Hi,
I noticed that mounts slowed down considerably on my system which uses
512 byte NAND last week when I upgraded to Linux 2.6.19-rc6.
I believe the attached patch fixes the situation. The yaffs_mtdif.c
translate_oob2spare() function was added to deal with recent MTD rework,
but it doesn't modify the spare ecc1 and ecc2 fields, so they end up
with garbage values when the structure is returned to
yaffs_TagsCompatabilityReadChunkWithTagsFromNAND().
yaffs_TagsCompatabilityReadChunkWithTagsFromNAND() then compares spare
to a buffer of 0xff's to see if the chunk is used, and the comparison
always indicates that the chunk is in use, even if the nand flash was
just erased.
I'm not sure if this fix has any side-effects (I haven't noticed any) or
if it is the most efficient way to deal with this, but it does reduce
the mount time of a freshly erased 58MB partition from over a minute to
around 2 seconds.
Regards,
../fam
--
Frank Mandarino fmandarino(a)endrelia.com
Endrelia Technologies Inc.
Toronto, Ontario, Canada
Index: yaffs_mtdif.c
===================================================================
RCS file: /home/aleph1/cvs/yaffs2/yaffs_mtdif.c,v
retrieving revision 1.16
diff -u -r1.16 yaffs_mtdif.c
--- yaffs_mtdif.c 8 Nov 2006 06:24:34 -0000 1.16
+++ yaffs_mtdif.c 29 Nov 2006 14:24:40 -0000
@@ -64,8 +64,10 @@
spare->tagByte5 = oob[5] == 0xff ? 0xff : oob[5] & 0x3f;
spare->blockStatus = oob[5] & 0x80 ? 0xff : 'Y';
spare->pageStatus = oob[5] & 0x40 ? 0xff : 0;
+ spare->ecc1[0] = spare->ecc1[1] = spare->ecc1[2] = 0xff;
spare->tagByte6 = oob[6];
spare->tagByte7 = oob[7];
+ spare->ecc2[0] = spare->ecc2[1] = spare->ecc2[2] = 0xff;
nspare->eccres1 = nspare->eccres2 = 0; /* FIXME */
}
(text/plain)
