Skip to contentSkip to conversation

2.4.15aa1

DISCUSSION Andrea Arcangeli 5 participants 7 messages Nov 24, 2001 – Nov 26, 2001

lore
Share & tools
lore
#0 lore raw

Note: the "00_read_super-stale-inode-1" fix is under discussion with Al, but overall it should be just ok for public consumation (even if that are may change if we find any better alternative, at the moment I think it is better (cleaner, simpler and faster) alternative than the iput changes in function of the MS_ACTIVE info).

URL:

	ftp://ftp.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.15aa1.bz2
	ftp://ftp.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.15aa1/

Only in 2.4.15aa1: 00_iput-unmount-corruption-fix-1

Fix iput umount corruption.

Only in 2.4.15aa1: 00_read_super-stale-inode-1

If read_super fails avoid lefting stale inodes queued into the superblock.

Only in 2.4.15pre9aa1: 10_vm-16
Only in 2.4.15aa1: 10_vm-17

Dropped a leftover touch_buffer in bread (there's just one in getblk in -aa, and we need it in getblk [not only for reiserfs]).

Andrea

#1 lore raw

Will this fix the potential possibility of 2.4.15 screwing my system? (inode problem as people have been mentioning it, etc?)

Best regards,

Paulo

Quoted 33 lines · Andrea Arcangeli

Note: the "00_read_super-stale-inode-1" fix is under discussion with Al, but overall it should be just ok for public consumation (even if that are may change if we find any better alternative, at the moment I think it is better (cleaner, simpler and faster) alternative than the iput changes in function of the MS_ACTIVE info).

URL:

	ftp://ftp.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.15aa1.bz2
	ftp://ftp.us.kernel.org/pub/linux/kernel/people/andrea/kernels/v2.4/2.4.15aa1/

Only in 2.4.15aa1: 00_iput-unmount-corruption-fix-1

Fix iput umount corruption.

Only in 2.4.15aa1: 00_read_super-stale-inode-1

If read_super fails avoid lefting stale inodes queued into the superblock.

Only in 2.4.15pre9aa1: 10_vm-16
Only in 2.4.15aa1: 10_vm-17

Dropped a leftover touch_buffer in bread (there's just one in getblk in -aa, and we need it in getblk [not only for reiserfs]).

Andrea
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Signature · 7 lines
Paulo J. Matos aka PDestroy : pocm(_at_)rnl.ist.utl.pt
Instituto Superior Tecnico - Lisbon    
Software & Computer Engineering - A.I.
 - > http://www.rnl.ist.utl.pt/~pocm 
 ---	
	Yes, God had a deadline...
		So, He wrote it all in Lisp!
#2 lore raw
In reply to Andrea Arcangeli
Quoted 3 lines · Paulo J. Matos aka PDestroy

Will this fix the potential possibility of 2.4.15 screwing my system? (inode problem as people have been mentioning it, etc?)

yes.

Andrea

#3 lore raw
In reply to Andrea Arcangeli
Quoted 3 lines · Andrea Arcangeli

Only in 2.4.15aa1: 00_iput-unmount-corruption-fix-1

Fix iput umount corruption.

Is this the problem that Al put out a patch for yesterday? And is his patch been tested and working?

Quoted 4 lines

Only in 2.4.15aa1: 00_read_super-stale-inode-1

If read_super fails avoid lefting stale inodes queued into the superblock.

What is this? How dangerous is it?

Signature · 4 lines
Phil Sorber
AIM: PSUdaemon
IRC: irc.openprojects.net #psulug PSUdaemon
GnuPG: keyserver - pgp.mit.edu
#4 lore raw

In article <1006629351.1470.8.camel@praetorian> [email protected] asked:

| -=-=-=-=-=-
| 
| On Sat, 2001-11-24 at 02:50, Andrea Arcangeli wrote:
| > Only in 2.4.15aa1: 00_iput-unmount-corruption-fix-1
| > 
| > 	Fix iput umount corruption.
| 
| Is this the problem that Al put out a patch for yesterday? And is his
| patch been tested and working?
| 
| > Only in 2.4.15aa1: 00_read_super-stale-inode-1
| > 
| > 	If read_super fails avoid lefting stale inodes queued into
| > 	the superblock.
| 
| What is this? How dangerous is it?

Good question. As I read the patch, the problem occurs during umount, when dirty inodes are not (properly) written to the disk. Would a sync() help or eliminate this problem, assuming that all files were closed? Hopefully someone knows, I don't want to tell you it only happens at umount, but that's my impression.

In any case, since 2.4.16 is out (so much for "2.4.15 released without embarrassment") to fix the problem, I would go with that unless you have a reason to use whichever patch pleases you.

Signature · 3 lines
bill davidsen <[email protected]>
  His first management concern is not solving the problem, but covering
his ass. If he lived in the middle ages he'd wear his codpiece backward.
#5 lore raw
In reply to Andrea Arcangeli
Quoted 2 lines · Phil Sorber

Is this the problem that Al put out a patch for yesterday? And is his patch been tested and working?

I'm running it 2.4.15-viro three systems here with no detectable problems. One has been building lots of kernels, the other has been sitting rather idle, except for the occasional redhat package build, and has been rebooted a few times. These are using a mixture of ext2 and NFS.

The last system is my firewall, which I don't expect to show any problems with or without Al's fix, since it's running completely diskless (ie, NFS).

Naturally, they're all ARM boxen.

--
Russell King ([email protected]) The developer of ARM Linux
http://www.arm.linux.org.uk/personal/aboutme.html

#6 lore raw
In reply to Andrea Arcangeli
Quoted 7 lines · Phil Sorber
Quoted 3 lines · Andrea Arcangeli

Only in 2.4.15aa1: 00_iput-unmount-corruption-fix-1

Fix iput umount corruption.

Is this the problem that Al put out a patch for yesterday? And is his patch been tested and working?

it's an alternate fix (goes back to the pre8 way).

Quoted 7 lines
Quoted 5 lines

Only in 2.4.15aa1: 00_read_super-stale-inode-1

If read_super fails avoid lefting stale inodes queued into the superblock.

What is this? How dangerous is it?

pre8 and all previous kernels could left a stale inode if read_super failed after an iget [see the discussion with Al on l-k]. this patch should fix it in practice for all fs in the kernel.

Andrea