lkml.org 
[lkml]   [2005]   [Sep]   [13]   [last100]   RSS Feed
Views: [more markup]  [less markup]  [headers]  [forward] 
 
Messages in this thread
/
DateTue, 13 Sep 2005 09:04:32 +0200
FromEric Dumazet <>
SubjectRe: "Read my lips: no more merges" - aka Linux 2.6.14-rc1
Sonny Rao a écrit :
> On Mon, Sep 12, 2005 at 08:34:17PM -0700, Linus Torvalds wrote:
> <snip> 
> 
>>On the filesystem level, FUSE got merged, and ntfs and xfs got updated. In 
>>the core VFS layer, the "struct files" thing is now handled with RCU and 
>>has less expensive locking.
> 
> 
> I hope this means that people will be more accepting of multi-threaded
> benchmarks (who needs real apps... ;-)) which do open() and close().
> 
> 
> Yes, no?

If you look at RCU change, you discover they impact read()/write()/... (no 
more locking), but not open()/dup()/socket() and close() that still take a 
spinlock to modify the state.

And if your process has many files opened, the cost (read : latency) of open() 
can be very high, finding a zero bit in a large bit array.

So these RCU changes can help some benchmarks (or real apps... ;-) ), but not 
some others :)

I wish a process param could allow open() to take any free fd available, not 
the lowest one. One can always use fcntl(fd, F_DUPFD, slot) to move a fd on a 
specific high slot and always keep the 64 first fd slots free to speedup the 
kernel part at open()/dup()/socket() time.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-09-13 07:07    [from the cache]
©2003-2008