On Fri, 28 Apr 2000, Andi Kleen wrote:
[ file_fsync calls write_super without locking the super ]
> On Fri, Apr 28, 2000 at 12:32:26PM -0700, Chris Mason wrote:
> > Hmm, that makes sense I can see it being a slow down for ext2 if oracle or
> > mail servers pound on fsync (in reiserfs, they block on the commit).
> >
> > I'm not sure I understand why need the locking in sync_supers if we don't
> > need it in fsync though.
>
> At least the ext2 write_super does not seem to care (in theory there
> could be a small race against remounting read only, but that is probably
> harmless and would occur even with the lock)
>
> The locking protocol should be better documented ...
>
In truth, this isn't a huge issue, as reiserfs isn't using file_fsync. I
thought it was being called elsewhere but just did a find and it looks as
though it is only being used when the FS puts it in an operations struct.
The major reason I put it in the diff was to draw attention to
the differences in how write_super is called...
-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 : Sun Apr 30 2000 - 21:00:15 EST