|
| From: | Chris Westervelt |
| Subject: | [Ltib] RE: JFFS2 crc errors |
| Date: | Fri, 6 Aug 2010 07:27:03 -0500 |
|
Thanks Kevin and Giri. I have
looked at the pages sent by Giri and I have a much better understanding of the
whole thing now. From: Kevin Wells
[mailto:address@hidden Hi Chris, I think you’re ok and it’s just
an artifact of stopping the garbage collector daemon….but this question might
be better suited for the MTD/JFFS2 forums. In the link Giri provided, I think
you’re looking at unclean reboot case when the GC was working. I did run a test on my board
where I filled a JFFS2 partition with files, erased them all, and then refilled
them again with other files. And then shut down the system (safely) with the garbage
collector daemon still running and before it was complete without un-mounting
the partition. On my next boot, the thread found the same type of error. If I
un-mounted the partition and shut down, it was ok. During garbage collection,
you’ll see this running…
436 2 root
RWN 0 0% 39% [jffs2_gcd_mtd4] And here is the error on my
board. On my board, the error showed up about 20 seconds after the partition
was mounted and when gcd was running… address@hidden /]# JFFS2 notice:
(436) check_node_data: wrong data CRC in data node at 0x03200000: read
0xb0b6030f, calculated 0x95e2caad. You can also test out the MTD
subsystem using the kernel MTD tests if you want to check the MTD support.
These install as modules and will help verify NAND and the MTD driver are
working correctly. Kevin Subject: [Ltib] RE: JFFS2 crc errors Hi Chris, More info regarding JFFS2 messages http://www.linux-mtd.infradead.org/faq/jffs2.html Regards, Giri From: address@hidden
[mailto:address@hidden On Behalf Of Chris
Westervelt No I just made some
changes to the rootfs and saved them back. After a power down and reboot,
it started reporting this. The product cannot be software shut down and
must tolerate abrupt power loss. If that is something we must consider in
design then I need to get any appropriate changes made to the hardware/software
to correct or tolerate this condition. This is not the first time I have
seen this in fact it almost always happens when we change the rootfs in any
way. I just wanted to better understand it and make sure it’s not a
problem. From: Kevin Wells [mailto:address@hidden Hi Chris, This can happen if something is
writing to the NAND when the system is abruptly powered off or reset. Only part
of the data will have been written to NAND. Was the board shut down with the
‘reboot’ or ‘poweroff’ commands? This should give you a safe shutdown of JFFS2.
An abrupt power cycle or reset can lead to issues with JFFS2 data (similar to
any file system). thanks, Kevin Subject: [Ltib] JFFS2 crc errors All, I have started the phase of our project where we are weaning
off NFS and using JFFS2. Now my understanding is the flash should work
just like a normal drive but I am seeing errors about failed CRCs that I don’t
understand. Is this normal or do we have a real problem here? Also,
if it is normal, is there a way to suppress these messages and or automatic
checks. The following is what I see. address@hidden /]# JFFS2 notice: (309) check_node_data: wrong
data CRC in data node at 0x01dd4000: read 0x971bfcf, calculated 0xe401df4b. JFFS2 notice: (309) check_node_data: wrong data CRC in data
node at 0x01dd59b0: read 0x7299d564, calculated 0x4a339945. Chris Westervelt Senior Product Development Engineer Infrasafe, Inc. http://www.infrasafe.com Mobile: (407)
595-7023 |
| [Prev in Thread] | Current Thread | [Next in Thread] |