I've added Jim's AIX building patches, and I tweaked openslp.misc/debian/init.d to have the " -- -p /var/run/$NAME.pid" set correctly. Version 1694.
Other than that, I think the only potentially outstanding thing is Gavin's code to handle network changes. Which I have not looked at, and will probably require more testing. Does the group want to look at them, or move on with 2.0? Or are there other patches/bugs we've missed? --Nick On Thu, Nov 1, 2012 at 2:49 PM, Jim Marshall <jim.marsh...@wbemsolutions.com > wrote: > Are there any other changes that are being waited on? I've run our tests > and they all work, so I would like to propose we roll out 2.0. > > Thanks > > John Calcote wrote: > > Fixed in revision 1693 – SLP atomics now return the correct (new) value > on AIX.**** > > ** ** > > John**** > > ** ** > > *From:* John Calcote [mailto:john.calc...@gmail.com<john.calc...@gmail.com>] > > *Sent:* Tuesday, October 30, 2012 12:14 PM > *To:* Jim Marshall > *Cc:* openslp-devel@lists.sourceforge.net > *Subject:* Re: [Openslp-devel] slptool hangs on AIX?**** > > ** ** > > Hi Jim,**** > > On Tue, Oct 30, 2012 at 10:54 AM, Jim Marshall < > jim.marsh...@wbemsolutions.com> wrote:**** > > Hi, > Been testing the AIX build I have and the agent is working fine but when > I invoke slptool it basically hangs. Stepping into the code we get to > libslp_handle.c line 66 > > if (SLPAtomicInc(&s_OpenSLPHandleCount) == 1) > > The SLPAtomicInc function does this on AIX: > > return (intptr_t)fetch_and_add((atomic_p)pn, 1); > > This is the cause of the hang, because fetch_and_add "[...] returns the > original value of the variable" (from the man page), in this case '0'. When > this happens the code enters a while loop**** > > ** ** > > ** ** > > Well, that's kind of useless, isn't it? It should be replaced with:**** > > ** ** > > return (intptr_t)fetch_and_add((atomic_p)pn, 1) + 1;**** > > ** ** > > The idea here is to atomically increment the value stored at pn. If two > threads try to do it at once, then they could both read the value, > increment it and both write the same value back out. The use of atomic_add > should cause both threads to actually increment the value, regardless of > how in-sync they are on their attempts to update the value.**** > > ** ** > > It's only a side-effect (that we take advantage of) that most atomic incs > return the newly updated value. Since the AIX version returns the original > value, and we're guaranteed to read/update/write atomically when we add 1 > to that original value, then we can assume that the value we get back PLUS > ONE is our actual desired return value.**** > > ** ** > > I'll fix it.**** > > ** ** > > John**** > > > > -- > Jim Marshall Sr. Software Engineer > [image: WS] <http://ws-inc.com/> > > > > ------------------------------------------------------------------------------ > Everyone hates slow websites. So do we. > Make your web apps faster with AppDynamics > Download AppDynamics Lite for free today: > http://p.sf.net/sfu/appdyn_sfd2d_oct > _______________________________________________ > Openslp-devel mailing list > Openslp-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/openslp-devel > >
<<WSLogo.jpg>>
------------------------------------------------------------------------------ LogMeIn Central: Instant, anywhere, Remote PC access and management. Stay in control, update software, and manage PCs from one command center Diagnose problems and improve visibility into emerging IT issues Automate, monitor and manage. Do more in less time with Central http://p.sf.net/sfu/logmein12331_d2d
_______________________________________________ Openslp-devel mailing list Openslp-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/openslp-devel
