jffs2 tests
Frederic Giasson
fgiasson at mediatrix.com
Wed Aug 8 08:58:33 EDT 2001
Hi Xavier,
I remember that I had that UNKNOWN_INCOMPAT_NODETYPE error message quite
some time ago, and I did what's described next and then the problem was
gone. First, I peered into JFFS2 code and I added what follows to the
beggining of the default case of the switch block in jffs2_get_inode_nodes()
function, file nodelist.c:
default:
if (ref->flash_offset & 1)
{
/* Obsoleted */
break;
}
...
I remember that I had summited this to David, but I did'nt check if it has
been done. Try it and tell me if it solved the problem.
And about the garbage collector latency problem, I just figured out that the
same thing happens to me but it is very rare. I looked at the log of my
testcase started last week, and the first occurence I seen was at loop
8736... In my opinion though, it may not be due to garbage collector
latency because as soon as JFFS2 deletes an inode and all child nodes, it
marks them obsolete. When df is called, obsolete nodes are taken in account
as free space. But I may be wrong (anyone can correct me)? This problem
should definitely be addressed.
Frédéric Giasson
-----Original Message-----
From: Xavier DEBREUIL [mailto:xde at inventel.fr]
Sent: Wednesday, August 08, 2001 4:45 AM
To: linux-mtd
Subject: jffs2 tests
Hello,
I do a similar test on jffs2 (20010802) as Frederic Giasson did :
1- copy randomly generated files of different sizes several times to
fill the jffs2 partition
2- df
3- remove all files
4- df
at step 4, most of the time I have :
/dev/mtdblock2 3200 648 2552 20%
/mnt/jffs2/jffs2_0
and from time to time :
/dev/mtdblock2 3200 904 2296 28%
/mnt/jffs2/jffs2_0
Nevertheless, as the following copy is able to copy the same amount of
files, it should be due to the latency of the garbage collector work
(confirm).
The only thing that seems odd to me is the occurence from time to time
of the following :
rm -R /mnt/jffs2/jffs2_0/*
Unknown INCOMPAT nodetype C002 at 001B1588
jffs2_read_inode(): No data nodes found for ino #23289
Eep. read_inode() failed for ino #23289
Unknown INCOMPAT nodetype C002 at 000E1A8C
jffs2_read_inode(): No data nodes found for ino #23290
Eep. read_inode() failed for ino #23290
the unknown nodetype is always C002
What kind of debug is worth to be printed out to get more information
about that problem ?
Xavier
______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/
More information about the linux-mtd
mailing list