[Buildroot] [PATCH v2] valgrind: fix musl compile

[Peter Seiderer <ps.report@gmx.net>]

Hello Arnout,

(one more e-mail, hit the send button too early)

On Tue, 9 Feb 2016 23:16:30 +0100, Peter Seiderer <ps.report@gmx.net> wrote:

> Hello Arnout,
> 
> On Tue, 9 Feb 2016 22:43:46 +0100, Arnout Vandecappelle <arnout@mind.be> wrote:
> 
> > On 09-02-16 21:53, Peter Seiderer wrote:
> > > Add musl libc detection (based on suggestions by Romain Naour and
> > > Arnout Vandecappelle) and add some tweaks for musl support (based
> > > on OpenWRT patch [1]).
> > > 
> > > Fixes ([2]):
> > >   checking the glibc version... unsupported version
> > >   configure: error: Valgrind requires glibc version 2.2 or later
> > > 
> > > Patch suggested upstream (see [3]).
> > > 
> > > [1] https://dev.openwrt.org/browser/trunk/package/devel/valgrind/patches/200-musl_fix.patch?rev=46302
> > > [2] http://autobuild.buildroot.org/results/7b0/7b048ba58918f0a08498c61327fcf35a85a84837
> > > [3] https://bugs.kde.org/show_bug.cgi?id=359202
> > > 
> > > Signed-off-by: Peter Seiderer <ps.report@gmx.net>
> > > ---
> > > Changes v1 -> v2:
> > >   - add proper musl libc detection (suggested by Romain Naour and
> > >     Arnout Vandecappelle)
> > [snip]
> > > ++     musl)
> > > ++	AC_MSG_RESULT(Musl)
> > > ++	AC_DEFINE([MUSL_LIBC], 1, [Define to 1 if you're using Musl libc])
> > > ++	# no DEFAULT_SUPP file yet for musl libc.
> > > ++	;;
> > > +      2.0|2.1|*)
> > > + 	AC_MSG_RESULT([unsupported version ${GLIBC_VERSION}])
> > > + 	AC_MSG_ERROR([Valgrind requires glibc version 2.2 or later,])
> > > +-	AC_MSG_ERROR([Darwin libc, Bionic libc or Solaris libc])
> > > ++	AC_MSG_ERROR([musl libc, Darwin libc, Bionic libc or Solaris libc])
> > 
> >  Actuallly, uClibc was already supported as well :-)
> 
> O.k, will add uclibc as well...
> 
> > 
> > > + 	;;
> > > + esac
> > > + 
> > > +diff --git a/coregrind/vg_preloaded.c b/coregrind/vg_preloaded.c
> > > +index 2ea7a7a..e49c832 100644
> > > +--- a/coregrind/vg_preloaded.c
> > > ++++ b/coregrind/vg_preloaded.c
> > > +@@ -56,7 +56,7 @@
> > > + void VG_NOTIFY_ON_LOAD(freeres)( void );
> > > + void VG_NOTIFY_ON_LOAD(freeres)( void )
> > > + {
> > > +-#  if !defined(__UCLIBC__) \
> > > ++#  if !defined(__UCLIBC__) && defined(__GLIBC__) \
> > 
> >  Adding defined(__GLIBC__) effectively excludes android (bionic), doesn't it?

Not sure about this one, I think it is o.k. for the buildroot case:
- old behavior was __UCLIBC__ or __GLIBC__ is defined (!__UCLIBC__
  means __GLIBC__ is defined)
- new behavior is __UCLIBC__ or __GLIBC__ or MUSL_LIBC is defined,
  so from !__UCLIBC__ follows __GLIBC__ or MUSL_LIBC, so the need
  to check explicit for __GLIBC__ ...

Do I miss something? Not sure about the right solution for upstream...

Thanks for review...

Regards,
Peter

> > 
> > > +       && !defined(VGPV_arm_linux_android) \
> > > +       && !defined(VGPV_x86_linux_android) \
> > > +       && !defined(VGPV_mips32_linux_android) \
> > > +diff --git a/include/pub_tool_redir.h b/include/pub_tool_redir.h
> > > +index bac00d7..b8cd97e 100644
> > > +--- a/include/pub_tool_redir.h
> > > ++++ b/include/pub_tool_redir.h
> > > +@@ -242,8 +242,11 @@
> > > + /* --- Soname of the standard C library. --- */
> > > + 
> > > + #if defined(VGO_linux) || defined(VGO_solaris)
> > > ++# if defined(__GLIBC__) || defined(__UCLIBC__)
> > 
> >  AFAICS, on solaris neither of these is defined, so that's probably not good
> > either. I guess it's better to revert the condition and check on musl instead.
> 
> O.k, will do...
> 
> > 
> > > + #  define  VG_Z_LIBC_SONAME  libcZdsoZa              // libc.so*
> > > +-
> > > ++#else
> > > ++#  define  VG_Z_LIBC_SONAME  libcZdZa              // libc.*
> > > ++#endif
> > > + #elif defined(VGO_darwin) && (DARWIN_VERS <= DARWIN_10_6)
> > > + #  define  VG_Z_LIBC_SONAME  libSystemZdZaZddylib    // libSystem.*.dylib
> > > + 
> > > +@@ -274,7 +277,11 @@
> > > + /* --- Soname of the pthreads library. --- */
> > > + 
> > > + #if defined(VGO_linux)
> > > ++# if defined(__GLIBC__) || defined(__UCLIBC__)
> > 
> >  Same here.
> > 
> 
> O.k, will do...
> > 
> > 
> >  Regards,
> >  Arnout
> > 
> > 
> > > + #  define  VG_Z_LIBPTHREAD_SONAME  libpthreadZdsoZd0     // libpthread.so.0
> > > ++#else
> > > ++#  define  VG_Z_LIBPTHREAD_SONAME  libcZdZa              // libc.*
> > > ++#endif
> > > + #elif defined(VGO_darwin)
> > > + #  define  VG_Z_LIBPTHREAD_SONAME  libSystemZdZaZddylib  // libSystem.*.dylib
> > > + #elif defined(VGO_solaris)
> > > +-- 
> > > +2.1.4
> > > +
> > > diff --git a/package/valgrind/valgrind.mk b/package/valgrind/valgrind.mk
> > > index 5849b49..46ba13e 100644
> > > --- a/package/valgrind/valgrind.mk
> > > +++ b/package/valgrind/valgrind.mk
> > > @@ -12,6 +12,9 @@ VALGRIND_LICENSE_FILES = COPYING COPYING.DOCS
> > >  VALGRIND_CONF_OPTS = --disable-ubsan
> > >  VALGRIND_INSTALL_STAGING = YES
> > >  
> > > +# patch 0004-Fixes-for-musl-libc.patch touching configure.ac
> > > +VALGRIND_AUTORECONF = YES
> > > +
> > >  ifeq ($(BR2_GCC_ENABLE_TLS),y)
> > >  VALGRIND_CONF_OPTS += --enable-tls
> > >  else
> > > 
> > 
> > 
> 
> _______________________________________________
> buildroot mailing list
> buildroot at busybox.net
> http://lists.busybox.net/mailman/listinfo/buildroot