Coda File System

Re: gcc-2.95 patches (untested)

From: Jan Harkes <jaharkes_at_cs.cmu.edu>
Date: Mon, 30 Aug 1999 13:12:36 -0400
On Mon, Aug 30, 1999 at 02:53:49PM +0900, Stephen J. Turnbull wrote:
> >>>>> "Troy" == Troy Benjegerdes <hozer_at_drgw.net> writes:
> 
>     Troy> Attached is a patch which I just used to build coda with
>     Troy> GCC-2.95. However, I haven't had a chance to test this, so
>     Troy> someone let me know how this works.
> 
> I made very similar changes, and managed a build.
> 
> I cannot get venus-5.2.7 to initialize; it hangs up unable to get the
> RootVolume name.  (Hang up == every thirty seconds it tries again
> until killed.)  This is true both with a local venus and a venus
> across the net, and also with a not-yet-updated venus-5.2.0.
> 
> Shoulda known better than to upgrade file system code with the
> "just-before-vacation" release.  :-P

Actually, it is reasonable stable. We have at least 1 fileserver that
was started the day 5.2.7 was released that is still running, 65.8 days
by now. And it has been used quite a lot, it handles among others my
email folders, the webpages, and homedirectories for 5 users.

So I wonder what the reason for the failed initialization might be.
Could you try to get a trace of the rpc2 packets with either
rpc2tcpdump, or by turning up debuglevels in the client (vutil -d 10 10)
and server (volutil setdebug 10). RPC2 should then write gobs of
information into /usr/coda/etc/console and /vice/srv/SrvErr

>     Troy> As for adding the -fpermissive flag, here is what I use in
>     Troy> my RPM spec file:
> 
>     Troy> MYFLAGS="$RPM_OPT_FLAGS -fpermissive" make  
> 
> Under Debian, "CXX='g++ -fpermissive -fno-exceptions debian/rules" at
> a bourne shell prompt works.  "CXXFLAGS=-fpermissive -fno-exceptions"
> doesn't work.

Yes, some makefiles override the CFLAGS and CXXFLAGS variables. MYFLAGS
seems to be used everywhere, just like CC and CXX.

Jan
Received on 1999-08-30 13:16:10