reiser4 cryptcompress test setup: bug

reiser4 cryptcompress test setup: bug

[Danny Milosavljevic]

Hi Edward,

I finally tried your cryptcompress setup (2.6.18-mm3) and just did the
first evil thing I could think of:

the reiser4 partition with ccreg40 enabled is /dev/sda1 (/mnt/tmp/)
(mkfs'ed, thus empty).

dannym@pyramid: /mnt/tmp/dannym/> du -m ~/doc/literatur
2609    /home/dannym/doc/literatur/

dannym@pyramid: /mnt/tmp/dannym/> cp -R ~/doc/literatur .
(hangs indefinitely)

(note that 2609 MB is way too much to fit on the 250 MB memory stick, but
that was the point, right :))

dannym@pyramid: /mnt/tmp/dannym/> df /dev/sda1
/dev/sda1               242948    242824       124 100% /mnt/tmp

'ps aux |grep cp' hangs.

dmesg says:
[   84.597446] reiser4: sda1: found disk format 4.0.0.
[   88.544450] Bluetooth: Core ver 2.10
[   88.544529] NET: Registered protocol family 31
[   88.544531] Bluetooth: HCI device and connection manager initialized
[   88.544535] Bluetooth: HCI socket layer initialized
[   88.581518] Bluetooth: L2CAP ver 2.8
[   88.581523] Bluetooth: L2CAP socket layer initialized
[   88.648403] Bluetooth: RFCOMM socket layer initialized
[   88.648422] Bluetooth: RFCOMM TTY layer initialized
[   88.648424] Bluetooth: RFCOMM ver 1.8
[  372.564228] BUG: unable to handle kernel paging request at virtual address 4b1b5d0b
[  372.564235]  printing eip:
[  372.564237] c01c498b
[  372.564238] *pde = 00000000
[  372.564242] Oops: 0000 [#1]
[  372.564243] PREEMPT 
[  372.564246] last sysfs file: /block/sda/removable
[  372.564248] Modules linked in: rfcomm l2cap bluetooth binfmt_misc reiserfs hfs nls_utf8 hfsplus nls_cp437 ntfs vfat fat isofs udf capability commoncap cpufreq_userspace cpufreq_stats cpufreq_powersave cpufreq_ondemand freq_table cpufreq_conservative video thermal sony_acpi backlight processor fan container button battery asus_acpi ac reiser4 zlib_deflate ext2 configfs dm_mod md_mod fusion fuse lp snd_seq_dummy snd_seq_oss snd_seq_midi snd_seq_midi_event snd_seq tuner i2c_viapro sg sd_mod ipv6 af_packet tsdev usbhid generic snd_via82xx snd_ac97_codec snd_ac97_bus snd_pcm_oss snd_mixer_oss snd_pcm snd_timer snd_page_alloc snd_mpu401 snd_mpu401_uart usb_storage snd_rawmidi snd_seq_device bttv via_ircc video_buf ir_common snd ide_cd ehci_hcd irda analog compat_ioctl32 btcx_risc tveeprom parport_pc parport crc_ccitt aic7xxx rtc soundcore serio_raw cdrom floppy evdev psmouse pcspkr gameport
  videodev v4l1_compat v4l2_common via_agp agpgart scsi_transport_spi scsi_mod via_rhine uhci_hcd mii usbcore
[  372.564307] CPU:    0
[  372.564307] EIP:    0060:[<c01c498b>]    Not tainted VLI
[  372.564309] EFLAGS: 00010282   (2.6.18-mm3 #4)
[  372.564316] EIP is at journal_start+0x3b/0x100
[  372.564319] eax: 4b1b5d0b   ebx: f5fa0d20   ecx: f7dee400   edx: 00000002
[  372.564322] esi: f7c23960   edi: f5ab1990   ebp: 00000002   esp: f514dec0
[  372.564324] ds: 007b   es: 007b   ss: 0068
[  372.564327] Process cp (pid: 4186, ti=f514c000 task=f57cc550 task.ti=f514c000)
[  372.564329] Stack: 00000001 ffffffd8 c0170e86 f514df34 f514df34 f5fa0d20 f5ab1990 f5fa0d20 
[  372.564335]        f5ab1990 00000001 c01b6400 c016b4b9 f514df68 f5ab1990 f562b620 00030000 
[  372.564340]        c0186914 c011fb66 3b9aca00 f7dee400 f5ab1990 f562b620 00030000 f7dee400 
[  372.564346] Call Trace:
[  372.564350]  [<c0170e86>] may_open+0x66/0x260
[  372.564358]  [<c01b6400>] ext3_dirty_inode+0x30/0x90
[  372.564364]  [<c016b4b9>] cp_new_stat64+0xf9/0x110
[  372.564371]  [<c0186914>] __mark_inode_dirty+0x34/0x1c0
[  372.564377]  [<c011fb66>] current_fs_time+0x46/0x50
[  372.564386]  [<c017d4ad>] touch_atime+0x5d/0xb0
[  372.564394]  [<c014681b>] generic_file_mmap+0x3b/0x40
[  372.564399]  [<c01594b9>] do_mmap_pgoff+0x429/0x770
[  372.564414]  [<c01077ce>] sys_mmap2+0xbe/0xe0
[  372.564424]  [<c0103065>] sysenter_past_esp+0x56/0x79
[  372.564435]  =======================
[  372.564436] Code: 85 f6 89 5c 24 18 bb e2 ff ff ff 89 6c 24 24 89 d5 89 7c 24 20 8b 00 8b 80 bc 04 00 00 89 44 24 14 74 3e 85 c0 74 50 89 c3 8b 00 <3b> 30 74 2e c7 44 24 10 3c 9b 34 c0 c7 44 24 0c 0f 01 00 00 c7 
[  372.564456] EIP: [<c01c498b>] journal_start+0x3b/0x100 SS:ESP 0068:f514dec0
[  372.564461]  

I'll leave the /dev/sda1 untouched in case we need it
(well, I'll unmount it since I can't sleep with the fan noise :))

Tell me if you need anything...

cheers,
  Danny

Re: reiser4 cryptcompress test setup: bug

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 561 bytes --]

On Mon, 06 Nov 2006 19:27:57 GMT, Danny Milosavljevic said:
> Hi Edward,
> 
> I finally tried your cryptcompress setup (2.6.18-mm3) and just did the
> first evil thing I could think of:
> 
> the reiser4 partition with ccreg40 enabled is /dev/sda1 (/mnt/tmp/)
> (mkfs'ed, thus empty).

> [  372.564358]  [<c01b6400>] ext3_dirty_inode+0x30/0x90
> [  372.564364]  [<c016b4b9>] cp_new_stat64+0xf9/0x110

You might want to check /home sometime - this looks like an ext3 botch
rather than a reiserfs botch, unless reiser4 is stomping on something
behind ext3's back.

[-- Attachment #2: Type: application/pgp-signature, Size: 226 bytes --]

Re: reiser4 cryptcompress test setup: bug

[Danny Milosavljevic]

Hi,

On Mon, 06 Nov 2006 17:07:16 -0500, Valdis.Kletnieks wrote:

> On Mon, 06 Nov 2006 19:27:57 GMT, Danny Milosavljevic said:
>> Hi Edward,
>> 
>> I finally tried your cryptcompress setup (2.6.18-mm3) and just did the
>> first evil thing I could think of:
>> 
>> the reiser4 partition with ccreg40 enabled is /dev/sda1 (/mnt/tmp/)
>> (mkfs'ed, thus empty).
> 
>> [  372.564358]  [<c01b6400>] ext3_dirty_inode+0x30/0x90
>> [  372.564364]  [<c016b4b9>] cp_new_stat64+0xf9/0x110
> 
> You might want to check /home sometime - this looks like an ext3 botch
> rather than a reiserfs botch, unless reiser4 is stomping on something
> behind ext3's back.

Well, I thought so at first, but:
/home is reiser4
/ is ext3 
no other ext3s

~/doc/literatur is on /home (checked via 'df .').

So, well, although it looks like it ext3 is the culprit, it is improbable
that it actually was the one triggering first (I never saw that
backtrace before, too). Naturally once Edwards says he doesn't need the
partition for analyzing, I'll nuke it and try to reproduce it again :)

cheers,
  Danny