Stephen Berry wrote:
Dirk Behme wrote:
Stephen Berry wrote:
Ok - first let me say thanks!
Second, there was a slight issue with the patch, but it did compile more
or less cleanly.
Thirdly, it appears that there is some incompatibility between the old
and new yaffs filesystems - since it gaak'd all over the existing one.
So after jumping through a few hoops, I was able to integrate the latest
into the kernel and mount the partition. BUT, when I tried to replace
the filesystem in NAND, I started seeing some errors (ok lots of errors)
**>> Block 2427 retired
Block 2427 is in state 9 after gc, should be erased
yaffs: Block struck out
nand_update_bbt: Out of memory
Doesn't look like it will work out of the box....
I supposed something like this. There is a lot of time between the
yaffs2 in 2.6.10 and the recent one. So most probably some
incompatible changes.
This is why I proposed to use a new board (or at least a low level
formatted NAND) and establish a complete new yaffs2 fs, i.e. starting
from scratch with the new yaffs2 code. Then put your original content
into the new yaffs2 fs and try if it behaves better.
Dirk
I have no problem trashing the image that was originally there - so I
*did* erase the partition before mounting it and un-tarring my
filesystem image.
I used flash_eraseall /dev/mtd4 to do this... is there a better way?
No, this is what I meant ;)
So you started from scratch and you get above "Block 2427 retired ..."
error? Uups, then something else is incompatible, maybe the NAND/MTD
subsystem and the new yaffs2 code? Don't know. Seems it is time then
to stop this test and mark it as "don't work" :(
Just to get a feeling if new yaffs2 code behaves better than the old
stuff in 2.6.10, if you like to spent one or two additional hours:
Take recent 2.6.24 DaVinci git, apply yaffs2 patch, and let it boot
into your yaffs root. You don't need a lot (maybe still missing)
drivers for this. Simply try to build a small test kernel with enough
to boot into root yaffs fs.
Don't get me wrong: I don't propose this for your production devices,
as you may have reasons to stay with 2.6.10. But it would be
interesting if new kernel & yaffs behaves better.
Regards
Dirk
Dirk Behme wrote:
Narnakaje, Snehaprabha wrote:
Steve,
Unfortunately, this is a known issue in the 2.6.10 Kernel. YAFFS2
in the
2.6.10 kernel does not have "checkpoint" support.
I'm really not sure about the features of a more recent YAFFS2. But if
you like you could try to update the old yaffs2 to a recent one. In
attachment is a yaffs2 patch containing yaffs2 CVS snapshot from today.
If you like, you could try to exchange your exisiting yaffs2 with this
most recent version. To apply this patch, back up your existing
fs/yaffs2 directory (e.g "mv fs/yaffs2 fs/yaffs2_original"),
fs/Kconfig and fs/Makefile. Then manually remove existing entries
# Patched by YAFFS
obj-$(CONFIG_YAFFS_FS) += yaffs2/
from Makefile and
# Patched by YAFFS
source "fs/yaffs2/Kconfig"
from Kconfig.
Now you should be able to apply patch in attachment.
Not sure if it compiles with a 2.6.10 or works, though. Just try it.
Btw.: A backup of the exisiting target yaffs2 file system or using an
other board where you can establish a new and fresh yaffs2 would be a
good idea as well ;)
Does it work?
Dirk
On each reboot, it
assumes to be starting from an "unclean shutdown" system, thus
spending
time doing the house keeping stuff for all the files. The mount
time is
proportional to the number of files (as well as the large size) in the
YAFFS2 filesystem.
Is your "minimal" filesystem busybox-based? You might have to
update the
busybox application in this filesystem.
Thanks
Sneha
-----Original Message-----
From:
[EMAIL PROTECTED]
om
[mailto:[EMAIL PROTECTED]
ncidsp.com] On Behalf Of Stephen Berry
Sent: Sunday, March 16, 2008 9:22 AM
To: davinci-linux-open-source@linux.davincidsp.com
Subject: Yaffs2 boot times
I've noticed some odd behavior with booting from nand, specifically
that
the time it takes to boot (or mount) the partition is directly related
to the number of files in the filesystem.
I have two partitions, one with the same minimal filesystem of 50mB in
size and the other of ~400mB. Both nand partitions are the same
physical
size. The minimal filesystem boots fairly quickly, under 10 seconds or
so, while the *big* filesystem takes several minutes. If it weren't
for
the fact the the minimal filesystem is a pain in the butt to work with
(control-c, telnet, ftp, seem to NOT work among other things) I'd
just
stick with it.
Does anyone have any ideas as to what is going on with mounting
yaffs2 ?
Steve
_______________________________________________
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
_______________________________________________
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
_______________________________________________
Davinci-linux-open-source mailing list
Davinci-linux-open-source@linux.davincidsp.com
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
