LAM/MPI logo

LAM/MPI General User's Mailing List Archives

  |   Home   |   Download   |   Documentation   |   FAQ   |   all just in this list

From: Brian Barrett (brbarret_at_[hidden])
Date: 2005-08-19 15:27:12


On Aug 19, 2005, at 11:53 AM, Grismer Matthew J Civ AFRL/VAAC wrote:
> I’m kind of amazed by this, but the problems with sysv and usysv
> resolved themselves after installing the latest
> SecurityUpdate2005-007 v1.1 from Apple and rebooting the machine.
> I’m not sure if it was the security patches or rebooting the
> machine, but MPI codes now run fine with the shared memory rpi’s.
It was probably actually the reboot that fixed things for you. SysV
shared memory (what LAM uses for the sysv and usysv rpis) doesn't get
automatically released when all processes using it exit. LAM takes
great pains not to leak sysv shared memory, but if a lamd crashes or
the universe is never taken down, sometimes memory is left
allocated. Using the lamclean command usually clears that up.
Apache and MySQL both can use large chunks of sysv shared memory, so
if they are running, this can be more noticeable.

OS X has a really low default amount of sysv shared memory (just
barely enough to meet the minimums for LAM), so if any gets "leaked",
it can mean the RPIs don't select, just as you've seen.
> Regarding the problems with building the Totalview support, here is
> the output from the build process when it fails
Can you try compiling without the -fast CFLAG? The error you are
seeing is deep voodoo magic with how code is laid out for shared
libraries. I'm not sure if -fast is causing this, but for some
reason, you aren't compiling properly for dynamically loaded shared
libraries. Apple doesn't do quite as well regression testing the -
fast compiler (it's actually a different compiler then the normal
one, and can do some really unsafe things - I generally don't
recommend using the -fast option on OS X), so I'm wondering if they
missed a case here.

Thanks,

Brian

> Making all in totalview
>
> Making all in src
>
> if /bin/sh ../libtool --tag=CC --mode=compile gcc -DPACKAGE_NAME=
> \"\" -DPACKAGE_TARNAME=\"\" -DPACKAGE_VERSION=\"\" -DPACKAGE_STRING=
> \"\" -DPACKAGE_BUGREPORT=\"\" -DLAM_WANT_DEBUG=0 -DLAM_CC=\"gcc\" -
> Dlam_sizeof_void_p=4 -DSTDC_HEADERS=1 -DHAVE_SYS_TYPES_H=1 -
> DHAVE_SYS_STAT_H=1 -DHAVE_STDLIB_H=1 -DHAVE_STRING_H=1 -
> DHAVE_MEMORY_H=1 -DHAVE_STRINGS_H=1 -DHAVE_INTTYPES_H=1 -
> DHAVE_STDINT_H=1 -DHAVE_UNISTD_H=1 -DHAVE_DLFCN_H=1 -I. -I. -
> I../../../share/include -I../../../share/include -DLAM_BUILDING=1 -
> DLAM_BUILDING=1 -D_REENTRANT -DLAM_BUILDING=1 -fast -MT dll_lam.lo
> -MD -MP -MF ".deps/dll_lam.Tpo" -c -o dll_lam.lo dll_lam.c; \
>
> then mv -f ".deps/dll_lam.Tpo" ".deps/dll_lam.Plo"; else rm -f
> ".deps/dll_lam.Tpo"; exit 1; fi
>
> gcc -DPACKAGE_NAME=\"\" -DPACKAGE_TARNAME=\"\" -DPACKAGE_VERSION=
> \"\" -DPACKAGE_STRING=\"\" -DPACKAGE_BUGREPORT=\"\" -
> DLAM_WANT_DEBUG=0 -DLAM_CC=\"gcc\" -Dlam_sizeof_void_p=4 -
> DSTDC_HEADERS=1 -DHAVE_SYS_TYPES_H=1 -DHAVE_SYS_STAT_H=1 -
> DHAVE_STDLIB_H=1 -DHAVE_STRING_H=1 -DHAVE_MEMORY_H=1 -
> DHAVE_STRINGS_H=1 -DHAVE_INTTYPES_H=1 -DHAVE_STDINT_H=1 -
> DHAVE_UNISTD_H=1 -DHAVE_DLFCN_H=1 -I. -I. -I../../../share/include -
> I../../../share/include -DLAM_BUILDING=1 -DLAM_BUILDING=1 -
> D_REENTRANT -DLAM_BUILDING=1 -fast -MT dll_lam.lo -MD -MP -MF .deps/
> dll_lam.Tpo -c dll_lam.c -fno-common -DPIC -o .libs/dll_lam.o
>
> /bin/sh ../libtool --tag=CC --mode=link gcc -fast -o
> liblam_totalview.la -rpath /usr/local/lib/lam dll_lam.lo
>
> gcc -dynamiclib ${wl}-flat_namespace ${wl}-undefined ${wl}suppress -
> o .libs/liblam_totalview.0.0.0.dylib .libs/dll_lam.o -
> install_name /usr/local/lib/lam/liblam_totalview.0.dylib -Wl,-
> compatibility_version -Wl,1 -Wl,-current_version -Wl,1.0
>
> ld: .libs/dll_lam.o has local relocation entries in non-writable
> section (__TEXT,__text)
>
> /usr/bin/libtool: internal link edit command failed
>
> make[3]: *** [liblam_totalview.la] Error 1
>
> make[2]: *** [all-recursive] Error 1
>
> make[1]: *** [all-recursive] Error 1
>
> make: *** [all-recursive] Error 1

-- 
   Brian Barrett
   LAM/MPI developer and all around nice guy
   Have a LAM/MPI day: http://www.lam-mpi.org/