libffi 3.0.12 rc2 ready for testing

libffi 3.0.12 rc2 ready for testing

[Anthony Green]

Thanks for the feedback so far.  I've just uploaded the tarball of
release candidate 2 here:

 ftp://sourceware.org/pub/libffi/libffi-3.0.12-rc2.tar.gz

Again... feedback on the wiki welcome...
http://moxielogic.org/wiki/index.php/Libffi_3.0.12

I'm hoping to make this the most widely tested version of libffi yet.
Here's what changed from yesterday...


2013-02-07  Anthony Green <green@moxielogic.com>

        * configure.ac: Update bug reporting address.
        * configure.in: Rebuild.

        * src/sparc/v8.S (ffi_flush_icache): Out-of-line cache flusher for
        Sun compiler.
        * src/sparc/ffi.c (ffi_call): Remove warning.
        Call ffi_flush_icache for non-GCC builds.
        (ffi_prep_closure_loc): Use ffi_flush_icache.

        * Makefile.am (EXTRA_DIST): Add libtool-ldflags.
        * Makefile.in: Rebuilt.
        * libtool-ldflags: New file.

2013-02-07  Daniel Schepler <dschepler@gmail.com>

        * configure.ac: Correctly identify x32 systems as 64-bit.
        * m4/libtool.m4: Remove libtool expr error.
        * aclocal.m4, configure: Rebuilt.

2013-02-07  Anthony Green <green@moxielogic.com>

        * configure.ac: Fix GCC usage test.
        * configure: Rebuilt.
        * README: Mention LLVM/GCC x86_64 issue.
        * testsuite/Makefile.in: Rebuilt.

2013-02-07  Anthony Green <green@moxielogic.com>

        * testsuite/libffi.call/cls_double_va.c (main): Replace // style
        comments with /* */ for xlc compiler.
        * testsuite/libffi.call/stret_large.c (main): Ditto.
        * testsuite/libffi.call/stret_large2.c (main): Ditto.
        * testsuite/libffi.call/nested_struct1.c (main): Ditto.
        * testsuite/libffi.call/huge_struct.c (main): Ditto.
        * testsuite/libffi.call/float_va.c (main): Ditto.
        * testsuite/libffi.call/cls_struct_va1.c (main): Ditto.
        * testsuite/libffi.call/cls_pointer_stack.c (main): Ditto.
        * testsuite/libffi.call/cls_pointer.c (main): Ditto.
        * testsuite/libffi.call/cls_longdouble_va.c (main): Ditto.

2013-02-06  Anthony Green <green@moxielogic.com>

        * man/ffi_prep_cif.3: Clean up for debian lintian checker.

Re: libffi 3.0.12 rc2 ready for testing

[Matthias Klose]

Am 08.02.2013 00:10, schrieb Anthony Green:
> Thanks for the feedback so far.  I've just uploaded the tarball of
> release candidate 2 here:
> 
>  ftp://sourceware.org/pub/libffi/libffi-3.0.12-rc2.tar.gz

the machine isn't reachable by ftp since yesterday. could you place it somewhere
else too, and post the md5sum?

thanks, Matthias

RE: libffi 3.0.12 rc2 ready for testing

[James Greenhalgh]

> Thanks for the feedback so far.  I've just uploaded the tarball of
> release candidate 2 here:
> 
>  ftp://sourceware.org/pub/libffi/libffi-3.0.12-rc2.tar.gz
> 
> Again... feedback on the wiki welcome...
> http://moxielogic.org/wiki/index.php/Libffi_3.0.12

Hi Anthony,

It looks like the RC1 and RC2 tarballs haven't picked up
src/aarch64/sysv.S so the build is broken for the AArch64 target:

> libtool: link: `src/aarch64/sysv.lo' is not a valid libtool object

While I have the diff around, these are some other files which
I spotted missing from the RC tar balls which are in git:

build-ios.sh
stamp-h.in
testsuite/libffi.call/cls_struct_va1.c
testsuite/libffi.call/cls_uchar_va.c
testsuite/libffi.call/cls_uint_va.c
testsuite/libffi.call/cls_ulong_va.c
testsuite/libffi.call/cls_ushort_va.c
testsuite/libffi.call/nested_struct11.c
testsuite/libffi.call/uninitialized.c
testsuite/libffi.call/va_1.c
testsuite/libffi.call/va_struct1.c
testsuite/libffi.call/va_struct2.c
testsuite/libffi.call/va_struct3.c
.travis.yml

Thanks,
James Greenhalgh

Re: libffi 3.0.12 rc2 ready for testing

[Peter Bergner]

Hi Anthony,
================================================================================
Here's a patch to fix the libffi.call/many.c testsuite failures on
powerpc-linux due to an over eager assert which I first pointed out here:

    http://sourceware.org/ml/libffi-discuss/2012/msg00093.html

Having looked at the code again, I believe the following patch is
the correct fix for this problem.  With the following patch, we're
now down to zero unexpected testsuite failures on powerpc-linux.

Peter


2013-02-08  Peter Bergner  <bergner@vnet.ibm.com>

	* src/powerpc/ffi.c (ffi_prep_args_SYSV): Account for FP args pushed
	on the stack.

diff -urpN -X /home/bergner/cvs/dontdiff libffi-3.0.12-rc1.orig/src/powerpc/ffi.c libffi-3.0.12-rc1/src/powerpc/ffi.c
--- libffi-3.0.12-rc1.orig/src/powerpc/ffi.c	2013-02-06 16:36:41.000000000 -0600
+++ libffi-3.0.12-rc1/src/powerpc/ffi.c	2013-02-08 12:07:53.000000000 -0600
@@ -370,6 +370,12 @@ ffi_prep_args_SYSV (extended_cif *ecif,
 #ifndef __NO_FPRS__
   FFI_ASSERT (fpr_base.u
 	      <= stacktop.u - ASM_NEEDS_REGISTERS - NUM_GPR_ARG_REGISTERS);
+  /* The assert below is testing that the number of integer arguments agrees
+     with the number found in ffi_prep_cif_machdep().  However, intarg_count
+     is incremented whenever we place an FP arg on the stack, so account for
+     that before our assert test.  */
+  if (fparg_count > NUM_FPR_ARG_REGISTERS)
+    intarg_count -= fparg_count - NUM_FPR_ARG_REGISTERS;
 #endif
   FFI_ASSERT (flags & FLAG_4_GPR_ARGUMENTS || intarg_count <= 4);
 }

Re: libffi 3.0.12 rc2 ready for testing

[Anthony Green]

On Fri, Feb 8, 2013 at 9:47 AM, James Greenhalgh
<james.greenhalgh@arm.com> wrote:
> It looks like the RC1 and RC2 tarballs haven't picked up
> src/aarch64/sysv.S so the build is broken for the AArch64 target:
>

Thanks James.  I've committed a fix that should make the next tarball work.

AG

Re: libffi 3.0.12 rc2 ready for testing

[Matthias Klose]

Am 08.02.2013 00:10, schrieb Anthony Green:
> Thanks for the feedback so far.  I've just uploaded the tarball of
> release candidate 2 here:
> 
>  ftp://sourceware.org/pub/libffi/libffi-3.0.12-rc2.tar.gz
> 
> Again... feedback on the wiki welcome...
> http://moxielogic.org/wiki/index.php/Libffi_3.0.12
> 
> I'm hoping to make this the most widely tested version of libffi yet.
> Here's what changed from yesterday...

one of the sparc related patches did break the build on sparc (32bit, using gcc
-m32 -mcpu=v9). Looks like some tests should be

 #if defined(SPARC64) || defined(__sparc_v9__)

instead of

 #ifdef SPARC64

Matthias


libtool: link: sparc-linux-gnu-gcc -shared  -fPIC -DPIC  src/.libs/prep_cif.o
src/.libs/types.o src/.libs/raw_api.o src/.libs/java_raw_api.o
src/.libs/closures.o src/sparc/.libs/ffi.o src/sparc/.libs/v8.o
src/sparc/.libs/v9.o    -O2 -Wl,-z -Wl,defs -Wl,-z -Wl,relro   -Wl,-soname
-Wl,libffi.so.6 -o .libs/libffi.so.6.0.1
src/sparc/.libs/ffi.o: In function `ffi_call':
/build/buildd-libffi_3.0.12~rc2-1-sparc-56QJre/libffi-3.0.12~rc2/build/../src/sparc/ffi.c:459:
undefined reference to `ffi_call_v8'
/build/buildd-libffi_3.0.12~rc2-1-sparc-56QJre/libffi-3.0.12~rc2/build/../src/sparc/ffi.c:447:
undefined reference to `ffi_call_v8'
/build/buildd-libffi_3.0.12~rc2-1-sparc-56QJre/libffi-3.0.12~rc2/build/../src/sparc/ffi.c:453:
undefined reference to `ffi_call_v8'
collect2: ld returned 1 exit status
make[3]: *** [libffi.la] Error 1

Re: libffi 3.0.12 rc2 ready for testing

[Andreas Tobler]

On 08.02.13 21:41, Matthias Klose wrote:
> Am 08.02.2013 00:10, schrieb Anthony Green:
>> Thanks for the feedback so far.  I've just uploaded the tarball of
>> release candidate 2 here:
>>
>>  ftp://sourceware.org/pub/libffi/libffi-3.0.12-rc2.tar.gz
>>
>> Again... feedback on the wiki welcome...
>> http://moxielogic.org/wiki/index.php/Libffi_3.0.12
>>
>> I'm hoping to make this the most widely tested version of libffi yet.
>> Here's what changed from yesterday...
> 
> one of the sparc related patches did break the build on sparc (32bit, using gcc
> -m32 -mcpu=v9). Looks like some tests should be
> 
>  #if defined(SPARC64) || defined(__sparc_v9__)
> 
> instead of
> 
>  #ifdef SPARC64
> 
> Matthias
> 
> 
> libtool: link: sparc-linux-gnu-gcc -shared  -fPIC -DPIC  src/.libs/prep_cif.o
> src/.libs/types.o src/.libs/raw_api.o src/.libs/java_raw_api.o
> src/.libs/closures.o src/sparc/.libs/ffi.o src/sparc/.libs/v8.o
> src/sparc/.libs/v9.o    -O2 -Wl,-z -Wl,defs -Wl,-z -Wl,relro   -Wl,-soname
> -Wl,libffi.so.6 -o .libs/libffi.so.6.0.1
> src/sparc/.libs/ffi.o: In function `ffi_call':
> /build/buildd-libffi_3.0.12~rc2-1-sparc-56QJre/libffi-3.0.12~rc2/build/../src/sparc/ffi.c:459:
> undefined reference to `ffi_call_v8'
> /build/buildd-libffi_3.0.12~rc2-1-sparc-56QJre/libffi-3.0.12~rc2/build/../src/sparc/ffi.c:447:
> undefined reference to `ffi_call_v8'
> /build/buildd-libffi_3.0.12~rc2-1-sparc-56QJre/libffi-3.0.12~rc2/build/../src/sparc/ffi.c:453:
> undefined reference to `ffi_call_v8'
> collect2: ld returned 1 exit status
> make[3]: *** [libffi.la] Error 1

This should be fixed in the git repo but it is missing in the rc2. I had
something similar here.
https://github.com/atgreen/libffi/commit/9db7e1a958fc484ba149efe027008b9a170395fb

Have a look for the v8.s, there an endif was on the wrong place.

Andreas