Hi Andreas, On 08/08/2014 02:20 PM, Andreas Beckmann wrote: > On Thu, 12 Jun 2014 17:30:36 +0200 "Helge Deller" <del...@gmx.de> wrote: >> The attached trivial patch fixes the build on the Debian hppa/linux arch. >> Please apply for the next version. > > I'll add your patch in the 5.3.2 upload that I'm currently preparing.
Sadly the patch was wrong as I wrote already... In configure.in it should read "parisc*", not "hppa*" so the build failed again. > Please report this problem + patch in the upstream bugtracker, too. > http://icl.cs.utk.edu/trac/papi/ As suggested I just opened a ticket upstream in trac: http://icl.cs.utk.edu/trac/papi/ticket/204 I've attached the new (fixed) hppa patch again to this mail. Maybe you can replace the existing one with this one for the next upload? Thanks, Helge
Author: Helge Deller <del...@gmx.de> Description: fix build on hppa/linux Bug-Debian: https://bugs.debian.org/751411 diff -up ./src/configure.in.org ./src/configure.in --- ./src/configure.in.org 2014-06-12 16:36:43.136183226 +0200 +++ ./src/configure.in 2014-06-12 16:36:55.312184354 +0200 @@ -384,7 +384,7 @@ AC_DEFINE_UNQUOTED(CPU,$CPU,[cpu type]) # First set pthread-mutexes based on arch case $arch in - aarch64|arm*) + aarch64|arm*|parisc*) pthread_mutexes=yes CFLAGS="$CFLAGS -DUSE_PTHREAD_MUTEXES" echo "forcing use of pthread mutexes... " >&6 diff -up ./src/linux-context.h.org ./src/linux-context.h --- ./src/linux-context.h.org 2014-06-12 16:38:34.792193369 +0200 +++ ./src/linux-context.h 2014-06-12 16:43:47.632219446 +0200 @@ -35,6 +35,8 @@ typedef ucontext_t hwd_ucontext_t; #define OVERFLOW_ADDRESS(ctx) ctx.ucontext->uc_mcontext.pc #elif defined(__mips__) #define OVERFLOW_ADDRESS(ctx) ctx.ucontext->uc_mcontext.pc +#elif defined(__hppa__) +#define OVERFLOW_ADDRESS(ctx) ctx.ucontext->uc_mcontext.sc_iaoq[0] #else #error "OVERFLOW_ADDRESS() undefined!" #endif diff -up ./src/linux-timer.c.org ./src/linux-timer.c --- ./src/linux-timer.c.org 2014-06-12 16:49:06.096242651 +0200 +++ ./src/linux-timer.c 2014-06-12 16:49:35.744244649 +0200 @@ -215,7 +215,7 @@ get_cycles( void ) /* POWER get_cycles() */ /************************/ -#elif (defined(__powerpc__) || defined(__arm__) || defined(__mips__) || defined(__aarch64__)) +#elif (defined(__powerpc__) || defined(__arm__) || defined(__mips__) || defined(__aarch64__) || defined(__hppa__)) /* * It's not possible to read the cycles from user space on ppc970. * There is a 64-bit time-base register (TBU|TBL), but its
