Re: speed of ffs (OpenBSD / 4.3BSD)

[Marc Espie <Marc_(_dot_)_Espie_(_at_)_liafa_(_dot_)_jussieu_(_dot_)_fr>]

On Sun, May 17, 1998 at 02:27:44AM -0600, Theo de Raadt wrote:

> The Linux buffer cache has NO ordering constraints at all, and hence
> it is obviously faster at doing directory operations.  What does this
> mean?  It means that if you crash that machine at the right time, your
> filesystem will be utterly fucked.

With regards to design decisions, I remember Linus saying:
`fsck should NOT be considered as a good substitute for regular
backups'... that was when they removed the i-node clearing code
for rm.

You can't always have your cake and eat it too. :-)

What annoys me somewhat there is that considering the differing
design decisions, we can't easily benchmark OpenBSD FFS vs.
linux ext2fs, and see whether there's any `secure' improvement
we missed, can we ?
-- 
	Marc Espie		
|anime, sf, juggling, unicycle, acrobatics, comics...
|AmigaOS, OpenBSD, C++, perl, Icon, PostScript...