libffi changes for 3.0.11

libffi changes for 3.0.11

[Anthony Green]

I just released 3.0.10.  Thanks to everybody who contributed!

My plan is to release 3.0.11 within a couple of weeks.  It will contain 
the soname bump (since I've removed debug symbols, again), and I also 
want to get David Gilbert's variadic function support in there.

I am not able to personally test every platform that libffi purports to 
support, so please test and post your results here:
  http://moxielogic.org/wiki/index.php?title=Libffi_3.0.11

Thanks!

AG

Re: libffi changes for 3.0.11

[David Gilbert]

On 23 August 2011 17:37, Anthony Green <green@moxielogic.com> wrote:
> I just released 3.0.10.  Thanks to everybody who contributed!
>
> My plan is to release 3.0.11 within a couple of weeks.  It will contain the
> soname bump (since I've removed debug symbols, again), and I also want to
> get David Gilbert's variadic function support in there.
>
> I am not able to personally test every platform that libffi purports to
> support, so please test and post your results here:
>  http://moxielogic.org/wiki/index.php?title=Libffi_3.0.11

Hi Anthony,
  I've just added the i686 and arm (soft float) - seems OK; a few xpass on arm.

It'll be good to get the variadic stuff in!

Dave

Re: libffi changes for 3.0.11

[Bobby Holley]

I have another fix you might want to include, which I haven't had time
to formally submitting for upstream inclusion.

The OS detection within libffi doesn't handle darwin11 (OSX 10.7
Lion), making it crash mysteriously when doing anything with closures.

The fix is very safe - it just bumps the list of platforms that
allocate NOEXEC to include everything through darwin19.

The patch is attached to the bug here:
https://bugzilla.mozilla.org/show_bug.cgi?id=682180 - you probably
only want the one-line configure.ac change.

Cheers,
-Bobby Holley, Mozilla Platform Dev

On Fri, Sep 2, 2011 at 5:32 AM, David Gilbert <david.gilbert@linaro.org> wrote:
>
> On 23 August 2011 17:37, Anthony Green <green@moxielogic.com> wrote:
> > I just released 3.0.10.  Thanks to everybody who contributed!
> >
> > My plan is to release 3.0.11 within a couple of weeks.  It will contain the
> > soname bump (since I've removed debug symbols, again), and I also want to
> > get David Gilbert's variadic function support in there.
> >
> > I am not able to personally test every platform that libffi purports to
> > support, so please test and post your results here:
> >  http://moxielogic.org/wiki/index.php?title=Libffi_3.0.11
>
> Hi Anthony,
>  I've just added the i686 and arm (soft float) - seems OK; a few xpass on arm.
>
> It'll be good to get the variadic stuff in!
>
> Dave

Re: libffi changes for 3.0.11

[Bobby Holley]

[-- Attachment #1: Type: text/plain, Size: 1779 bytes --]

This turned out to be not quite right. The final patch that we applied
is attached.

You can also find the changeset here:
http://hg.mozilla.org/mozilla-central/rev/def68a106743

Cheers,
-Bobby Holley, Mozilla Platform Dev

On Fri, Sep 2, 2011 at 9:43 AM, Bobby Holley <bobbyholley@gmail.com> wrote:
>
> I have another fix you might want to include, which I haven't had time
> to formally submitting for upstream inclusion.
>
> The OS detection within libffi doesn't handle darwin11 (OSX 10.7
> Lion), making it crash mysteriously when doing anything with closures.
>
> The fix is very safe - it just bumps the list of platforms that
> allocate NOEXEC to include everything through darwin19.
>
> The patch is attached to the bug here:
> https://bugzilla.mozilla.org/show_bug.cgi?id=682180 - you probably
> only want the one-line configure.ac change.
>
> Cheers,
> -Bobby Holley, Mozilla Platform Dev
>
> On Fri, Sep 2, 2011 at 5:32 AM, David Gilbert <david.gilbert@linaro.org> wrote:
> >
> > On 23 August 2011 17:37, Anthony Green <green@moxielogic.com> wrote:
> > > I just released 3.0.10.  Thanks to everybody who contributed!
> > >
> > > My plan is to release 3.0.11 within a couple of weeks.  It will contain the
> > > soname bump (since I've removed debug symbols, again), and I also want to
> > > get David Gilbert's variadic function support in there.
> > >
> > > I am not able to personally test every platform that libffi purports to
> > > support, so please test and post your results here:
> > >  http://moxielogic.org/wiki/index.php?title=Libffi_3.0.11
> >
> > Hi Anthony,
> >  I've just added the i686 and arm (soft float) - seems OK; a few xpass on arm.
> >
> > It'll be good to get the variadic stuff in!
> >
> > Dave

[-- Attachment #2: 02-bug-682180.patch --]
[-- Type: application/octet-stream, Size: 1169 bytes --]

diff --git a/js/src/ctypes/libffi/configure b/js/src/ctypes/libffi/configure
index 2c08e1b..37e3055 100755
--- a/js/src/ctypes/libffi/configure
+++ b/js/src/ctypes/libffi/configure
@@ -12362,7 +12362,7 @@ $as_echo "#define HAVE_AS_STRING_PSEUDO_OP 1" >>confdefs.h
 fi
 
 case "$target" in
-     *-apple-darwin10* | *-*-freebsd* | *-*-openbsd* | *-pc-solaris*)
+     *-apple-darwin1* | *-*-freebsd* | *-*-openbsd* | *-pc-solaris*)
 
 $as_echo "#define FFI_MMAP_EXEC_WRIT 1" >>confdefs.h
 
diff --git a/js/src/ctypes/libffi/configure.ac b/js/src/ctypes/libffi/configure.ac
index e85cff1..1db02ce 100644
--- a/js/src/ctypes/libffi/configure.ac
+++ b/js/src/ctypes/libffi/configure.ac
@@ -316,7 +316,8 @@ if test x$TARGET = xX86 || test x$TARGET = xX86_WIN32 || test x$TARGET = xX86_64
 fi
 
 case "$target" in
-     *-apple-darwin10* | *-*-freebsd* | *-*-openbsd* | *-pc-solaris*)
+     # Darwin 10 (OSX 10.6) and beyond allocate non-executable pages
+     *-apple-darwin1* | *-*-freebsd* | *-*-openbsd* | *-pc-solaris*)
        AC_DEFINE(FFI_MMAP_EXEC_WRIT, 1,
                  [Cannot use malloc on this target, so, we revert to
                    alternative means])