UBIFS reboots issue

UBIFS reboots issue

[Boaz Ben-David]

Hi all,

We have an embedded system using a Samsung 2GB MLC NAND and are using
UBIFS as the file system on it.
I have a question regarding UBIFS and hard reboots.
After some work with the flash (doing reads and writes to it), some
times one of our partitions would fail to mount.
I suspected this has something to do with a reboot which happens during
a write to the flash.
In order to test this I created a simple script which copies a file from
to the partition, syncs, then deletes it and does this repeatedly until 
a hard reboot occurs (using the watch dog to cause the boot).
This process repeats itself until the partition fails to mount.
When I performed the test mentioned above, after about 3-4 times the
partition would fail.
I am not sure if this behaviour is normal or not.
Did someone encounter similiar behaviour?

Thanks,

Boaz. 

Re: UBIFS reboots issue

[Steve Iribarne]

On 12/22/2010 06:36 AM, Boaz Ben-David wrote:
> Hi all,
>
> We have an embedded system using a Samsung 2GB MLC NAND and are using
> UBIFS as the file system on it.
> I have a question regarding UBIFS and hard reboots.
> After some work with the flash (doing reads and writes to it), some
> times one of our partitions would fail to mount.
> I suspected this has something to do with a reboot which happens during
> a write to the flash.
> In order to test this I created a simple script which copies a file from
> to the partition, syncs, then deletes it and does this repeatedly until
> a hard reboot occurs (using the watch dog to cause the boot).
> This process repeats itself until the partition fails to mount.
> When I performed the test mentioned above, after about 3-4 times the
> partition would fail.
> I am not sure if this behaviour is normal or not.
> Did someone encounter similiar behaviour?

What is the error when it tries to remount?

Re: UBIFS reboots issue

[Boaz Ben-David]

I can't seem to post back...

On Wed, 2010-12-22 at 16:36 +0200, Boaz Ben-David wrote:
> Hi all,
> 
> We have an embedded system using a Samsung 2GB MLC NAND and are using
> UBIFS as the file system on it.
> I have a question regarding UBIFS and hard reboots.
> After some work with the flash (doing reads and writes to it), some
> times one of our partitions would fail to mount.
> I suspected this has something to do with a reboot which happens during
> a write to the flash.
> In order to test this I created a simple script which copies a file from
> to the partition, syncs, then deletes it and does this repeatedly until 
> a hard reboot occurs (using the watch dog to cause the boot).
> This process repeats itself until the partition fails to mount.
> When I performed the test mentioned above, after about 3-4 times the
> partition would fail.
> I am not sure if this behaviour is normal or not.
> Did someone encounter similiar behaviour?
> 
> Thanks,
> 
> Boaz. 
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: UBIFS reboots issue

[Boaz Ben-David]

Ok, got it.

Here is the ubiattach and mount outputs:

attaching mtd7 to ubi2
UBI: physical eraseblock size:   524288 bytes (512 KiB)
UBI: logical eraseblock size:    516096 bytes
UBI: smallest flash I/O unit:    4096
UBI: VID header offset:          4096 (aligned 4096)
UBI: data offset:                8192
UBI: attached mtd7 to ubi2
UBI: MTD device name:            "nand.db"
UBI: MTD device size:            1648 MiB
UBI: number of good PEBs:        3291
UBI: number of bad PEBs:         5
UBI: max. allowed volumes:       128
UBI: wear-leveling threshold:    4096
UBI: number of internal volumes: 1
UBI: number of user volumes:     1
UBI: available PEBs:             0
UBI: total number of reserved PEBs: 3291
UBI: number of PEBs reserved for bad PEB handling: 32
UBI: max/mean erase counter: 237/11
UBI: image sequence number: 0
UBI: background thread "ubi_bgt2d" started, PID 1326
UBI device number 2, total 3291 LEBs (1698471936 bytes, 1.6 GiB),
available 0 LEBs (0 bytes), LEB size 516096 bytes (504.0 KiB)
UBIFS: recovery needed
UnCorrectable RS-ECC Error
UBI error: ubi_io_read: error -74 while reading 516096 bytes from PEB
2237:8192, read 516096 bytes
UBIFS error (pid 1328): ubifs_check_node: bad node length 9
UBIFS error (pid 1328): ubifs_check_node: bad node at LEB 244:360448
Unable to handle kernel paging request at virtual address c93d5000
pgd = c56a8000
[c93d5000] *pgd=85618011, *pte=00000000, *ppte=00000000
Internal error: Oops: 7 [#1] PREEMPT
Modules linked in:
CPU: 0    Not tainted  (2.6.31-207-g7286c01-hg639121a71164-dirty #49)
PC is at hex_dump_to_buffer+0x164/0x270
LR is at hex_dump_to_buffer+0x48/0x270
pc : [<c03bb9a4>]    lr : [<c03bb888>]    psr: 80000013
sp : c56c9a80  ip : 0000000f  fp : 00000000
r10: 000000c8  r9 : 00000001  r8 : c56c9ac8
r7 : 00000020  r6 : c93d4ff0  r5 : 00000020  r4 : 00000030
r3 : 00000010  r2 : 00000033  r1 : c56c9afb  r0 : 00000020
Flags: Nzcv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment user
Control: 00c5387d  Table: 856a8008  DAC: 00000015
Process mount (pid: 1328, stack limit = 0xc56c8268)
Stack: (0xc56c9a80 to 0xc56ca000)
9a80: c93d4ff0 fffda019 00000020 00000020 00025fc0 c93d4ff0 c56c9ac8
c0627fe8 
9aa0: fffda019 c06450d3 00000002 c03bbb20 c56c9ac8 000000c8 00000000
c062e3f8 
9ac0: ffffffd8 00000000 66206666 66662066 20666620 66206666 66662066
20666620 
9ae0: 66206666 66662066 20666620 66206666 66662066 20666620 66206666
66662066 
9b00: 20666620 66206666 66662066 20666620 66206666 66662066 20666620
66206666 
9b20: 66662066 00666620 c56c9b30 c025630c ffffffff 60000013 ffffffff
ffffffff 
9b40: 00000000 00000000 ffffffff 0000000a ffffffff ffffffff 00000000
c56c9bf0 
9b60: 0000a78f c06f81dc 00058000 000000f4 c56ad3e0 c93af030 ffffffd9
c5312000 
9b80: 00058000 000000f4 c56ad3e0 00000000 00000009 c93af030 ffffffd9
c5312000 
9ba0: 00058000 000000f4 c56ad3e0 00000000 00000009 c03903ac 00000001
c93af030 
9bc0: ffffffd9 00000000 00000050 000000f4 c0694378 c93af000 ffffffea
c5312000 
9be0: 00058000 000000f4 c56ad3e0 00000000 0000004b 2000a78f c93af000
ffffffea 
9c00: c5312000 00058000 000000f4 c56ad3e0 00000000 c0366dc8 00058000
c93af000 
9c20: 06101831 c93af000 00058000 c5312000 000000f4 c036edf0 00000000
00000001 
9c40: c5312000 c93af000 c56c8000 000000f4 fffffffd c56ad3e0 00058000
c03865b0 
9c60: 00058000 00000000 00000000 00000000 c9357000 00026000 00000001
00058000 
9c80: c93af000 00026000 c0694378 c56c8000 00000000 c5312000 00000000
c59615a4 
9ca0: 00000002 000000f4 c56ad520 c036f7b8 00000001 c069f3b8 004f4dfa
00000000 
9cc0: 00000000 c56c9d44 00000000 00000000 0a78dace 00000000 00000020
c04348f4 
9ce0: 00000001 c596159c 00000040 00000000 004f4cbf 00000000 c5312844
00000000 
9d00: c56ad3e8 c5312850 00000002 c5960c9c 004f4daf 00000000 00000005
0000004b 
9d20: 0005a000 00000040 00078380 00000000 00000000 00000000 c56c8000
c036f21c 
9d40: 00000000 00000000 00000000 c5312000 c56c8000 c56ad3e8 c56ad3e0
00000005 
9d60: c56ad540 00000001 c5312850 c0370c5c 00000032 00000000 c56c9dec
c57c1400 
9d80: 00000400 c02b8898 c56c9da4 c02b8900 c56ad3a0 c02b8a04 00000400
00000000 
9da0: c56ad3e0 c57c15f8 c531276c c56ad3e8 c94e0000 00000001 00079000
00000002 
9dc0: c56c9de8 c02b8be8 00000020 c5312850 00000165 00000000 004f4dfc
00000000 
9de0: 00000000 00000001 c945a00c 00000000 00001000 c53126e0 c5312000
00000000 
9e00: c5320800 00000000 00000001 00000000 c531271c c0362f88 00000000
00000005 
9e20: c5320964 c0626c44 c06ed480 c5320964 00000005 c03b8bd8 c56c9e50
c56c8000 
9e40: 00000000 c06a80b0 c5320800 c02c77d0 c5216800 00000002 00000007
c5312724 
9e60: 00000003 c5312008 c5216800 c56ad1c0 c5320800 00008000 c06a80b0
c56c9edc 
9e80: c56b0000 c06a97f4 c56c9f48 c036448c c56c8000 c03b426c 00000002
00000000 
9ea0: 00000cb7 c02dca68 64212000 00000000 c569e000 00000003 00000000
00000000 
9ec0: 00000001 0007e000 00000004 c5320524 0f600001 c06eda30 c569e000
c569e005 
9ee0: c7f6c900 c569e000 c56b0000 00008000 c06a80b0 c7f06000 00008000
c02c6a30 
9f00: c7f6c900 00008000 c06a80b0 c7f06000 c56b0000 c569e000 00008000
c02c6ac0 
9f20: 00000000 00000020 c569e000 00008000 c56b0000 c02ddc40 c56c9f8c
c56c8000 
9f40: 00000003 00000000 c7e20000 c627c080 c029e408 c5317000 bef7ce39
00008000 
9f60: 00000000 c0232f44 c56c8000 00000003 00000000 c02ddd38 c56b0000
000a1000 
9f80: 000c2000 c569e000 c7f06000 c56b0000 bef7cb18 000a1018 bef7cb18
400214d0 
9fa0: 00000015 c0232dc0 000a1018 bef7cb18 bef7ce39 bef7ce43 bef7ce33
00008000 
9fc0: 000a1018 bef7cb18 400214d0 00000015 00000000 00000000 00000003
00000000 
9fe0: 000a1018 bef7ca34 0004076c 40101634 60000010 bef7ce39 00000000
00000000 
[<c03bb9a4>] (hex_dump_to_buffer+0x164/0x270) from [<c03bbb20>]
(print_hex_dump+0x70/0xf8)
[<c03bbb20>] (print_hex_dump+0x70/0xf8) from [<c03903ac>] (dbg_dump_node
+0xf50/0x13cc)
[<c03903ac>] (dbg_dump_node+0xf50/0x13cc) from [<c0366dc8>]
(ubifs_check_node+0x290/0x2e8)
[<c0366dc8>] (ubifs_check_node+0x290/0x2e8) from [<c036edf0>]
(ubifs_scan_a_node+0x21c/0x3d4)
[<c036edf0>] (ubifs_scan_a_node+0x21c/0x3d4) from [<c03865b0>]
(ubifs_recover_leb+0x16c/0xacc)
[<c03865b0>] (ubifs_recover_leb+0x16c/0xacc) from [<c036f7b8>]
(replay_buds+0xdc/0xb50)
[<c036f7b8>] (replay_buds+0xdc/0xb50) from [<c0370c5c>]
(ubifs_replay_journal+0xa30/0x127c)
[<c0370c5c>] (ubifs_replay_journal+0xa30/0x127c) from [<c0362f88>]
(ubifs_fill_super+0xc50/0x1e70)
[<c0362f88>] (ubifs_fill_super+0xc50/0x1e70) from [<c036448c>]
(ubifs_get_sb+0x2e4/0x358)
[<c036448c>] (ubifs_get_sb+0x2e4/0x358) from [<c02c6a30>]
(vfs_kern_mount+0x4c/0x98)
[<c02c6a30>] (vfs_kern_mount+0x4c/0x98) from [<c02c6ac0>] (do_kern_mount
+0x34/0xd8)
[<c02c6ac0>] (do_kern_mount+0x34/0xd8) from [<c02ddc40>] (do_mount
+0x710/0x784)
[<c02ddc40>] (do_mount+0x710/0x784) from [<c02ddd38>] (sys_mount
+0x84/0xcc)
[<c02ddd38>] (sys_mount+0x84/0xcc) from [<c0232dc0>] (ret_fast_syscall
+0x0/0x2c)
Code: e1570009 bafffff0 e3a07005 ea000013 (e7d6c003) 
---[ end trace 977c7948905d1436 ]---
note: mount[1328] exited with preempt_count 2
BUG: scheduling while atomic: mount/1328/0x40000002
Modules linked in:
[<c02375f8>] (unwind_backtrace+0x0/0xd0) from [<c0563f80>] (schedule
+0x78/0x3e4)
[<c0563f80>] (schedule+0x78/0x3e4) from [<c024f4d8>] (__cond_resched
+0x24/0x40)
[<c024f4d8>] (__cond_resched+0x24/0x40) from [<c05643b8>] (_cond_resched
+0x30/0x40)
[<c05643b8>] (_cond_resched+0x30/0x40) from [<c02b0238>] (unmap_vmas
+0x528/0x5d8)
[<c02b0238>] (unmap_vmas+0x528/0x5d8) from [<c02b2ce8>] (exit_mmap
+0xac/0x198)
[<c02b2ce8>] (exit_mmap+0xac/0x198) from [<c0252ff4>] (mmput+0x38/0xf8)
[<c0252ff4>] (mmput+0x38/0xf8) from [<c0256bb4>] (exit_mm+0x140/0x144)
[<c0256bb4>] (exit_mm+0x140/0x144) from [<c0258260>] (do_exit
+0x15c/0x6bc)
[<c0258260>] (do_exit+0x15c/0x6bc) from [<c0236500>] (die+0x168/0x188)
[<c0236500>] (die+0x168/0x188) from [<c02385c4>] (__do_kernel_fault
+0x64/0x74)
[<c02385c4>] (__do_kernel_fault+0x64/0x74) from [<c02387e8>]
(do_page_fault+0x214/0x230)
[<c02387e8>] (do_page_fault+0x214/0x230) from [<c023225c>] (do_DataAbort
+0x30/0x94)
[<c023225c>] (do_DataAbort+0x30/0x94) from [<c02329ec>] (__dabt_svc
+0x4c/0x60)
Exception stack(0xc56c9a38 to 0xc56c9a80)
9a20:                                                       00000020
c56c9afb 
9a40: 00000033 00000010 00000030 00000020 c93d4ff0 00000020 c56c9ac8
00000001 
9a60: 000000c8 00000000 0000000f c56c9a80 c03bb888 c03bb9a4 80000013
ffffffff 
[<c02329ec>] (__dabt_svc+0x4c/0x60) from [<c03bb9a4>]
(hex_dump_to_buffer+0x164/0x270)
[<c03bb9a4>] (hex_dump_to_buffer+0x164/0x270) from [<c03bbb20>]
(print_hex_dump+0x70/0xf8)
[<c03bbb20>] (print_hex_dump+0x70/0xf8) from [<c03903ac>] (dbg_dump_node
+0xf50/0x13cc)
[<c03903ac>] (dbg_dump_node+0xf50/0x13cc) from [<c0366dc8>]
(ubifs_check_node+0x290/0x2e8)
[<c0366dc8>] (ubifs_check_node+0x290/0x2e8) from [<c036edf0>]
(ubifs_scan_a_node+0x21c/0x3d4)
[<c036edf0>] (ubifs_scan_a_node+0x21c/0x3d4) from [<c03865b0>]
(ubifs_recover_leb+0x16c/0xacc)
[<c03865b0>] (ubifs_recover_leb+0x16c/0xacc) from [<c036f7b8>]
(replay_buds+0xdc/0xb50)
[<c036f7b8>] (replay_buds+0xdc/0xb50) from [<c0370c5c>]
(ubifs_replay_journal+0xa30/0x127c)
[<c0370c5c>] (ubifs_replay_journal+0xa30/0x127c) from [<c0362f88>]
(ubifs_fill_super+0xc50/0x1e70)
[<c0362f88>] (ubifs_fill_super+0xc50/0x1e70) from [<c036448c>]
(ubifs_get_sb+0x2e4/0x358)
[<c036448c>] (ubifs_get_sb+0x2e4/0x358) from [<c02c6a30>]
(vfs_kern_mount+0x4c/0x98)
[<c02c6a30>] (vfs_kern_mount+0x4c/0x98) from [<c02c6ac0>] (do_kern_mount
+0x34/0xd8)
[<c02c6ac0>] (do_kern_mount+0x34/0xd8) from [<c02ddc40>] (do_mount
+0x710/0x784)
[<c02ddc40>] (do_mount+0x710/0x784) from [<c02ddd38>] (sys_mount
+0x84/0xcc)
[<c02ddd38>] (sys_mount+0x84/0xcc) from [<c0232dc0>] (ret_fast_syscall
+0x0/0x2c)

Thanks!



On Wed, 2010-12-22 at 19:32 +0200, Boaz Ben-David wrote:
> I can't seem to post back...
> 
> On Wed, 2010-12-22 at 16:36 +0200, Boaz Ben-David wrote:
> > Hi all,
> > 
> > We have an embedded system using a Samsung 2GB MLC NAND and are using
> > UBIFS as the file system on it.
> > I have a question regarding UBIFS and hard reboots.
> > After some work with the flash (doing reads and writes to it), some
> > times one of our partitions would fail to mount.
> > I suspected this has something to do with a reboot which happens during
> > a write to the flash.
> > In order to test this I created a simple script which copies a file from
> > to the partition, syncs, then deletes it and does this repeatedly until 
> > a hard reboot occurs (using the watch dog to cause the boot).
> > This process repeats itself until the partition fails to mount.
> > When I performed the test mentioned above, after about 3-4 times the
> > partition would fail.
> > I am not sure if this behaviour is normal or not.
> > Did someone encounter similiar behaviour?
> > 
> > Thanks,
> > 
> > Boaz. 
> > ______________________________________________________
> > Linux MTD discussion mailing list
> > http://lists.infradead.org/mailman/listinfo/linux-mtd/
> 
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: UBIFS reboots issue

[Steve Iribarne]

> Ok, got it.
>
> Here is the ubiattach and mount outputs:
>

I know we've seen this error... I'm asking someone else here what caused 
it...

UnCorrectable RS-ECC Error
UBI error: ubi_io_read: error -74 while reading 516096 bytes from PEB



I thought it was a bad flash part but not sure... I'll get back to you 
as soon as he answers.  He works remote today.

-stv

RE: UBIFS reboots issue

[Boaz Ben-David]

Thanks for the quick reply.

About the bad flash part, I am not too sure this is the case.
We can reproduce this on every one of our devices, and also on the development board we used which has another (different part number) 
Samsung chip, also MLC.

Boaz.

________________________________________
From: linux-mtd-bounces@lists.infradead.org [linux-mtd-bounces@lists.infradead.org] On Behalf Of Steve Iribarne [siribarne@grid-net.com]
Sent: Wednesday, December 22, 2010 8:09 PM
To: linux-mtd@lists.infradead.org
Subject: Re: UBIFS reboots issue

> Ok, got it.
>
> Here is the ubiattach and mount outputs:
>

I know we've seen this error... I'm asking someone else here what caused
it...

UnCorrectable RS-ECC Error
UBI error: ubi_io_read: error -74 while reading 516096 bytes from PEB



I thought it was a bad flash part but not sure... I'll get back to you
as soon as he answers.  He works remote today.

-stv

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: UBIFS reboots issue

[Artem Bityutskiy]

On Wed, 2010-12-22 at 16:36 +0200, Boaz Ben-David wrote:
> Hi all,
> 
> We have an embedded system using a Samsung 2GB MLC NAND and are using
> UBIFS as the file system on it.
> I have a question regarding UBIFS and hard reboots.
> After some work with the flash (doing reads and writes to it), some
> times one of our partitions would fail to mount.
> I suspected this has something to do with a reboot which happens during
> a write to the flash.
> In order to test this I created a simple script which copies a file from
> to the partition, syncs, then deletes it and does this repeatedly until 
> a hard reboot occurs (using the watch dog to cause the boot).
> This process repeats itself until the partition fails to mount.
> When I performed the test mentioned above, after about 3-4 times the
> partition would fail.
> I am not sure if this behaviour is normal or not.
> Did someone encounter similiar behaviour?

Just curious, did you see this:

http://linux-mtd.infradead.org/faq/ubifs.html#L_ubifs_mlc

I do not think you can just use UBI/UBIFS on MLC as is, you'd need to
make sure MLC-specific behavior is handled properly.

-- 
Best Regards,
Artem Bityutskiy (Битюцкий Артём)

Re: UBIFS reboots issue

[Boaz Ben-David]

I am using kernel 2.6.31 for iMX35 and I'm not sure what version of
UBIFS is patched in there.
What do I need to do in order to patch to the latest version? 
Is is enough to clone UBIFS 2.6.31 and copy over the fs/ubifs and
mtd/ubi directories or are there any other directories I need to take
into account?

Thanks,

Boaz.

On Wed, 2010-12-22 at 22:20 +0200, Artem Bityutskiy wrote:
> On Wed, 2010-12-22 at 16:36 +0200, Boaz Ben-David wrote:
> > Hi all,
> > 
> > We have an embedded system using a Samsung 2GB MLC NAND and are using
> > UBIFS as the file system on it.
> > I have a question regarding UBIFS and hard reboots.
> > After some work with the flash (doing reads and writes to it), some
> > times one of our partitions would fail to mount.
> > I suspected this has something to do with a reboot which happens during
> > a write to the flash.
> > In order to test this I created a simple script which copies a file from
> > to the partition, syncs, then deletes it and does this repeatedly until 
> > a hard reboot occurs (using the watch dog to cause the boot).
> > This process repeats itself until the partition fails to mount.
> > When I performed the test mentioned above, after about 3-4 times the
> > partition would fail.
> > I am not sure if this behaviour is normal or not.
> > Did someone encounter similiar behaviour?
> 
> Just curious, did you see this:
> 
> http://linux-mtd.infradead.org/faq/ubifs.html#L_ubifs_mlc
> 
> I do not think you can just use UBI/UBIFS on MLC as is, you'd need to
> make sure MLC-specific behavior is handled properly.
> 
> -- 
> Best Regards,
> Artem Bityutskiy (Битюцкий Артём)
> 

RE: UBIFS reboots issue

[ZHOU Zhongjun]

Why do you not to use YAFFS2 for NAND flash?  It seems that Yaffs2 is more popularity.


-----Original Message-----
From: linux-mtd-bounces@lists.infradead.org [mailto:linux-mtd-bounces@lists.infradead.org] On Behalf Of Boaz Ben-David
Sent: Thursday, December 23, 2010 8:52 PM
To: linux-mtd@lists.infradead.org
Subject: Re: UBIFS reboots issue

I am using kernel 2.6.31 for iMX35 and I'm not sure what version of UBIFS is patched in there.
What do I need to do in order to patch to the latest version? 
Is is enough to clone UBIFS 2.6.31 and copy over the fs/ubifs and mtd/ubi directories or are there any other directories I need to take into account?

Thanks,

Boaz.

On Wed, 2010-12-22 at 22:20 +0200, Artem Bityutskiy wrote:
> On Wed, 2010-12-22 at 16:36 +0200, Boaz Ben-David wrote:
> > Hi all,
> > 
> > We have an embedded system using a Samsung 2GB MLC NAND and are 
> > using UBIFS as the file system on it.
> > I have a question regarding UBIFS and hard reboots.
> > After some work with the flash (doing reads and writes to it), some 
> > times one of our partitions would fail to mount.
> > I suspected this has something to do with a reboot which happens 
> > during a write to the flash.
> > In order to test this I created a simple script which copies a file 
> > from to the partition, syncs, then deletes it and does this 
> > repeatedly until a hard reboot occurs (using the watch dog to cause the boot).
> > This process repeats itself until the partition fails to mount.
> > When I performed the test mentioned above, after about 3-4 times the 
> > partition would fail.
> > I am not sure if this behaviour is normal or not.
> > Did someone encounter similiar behaviour?
> 
> Just curious, did you see this:
> 
> http://linux-mtd.infradead.org/faq/ubifs.html#L_ubifs_mlc
> 
> I do not think you can just use UBI/UBIFS on MLC as is, you'd need to 
> make sure MLC-specific behavior is handled properly.
> 
> --
> Best Regards,
> Artem Bityutskiy (Битюцкий Артём)
> 

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

RE: UBIFS reboots issue

[Boaz Ben-David]

I would very much want to try yaffs2. I cloned the git repo and patched my kernel using the script provided
with the yaffs code. The kernel compiled with no errors but after I mounted a yaffs partition I could not create any new files on it.
Each time I try to create a file on the partition I get "cannot allocate memory",,,
I haven't had much time to investigate this so if anyone has any idea why this is happening, this is also a very good direction for me to go with (using yaffs).
________________________________________
From: ZHOU Zhongjun [Zhongjun.Zhou@alcatel-sbell.com.cn]
Sent: Friday, December 24, 2010 5:23 AM
To: Boaz Ben-David; linux-mtd@lists.infradead.org
Subject: RE: UBIFS reboots issue

Why do you not to use YAFFS2 for NAND flash?  It seems that Yaffs2 is more popularity.


-----Original Message-----
From: linux-mtd-bounces@lists.infradead.org [mailto:linux-mtd-bounces@lists.infradead.org] On Behalf Of Boaz Ben-David
Sent: Thursday, December 23, 2010 8:52 PM
To: linux-mtd@lists.infradead.org
Subject: Re: UBIFS reboots issue

I am using kernel 2.6.31 for iMX35 and I'm not sure what version of UBIFS is patched in there.
What do I need to do in order to patch to the latest version?
Is is enough to clone UBIFS 2.6.31 and copy over the fs/ubifs and mtd/ubi directories or are there any other directories I need to take into account?

Thanks,

Boaz.

On Wed, 2010-12-22 at 22:20 +0200, Artem Bityutskiy wrote:
> On Wed, 2010-12-22 at 16:36 +0200, Boaz Ben-David wrote:
> > Hi all,
> >
> > We have an embedded system using a Samsung 2GB MLC NAND and are
> > using UBIFS as the file system on it.
> > I have a question regarding UBIFS and hard reboots.
> > After some work with the flash (doing reads and writes to it), some
> > times one of our partitions would fail to mount.
> > I suspected this has something to do with a reboot which happens
> > during a write to the flash.
> > In order to test this I created a simple script which copies a file
> > from to the partition, syncs, then deletes it and does this
> > repeatedly until a hard reboot occurs (using the watch dog to cause the boot).
> > This process repeats itself until the partition fails to mount.
> > When I performed the test mentioned above, after about 3-4 times the
> > partition would fail.
> > I am not sure if this behaviour is normal or not.
> > Did someone encounter similiar behaviour?
>
> Just curious, did you see this:
>
> http://linux-mtd.infradead.org/faq/ubifs.html#L_ubifs_mlc
>
> I do not think you can just use UBI/UBIFS on MLC as is, you'd need to
> make sure MLC-specific behavior is handled properly.
>
> --
> Best Regards,
> Artem Bityutskiy (Битюцкий Артём)
>

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

RE: UBIFS reboots issue

[Boaz Ben-David]

I'm sorry for changing the thread subject in the previous mail..

I guess it sums up to one main question (and I would be very greatful if
someone could answer):

	Overall, for use on MLC NAND, is UBIFS or JFFS2 more suitable?
	From what I understand, JFFS2 doesn't do wear-leveling very
	good which will shorten the flash lifetime, but on the other
	hand, as Artem said and is written in the UBIFS web-site, UBIFS
	needs some work to be used on MLC. 

On another note, I would like to also try yaffs2 but I can't get it to
work (can't seem to create any files). Does anyone has any experience
using it on the Freescale iMX35 PDK?

Thanks.



On Fri, 2010-12-24 at 12:40 +0200, Boaz Ben-David wrote:
> I would very much want to try yaffs2. I cloned the git repo and patched my kernel using the script provided
> with the yaffs code. The kernel compiled with no errors but after I mounted a yaffs partition I could not create any new files on it.
> Each time I try to create a file on the partition I get "cannot allocate memory",,,
> I haven't had much time to investigate this so if anyone has any idea why this is happening, this is also a very good direction for me to go with (using yaffs).
> ________________________________________
> From: ZHOU Zhongjun [Zhongjun.Zhou@alcatel-sbell.com.cn]
> Sent: Friday, December 24, 2010 5:23 AM
> To: Boaz Ben-David; linux-mtd@lists.infradead.org
> Subject: RE: UBIFS reboots issue
> 
> Why do you not to use YAFFS2 for NAND flash?  It seems that Yaffs2 is more popularity.
> 
> 
> -----Original Message-----
> From: linux-mtd-bounces@lists.infradead.org [mailto:linux-mtd-bounces@lists.infradead.org] On Behalf Of Boaz Ben-David
> Sent: Thursday, December 23, 2010 8:52 PM
> To: linux-mtd@lists.infradead.org
> Subject: Re: UBIFS reboots issue
> 
> I am using kernel 2.6.31 for iMX35 and I'm not sure what version of UBIFS is patched in there.
> What do I need to do in order to patch to the latest version?
> Is is enough to clone UBIFS 2.6.31 and copy over the fs/ubifs and mtd/ubi directories or are there any other directories I need to take into account?
> 
> Thanks,
> 
> Boaz.
> 
> On Wed, 2010-12-22 at 22:20 +0200, Artem Bityutskiy wrote:
> > On Wed, 2010-12-22 at 16:36 +0200, Boaz Ben-David wrote:
> > > Hi all,
> > >
> > > We have an embedded system using a Samsung 2GB MLC NAND and are
> > > using UBIFS as the file system on it.
> > > I have a question regarding UBIFS and hard reboots.
> > > After some work with the flash (doing reads and writes to it), some
> > > times one of our partitions would fail to mount.
> > > I suspected this has something to do with a reboot which happens
> > > during a write to the flash.
> > > In order to test this I created a simple script which copies a file
> > > from to the partition, syncs, then deletes it and does this
> > > repeatedly until a hard reboot occurs (using the watch dog to cause the boot).
> > > This process repeats itself until the partition fails to mount.
> > > When I performed the test mentioned above, after about 3-4 times the
> > > partition would fail.
> > > I am not sure if this behaviour is normal or not.
> > > Did someone encounter similiar behaviour?
> >
> > Just curious, did you see this:
> >
> > http://linux-mtd.infradead.org/faq/ubifs.html#L_ubifs_mlc
> >
> > I do not think you can just use UBI/UBIFS on MLC as is, you'd need to
> > make sure MLC-specific behavior is handled properly.
> >
> > --
> > Best Regards,
> > Artem Bityutskiy (Битюцкий Артём)
> >
> 
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/
> 
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: UBIFS reboots issue

[Artem Bityutskiy]

On Thu, 2010-12-23 at 14:51 +0200, Boaz Ben-David wrote:
> I am using kernel 2.6.31 for iMX35 and I'm not sure what version of
> UBIFS is patched in there.
> What do I need to do in order to patch to the latest version? 
> Is is enough to clone UBIFS 2.6.31 and copy over the fs/ubifs and
> mtd/ubi directories or are there any other directories I need to take
> into account?

You need to 'git pull' the ubifs-v2.6.31 tree.

-- 
Best Regards,
Artem Bityutskiy (Битюцкий Артём)

RE: UBIFS reboots issue

[Artem Bityutskiy]

On Tue, 2010-12-28 at 17:47 +0200, Boaz Ben-David wrote:
> I'm sorry for changing the thread subject in the previous mail..
> 
> I guess it sums up to one main question (and I would be very greatful if
> someone could answer):
> 
> 	Overall, for use on MLC NAND, is UBIFS or JFFS2 more suitable?

None, because no one with clue analyzed them WRT MLC-specific aspects.
And no-one seriously stress-tested them on MLC, at least no-one sent any
public report about this. UBIFS may appear to work, but then fail
horribly when you have an unclean power cut.

So again, you cannot use UBIFS as-is on MLC and sleep well. You need to
spend time and test/analyze them.

-- 
Best Regards,
Artem Bityutskiy (Битюцкий Артём)