On Wed, 29 Mar 2000, Andrea Arcangeli wrote:
> On Wed, 29 Mar 2000, Chris Mason wrote:
>
> >[ for protecting files from corrupt data ]
>
> Yep. More precisely what I meant was to protect files from having random
> contents that was previously on disk before the first data write to the
> file (that's also _the_ security issue).
>
Ok. The security issue is easier to fix. When adding blocks onto the end
of the file, flush them to disk before allowing the transaction that added
them to commit. This won't be a huge penalty with transaction ages of 30
seconds or so. I'm already doing this in 2.3 for direct to indirect
conversions, which is the one place in our 2.2 code that logs data blocks.
This won't solve the kernel.org problem, as it only promises that what is
in the file has been written at least once, instead of making sure the
entire file is there.
-chris
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Mar 31 2000 - 21:00:25 EST