fcntl F_SETLKW64 failing on Solaris

Tristan Ball tristanb at vsl.com.au
Wed Jan 9 16:27:03 GMT 2002


Would it be so bad to count the number of times we spin on a lock, and once
it reaches a threshold, check for a dead process? For it to be workable, it
would probably need to track the PID that owns a lock as well....

----- Original Message -----
From: "Anton Blanchard" <anton at samba.org>
To: "Jeremy Allison" <jra at samba.org>
Cc: "Tristan Ball" <tristanb at vsl.com.au>; <samba-technical at samba.org>
Sent: Thursday, January 10, 2002 11:11 AM
Subject: Re: fcntl F_SETLKW64 failing on Solaris


>
> > Just a warning - using --with-spinlocks can be very dangerous. If a smbd
> > process abends holding a spinlock, all the smbd's will have to be
restarted
> > and the tdb cleared to remove it.
> >
> > The same won't happen with fcntl tdb's.
> >
> > Just a little touch of paranoia.... :-).
>
> But samba never SEGVs Jeremy :) Yes there is no way out if something
> dies with a spinlock. But if a process dies inside a critical region its
> probably better to start from scratch, who knows what ugly things have
> happened to the tdb by that stage.
>
> Anton
>
>





More information about the samba-technical mailing list