Re: matlab

["Eric G. Miller" <egm2@jps.net>]

On Tue, Feb 20, 2001 at 04:04:16PM -0800, Rick Rezinas wrote:
> if you can check out the lines shortly preceding the first
> 
> read(4, 0xbfffdc0c, 32)                 = -1 EAGAIN (Resource temporarily
> unavailable)
> 
> there may be some useful information there.  I don't know nearly enough to
> do anything with this stuff ;)

That doesn't look to be the problem since the program then goes into an
indefinite sleep using select() to wait for data to become available on
the read() file descriptor (4). In the fragments below, this resource
finally does become readable. Though, calling select() with the timeout
arg as NULL, could lead to starvation if the resource being waited on
never becomes available.  

The sigsuspend call might be where it's hanging since it's basically
saying sleep until I get "some" signal(s).  Is there more to this trace?
I missed the beginning of this thread...

But sinced you mentioned you're "su" symptom, is it possible something
is clobbering your environment variables ($USER, $HOME, $SHELL,
$LOGNAME)?  Don't know why it'd hang on such things...???

> 
> rick
> 
> On Tue, 20 Feb 2001, Pascal Hos wrote:
> 
> > Well the same happens when using ssh or rsh or any variant of remote login. 
> > The environments are exactly the same when logged in locally or remote.
> > 
> > strace matlab hangs on the following statement:
> > bash-2.03$ strace matlab
> > execve("/usr/local/bin/matlab", ["matlab"], [/* 23 vars */]) = 0
> > brk(0)                                  = 0x80bb3ec
> > .......
> > 
> > <snip>
> > 
> > .......
> > read(4, 0xbfffdc0c, 32)                 = -1 EAGAIN (Resource temporarily 
> > unavailable)
> > select(5, [4], NULL, NULL, NULL)        = 1 (in [4])
> > read(4, "\1\0:\1\0\0\0\0\1\0\0\0\2\0\0\0t\370\377\277\334\234*\10"..., 32) = 
> > 32
> > write(4, "\217\16\2\0\0\0\0\0", 8)      = 8
> > read(4, 0xbfffdbc8, 32)                 = -1 EAGAIN (Resource temporarily 
> > unavailable)
> > select(5, [4], NULL, NULL, NULL)        = 1 (in [4])
> > read(4, "\1\370;\1\200\0\0\0\4\0\0\0 \0\0\0\0\20\0\0\0\0\0\0\0\0"..., 32) = 32
> > read(4, "#\0\0\0\4\0\0\0\1\0\0\0\10\0\0\0\10\0\0\0\10\0\0\0\0\0"..., 128) = 
> > 128
> > rt_sigprocmask(SIG_SETMASK, NULL, [RT_0], 8) = 0
> > rt_sigsuspend([]
> > 
> > I have no idea what that means. Maybe one of you knows. Anyway, thanks for 
> > replying.
> > 
> > Pascal
> > 
> > On Tuesday 20 February 2001 17:46, Brian May wrote:
> > > >>>>> "Erik" == Erik Steffl <steffl@bigfoot.com> writes:
> > >
> > >     Erik> Pascal Hos wrote:
> > >     >> I just received my matlab R12 student version and I'm having
> > >     >> some difficulties getting it to work correctly on potato. When
> > >     >> I start it from an xterm the splash screen shows up and it then
> > >     >> the program hangs. The strange part is that if I su as myself
> > >     >> and then execute matlab, everything works just fine!? The same
> > >     >> problem occurs as root. What could be happening here?
> > >
> > > Try strace matlab?
> > >
> > >     Erik>   I suspect that program tries to figure out who's ruinning
> > >     Erik> it and does not succeed - perhaps it uses entry in utmp/wtmp
> > >     Erik> and xterm does not log entry there but su does. also check
> > >     Erik> the differences between environments (when you open xterm
> > >     Erik> and then when you su)
> > >
> > > Is Matlab still based on libc5?
> > >
> > > If so that might be a problem - the format of utmp/wtmp IIRC has
> > > changed from libc5 to libc6.
> > >
> > > (then again, if that was the case it shouldn't work with su...).
 

-- 
Eric G. Miller <egm2@jps.net>