svn commit: samba r8512 - in trunk/source: include smbd

Jeremy Allison jra at samba.org
Sat Jul 16 09:50:30 GMT 2005


On Sat, Jul 16, 2005 at 11:45:25AM +0200, Volker Lendecke wrote:
> On Sat, Jul 16, 2005 at 02:30:08AM -0700, Jeremy Allison wrote:
> > It simplifies the logic but at a cost of merging two
> > different actions (level2 break and kernel break) that
> > happen to have the same code paths.
> 
> Hmmm. The code I left in still does distinguish between kernel induced breaks
> and ones triggered by another open. I just could not tell why a message
> between two smbds should carry information what kind of oplock the target smbd
> had left in the share mode db. It should know better itself what kind of
> oplock it currently holds.

Probably paranoia. The dept. of redundancy dept. :-). It can
help to debug - just in case we have an issue where one
smbd reads a record entry and sends a break message and
there is a locking bug where someone else modifies it...

Paranoia I know, but it's just defensive programming to
me - helps me track this sort of thing down.

Jeremy.


More information about the samba-technical mailing list