RE: ext2fs or ramdisk problems......

RE: ext2fs or ramdisk problems......

[Whitney, John]

I have had odd results using ext2 on the ramdisk (in my case, a 16MB
ramdisk).  Using the 2.2.13, ramdisks I created using ext2 would boot
up just fine on the system, but after use the filesystem would declare
that it was out of space, even though "df" showed several megabytes
free on that filesystem.

I never did find out where the oddity was.  I moved forward to 2.2.15
in an effort to correct the problem, but that didn't help either.
Eventually, I moved on to using the Minix filesystem (which from what
I understand, uses diskspace better for what I need anyway) and have
not had similar problems.  The one downside is that Minix is not
byte-ordering independant, so currently I build the ramdisk image on
a iMac running YellowDog Linux.  Someday I'll go into the Minix source
code on my PC and modify it to be big-endian, so I don't have to worry
about it anymore.

Has anyone else seen problems like this using ext2?

John Whitney
johnw@aiinet.com

-----Original Message-----
From: greyham@research.canon.com.au
[mailto:greyham@research.canon.com.au]
Sent: Thursday, August 24, 2000 8:19 PM
To: kd@flaga.is
Cc: linuxppc-embedded@lists.linuxppc.org
Subject: Re: ext2fs or ramdisk problems......



kd@flaga.is writes:
> Or is this still the ramdisk driver teesing me?

If you haven't solved your problem with the ramdisk crashing, I very much
doubt that it will cope with having a filesystem mounted on it.  As it's a
custom board, how do you know your UPM tables are correct?  Have you tried
running the same kernel on a known-good board from Embedded Planet,
Bright Star or TQM?

Regards,
Graham


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: ext2fs or ramdisk problems......

[Scott Anderson]

"Whitney, John" wrote:
>
> I have had odd results using ext2 on the ramdisk (in my case, a 16MB
> ramdisk).  Using the 2.2.13, ramdisks I created using ext2 would boot
> up just fine on the system, but after use the filesystem would declare
> that it was out of space, even though "df" showed several megabytes
> free on that filesystem.

My guess is that you ran out of inodes.  Examine the output of dumpe2fs
and see if it gives you any clues.

    Scott Anderson
    scott_anderson@mvista.com	MontaVista Software Inc.
    (408)328-9214		490 Potrero Ave.
    http://www.mvista.com	Sunnyvale, CA  94086

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: ext2fs or ramdisk problems......

[Graham Stoney]

Whitney, John writes:
> I have had odd results using ext2 on the ramdisk (in my case, a 16MB
> ramdisk).  Using the 2.2.13, ramdisks I created using ext2 would boot
> up just fine on the system, but after use the filesystem would declare
> that it was out of space, even though "df" showed several megabytes
> free on that filesystem.

Scott's suggestion re: inodes sounds like the explanation here.
Unfortunately the error message you get for being out of inodes is the same
as what you get for running out of available blocks.

> Someday I'll go into the Minix source code on my PC and modify it to be
> big-endian, so I don't have to worry about it anymore.

This would be a really worthwhile enhancement.  I'm a little suprised noone's
bothered to fix this, given that ext2 gets it right.  It should be possible
to auto-detect the byte order it was constructed with and have it "just work"
on either architecture, even if mkfs.minix continues to write the fields in
the native byte order.

Regards,
Graham
--
Graham Stoney
Principal Hardware/Software Engineer
Canon Information Systems Research Australia
Ph: +61 2 9805 2909  Fax: +61 2 9805 2929

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

RE: ext2fs or ramdisk problems......

[Whitney, John]

Well, I'd believe that, but...

If I copy a file, such as /bin/vim, into new files until they fill up
the unused ramdisk space, delete the copies, and start copying them
again, I get the "out of space" error (actually, I don't think I even
have to fill up the space before I can get this to happen).  If it were
inode related, I'd think that it was the *number* of files and that for
each file I deleted, I could create a new one (which isn't the case).

I saw the problem in each kernel from the standard 2.2.13 up to 2.2.15
(I was trying to see if newer kernels fixed the problem).  I'm not sure
what is causing it or if anyone else has been troubled by it, but I
thought I'd bring it up to the group.

John Whitney
johnw@aiinet.com

-----Original Message-----
From: Scott Anderson [mailto:sanders@mvista.com]
Sent: Friday, August 25, 2000 1:35 PM
To: Whitney, John
Cc: 'greyham@research.canon.com.au'; kd@flaga.is;
linuxppc-embedded@lists.linuxppc.org
Subject: Re: ext2fs or ramdisk problems......



"Whitney, John" wrote:
>
> I have had odd results using ext2 on the ramdisk (in my case, a 16MB
> ramdisk).  Using the 2.2.13, ramdisks I created using ext2 would boot
> up just fine on the system, but after use the filesystem would declare
> that it was out of space, even though "df" showed several megabytes
> free on that filesystem.

My guess is that you ran out of inodes.  Examine the output of dumpe2fs
and see if it gives you any clues.

    Scott Anderson
    scott_anderson@mvista.com	MontaVista Software Inc.
    (408)328-9214		490 Potrero Ave.
    http://www.mvista.com	Sunnyvale, CA  94086


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: ext2fs or ramdisk problems......

[Graham Stoney]

Whitney, John writes:
> If I copy a file, such as /bin/vim, into new files until they fill up
> the unused ramdisk space, delete the copies, and start copying them
> again, I get the "out of space" error (actually, I don't think I even
> have to fill up the space before I can get this to happen).  If it were
> inode related, I'd think that it was the *number* of files and that for
> each file I deleted, I could create a new one (which isn't the case).

True.  I just tried both these cases, and I get the results you expect.
Are you sure your memory subsystem is working properly?
Did you pass an explicit size to mke2fs, and if so, does it match the size of
your ramdisk?

Regards,
Graham

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

ext2fs or ramdisk problems......

[kd]

Hi,

Is it for shure that the ext2fs is endian independent?
fs created with the commands (issued in Debian 2.2. (potato) on an i386
machine) :

dd if=/dev/zero of=fs bs=1k count=4096
mke2fs -m0   fs 4096
mount fs /mnt/ramdisk/ -o loop
cd /mnt/ramdisk/
tar -xvf /tmp/fs.tar
cd /tmp
umount /mnt/ramdisk/
gzip -9 fs

results in the following......

loaded at:     FE020000 FE02C204
relocated to:  00100000 0010C204
board data at: 001001C4 001001E8
relocated to:  00200100 00200124
zimage at:     FE027000 FE0758CB
initrd at:     FE0758CB FE167ECB
relocated to:  00F0D000 00FFF600
avail ram:     00201000 00F0D000

Linux/PPC load: root=/dev/ram
Uncompressing Linux...done.
Now booting the kernel
Linux version 2.2.13 (kd@BigTits) (gcc version 2.95.2 19991024 (release))
#31 Thu Aug 24 09:40:59 GMT 2000
Boot arguments: root=/dev/ram
time_init: decrementer frequency = 90000000/60
Calibrating delay loop... 23.71 BogoMIPS
Memory: 14400k available (572k kernel code, 412k data, 28k init)
[c0000000,c1000000]
DENTRY hash table entries: 262144 (order: 9, 2097152 bytes)
Buffer-cache hash table entries: 32768 (order: 5, 131072 bytes)
Page-cache hash table entries: 4096 (order: 2, 16384 bytes)
POSIX conformance testing by UNIFIX
Linux NET4.0 for Linux 2.2
Based upon Swansea University Computer Society NET3.039
NET4: Unix domain sockets 1.0 for Linux NET4.0.
NET4: Linux TCP/IP 1.0 for NET4.0
IP Protocols: ICMP, UDP, TCP
TCP: Hash tables configured (ehash 16384 bhash 16384)
Starting kswapd v 1.5
CPM UART driver version 0.03
ttyS00 at 0x0280 is a SMC
pty: 256 Unix98 ptys configured
RAM disk driver initialized:  1 RAM disks of 4096K size
loop: registered device at major 7
eth0: CPM ENET Version 0.2, 00:01:02:03:04:07
RAMDISK: Compressed image found at block 0
VFS: Mounted root (ext2 filesystem).
Freeing unused kernel memory: 28k init
attempt to access beyond end of device
01:00: rw=0, want=2036662601, limit=4096
dev 01:00 blksize=1024 blocknr=-110821048 sector=-221642096 size=1024
count=1
attempt to access beyond end of device
01:00: rw=0, want=1879073169, limit=4096
dev 01:00 blksize=1024 blocknr=1879073168 sector=-536820960 size=1024
count=1
attempt to access beyond end of device
01:00: rw=0, want=201441084, limit=4096
dev 01:00 blksize=1024 blocknr=-1946042565 sector=402882166 size=1024
count=1
attempt to access beyond end of device
etc.......
and then it hangs.

Or is this still the ramdisk driver teesing me?

Board is custom board with mpc823.
Kernel is the hard hat 2.2.13 with no nfs support, bootp or rarp support
built in.

Totally pussled!
K.D.


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: ext2fs or ramdisk problems......

[Ron Flory]

kd@flaga.is wrote:
>
> Hi,
>
> Is it for shure that the ext2fs is endian independent?
> fs created with the commands (issued in Debian 2.2. (potato) on an i386
> machine) :

 unless something changed in the last few weeks, it is.  I use e2fs on
x86 (little-endian) and m68k (big-endian) on the same physical disk
drive all the time, but that doesn't mean the partitioning schemes are
identical on all systems...

ron

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: ext2fs or ramdisk problems......

[Graham Stoney]

kd@flaga.is writes:
> Or is this still the ramdisk driver teesing me?

If you haven't solved your problem with the ramdisk crashing, I very much
doubt that it will cope with having a filesystem mounted on it.  As it's a
custom board, how do you know your UPM tables are correct?  Have you tried
running the same kernel on a known-good board from Embedded Planet,
Bright Star or TQM?

Regards,
Graham

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/