JFFS2 Mount issue

Wed Aug 24 03:52:12 EDT 2005

> Can you send me more information about it? For example kernel log with
> debug level 1. And please make sure that you use the latest mtd            > snapshot
> with newest summary patch - use only erase block summary, centralized
> summary is now under construction.

   Initially I have used the EBS patch and mtd-snapshots released on
11/08/2005. I faced the segmentation fault issue with this patches
used in 2.6.10 kernel.

   Today I have downloaded the latest EBS patch and mtd-snapshot
released on 23/08/2005 from the below sites

   http://www.inf.u-szeged.hu/jffs2/mount.php - EBS patch
   ftp://ftp.uk.linux.org/pub/people/dwmw2/mtd/cvs/ - MTD 23/08/05

Applied the EBS patches to JFFS2 utilities in MTD snapshots, patched
the 2.6.10 kernel with the command /mtd/kernelpatch.sh -2
/usr/src/linux-2.6.10.
Created the jffs2 image with new mkfs.jffs2 utility. Created the jffs2
image with sumtool utility. Mounted the same. Copied the 1.5MB file to
JFFS2 partition, still i faced the Segmentation fault issue. The
following are the debug messages posted by the kernel.

/dev/hdc13 (120MB) - jffs2 image created by mkfs.jffs2 utility
/dev/hdc14 (120MB) - jffs2 image created with sumtool utility after
mkfs.jffs2 utility.

Only 472K of 1.5MB file is copied into the JFFS2 partition in both the
cases. This is FYI.

-----------------------------Kernel Message Starts----------------------------
blkmtd: version $Revision: 1.26 $
blkmtd: mtd0: [/dev/hda14] erase_size = 256KiB 
JFFS2 version 2.2. (NAND) (SUMMARY)  (C) 2001-2003 Red Hat, Inc.
jffs2: Flash size not aligned to erasesize, reducing to 128256KiB
JFFS2: jffs2_sum_write_sumnode(): empty summary info!!!
------------[ cut here ]------------
kernel BUG at fs/jffs2/summary.c:661!
invalid operand: 0000 [#1]
PREEMPT SMP 
Modules linked in: jffs2 zlib_deflate zlib_inflate blkmtd via_rhine ehci_hcd
CPU:    0
EIP:    0060:[<de90b7ac>]    Not tainted VLI
EFLAGS: 00010202   (2.6.10) 
EIP is at jffs2_sum_write_sumnode+0x2c/0x170 [jffs2]
eax: 0000003b   ebx: 0000003c   ecx: d4553c9c   edx: de90f7e0
esi: d98f4890   edi: dd420a00   ebp: d98f4890   esp: d4553c98
ds: 007b   es: 007b   ss: 0068
Process cp (pid: 2492, threadinfo=d4552000 task=d629a5e0)
Stack: de90f7e0 0000003c d98f4890 dd420a00 00000012 de8ff0eb dd420a00 00000012 
       dd420a00 00000000 dd420a00 de8fec9a dd420a00 000000c4 d4553d3c d4553d40 
       00000012 000000b4 d4552000 d94491b0 000000b4 00000000 dd420aec 0000000d 
Call Trace:
 [<de8ff0eb>] jffs2_do_reserve_space+0x15b/0x190 [jffs2]
 [<de8fec9a>] jffs2_reserve_space+0x14a/0x1d0 [jffs2]
 [<de901581>] jffs2_write_inode_range+0x51/0x2a0 [jffs2]
 [<de8fce99>] jffs2_commit_write+0xc9/0x1be [jffs2]
 [<c013b1a5>] generic_file_buffered_write+0x195/0x580
 [<c010400c>] apic_timer_interrupt+0x1c/0x30
 [<c01766e3>] __mark_inode_dirty+0x193/0x1a0
 [<c0139ba6>] do_generic_mapping_read+0x336/0x4a0
 [<c013b814>] __generic_file_aio_write_nolock+0x284/0x4a0
 [<c013bb5a>] __generic_file_write_nolock+0x9a/0xc0
 [<c012e3f0>] autoremove_wake_function+0x0/0x40
 [<c013bd50>] generic_file_write+0x40/0xb0
 [<c0156702>] vfs_write+0x92/0xe0
 [<c01567fd>] sys_write+0x3d/0x70
 [<c0102fd9>] sysenter_past_esp+0x52/0x79
Code: 57 56 53 8b 44 24 14 8b 90 58 01 00 00 8b a8 8c 00 00 00 8b 42
04 85 c0 74 07 8b 42 0c 85 c0 75 1d 68 e0 f7 90 de e8 64 fe 80 e1 <0f>
0b 95 02 27 fb 90 de 58 8b 44 24 14 8b 90 58 01 00 00 8b 02
 <6>note: cp[2492] exited with preempt_count 1
scheduling while atomic: cp/0x00000001/2492
 [<c038afd6>] schedule+0xb66/0xb70
 [<c01470fa>] unmap_page_range+0x3a/0x60
 [<c0147327>] unmap_vmas+0x207/0x250
 [<c014b690>] exit_mmap+0x80/0x170
 [<c0119184>] mmput+0x24/0x80
 [<c011db45>] do_exit+0x195/0x450
 [<c010480a>] die+0x16a/0x170
 [<c0104a90>] do_invalid_op+0x0/0x90
 [<c0104b0a>] do_invalid_op+0x7a/0x90
 [<de90b7ac>] jffs2_sum_write_sumnode+0x2c/0x170 [jffs2]
 [<c012e40b>] autoremove_wake_function+0x1b/0x40
 [<c01172f6>] __wake_up_common+0x36/0x60
 [<c0117348>] __wake_up+0x28/0x40
 [<c01040eb>] error_code+0x2b/0x30
 [<de90b7ac>] jffs2_sum_write_sumnode+0x2c/0x170 [jffs2]
 [<de8ff0eb>] jffs2_do_reserve_space+0x15b/0x190 [jffs2]
 [<de8fec9a>] jffs2_reserve_space+0x14a/0x1d0 [jffs2]
 [<de901581>] jffs2_write_inode_range+0x51/0x2a0 [jffs2]
 [<de8fce99>] jffs2_commit_write+0xc9/0x1be [jffs2]
 [<c013b1a5>] generic_file_buffered_write+0x195/0x580
 [<c010400c>] apic_timer_interrupt+0x1c/0x30
 [<c01766e3>] __mark_inode_dirty+0x193/0x1a0
 [<c0139ba6>] do_generic_mapping_read+0x336/0x4a0
 [<c013b814>] __generic_file_aio_write_nolock+0x284/0x4a0
 [<c013bb5a>] __generic_file_write_nolock+0x9a/0xc0
 [<c012e3f0>] autoremove_wake_function+0x0/0x40
 [<c013bd50>] generic_file_write+0x40/0xb0
 [<c0156702>] vfs_write+0x92/0xe0
 [<c01567fd>] sys_write+0x3d/0x70
 [<c0102fd9>] sysenter_past_esp+0x52/0x79
 Segmentation Fault
-----------------------------Kernel Message Ends----------------------------

I am not facing this kind of segmentation fault with 2.4.25 kernel. It
is working fine in 2.4 kernel.

> Another question: is there any reason that you use 2.6.10 instead of the
> newest kernel?

   Yet to try with higher kernel version. It is not mandatory to use
2.6.10 kernel.

Provide ur comments on this. Kindly help me in this issue.

Regards,
K. Suresh

On 8/23/05, Ferenc Havasi <havasi at inf.u-szeged.hu> wrote:
> Hi,
> 
> >  1. Compiled the kernel with JFFS2 summary support. Created the JFFS2
> >image with the following command.
> >      mkfs.jffs2 -d testjffs2 -e 0x40000 --pad=0x7800000 -o jffs2.img -v
> >
> >      This comand created the JFFS2 image of size 120MB.
> >      Copied the jffs2.img to one of the partition in Hard Disk.
> >Mounted
> >      the partition using blkmtd support. It takes around 27 seconds
> >to
> >      mount.
> >
> >      Why the mount time is high even if the kernel having the "JFFS2
> >summary" support?.
> >
> >
> Because there was no summary support in your image, so the original
> scanning method had to be applied. Summary support means to process this
> summary information and generate it automatically if you write date onto
> the filesystem.
> 
> >   2. Created the JFFS2 image with summary information inbuilt support
> >using sumtool2 utility.
> >         sumtool2 -e 0x40000 -i jffs2.img -o jffs2-sum.img
> >
> >       Copied the jffs2-sum.img to other partition in Hard Disk.
> >Mounted
> >       the partition using blkmtd support. It takes around 6 seconds
> >to
> >       mount.
> >Another problem is that "Segmentation fault" occurs when copying a
> >directory of size 10MB to JFFS2 partition after mount.
> >
> >Following are the error posted when copying the directory.
> >
> >
> Can you send me more information about it? For example kernel log with
> debug level 1. And please make sure that you use the latest mtd snapshot
> with newest summary patch - use only erase block summary, centralized
> summary is now under construction.
> 
> It would be also good to know that if it works without summary.
> 
> Another question: is there any reason that you use 2.6.10 instead of the
> newest kernel?
> 
> Bye,
> Ferenc
> 
>