|
»
»
Re: libjava broken on the trunk on i386-darwin8.11
|
View:
New views
11 Messages
—
Rating Filter:
Alert me
|
 |
|
|
|
Re: libjava broken on the trunk on i386-darwin8.11
by Ralf Wildenhues
::
Rate this Message:
Hi Dave,
* Dave Korn wrote on Wed, Sep 30, 2009 at 03:47:51AM CEST: > --- configure.host (revision 152310) > +++ configure.host (working copy) > @@ -347,6 +349,7 @@ case "${host}" in > BACKTRACESPEC= > # Win32 DLLs are limited to 64k exported symbols each. > enable_libgcj_sublibs_default=yes > + libgcj_sublib_ltflags=-no-undefined -bindir $(bindir) This is shell script, so you need to quote spaces on the right hand side of the assignment, otherwise this is a temporary variable setting for the duration of invoking a "-bindir" command. More generally though, why don't you simply pass these flags unconditionally on all systems, not just w32 ones? Cheers, Ralf |
|
|
Re: libjava broken on the trunk on i386-darwin8.11
by Andrew Pinski-2
::
Rate this Message:
On Tue, Sep 29, 2009 at 10:34 PM, Ralf Wildenhues
<Ralf.Wildenhues@...> wrote: > Hi Dave, > > * Dave Korn wrote on Wed, Sep 30, 2009 at 03:47:51AM CEST: >> --- configure.host (revision 152310) >> +++ configure.host (working copy) > >> @@ -347,6 +349,7 @@ case "${host}" in >> BACKTRACESPEC= >> # Win32 DLLs are limited to 64k exported symbols each. >> enable_libgcj_sublibs_default=yes >> + libgcj_sublib_ltflags=-no-undefined -bindir $(bindir) > > This is shell script, so you need to quote spaces on the right hand side > of the assignment, otherwise this is a temporary variable setting for > the duration of invoking a "-bindir" command. > > More generally though, why don't you simply pass these flags > unconditionally on all systems, not just w32 ones? > They were passed unconditionally and that broke darwin. Which is the reason for the change :). Thanks, Andrew Pinski |
|
|
Re: libjava broken on the trunk on i386-darwin8.11
by Ralf Wildenhues
::
Rate this Message:
* Andrew Pinski wrote on Wed, Sep 30, 2009 at 07:50:24AM CEST:
> On Tue, Sep 29, 2009 at 10:34 PM, Ralf Wildenhues wrote: > > > > More generally though, why don't you simply pass these flags > > unconditionally on all systems, not just w32 ones? > > They were passed unconditionally and that broke darwin. Which is the > reason for the change :). Oh, sorry about that. In that case however, why doesn't libjava on Darwin have a link dependency on libiconv (or whatever library that defines these symbols)? Seems like the better fix to me. (Did we already go through this? I can't remember.) Thanks, Ralf |
|
|
Re: libjava broken on the trunk on i386-darwin8.11
by Andrew Pinski-2
::
Rate this Message:
On Tue, Sep 29, 2009 at 10:58 PM, Ralf Wildenhues
<Ralf.Wildenhues@...> wrote: > * Andrew Pinski wrote on Wed, Sep 30, 2009 at 07:50:24AM CEST: >> On Tue, Sep 29, 2009 at 10:34 PM, Ralf Wildenhues wrote: >> > >> > More generally though, why don't you simply pass these flags >> > unconditionally on all systems, not just w32 ones? >> >> They were passed unconditionally and that broke darwin. Which is the >> reason for the change :). > > Oh, sorry about that. In that case however, why doesn't libjava on > Darwin have a link dependency on libiconv (or whatever library that > defines these symbols)? Seems like the better fix to me. > (Did we already go through this? I can't remember.) it is not just libiconv but _environ also which is only supplied by the dynamic linker and there is no library which contains that symbol. Thanks, Andrew Pinski |
|
|
Re: libjava broken on the trunk on i386-darwin8.11
by Dave Korn-6
::
Rate this Message:
Ralf Wildenhues wrote:
> Hi Dave, > > * Dave Korn wrote on Wed, Sep 30, 2009 at 03:47:51AM CEST: >> --- configure.host (revision 152310) >> +++ configure.host (working copy) > >> @@ -347,6 +349,7 @@ case "${host}" in >> BACKTRACESPEC= >> # Win32 DLLs are limited to 64k exported symbols each. >> enable_libgcj_sublibs_default=yes >> + libgcj_sublib_ltflags=-no-undefined -bindir $(bindir) > > This is shell script, so you need to quote spaces on the right hand side > of the assignment, otherwise this is a temporary variable setting for > the duration of invoking a "-bindir" command. That's just one of the many hazards of writing a patch at 4 am and hoping it'll be right so that you'll have a successful build available when you wake up! Sigh... > checking for fastjar... no > checking for gjar... no > checking for zip... /usr/bin/zip > checking for unzip... /usr/bin/unzip > checking whether to enable maintainer-specific portions of Makefiles... no > /gnu/gcc/gcc-unpatched/libjava/configure.host: line 352: bindir: command not found > /gnu/gcc/gcc-unpatched/libjava/configure.host: line 352: -bindir: command not found > checking if the linker (/gnu/gcc/obj-java-pinski/./gcc/collect-ld) is GNU ld... ... also looks like I need to escape that dollar sign, rather than trying to expand a makefile variable in a bash script... I'll have the respin up shortly. cheers, DaveK |
|
|
[PATCH] Re: libjava broken on the trunk on i386-darwin8.11
by Dave Korn-6
::
Rate this Message:
Dave Korn wrote:
> I'm now bootstrapping this. Andrew, would you like to check if it fixes > your problem on Darwin? As we see upthread, the first spin lacked some quoting. Here's the fixed version, tested by blowing away the $target/libjava directory and rebuilding with "make configure-target-libjava all-target-libjava install-target-libjava" (using DESTDIR for the install stage). Build completed successfully and verified manually that the correct definition of LIBGCJ_SUBLIB_LTFLAGS made it into the generated $target/libjava/Makefile and the correct files got installed into the correct locations in the DESTDIR, for target=i686-pc-cygwin. Same ChangeLog as before: > libjava/ChangeLog: > > * configure.host (libgcj_sublib_ltflags): New variable. > * configure.ac (LIBGCJ_SUBLIB_LTFLAGS): AC_SUBST variable for > the above in makefiles. > * configure: Regenerate. > > * Makefile.am (libgcj_la_LDFLAGS_NOUNDEF): Rename from this ... > (LIBJAVA_LDFLAGS_NOUNDEF): ... to this, initialise from > LIBGCJ_SUBLIB_LTFLAGS (only when building sublibs), and ... > (libgij_la_LDFLAGS, libgcj_tools_la_LDFLAGS, libjvm_la_LDFLAGS, > lib_gnu_awt_xlib_la_LDFLAGS, libgcj_bc_la_LDFLAGS): ... use it in > all these places to replace hard-coded -no-undefined and -bindir > flags, as well as ... > (libgcj_la_LDFLAGS): ... updating existing reference here. > * Makefile.in: Regenerate. > * include/Makefile.in: Likewise. > * testsuite/Makefile.in: Likewise. > * gcj/Makefile.in: Likewise. cheers, DaveK Index: libjava/configure.host =================================================================== --- libjava/configure.host (revision 152317) +++ libjava/configure.host (working copy) @@ -18,6 +18,7 @@ # libgcj_cflags Special CFLAGS to use when building # libgcj_cxxflags Special CXXFLAGS to use when building # libgcj_javaflags Special JAVAFLAGS to use when building +# libgcj_sublib_ltflags Special Libtool flags to use when building. # libgcj_interpreter If the bytecode interpreter supports this platform. # enable_java_net_default If java.net native code should be enabled by # default. @@ -41,6 +42,7 @@ libgcj_flags= libgcj_cflags= libgcj_cxxflags= libgcj_javaflags= +libgcj_sublib_ltflags= libgcj_interpreter= enable_java_net_default=yes enable_hash_synchronization_default=no @@ -347,6 +349,7 @@ case "${host}" in BACKTRACESPEC= # Win32 DLLs are limited to 64k exported symbols each. enable_libgcj_sublibs_default=yes + libgcj_sublib_ltflags='-no-undefined -bindir $(bindir)' ;; esac Index: libjava/configure.ac =================================================================== --- libjava/configure.ac (revision 152317) +++ libjava/configure.ac (working copy) @@ -206,11 +206,14 @@ AC_EXEEXT # libgcj_cflags - host specific C compiler flags # libgcj_cxxflags - host specific C++ compiler flags # libgcj_javaflags - host specific Java compiler flags +# libgcj_sublib_ltflags - host specific Libtool flags +# (only used when building sublibs) # and a number of others; see the list at the start of the file. libgcj_cflags= libgcj_cxxflags= libgcj_javaflags= +libgcj_sublib_ltflags= . ${srcdir}/configure.host @@ -223,11 +226,13 @@ fi LIBGCJ_CFLAGS="${libgcj_cflags}" LIBGCJ_CXXFLAGS="${libgcj_cxxflags}" LIBGCJ_JAVAFLAGS="${libgcj_javaflags}" +LIBGCJ_SUBLIB_LTFLAGS="${libgcj_sublib_ltflags}" LIBGCJ_LD_SYMBOLIC="${libgcj_ld_symbolic}" LIBGCJ_LD_SYMBOLIC_FUNCTIONS="${libgcj_ld_symbolic_functions}" AC_SUBST(LIBGCJ_CFLAGS) AC_SUBST(LIBGCJ_CXXFLAGS) AC_SUBST(LIBGCJ_JAVAFLAGS) +AC_SUBST(LIBGCJ_SUBLIB_LTFLAGS) AC_SUBST(LIBGCJ_LD_SYMBOLIC) AC_SUBST(LIBGCJ_LD_SYMBOLIC_FUNCTIONS) Index: libjava/Makefile.am =================================================================== --- libjava/Makefile.am (revision 152317) +++ libjava/Makefile.am (working copy) @@ -201,12 +201,12 @@ if BUILD_SUBLIBS toolexeclib_LTLIBRARIES += libgcj-noncore.la libgcj_noncore_la_LIBADD_SUBOBJECTS = $(NONCORE_PACKAGE_SOURCE_FILES_LO) libgcj_la_LIBADD_SUBOBJECTS = $(CORE_PACKAGE_SOURCE_FILES_LO) -libgcj_la_LDFLAGS_NOUNDEF = -no-undefined -bindir $(bindir) +LIBJAVA_LDFLAGS_NOUNDEF = $(LIBGCJ_SUBLIB_LTFLAGS) else # If not building sublibraries, everything goes in libgcj, # and it cannot be usefully built shared on PE platforms. libgcj_la_LIBADD_SUBOBJECTS = $(ALL_PACKAGE_SOURCE_FILES_LO) -libgcj_la_LDFLAGS_NOUNDEF = +LIBJAVA_LDFLAGS_NOUNDEF = endif if USE_LIBGCJ_BC @@ -388,7 +388,7 @@ endif ## See jv_convert_LDADD. libgij_la_LIBADD = -L$(here)/.libs libgcj.la ## The mysterious backslash in the grep pattern is consumed by make. -libgij_la_LDFLAGS = -rpath $(toolexeclibdir) -no-undefined -bindir $(bindir) \ +libgij_la_LDFLAGS = -rpath $(toolexeclibdir) $(LIBJAVA_LDFLAGS_NOUNDEF) \ -version-info `grep -v '^\#' $(srcdir)/libtool-version` $(LIBGCJ_LD_SYMBOLIC) libgij_la_LINK = $(CXXLINK) $(libgij_la_LDFLAGS) @@ -445,7 +445,7 @@ xlib_nat_files = $(xlib_nat_source_files:.cc=.lo) # certain linuxthread functions get linked: ## The mysterious backslash in the grep pattern is consumed by make. libgcj_la_LDFLAGS = -rpath $(toolexeclibdir) $(THREADLDFLAGS) $(extra_ldflags) $(THREADLIBS) \ - $(LIBLTDL) $(SYS_ZLIBS) $(libgcj_la_LDFLAGS_NOUNDEF) \ + $(LIBLTDL) $(SYS_ZLIBS) $(LIBJAVA_LDFLAGS_NOUNDEF) \ -version-info `grep -v '^\#' $(srcdir)/libtool-version` \ $(LIBGCJ_LD_SYMBOLIC_FUNCTIONS) libgcj_la_LIBADD = \ @@ -495,7 +495,7 @@ libgcj_tools_la_GCJFLAGS = $(AM_GCJFLAGS) -findire -fsource-filename=$(here)/classpath/tools/all-classes.lst libgcj_tools_la_LDFLAGS = -rpath $(toolexeclibdir) \ -version-info `grep -v '^\#' $(srcdir)/libtool-version` \ - $(LIBGCJ_LD_SYMBOLIC_FUNCTIONS) -no-undefined -bindir $(bindir) + $(LIBGCJ_LD_SYMBOLIC_FUNCTIONS) $(LIBJAVA_LDFLAGS_NOUNDEF) libgcj_tools_la_LIBADD = libgcj.la -lm libgcj_tools_la_DEPENDENCIES = libgcj.la libgcj.spec if BUILD_SUBLIBS @@ -512,7 +512,7 @@ endif ## See jv_convert_LDADD. libjvm_la_LIBADD = -L$(here)/.libs libgcj.la libjvm_la_LDFLAGS = -avoid-version $(LIBGCJ_LD_SYMBOLIC) \ - -no-undefined -bindir $(bindir) + $(LIBJAVA_LDFLAGS_NOUNDEF) libjvm_la_LINK = $(CXXLINK) $(libjvm_la_LDFLAGS) ## The .db file. This rule is only used for native builds, so it is @@ -542,7 +542,7 @@ lib_gnu_awt_xlib_la_CPPFLAGS = \ ## The mysterious backslash in the grep pattern is consumed by make. lib_gnu_awt_xlib_la_LDFLAGS = ../libstdc++-v3/src/libstdc++.la \ @X_PRE_LIBS@ @X_LIBS@ -lX11 @X_EXTRA_LIBS@ \ - -rpath $(toolexeclibdir) -no-undefined -bindir $(bindir) \ + -rpath $(toolexeclibdir) $(LIBJAVA_LDFLAGS_NOUNDEF) \ -version-info `grep -v '^\#' $(srcdir)/libtool-version` $(LIBGCJ_LD_SYMBOLIC) lib_gnu_awt_xlib_la_LINK = $(LIBLINK) $(lib_gnu_awt_xlib_la_LDFLAGS) @@ -552,7 +552,7 @@ lib_gnu_awt_xlib_la_LINK = $(LIBLINK) $(lib_gnu_aw ## This library is not linked against libgcj. libgcj_bc_la_SOURCES = libgcj_bc.c libgcj_bc_la_LDFLAGS = -rpath $(toolexeclibdir) -no-static -version-info 1:0:0 \ - $(LIBGCJ_LD_SYMBOLIC_FUNCTIONS) -no-undefined -bindir $(bindir) + $(LIBGCJ_LD_SYMBOLIC_FUNCTIONS) $(LIBJAVA_LDFLAGS_NOUNDEF) libgcj_bc_la_DEPENDENCIES = libgcj.la libgcj_bc_la_LINK = $(LIBLINK) $(libgcj_bc_la_LDFLAGS) ## This is specific to Linux/{Free,Net,Open}BSD/Hurd and perhaps few others. |
|
|
Re: [PATCH] Re: libjava broken on the trunk on i386-darwin8.11
by Andrew Haley
::
Rate this Message:
Dave Korn wrote:
> Dave Korn wrote: > >> I'm now bootstrapping this. Andrew, would you like to check if it fixes >> your problem on Darwin? > > As we see upthread, the first spin lacked some quoting. Here's the fixed > version, tested by blowing away the $target/libjava directory and rebuilding > with "make configure-target-libjava all-target-libjava install-target-libjava" > (using DESTDIR for the install stage). Build completed successfully and > verified manually that the correct definition of LIBGCJ_SUBLIB_LTFLAGS made it > into the generated $target/libjava/Makefile and the correct files got > installed into the correct locations in the DESTDIR, for > target=i686-pc-cygwin. Same ChangeLog as before: > >> libjava/ChangeLog: >> >> * configure.host (libgcj_sublib_ltflags): New variable. >> * configure.ac (LIBGCJ_SUBLIB_LTFLAGS): AC_SUBST variable for >> the above in makefiles. >> * configure: Regenerate. >> >> * Makefile.am (libgcj_la_LDFLAGS_NOUNDEF): Rename from this ... >> (LIBJAVA_LDFLAGS_NOUNDEF): ... to this, initialise from >> LIBGCJ_SUBLIB_LTFLAGS (only when building sublibs), and ... >> (libgij_la_LDFLAGS, libgcj_tools_la_LDFLAGS, libjvm_la_LDFLAGS, >> lib_gnu_awt_xlib_la_LDFLAGS, libgcj_bc_la_LDFLAGS): ... use it in >> all these places to replace hard-coded -no-undefined and -bindir >> flags, as well as ... >> (libgcj_la_LDFLAGS): ... updating existing reference here. >> * Makefile.in: Regenerate. >> * include/Makefile.in: Likewise. >> * testsuite/Makefile.in: Likewise. >> * gcj/Makefile.in: Likewise. > > Tests running now, assuming no regressions OK for head? Sure. Andrew Pinski, is this OK? Andrew. |
|
|
Re: [PATCH] Re: libjava broken on the trunk on i386-darwin8.11
by Dave Korn-6
::
Rate this Message:
Andrew Haley wrote:
> Dave Korn wrote: >> Tests running now, assuming no regressions OK for head? > > Sure. Andrew Pinski, is this OK? Tests passed with usual results: http://gcc.gnu.org/ml/gcc-testresults/2009-09/msg02784.html Just waiting to hear from AP that it fixes his problem before I commit. cheers, DaveK |
|
|
Re: [PATCH] Re: libjava broken on the trunk on i386-darwin8.11
by Andrew Pinski-2
::
Rate this Message:
On Wed, Sep 30, 2009 at 6:56 AM, Andrew Haley <aph@...> wrote:
> Sure. Andrew Pinski, is this OK? Sorry I did not get back until now, I had to update my autoconf and automake. YES this works on i386-darwin8.11. Thanks, Andrew Pinski |
|
|
Re: [PATCH] Re: libjava broken on the trunk on i386-darwin8.11
by Dave Korn-6
::
Rate this Message:
Andrew Pinski wrote:
> On Wed, Sep 30, 2009 at 6:56 AM, Andrew Haley <aph@...> wrote: >> Sure. Andrew Pinski, is this OK? > > Sorry I did not get back until now, I had to update my autoconf and automake. > > YES this works on i386-darwin8.11. Thanks for verifying! Committed at r.152347. cheers, DaveK |
| Free embeddable forum powered by Nabble | Forum Help |
