BSC passes fine - I have no objections.
We should probably make sure this is in before 2.0.3

On 1/19/06, Andrew Beekhof <[EMAIL PROTECTED]> wrote:
> testing on OSX...
>
> On 1/19/06, David Lee <[EMAIL PROTECTED]> wrote:
> > At long, long last, I think I've got "cl_status" working on Solaris.  It
> > had never worked in the past (anyone else ever had any luck?)
> >
> > I propose applying the attached patch in the next day or so, but it would
> > be good if folk could sanity check it, primarily to ensure I haven't
> > broken any existing functionality (and secondarily, if they wish, to check
> > the new bits).
> >
> > The first key to this was a small patch to ipcsocket.c a few days ago,
> > which cured a word-alignment problem in a couple of assignments.  This had
> > been inducing silent crashes.
> >
> > The second is the proposed patch, again contained within "ipcsocket.c".
> > It looks alarmingly big, but is logically very small.
> >
> > In effect, all it does is to place the few lines of code that are specific
> > to "sockets" technology (accept(), listen(), bind(), etc.) inside a
> > branching "#if", and alongside it place another branch that can drive
> > "streams" technology.
> >
> > At the top of the patch is "#if" logic that says:
> >
> >    if any of the known credentials-via-sockets mechanisms exist
> >    then
> >       #define to use "sockets" technology /* preferred; as always */
> >    else
> >       #define to use "streams" technology /* fallback; new */
> >    fi
> >
> > Thus sockets are stongly preferred: streams (which carries uid/gid) is
> > merely the fallback option (necessary for Solaris releases up to "9").
> >
> > I understand that Solaris 10 introduces a credentials option available to
> > both sockets and streams.  I have sketched in (but am unable to test) the
> > logic to allow for this, so that it ought to choose our preferred
> > mechanism of sockets.
> >
> > (The curious and adventurous may wish to set the relevant "#define" to
> > force use of streams...)
> >
> > The revised code contines to pass "ipctest" on Linux/Intel 32-bit and
> > 64-bit, and now, for the first time ever, passes on Solaris/sparc.
> >
> > I propose to apply the patch in the next day or so.
> >
> > --
> >
> > :  David Lee                                I.T. Service          :
> > :  Senior Systems Programmer                Computer Centre       :
> > :                                           Durham University     :
> > :  http://www.dur.ac.uk/t.d.lee/            South Road            :
> > :                                           Durham DH1 3LE        :
> > :  Phone: +44 191 334 2752                  U.K.                  :
> >
> > _______________________________________________________
> > Linux-HA-Dev: [email protected]
> > http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
> > Home Page: http://linux-ha.org/
> >
> >
> >
> >
>
_______________________________________________________
Linux-HA-Dev: [email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/

Reply via email to