Burton, the babblefish english translation of HP�s german saying is very interesting...
Your translation is better: libc_r is an 10.20 compatiblity library which must ot be used for native 11.00 programs. Within 11.00 the functionality of real-time semaphores "sem_init" has been moved into librt and we have to link native 11.00 programs with lib_rt and not libc_r. (1st I linked ntop with libc_r by accident ---> core dump) Back to the facts: >From my knowledge (and rumors from HP Support centre), multithreading was experimental introduced into HP UX 10.20 and kernel threads were experimental added into 10.30. 10.30 was an experimental relase only. The 1st release with official thread support is 11.00. Therefore I would not encourage anyone to try out threads on 10.20 *and* 10.20 was discontinued in June 2002. No further patches will be available. My proposal is to use multithreading on 11.00 and no threads on 10.20 unless someone will test it on older 10.20 machines. I moved my installation to 11.00 in 2001. > Felk: > > I Babblefished HP's comment. I hope it makes more sense in German :-( > > libc_r one does not indicate, because it is a left on libc.1, which is > indicated again of ldd. That is a 10.-20-compatible Lib, which may not be > used for 11.0er of programs. Functionality real-time semaphores > after/usr/lib/librt.sl shifted, there gives it also the symbol sem_init. > Please you link the program with lrt and take you to lc_r raus. > > I *think* they are saying that -- at least in HP land > > libc_r is the 10.20 compatiblity library which may (should?) not be used > on > 11.0 and that they moved the functionality of real-time semaphores into > librt and we should link with that, not libc_r. > > Phew... > > What happens if you add another if test ahead of the existing c_r one for > rt?? > > > dnl> > dnl> Thread Library (optional) > dnl> check for `pthread_create' in -lpthread > dnl> > dnl> Do not perform any check for the thread library, > dnl> if the user decided to compile without multithread support. > dnl> > if test ".$ac_disable_multithread" = ".no"; then > AC_CHECK_LIB(pthread, pthread_create) > if test ".`echo $LIBS | grep pthread`" != .; then > dnl> > dnl> check for `sem_init' in -lposix4 > dnl> > AC_CHECK_LIB(posix4, sem_init) > AC_DEFINE(MULTITHREADED) > AC_DEFINE(ASYNC_ADDRESS_RESOLUTION) > else > AC_CHECK_LIB(pthreads, pthread_create) > if test ".`echo $LIBS | grep pthread`" != .; then > dnl> > dnl> check for `sem_init' in -lposix4 > dnl> > AC_CHECK_LIB(posix4, sem_init) > AC_DEFINE(MULTITHREADED) > AC_DEFINE(ASYNC_ADDRESS_RESOLUTION) > else > dnl> HPUX 11 > AC_CHECK_LIB(rt, pthread_create) > if test ".`echo $LIBS | grep rt`" != .; then > AC_DEFINE(MULTITHREADED) > AC_DEFINE(ASYNC_ADDRESS_RESOLUTION) > else > dnl> HPUX 10.20 > AC_CHECK_LIB(c_r, pthread_create) > if test ".`echo $LIBS | grep c_r`" != .; then > AC_DEFINE(MULTITHREADED) > AC_DEFINE(ASYNC_ADDRESS_RESOLUTION) > else > multithread_mode=no > fi > fi > fi > fi > else > AC_MSG_WARN([Multithread support disabled via command line option!]) > multithread_mode=no > fi > > > (I think that's right) > > > -----Burton > > > -----Original Message----- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf > Of Falk Hayn > Sent: Friday, September 27, 2002 11:53 AM > To: [EMAIL PROTECTED] > Cc: [EMAIL PROTECTED] > Subject: RE: [Ntop-dev] no multithread on HP-UX 11.00 - resolved - > codechange request > > > Burton, > > regarding Your question: > If pthread_create is a wrapper around the function > __pthread_create_system, > we still should test for the wrapper, because that's what ntop uses. > --> I tried this first, but autoconf was unable to find pthread_create > --> But autoconf can find __pthread_create_system. That is the only reason > > > > What it looks like is that you have a semaphore/thread library under HPUX > that isn't Posix nor any of the standard locations. So what makes you > think it will work like the Posix threads that ntop is expecting?? > --> I had a case open with HP regarding this and HP named the right libs > --> Thread apis exist in 3 flavours > --> HP DCE (propritary) > --> HP posix libs compatibile for 10.20 > --> HP posix libs for 11.00 > > This was the answer of HP GRC: > > libc_r wird deshalb nicht angezeigt, weil es ein Link auf libc.1 ist, die > wiederum vom ldd angezeigt wird. Das ist eine 10.20-kompatible Lib, die > f�r > 11.0er Programme nicht verwendet werden darf. > > Die Funktionalit�t der Realtime-Semaphoren wurde nach /usr/lib/librt.sl > verschoben, dort gibt es auch das Symbol sem_init. Bitte linken Sie das > Programm mit -lrt und nehmen Sie -lc_r raus. > > --> I compiled and tried it out. It seems to work. > > > > > > > Your English is a lot better than my Spanish (I can maybe order a beer) > > which is a lot better than my German... > > > > OK, let's try again. > > > > I was trying to confirm that the entire block of tests is only for the > > HP-UX > > environment. So that it would need to be set off by some type of "if > hpux > > ... endif" block. > > > > What I don't understand is why change the test? If pthread_create is a > > wrapper around the function __pthread_create_system, we still should > test > > for the wrapper, because that's what ntop uses. Otherwise we risk the > > test > > you propose could work and yet the wrapper would not be present and ntop > > would fail to compile... > > > > What it looks like is that you have a semaphore/thread library under > HPUX > > that isn't Posix nor any of the standard locations. So what makes you > > think > > it will work like the Posix threads that ntop is expecting?? > > > > If it is Posix compatible, then the proposed change to configure.in > would > > be > > this: > > > > dnl> > > dnl> Thread Library (optional) > > dnl> check for `pthread_create' in -lpthread > > dnl> > > dnl> Do not perform any check for the thread library, > > dnl> if the user decided to compile without multithread support. > > dnl> > > if test ".$ac_disable_multithread" = ".no"; then > > AC_CHECK_LIB(pthread, pthread_create) > > if test ".`echo $LIBS | grep pthread`" != .; then > > dnl> > > dnl> check for `sem_init' in -lposix4 > > dnl> > > AC_CHECK_LIB(posix4, sem_init) > > AC_DEFINE(MULTITHREADED) > > AC_DEFINE(ASYNC_ADDRESS_RESOLUTION) > > else > > AC_CHECK_LIB(pthreads, pthread_create) > > if test ".`echo $LIBS | grep pthread`" != .; then > > dnl> > > dnl> check for `sem_init' in -lposix4 > > dnl> > > AC_CHECK_LIB(posix4, sem_init) > > AC_DEFINE(MULTITHREADED) > > AC_DEFINE(ASYNC_ADDRESS_RESOLUTION) > > else > > AC_CHECK_LIB(c_r, pthread_create) > > if test ".`echo $LIBS | grep c_r`" != .; then > > AC_DEFINE(MULTITHREADED) > > AC_DEFINE(ASYNC_ADDRESS_RESOLUTION) > > else > > case "$host" in > > *-hpux*) > > dnl> Test for sem_init in a library that's unique to > HPUX... > > AC_CHECK_LIB(rt, sem_init) > > if test ".`echo $LIBS | grep rt `" != .; then > > AC_DEFINE(MULTITHREADED) > > AC_DEFINE(ASYNC_ADDRESS_RESOLUTION) > > else > > multithread_mode=no > > fi > > ;; > > *) > > multithread_mode=no > > ;; > > esac > > fi > > fi > > fi > > else > > AC_MSG_WARN([Multithread support disabled via command line option!]) > > multithread_mode=no > > fi > > > > > > > > > > > > > > > > > > -----Burton > > > > -----Original Message----- > > From: Falk Hayn [mailto:[EMAIL PROTECTED]] > > Sent: Friday, September 27, 2002 10:27 AM > > To: Burton M. Strauss III > > Cc: [EMAIL PROTECTED] > > Subject: RE: [Ntop-dev] no multithread on HP-UX 11.00 - resolved - > > codechange request > > > > > > Burton, > > > > my english is not good enough to understand Your comment. > > > > These are additional lines to detect the libs for > > multithreading on the HP-UX 11.00 platform > > > > # check __pthread_create_system for > > # HP-UX instead off pthread_create > > # pthread_create is a wrapper to __pthread_create_system > > # pls see /usr/lib/sys/pthread.h on any HP-UX 11.00 system > > AC_CHECK_LIB(pthread, __pthread_create_system) > > if test ".`echo $LIBS | grep pthread `" != .; then > > AC_CHECK_LIB(rt, sem_init) > > if test ".`echo $LIBS | grep rt `" != .; then > > AC_DEFINE(MULTITHREADED) > > AC_DEFINE(ASYNC_ADDRESS_RESOLUTION) > > else > > multithread_mode=no > > fi > > else > > multithread_mode=no > > fi > > > > > > > > > > > > > So this would all have to be HPUX only, right? > > > > > > -----Burton > > > > > > -----Original Message----- > > > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On > Behalf > > > Of Falk Hayn > > > Sent: Tuesday, September 24, 2002 9:55 AM > > > To: [EMAIL PROTECTED] > > > Subject: [Ntop-dev] no multithread on HP-UX 11.00 - resolved - > > > codechange request > > > > > > > > > Dear Burton, dear Ladies and Sirs, > > > > > > to have multithread support enabled for HP-UX 11.00, it is necessary > to > > > include some checkings for additional libs into configure.in near line > > 810 > > > > > > This change was tested with daily snapshot as ofSeptember 23th. > > > (I removed comments to better understand the if the hirarchy) > > > > > > Pls. notify me, after the change has been included into daily > snapshots. > > > > > > configure.in near line 810 > > > dnl> Do not perform any check for the thread library, > > > dnl> if the user decided to compile without multithread support. > > > dnl> > > > if test ".$ac_disable_multithread" = ".no"; then > > > AC_CHECK_LIB(pthread, pthread_create) > > > if test ".`echo $LIBS | grep pthread`" != .; then > > > AC_CHECK_LIB(posix4, sem_init) > > > AC_DEFINE(MULTITHREADED) > > > AC_DEFINE(ASYNC_ADDRESS_RESOLUTION) > > > else > > > AC_CHECK_LIB(pthreads, pthread_create) > > > if test ".`echo $LIBS | grep pthread`" != .; then > > > AC_CHECK_LIB(posix4, sem_init) > > > AC_DEFINE(MULTITHREADED) > > > AC_DEFINE(ASYNC_ADDRESS_RESOLUTION) > > > else > > > AC_CHECK_LIB(c_r, pthread_create) > > > if test ".`echo $LIBS | grep c_r`" != .; then > > > AC_DEFINE(MULTITHREADED) > > > AC_DEFINE(ASYNC_ADDRESS_RESOLUTION) > > > else > > > # check __pthread_create_system for > > > # HP-UX instead off pthread_create > > > # pthread_create is a wrapper to __pthread_create_system > > > # pls see /usr/lib/sys/pthread.h on any HP-UX 11.00 system > > > AC_CHECK_LIB(pthread, __pthread_create_system) > > > if test ".`echo $LIBS | grep pthread `" != .; then > > > AC_CHECK_LIB(rt, sem_init) > > > if test ".`echo $LIBS | grep rt `" != .; then > > > AC_DEFINE(MULTITHREADED) > > > AC_DEFINE(ASYNC_ADDRESS_RESOLUTION) > > > else > > > multithread_mode=no > > > fi > > > else > > > multithread_mode=no > > > fi > > > fi > > > fi > > > fi > > > else > > > AC_MSG_WARN([Multithread support disabled via command line option!]) > > > multithread_mode=no > > > fi > > > > > > -- > > > Werden Sie mit uns zum "OnlineStar 2002"! Jetzt GMX w�hlen - > > > und tolle Preise absahnen! http://www.onlinestar.de > > > > > > _______________________________________________ > > > Ntop-dev mailing list > > > [EMAIL PROTECTED] > > > http://lists.ntop.org/mailman/listinfo/ntop-dev > > > > > > > -- > > Werden Sie mit uns zum "OnlineStar 2002"! Jetzt GMX w�hlen - > > und tolle Preise absahnen! http://www.onlinestar.de > > > > > > _______________________________________________ > > Ntop-dev mailing list > > [EMAIL PROTECTED] > > http://lists.ntop.org/mailman/listinfo/ntop-dev > > > > -- > Werden Sie mit uns zum "OnlineStar 2002"! Jetzt GMX w�hlen - > und tolle Preise absahnen! http://www.onlinestar.de > > _______________________________________________ > Ntop-dev mailing list > [EMAIL PROTECTED] > http://lists.ntop.org/mailman/listinfo/ntop-dev > > _______________________________________________ > Ntop-dev mailing list > [EMAIL PROTECTED] > http://lists.ntop.org/mailman/listinfo/ntop-dev > -- Werden Sie mit uns zum "OnlineStar 2002"! Jetzt GMX w�hlen - und tolle Preise absahnen! http://www.onlinestar.de _______________________________________________ Ntop-dev mailing list [EMAIL PROTECTED] http://lists.ntop.org/mailman/listinfo/ntop-dev
