gjournal and Softupdates

[Pawel Jakub Dawidek]

On Wed, Sep 13, 2006 at 05:28:49PM +0200, Teufel wrote:
> Pawel Jakub Dawidek wrote:
> >>>- todays desktop drives can lie about writing data. SoftUpdates relies
> >>>on some assumptions about when the data is physically written to
> >>>media, and those are not always valid today
> >>>      
> >>I think journaling relies on the same assumptions.
> >>    
> >
> >Not gjournal, because it uses BIO_FLUSH I/O requests which flushes disk
> >write cache when needed
> so when the crash occur exactly when BIO_FLUSH is sent or while the
> cache is flushing, there is still no corruption possbile? [...]

That's right. One BIO_FLUSH is send to ensure the data are safely
stored, and another one is send when metadata is updated to point at
the last consistent journal.

> [...] If so, this would be an advantage over SU, as 
> it does surely not use the new introduced BIO_FLUSH. [...]

Soft-updates doesn't handle disk write caches at all.

> [...] In the other hand i've seen couple of other JFS that went corrupt for "no reason". I don't want to be paranoid, but i 
> really want to be "sure" that the design is trustable.

Of course a bug in file system (or gjournal) implementation is still
possible and can lead to file system corruption, but such a bug can
still corrupt file system in the way it will not be fixable by fsck.

From what I saw, file systems with journaling still enforce fsck every X
reboots or on the next reboot after Y days of uptime, whatever comes first.

-- 
Pawel Jakub Dawidek                       http://www.wheel.pl
pjd at FreeBSD.org                           http://www.FreeBSD.org
FreeBSD committer                         Am I Evil? Yes, I Am!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20060913/06d5a430/attachment.pgp