa ba 70 35 2a c0 89 f0 e8 21 0c 00 00 8b 46 6c 8b 40 08 8b 38 85 ff 74 08 <0f> 0b 68 05 f7 34 2a c0 c7 40 08 01 00 00 00 53 9d 89 f0 e8 f5
Signature · 4 lines
Ralf Hildebrandt (i.A. des IT-Zentrums)         [email protected]
Charite - Universitätsmedizin Berlin            Tel.  +49 (0)30-450 570-155
Gemeinsame Einrichtung von FU- und HU-Berlin    Fax.  +49 (0)30-450 570-962
IT-Zentrum Standort CBF                 send no mail to [email protected]
#1 lore raw

* Ralf Hildebrandt <[email protected]>:

Quoted 7 lines

I invoked "hdparm -w /dev/hda"

# uname -a
Linux hummus.charite.de 2.6.15 #1 Tue Jan 3 09:30:04 CET 2006 i686 GNU/Linux

Before you flame away at me for using the nvidia kernel module: I will reproduce this WITHOUT the nvidia kernel module. At least I'll try.

And alas, it doens't reproduce. I'll now crawl under a rock and be ashamed of myself.

Signature · 4 lines
Ralf Hildebrandt (i.A. des IT-Zentrums)         [email protected]
Charite - Universitätsmedizin Berlin            Tel.  +49 (0)30-450 570-155
Gemeinsame Einrichtung von FU- und HU-Berlin    Fax.  +49 (0)30-450 570-962
IT-Zentrum Standort CBF                 send no mail to [email protected]
#2 lore raw
In reply to Ralf Hildebrandt
Quoted 7 lines · Ralf Hildebrandt

I invoked "hdparm -w /dev/hda"

# uname -a
Linux hummus.charite.de 2.6.15 #1 Tue Jan 3 09:30:04 CET 2006 i686 GNU/Linux

Before you flame away at me for using the nvidia kernel module: I will reproduce this WITHOUT the nvidia kernel module. At least I'll try.

You should be able to reproduce it without the Nvidia code loaded if you do I/O on the disk and run hdparm -w in a tight loop while doing so.

You might want to do it on a disk you have a backup of

#3 lore raw

* Alan Cox <[email protected]>:

Quoted 4 lines

You should be able to reproduce it without the Nvidia code loaded if you do I/O on the disk and run hdparm -w in a tight loop while doing so.

You might want to do it on a disk you have a backup of

I think I'll pass :)

Signature · 4 lines
Ralf Hildebrandt (i.A. des IT-Zentrums)         [email protected]
Charite - Universitätsmedizin Berlin            Tel.  +49 (0)30-450 570-155
Gemeinsame Einrichtung von FU- und HU-Berlin    Fax.  +49 (0)30-450 570-962
IT-Zentrum Standort CBF                 send no mail to [email protected]
#4 lore raw
Quoted 5 lines · Ralf Hildebrandt

* Alan Cox <[email protected]>:

Quoted 3 lines

You should be able to reproduce it without the Nvidia code loaded if you do I/O on the disk and run hdparm -w in a tight loop while doing so.

Quoted 8 lines
Quoted 1 line

You might want to do it on a disk you have a backup of

I think I'll pass :)
--
Ralf Hildebrandt (i.A. des IT-Zentrums) [email protected]
Charite - Universitätsmedizin Berlin Tel. +49 (0)30-450 570-155
Gemeinsame Einrichtung von FU- und HU-Berlin Fax. +49 (0)30-450 570-962
IT-Zentrum Standort CBF send no mail to [email protected]

He might be able to do the same thing with cat /dev/urandom > /dev/hda. There are not a lot of file-system checks against bad data in file-systems. The '-w' reset will guarantee bad data, which may propagate to the file-system(s) in use, causing all kinds of problems. So, the OOPS under these conditions is expected.

Cheers,
Dick Johnson
Penguin : Linux version 2.6.13.4 on an i686 machine (5589.71 BogoMips). Warning : 98.36% of all statistics are fiction.

****************************************************************
The information transmitted in this message is confidential and may be privileged. Any review, retransmission, dissemination, or other use of this information by persons or entities other than the intended recipient is prohibited. If you are not the intended recipient, please notify Analogic Corporation immediately - by replying to this message or by sending an email to [email protected] - and destroy all copies of this information, including any attachments, without reading or disclosing them.

Thank you.

#5 lore raw
Quoted 1 line · linux-os (Dick Johnson)

He might be able to do the same thing with cat /dev/urandom > /dev/hda.

Or by taking the lid off and painting it purple with pink spots, but neither are relevant to the problem at hand.

Quoted 1 line

There are not a lot of file-system checks against bad data in file-systems.

Nothing to do with the file system

Quoted 3 lines

The '-w' reset will guarantee bad data, which may propagate to the file-system(s) in use, causing all kinds of problems. So, the OOPS under these conditions is expected.

No. If the locking in the drivers is correct then -w is totally boringly safe. But the locking isn't. Follow ups should probably go to the linux-ide list.

Alan