On Wed, 18 Aug 1999, Peter J. Braam wrote:
> I've recently learned that disks sometimes reorder writes (I think the
> SCSI models do it).
>
> If we look at how the buffers are flushed for ext2 files, then the
> system is (fs/buffer.c)
>
> flush data buffers
> write super block to buffers
> write inodes to buffers
> flush data buffers again
>
> The bdflush daemon does NOT require a wait after the first flush of
> buffers, so a smart disk, i.e. one that re-orders writes, may write an
> inode for a file while new block have not yet been written.
>
> This means that this file could see data that once belonged to someone
> else -- a true nono!
Only if the power goes out, I think.
I may be wrong, but I'd guess that the kernel wouldn't release the
buffers it was sending to disk until the driver (controller/disk)
reported success.
Any metadata (and data) which might be incorrect on disk will be
locked in kernel memory as buffers/caches won't it?
Or did I misread your question?
Matthew.
