Nabble - gcc - patches

Re: 4.5 release notes for target-specific improvements

2009-12-18T02:26:58Z

On Tue, 8 Dec 2009, Joseph S. Myers wrote:
> I've applied this patch to add 4.5 release notes for various
> target-specific improvements for ARM, M68K/ColdFire, MIPS and PowerPC.

Very nice, thanks a lot, Joseph!

> I'm sure there are more such improvements, both for these targets and
> for other targets; target maintainers need to look at changes for their
> targets and add release notes as appropriate.

Ping. :-)

Gerald

Re: Reduce libstdc++-v3 locale::locale() lock contention

2009-12-18T01:48:15Z

On 12/17/2009 09:11 PM, Jimmy Guo wrote:

>>> Yes you are right. Cannot do this without ABI breakage. The old object
>>> handling destruction will cause problem.
>>>
>>>
>> Yeah, I also think so, unfortunately :( Would improve the performance a
>> lot, eh? Any idea how much?
>>
> Indeed a lot ... so tempted to apply this to my custom gcc and ask all of
> our affected projects to rebuild their libraries ...
>
> a. Current:
> nthreads = 1: 1585
> nthreads = 2: 4501
> nthreads = 4: 26203
> nthreads = 8: 31745
>
> b. locale::locale() lock contention reduction:
> nthreads = 1: 1237
> nthreads = 2: 2028
> nthreads = 4: 8371
> nthreads = 8: 21262
>
> c. 'b' + C locale ref counting removal:
> nthreads = 1: 1159
> nthreads = 2: 1523
> nthreads = 4: 1742
> nthreads = 8: 1902
>

I see, grunt.

For now I committed the below (tested x86_64-linux multilib) which will
probably remain our main improvement for 4.5.0, we'll see...

Thanks again,
Paolo.

////////////////////

2009-12-18 Jimmy Guo <jguo@...>

PR libstdc++/40088
* src/locale_init.cc (locale::locale()): Optimize the common case
where _S_global still points to _S_classic.

Index: src/locale_init.cc
===================================================================
--- src/locale_init.cc (revision 155309)
+++ src/locale_init.cc (working copy)
@@ -208,9 +208,25 @@
locale::locale() throw() : _M_impl(0)
{
_S_initialize();
- __gnu_cxx::__scoped_lock sentry(get_locale_mutex());
- _S_global->_M_add_reference();
+
+ // Checked locking to optimize the common case where _S_global
+ // still points to _S_classic (locale::_S_initialize_once()):
+ // - If they are the same, just increment the reference count and
+ // we are done. This effectively constructs a C locale object
+ // identical to the static c_locale.
+ // - Otherwise, _S_global can and may be destroyed due to
+ // locale::global() call on another thread, in which case we
+ // fall back to lock protected access to both _S_global and
+ // its reference count.
_M_impl = _S_global;
+ if (_M_impl == _S_classic)
+ _M_impl->_M_add_reference();
+ else
+ {
+ __gnu_cxx::__scoped_lock sentry(get_locale_mutex());
+ _S_global->_M_add_reference();
+ _M_impl = _S_global;
+ }
}

locale

Re: Reduce libstdc++-v3 locale::locale() lock contention

2009-12-18T01:47:33Z

On 12/17/2009 09:11 PM, Jimmy Guo wrote:

I see, grunt.

New Chinese (simplified) PO file for 'gcc' (version 4.5-b20091203)

2009-12-18T01:47:17Z

Hello, gentle maintainer.

This is a message from the Translation Project robot.

A revised PO file for textual domain 'gcc' has been submitted
by the Chinese (simplified) team of translators. The file is available at:

http://translationproject.org/latest/gcc/zh_CN.po

(This file, 'gcc-4.5-b20091203.zh_CN.po', has just now been sent to you in
a separate email.)

All other PO files for your package are available in:

http://translationproject.org/latest/gcc/

Please consider including all of these in your next release, whether
official or a pretest.

Whenever you have a new distribution with a new version number ready,
containing a newer POT file, please send the URL of that distribution
tarball to the address below. The tarball may be just a pretest or a
snapshot, it does not even have to compile. It is just used by the
translators when they need some extra translation context.

The following HTML page has been updated:

http://translationproject.org/domain/gcc.html

If any question arises, please contact the translation coordinator.

Thank you for all your work,

The Translation Project robot, in the
name of your translation coordinator.
<coordinator@...>

Re: [wwwdocs] Buildstat update for 4.3

2009-12-18T01:41:04Z

On Fri, 4 Dec 2009, Tom G. Christensen wrote:
> Just a single result posted in the last 6 weeks.
>
> Testresults for 4.3.4:
> hppa-unknown-linux-gnu

Thanks, Tom. I applied the patch, sorry for the delay.

Gerald

Re: [PATCH,Cygwin] Add -tsaware linker flag to Cygwin linker specs.

2009-12-18T01:24:51Z

Kai Tietz wrote:

> Ok, for trunk. If you like to merge this fix back to 4.4 branch, it is
> fine for me, too.

Since we had just missed a release, I wasn't in any hurry about this, but
I've now finally checked it into the branch. JFTR.

cheers,
DaveK

Re: [wwwdocs] Buildstat update for 4.4

2009-12-18T00:25:36Z

On Fri, 4 Dec 2009, Tom G. Christensen wrote:
> Testresults for 4.4.2:
> hppa1.1-hp-hpux10.20
> hppa2.0w-hp-hpux11.00
> hppa2.0w-hp-hpux11.11
> hppa64-hp-hpux11.11
> i386-pc-solaris2.11

Thanks, Tom. I applied this on your behalf.

Gerald

[patch][graphite] Fix PR42221 and PR42205

2009-12-17T22:10:20Z

Hi,

I have committed the attached two patches to the graphite branch for
further test.
They fix two PRs as follows:

2009-12-17 Sebastian Pop <sebastian.pop@...>

PR middle-end/42205
* graphite-sese-to-poly.c (translate_scalar_reduction_to_array_for_stmt):
Insert the reduction copy in the same block as the phi node.
(follow_ssa_with_commutative_ops): Handle GIMPLE_NOPs.

* testsuite/gcc.dg/graphite/pr42205-1.c: New.
* testsuite/gcc.dg/graphite/pr42205-2.c: New.

2009-12-17 Sebastian Pop <sebastian.pop@...>

PR middle-end/42221
* sese.c (expand_scalar_variables_expr): Follow the SSA links into
the array indexing of ADDR_EXPRs.

* testsuite/gcc.dg/graphite/pr42221.c: New.

I will commit these two patches to trunk after the graphite auto testers
will report no regressions.

Sebastian

[0001-Fix-PR42221.patch]

From 8eff928a8bf4e24be1445b041649f0be796c7ce9 Mon Sep 17 00:00:00 2001 From: spop <spop@138bc75d-0d04-0410-961f-82ee72b054a4> Date: Fri, 18 Dec 2009 05:30:56 +0000 Subject: [PATCH] Fix PR42221. 2009-12-17 Sebastian Pop <sebastian.pop@...> PR middle-end/42221 * sese.c (expand_scalar_variables_expr): Follow the SSA links into the array indexing of ADDR_EXPRs. * testsuite/gcc.dg/graphite/pr42221.c: New. git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/branches/graphite@155333 138bc75d-0d04-0410-961f-82ee72b054a4 --- gcc/ChangeLog.graphite | 8 ++++++++ gcc/graphite-clast-to-gimple.c | 2 +- gcc/sese.c | 15 ++++++++++++++- gcc/testsuite/gcc.dg/graphite/pr42221.c | 24 ++++++++++++++++++++++++ 4 files changed, 47 insertions(+), 2 deletions(-) create mode 100644 gcc/testsuite/gcc.dg/graphite/pr42221.c diff --git a/gcc/ChangeLog.graphite b/gcc/ChangeLog.graphite index c9ab283..33d3579 100644 --- a/gcc/ChangeLog.graphite +++ b/gcc/ChangeLog.graphite @@ -1,5 +1,13 @@ 2009-12-17 Sebastian Pop <sebastian.pop@...> + PR middle-end/42221 + * sese.c (expand_scalar_variables_expr): Follow the SSA links into + the array indexing of ADDR_EXPRs. + + * testsuite/gcc.dg/graphite/pr42221.c: New. + +2009-12-17 Sebastian Pop <sebastian.pop@...> + PR middle-end/42178 PR middle-end/42334 * graphite-interchange.c (lst_try_interchange): Do not modify OUTER diff --git a/gcc/graphite-clast-to-gimple.c b/gcc/graphite-clast-to-gimple.c index 93138b6..c8ffd19 100644 --- a/gcc/graphite-clast-to-gimple.c +++ b/gcc/graphite-clast-to-gimple.c @@ -931,7 +931,7 @@ translate_clast_guard (sese region, struct clast_guard *stmt, edge next_e, static edge translate_clast (sese region, struct clast_stmt *stmt, edge next_e, htab_t rename_map, VEC (tree, heap) **newivs, - htab_t newivs_index, htab_t bb_pbb_mapping, + htab_t newivs_index, htab_t bb_pbb_mapping, htab_t params_index) { if (!stmt) diff --git a/gcc/sese.c b/gcc/sese.c index 2c03393..4450a11 100644 --- a/gcc/sese.c +++ b/gcc/sese.c @@ -874,7 +874,20 @@ expand_scalar_variables_expr (tree type, tree op0, enum tree_code code, return expand_scalar_variables_ssa_name (op0, bb, region, map, gsi); if (code == ADDR_EXPR) - return op0; + { + tree op00 = TREE_OPERAND (op0, 0); + + if (handled_component_p (op00) + && TREE_CODE (op00) == ARRAY_REF) + { + tree e = expand_scalar_variables_expr (TREE_TYPE (op00), op00, + TREE_CODE (op00), + NULL, bb, region, map, gsi); + return fold_build1 (code, TREE_TYPE (op0), e); + } + + return op0; + } gcc_unreachable (); return NULL; diff --git a/gcc/testsuite/gcc.dg/graphite/pr42221.c b/gcc/testsuite/gcc.dg/graphite/pr42221.c new file mode 100644 index 0000000..da8daa1 --- /dev/null +++ b/gcc/testsuite/gcc.dg/graphite/pr42221.c @@ -0,0 +1,24 @@ +/* { dg-options "-Os -fgraphite-identity" } */ + +static void b2w(unsigned int *out, const unsigned char *in, unsigned int len) +{ + const unsigned char *bpend = in + len; + for (; in != bpend; in += 4, ++out) + { + *out = (unsigned int) (in[0] ) | + (unsigned int) (in[3] << 24); + } +} +static void md4step(unsigned int state[4], const unsigned char *data) +{ + unsigned int A, X[16]; + b2w(X, data, 64); + state[0] += A; +} +void md4sum(void) +{ + unsigned char final[128]; + unsigned int state[4]; + md4step(state, final); + md4step(state, final + 64); +} -- 1.6.0.4

[0002-Fix-PR42205.patch]

From 99bec941a113ff91465defb98ce11ff0bc549d86 Mon Sep 17 00:00:00 2001 From: Sebastian Pop <sebpop@...> Date: Thu, 17 Dec 2009 23:56:59 -0600 Subject: [PATCH] Fix PR42205. 2009-12-17 Sebastian Pop <sebastian.pop@...> PR middle-end/42205 * graphite-sese-to-poly.c (translate_scalar_reduction_to_array_for_stmt): Insert the reduction copy in the same block as the phi node. (follow_ssa_with_commutative_ops): Handle GIMPLE_NOPs. * testsuite/gcc.dg/graphite/pr42205-1.c: New. * testsuite/gcc.dg/graphite/pr42205-2.c: New. --- gcc/ChangeLog.graphite | 10 ++++++++++ gcc/graphite-sese-to-poly.c | 7 +++++-- gcc/testsuite/gcc.dg/graphite/pr42205-1.c | 17 +++++++++++++++++ gcc/testsuite/gcc.dg/graphite/pr42205-2.c | 11 +++++++++++ 4 files changed, 43 insertions(+), 2 deletions(-) create mode 100644 gcc/testsuite/gcc.dg/graphite/pr42205-1.c create mode 100644 gcc/testsuite/gcc.dg/graphite/pr42205-2.c diff --git a/gcc/ChangeLog.graphite b/gcc/ChangeLog.graphite index 33d3579..1f527a1 100644 --- a/gcc/ChangeLog.graphite +++ b/gcc/ChangeLog.graphite @@ -1,5 +1,15 @@ 2009-12-17 Sebastian Pop <sebastian.pop@...> + PR middle-end/42205 + * graphite-sese-to-poly.c (translate_scalar_reduction_to_array_for_stmt): + Insert the reduction copy in the same block as the phi node. + (follow_ssa_with_commutative_ops): Handle GIMPLE_NOPs. + + * testsuite/gcc.dg/graphite/pr42205-1.c: New. + * testsuite/gcc.dg/graphite/pr42205-2.c: New. + +2009-12-17 Sebastian Pop <sebastian.pop@...> + PR middle-end/42221 * sese.c (expand_scalar_variables_expr): Follow the SSA links into the array indexing of ADDR_EXPRs. diff --git a/gcc/graphite-sese-to-poly.c b/gcc/graphite-sese-to-poly.c index 1eb0696..f921da9 100644 --- a/gcc/graphite-sese-to-poly.c +++ b/gcc/graphite-sese-to-poly.c @@ -2493,6 +2493,9 @@ follow_ssa_with_commutative_ops (tree arg, tree lhs) stmt = SSA_NAME_DEF_STMT (arg); + if (gimple_code (stmt) == GIMPLE_NOP) + return NULL; + if (gimple_code (stmt) == GIMPLE_PHI) { if (phi_contains_arg (stmt, lhs)) @@ -2666,13 +2669,13 @@ static void translate_scalar_reduction_to_array_for_stmt (tree red, gimple stmt, gimple loop_phi) { - basic_block bb = gimple_bb (stmt); - gimple_stmt_iterator insert_gsi = gsi_after_labels (bb); + gimple_stmt_iterator insert_gsi = gsi_after_labels (gimple_bb (loop_phi)); tree res = gimple_phi_result (loop_phi); gimple assign = gimple_build_assign (res, red); gsi_insert_before (&insert_gsi, assign, GSI_SAME_STMT); + insert_gsi = gsi_after_labels (gimple_bb (stmt)); assign = gimple_build_assign (red, gimple_assign_lhs (stmt)); insert_gsi = gsi_for_stmt (stmt); gsi_insert_after (&insert_gsi, assign, GSI_SAME_STMT); diff --git a/gcc/testsuite/gcc.dg/graphite/pr42205-1.c b/gcc/testsuite/gcc.dg/graphite/pr42205-1.c new file mode 100644 index 0000000..413b9f7 --- /dev/null +++ b/gcc/testsuite/gcc.dg/graphite/pr42205-1.c @@ -0,0 +1,17 @@ +/* { dg-options "-O1 -ffast-math -floop-interchange" } */ + +int adler32(int adler, char *buf, int n) +{ + int sum = 0; + do { + adler += buf[0]; + sum += adler; + adler += buf[1]; + sum += adler; + adler += buf[2]; + sum += adler; + adler += buf[3]; + sum += adler; + } while (--n); + return adler | (sum << 16); +} diff --git a/gcc/testsuite/gcc.dg/graphite/pr42205-2.c b/gcc/testsuite/gcc.dg/graphite/pr42205-2.c new file mode 100644 index 0000000..595cedb --- /dev/null +++ b/gcc/testsuite/gcc.dg/graphite/pr42205-2.c @@ -0,0 +1,11 @@ +/* { dg-options "-O1 -funsafe-math-optimizations -floop-interchange" } */ + +double f(double x) +{ + double y = 0.0; + int i; + for (i = 0; i < 8; i++) { + y += x * i; + } + return y; +} -- 1.6.0.4

Re: [PATCH] GCC symdb

2009-12-17T19:16:48Z

Gcc symbol database (symdb)

The patch is used to collect gcc internal data (cpp token, tree node etc.) and
outputting them into database for further usage.

Untar the attachment, go symdb.txt for more. The patch is applied to recent
gcc 4.5 trunk.

symdb.tbz (35K) Download Attachment

PATCH PING, objc: silence unused warning

2009-12-17T14:46:45Z

Ping?
http://gcc.gnu.org/ml/gcc-patches/2009-12/msg00687.html

Thanks,
Ben

Re: [PATCH] Use DW_OP_addr for constant addresses in debuginfo (take 2)

2009-12-17T14:01:39Z

OK.

Jason

[patch, testsuite] refers PR42348

2009-12-17T13:22:56Z

Hi,

The problem reported in PR42348 revealed several underlying problems:

1/ that target supports are non-functional for ObjC/C++ when the
runtime flags are used (e.g. -fgnu-runtime).
This is because those flags cause a warning from cc1 or cc1plus.

This patch resolves this by changing the file suffix to .m or .mm when
the tool under test is ObjC or ObjC++ respectively.

2/ there is no support for detection of whether a target can support
OBJC2 or NeXT runtime.,
The patch adds support for detection of __OBJC2__ and next_runtime
which are needed if we are to tidy up the number of failing ObjC/C++
tests.

3/ The tests for the gnu-runtime library usage are failing in
testsuite/lib/objc.exp and obj-c++.exp.
[they only detect if -fgnu-runtime is placed in the dg-options,
rather than also checking the multilib options].

the patch modifies gcc/testsuite/lib/objc.exp and gcc/testsuite/lib/
obj-c++.exp to detect the runtime in force after all the options are
processed.

with these patches a more controlled skip-if can be applied [example
in the PR]
(which should enable a lot of cleaning up of spurious fails for
darwin9/10).

cheers,
Iain

2009-12-17 Iain Sandoe <iain.sandoe@...>

PR testsuite/42348

*gcc/testsuite/lib/target-supports.exp: Add support for ObjC and ObjC+
+ tools in standard tests,
(check_effective_target_objc2) New proc.
(check_effective_target_next_runtime) New proc.

*gcc/testsuite/lib/objc.exp: Walk through the gnu/next runtime flags
to determine which is actually in force, attach the appropriate
includes accordingly.
*gcc/testsuite/lib/obj-c++.exp: Ditto

===

Index: gcc/testsuite/lib/target-supports.exp
===================================================================
--- gcc/testsuite/lib/target-supports.exp (revision 155320)
+++ gcc/testsuite/lib/target-supports.exp (working copy)
@@ -29,10 +29,17 @@
# If ARGS is not empty, its first element is a string that
# should be added to the command line.
#
-# Assume by default that CONTENTS is C code. C++ code should contain
-# "// C++" and Fortran code should contain "! Fortran".
+# Assume by default that CONTENTS is C code.
+# Otherwise, code should contain:
+# "// C++" for c++,
+# "! Fortran" for Fortran code,
+# "/* ObjC", for ObjC
+# and "// ObjC++" for ObjC++
+# if the tool is ObjC/ObjC++ then we overide the extension to .m, .mm to allow for ObjC/ObjC++
+# specific flags.
proc check_compile {basename type contents args} {
global tool
+ verbose "check_compile tool: $tool for $basename"

if { [llength $args] > 0 } {
set options [list "additional_flags=[lindex $args 0]"]
@@ -42,8 +49,17 @@
switch -glob -- $contents {
"*! Fortran*" { set src ${basename}[pid].f90 }
"*// C++*" { set src ${basename}[pid].cc }
- default { set src ${basename}[pid].c }
+ "*// ObjC++*" { set src ${basename}[pid].mm }
+ "*/* ObjC*" { set src ${basename}[pid].m }
+ default {
+ switch -- $tool {
+ "objc" { set src ${basename}[pid].m }
+ "obj-c++" { set src ${basename}[pid].mm }
+ default { set src ${basename}[pid].c }
+ }
+ }
}
+
set compile_type $type
switch -glob $type {
assembly { set output ${basename}[pid].s }
@@ -1188,7 +1204,26 @@
}]
}

+proc check_effective_target_objc2 { } {
+ return [check_no_compiler_messages objc2 object {
+ #ifdef __OBJC2__
+ int dummy[1];
+ #else
+ #fail
+ #endif
+ }]
+}

+proc check_effective_target_next_runtime { } {
+ return [check_no_compiler_messages objc2 object {
+ #ifdef __NEXT_RUNTIME__
+ int dummy[1];
+ #else
+ #fail
+ #endif
+ }]
+}
+
# Return 1 if we're generating 32-bit code using default options, 0
# otherwise.

Index: gcc/testsuite/lib/objc.exp
===================================================================
--- gcc/testsuite/lib/objc.exp (revision 155320)
+++ gcc/testsuite/lib/objc.exp (working copy)
@@ -141,10 +141,46 @@
global ld_library_path
global objc_libgcc_s_path
global shlib_ext
+
+ set shlib_ext [get_shlib_extension]
+ set ld_library_path ".:${objc_libgcc_s_path}"
+
+ # we have to figure out which runtime will be used on darwin - because we
+ # need to add the include path for the gnu runtime if that is in effect.
+ # first set the default...
+ if { [istarget "*-*-darwin*"] } {
+ set nextruntime 1
+ } else {
+ set nextruntime 0
+ }
+ verbose "initial next runtime state : $nextruntime"
+ # next see if we define the option in dg-options...
+ foreach opt $options {
+ if [regexp ".*-fnext-runtime.*" $opt] {
+ set nextruntime 1
+ }
+ if [regexp ".*-fgnu-runtime.*" $opt] {
+ set nextruntime 0
+ }
+ }
+ verbose "next runtime state after dg opts: $nextruntime"
+
+ set tgt [target_info name]
+ if [board_info $tgt exists multilib_flags] {
+ set lb [board_info $tgt multilib_flags]
+ verbose "board multilib_flags $lb"
+ foreach opt $lb {
+ if [regexp ".*-fnext-runtime.*" $opt] {
+ set nextruntime 1
+ }
+ if [regexp ".*-fgnu-runtime.*" $opt] {
+ set nextruntime 0
+ }
+ }
+ }
+ verbose "next runtime state after any multilib opts: $nextruntime"

- set ld_library_path ".:${objc_libgcc_s_path}"
lappend options "libs=-lobjc"
- set shlib_ext [get_shlib_extension]
verbose "shared lib extension: $shlib_ext"

if { [target_info needs_status_wrapper]!="" && [info exists gluefile] } {
@@ -167,37 +203,38 @@
set options [concat "{additional_flags=$TOOL_OPTIONS}" $options]
}

- # If we have built libobjc along with the compiler (which usually
- # _is not_ the case on Mac OS X systems), point the test harness
+ # If we have built libobjc along with the compiler, point the test harness
# at it (and associated headers).

set objcpath "[get_multilibs]"

set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc.a]
- if { $libobjc_dir == ""} {
- verbose "see if we have -fgnu-runtime in options"
- if [regexp ".*-fgnu-runtime.*" "${options}"] {
- set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc-gnu.a]
- }
- }
- # Now check if we have a shared only build.
- if { $libobjc_dir == ""} {
- set libobjc_dir \
- [lookfor_file ${objcpath} libobjc/.libs/libobjc.${shlib_ext}]
- }
- if { $libobjc_dir == ""} {
- verbose "see if we have -fgnu-runtime in options"
- if [regexp ".*-fgnu-runtime.*" "${options}"] {
- set libobjc_dir \
- [lookfor_file ${objcpath} libobjc/.libs/libobjc-gnu.${shlib_ext}]
+ if { $libobjc_dir == "" } {
+ # on darwin there is a gnu runtime too.
+ set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc-gnu.a]
}
+ # perhaps we didn't build static libs
+ if { $libobjc_dir == "" } {
+ set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc.${shlib_ext}]
+ # on darwin there is a gnu runtime too.
+ if { $libobjc_dir == "" } {
+ set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc-gnu.${shlib_ext}]
+ }
}
+
if { $libobjc_dir != "" } {
- set objc_include_dir "${srcdir}/../../libobjc"
- lappend options "additional_flags=-I${objc_include_dir}"
+ # if we are using the gnu runtime - add its includes
+ if { $nextruntime == 0 } {
+ set objc_include_dir "${srcdir}/../../libobjc"
+ lappend options "additional_flags=-I${objc_include_dir}"
+ verbose "adding gnu runtime include dir: $objc_include_dir "
+ }
set libobjc_dir [file dirname ${libobjc_dir}]
- set objc_link_flags "-L${libobjc_dir}"
+ # allow for %s spec substitutions..
+ set objc_link_flags " -B${libobjc_dir} "
lappend options "additional_flags=${objc_link_flags}"
+ set objc_link_flags " -L${libobjc_dir} "
+ lappend options "additional_flags=${objc_link_flags}"
append ld_library_path ":${libobjc_dir}"
}
if { $type == "precompiled_header" } {
Index: gcc/testsuite/lib/obj-c++.exp
===================================================================
--- gcc/testsuite/lib/obj-c++.exp (revision 155320)
+++ gcc/testsuite/lib/obj-c++.exp (working copy)
@@ -115,50 +115,54 @@
append ld_library_path ":${gccpath}/lib"
}
if [file exists "${gccpath}/libg++/libg++.a"] {
- append flags "-L${gccpath}/libg++ "
+ append flags " -L${gccpath}/libg++ "
append ld_library_path ":${gccpath}/libg++"
}
if [file exists "${gccpath}/libstdc++/libstdc++.a"] {
- append flags "-L${gccpath}/libstdc++ "
+ append flags " -L${gccpath}/libstdc++ "
append ld_library_path ":${gccpath}/libstdc++"
}
if [file exists "${gccpath}/libstdc++-v3/src/.libs/libstdc++.a"] {
- append flags " -L${gccpath}/libstdc++-v3/src/.libs "
- append ld_library_path ":${gccpath}/libstdc++-v3/src/.libs"
+ # allow for %s spec substitutions
+ append flags " -B${gccpath}/libstdc++-v3/src/.libs "
+ append flags " -L${gccpath}/libstdc++-v3/src/.libs "
+ append ld_library_path ":${gccpath}/libstdc++-v3/src/.libs"
}
# Look for libstdc++.${shlib_ext}.
if [file exists "${gccpath}/libstdc++-v3/src/.libs/libstdc++.${shlib_ext}"] {
+ # allow for %s spec substitutions
+ append flags " -B${gccpath}/libstdc++-v3/src/.libs "
append flags " -L${gccpath}/libstdc++-v3/src/.libs "
append ld_library_path ":${gccpath}/libstdc++-v3/src/.libs"
}
if [file exists "${gccpath}/libiberty/libiberty.a"] {
- append flags "-L${gccpath}/libiberty "
+ append flags " -L${gccpath}/libiberty "
}
if [file exists "${gccpath}/librx/librx.a"] {
- append flags "-L${gccpath}/librx "
+ append flags " -L${gccpath}/librx "
}

set objcpath "[get_multilibs]"
set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc.a]
if { $libobjc_dir == ""} {
- set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc-gnu.a]
+ set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc-gnu.a]
}
# Now check if we have a shared only build.
if { $libobjc_dir == ""} {
- set libobjc_dir \
- [lookfor_file ${objcpath} libobjc/.libs/libobjc.${shlib_ext}]
- }
- if { $libobjc_dir == ""} {
- set libobjc_dir \
- [lookfor_file ${objcpath} libobjc/.libs/libobjc-gnu.${shlib_ext}]
- }
+ set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc.${shlib_ext}]
+ if { $libobjc_dir == ""} {
+ set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc-gnu.${shlib_ext}]
+ }
+ }

verbose "libobjc_dir: $libobjc_dir"

if { $libobjc_dir != "" } {
- set libobjc_dir [file dirname ${libobjc_dir}]
- append flags "-L${libobjc_dir}"
- append ld_library_path ":${libobjc_dir}"
+ set libobjc_dir [file dirname ${libobjc_dir}]
+ # allow for %s spec substitutions
+ append flags " -B${libobjc_dir} "
+ append flags " -L${libobjc_dir} "
+ append ld_library_path ":${libobjc_dir}"
}
append ld_library_path \
[gcc-set-multilib-library-path $OBJCXX_UNDER_TEST]
@@ -284,43 +288,73 @@
global OBJCXX_UNDER_TEST
global shlib_ext

- lappend options "libs=-lobjc"
set shlib_ext [get_shlib_extension]
- verbose "shared lib extension: $shlib_ext"
+ verbose "input-options: $options"

+ # we have to figure out which runtime will be used on darwin - because we
+ # need to add the include path for the gnu runtime if that is in effect.
+ # first set the default...
+ if { [istarget "*-*-darwin*"] } {
+ set nextruntime 1
+ } else {
+ set nextruntime 0
+ }
+ verbose "initial next runtime state : $nextruntime"
+ # next see if we define the option in dg-options...
+ foreach opt $options {
+ if [regexp ".*-fnext-runtime.*" $opt] {
+ set nextruntime 1
+ }
+ if [regexp ".*-fgnu-runtime.*" $opt] {
+ set nextruntime 0
+ }
+ }
+ verbose "next runtime state after dg opts: $nextruntime"
+
+ set tgt [target_info name]
+ if [board_info $tgt exists multilib_flags] {
+ set lb [board_info $tgt multilib_flags]
+ verbose "board multilib_flags $lb"
+ foreach opt $lb {
+ if [regexp ".*-fnext-runtime.*" $opt] {
+ set nextruntime 1
+ }
+ if [regexp ".*-fgnu-runtime.*" $opt] {
+ set nextruntime 0
+ }
+ }
+ }
+ verbose "next runtime state after any multilib opts: $nextruntime"
+
+ lappend options "libs=-lobjc"
+
if { [target_info needs_status_wrapper] != "" && [info exists gluefile] } {
lappend options "libs=${gluefile}"
lappend options "ldflags=${wrap_flags}"
}

- # If we have built libobjc along with the compiler (which usually
- # _is not_ the case on Mac OS X systems), point the test harness
+ # If we have built libobjc along with the compiler, point the test harness
# at it (and associated headers).

set objcpath "[get_multilibs]"

set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc.a]
if { $libobjc_dir == ""} {
- verbose "see if we have -fgnu-runtime in options"
- if [regexp ".*-fgnu-runtime.*" "${options}" ] {
- set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc-gnu.a]
- }
+ set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc-gnu.a]
}
if { $libobjc_dir == ""} {
- set libobjc_dir \
- [lookfor_file ${objcpath} libobjc/.libs/libobjc.${shlib_ext}]
- }
- if { $libobjc_dir == ""} {
- verbose "see if we have -fgnu-runtime in options"
- if [regexp ".*-fgnu-runtime.*" "${options}"] {
- set libobjc_dir \
- [lookfor_file ${objcpath} libobjc/.libs/libobjc-gnu.${shlib_ext}]
+ set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc.${shlib_ext}]
+ if { $libobjc_dir == ""} {
+ set libobjc_dir [lookfor_file ${objcpath} libobjc/.libs/libobjc-gnu.${shlib_ext}]
+ }
}
- }

if { $libobjc_dir != "" } {
+ # if we are using the gnu runtime - add its includes
+ if { $nextruntime == 0 } {
set objc_include_dir "${srcdir}/../../libobjc"
lappend options "additional_flags=-I${objc_include_dir}"
+ }
}

lappend options "additional_flags=[libio_include_flags]"

Re: [PATCH c/40885] Fix build_indirect_ref i18n problems.

2009-12-17T13:07:37Z

On Thu, Dec 17, 2009 at 6:43 AM, Shujing Zhao <pearly.zhao@...> wrote:
> Hi David and all,
>
> In order to avoid more error report for my old commit r155304. I have
> committed this trivial patch. It is only make RO_NULL is known to that file.

Okay.

Thanks, David

[patch, target] fix PR41605

2009-12-17T12:48:24Z

Hello,

This patch is approved in the PR thread by Mike Stump - subject to a
review of the testsuite changes by Janis.

cheers,
Iain

2009-12-17 Iain Sandoe <iain.sandoe@...>

PR target/41605

*gcc/config/darwin.h(LINK_COMMAND_SPEC): Resolve fopenmp specifically
for target
(LINK_SPEC): replace dynamic libraries by static equivalents where
necessary

*gcc/config/darwin9.h(LINK_COMMAND_SPEC): Resolve fopenmp
specifically for target

Index: gcc/config/darwin.h
===================================================================
--- gcc/config/darwin.h (revision 155320)
+++ gcc/config/darwin.h (working copy)
@@ -272,9 +272,10 @@
%{A} %{e*} %{m} %{r} %{x} \
%{o*}%{!o:-o a.out} \
%{!A:%{!nostdlib:%{!nostartfiles:%S}}} \
- %{L*} %{fopenmp:%:include(libgomp.spec)%(link_gomp)} \
- %(link_libgcc) %o %{fprofile-arcs|fprofile-generate*|coverage:-lgcov} \
- %{!nostdlib:%{!nodefaultlibs:%(link_ssp) %G %L}} \
+ %{L*} %(link_libgcc) %o %{fprofile-arcs|fprofile-generate|coverage:-lgcov} \
+ %{fopenmp|ftree-parallelize-loops=*: \
+ %{static|static-libgcc|static-libstdc++|static-libgfortran: libgomp.a%s; : -lgomp } } \
+ %{!nostdlib:%{!nodefaultlibs: %(link_ssp) %G %L }} \
%{!A:%{!nostdlib:%{!nostartfiles:%E}}} %{T*} %{F*} }}}}}}}\n\
%{!fdump=*:%{!fsyntax-only:%{!c:%{!M:%{!MM:%{!E:%{!S:\
%{.c|.cc|.C|.cpp|.cp|.c++|.cxx|.CPP|.m|.mm: \
@@ -293,8 +294,12 @@
so put a * after their names so all of them get passed. */
#define LINK_SPEC \
"%{static}%{!static:-dynamic} \
- %{fgnu-runtime:%:replace-outfile(-lobjc -lobjc-gnu)}\
- %{static|static-libgfortran:%:replace-outfile(-lgfortran libgfortran.a%s)}\
+ %{fgnu-runtime: %{static|static-libgcc: \
+ %:replace-outfile(-lobjc libobjc-gnu.a%s); \
+ :%:replace-outfile(-lobjc -lobjc-gnu ) } }\
+ %{static|static-libgcc|static-libgfortran:%:replace-outfile(-lgfortran libgfortran.a%s)}\
+ %{static|static-libgcc|static-libstdc++|static-libgfortran:%:replace-outfile(-lgomp libgomp.a%s)}\
+ %{static|static-libgcc|static-libstdc++:%:replace-outfile(-lstdc++ libstdc++.a%s)}\
%{!Zdynamiclib: \
%{Zforce_cpusubtype_ALL:-arch %(darwin_arch) -force_cpusubtype_ALL} \
%{!Zforce_cpusubtype_ALL:-arch %(darwin_subarch)} \
Index: gcc/config/darwin9.h
===================================================================
--- gcc/config/darwin9.h (revision 155320)
+++ gcc/config/darwin9.h (working copy)
@@ -31,9 +31,10 @@
%{A} %{e*} %{m} %{r} %{x} \
%{o*}%{!o:-o a.out} \
%{!A:%{!nostdlib:%{!nostartfiles:%S}}} \
- %{L*} %{fopenmp:%:include(libgomp.spec)%(link_gomp)} \
- %(link_libgcc) %o %{fprofile-arcs|fprofile-generate*|coverage:-lgcov} \
- %{!nostdlib:%{!nodefaultlibs:%(link_ssp) %G %L}} \
+ %{L*} %(link_libgcc) %o %{fprofile-arcs|fprofile-generate|coverage:-lgcov} \
+ %{fopenmp|ftree-parallelize-loops=*: \
+ %{static|static-libgcc|static-libstdc++|static-libgfortran: libgomp.a%s; : -lgomp } } \
+ %{!nostdlib:%{!nodefaultlibs: %(link_ssp) %G %L }} \
%{!A:%{!nostdlib:%{!nostartfiles:%E}}} %{T*} %{F*} }}}}}}}\n\
%{!fdump=*:%{!fsyntax-only:%{!c:%{!M:%{!MM:%{!E:%{!S:\
%{.c|.cc|.C|.cpp|.cp|.c++|.cxx|.CPP|.m|.mm: \

*gcc/testsuite/lib/g++.exp: Provide -B options to allow for link spec
%s substitutions for static libraries.
*libstdc++-v3/testsuite/libstdc++-abi/abi.exp: Ditto
*libstdc++-v3/testsuite/lib/libstdc++.exp: Ditto
*libgomp/testsuite/lib/libgomp.exp: Ditto

Index: gcc/testsuite/lib/g++.exp
===================================================================
--- gcc/testsuite/lib/g++.exp (revision 155320)
+++ gcc/testsuite/lib/g++.exp (working copy)
@@ -125,6 +125,9 @@
}
if [file exists "${gccpath}/libstdc++-v3/src/.libs/libstdc++.a"] {
append flags " -L${gccpath}/libstdc++-v3/src/.libs "
+ # Some targets use libstdc++.a%s in their specs, so they need a -B option
+ # for uninstalled testing.
+ append flags " -B${gccpath}/libstdc++-v3/src/.libs "
append ld_library_path ":${gccpath}/libstdc++-v3/src/.libs"
}
# Look for libstdc++.${shlib_ext}.
Index: libstdc++-v3/testsuite/libstdc++-abi/abi.exp
===================================================================
--- libstdc++-v3/testsuite/libstdc++-abi/abi.exp (revision 155320)
+++ libstdc++-v3/testsuite/libstdc++-abi/abi.exp (working copy)
@@ -18,7 +18,12 @@
# this test. Or, hey, if we don't support this kind of symbol
# versioning test: don't run it.

-set lib $blddir/src/.libs/libstdc++.so
+if { [string match "*-*-darwin*" $target_triplet] } {
+ set lib $blddir/src/.libs/libstdc++.dylib
+} else {
+ set lib $blddir/src/.libs/libstdc++.so
+}
+
set baseline_subdir "[eval exec $cxx --print-multi-dir]"

# Build the support objects.
Index: libstdc++-v3/testsuite/lib/libstdc++.exp
===================================================================
--- libstdc++-v3/testsuite/lib/libstdc++.exp (revision 155320)
+++ libstdc++-v3/testsuite/lib/libstdc++.exp (working copy)
@@ -91,6 +91,8 @@
global target_triplet
global flags_file
global tool_timeout
+ global DEFAULT_CXXFLAGS
+ global STATIC_LIBCXXFLAGS

# We set LC_ALL and LANG to C so that we get the same error
# messages as expected.
@@ -106,6 +108,7 @@

set blddir [lookfor_file [get_multilibs] libstdc++-v3]
set flags_file "${blddir}/scripts/testsuite_flags"
+ set shlib_ext [get_shlib_extension]
v3track flags_file 2

# If a test doesn't have special options, use DEFAULT_CXXFLAGS.
@@ -114,7 +117,7 @@
# 2) might need to be negated
# In particular, some tests have to be run without precompiled
# headers, or without assertions.
- global DEFAULT_CXXFLAGS
+
if ![info exists DEFAULT_CXXFLAGS] then {
set DEFAULT_CXXFLAGS ""
# Host specific goo here.
@@ -145,7 +148,7 @@

# Locate libgomp. This is only required for parallel mode.
set v3-libgomp 0
- set libgompdir [lookfor_file $blddir/../libgomp .libs/libgomp.so]
+ set libgompdir [lookfor_file $blddir/../libgomp .libs/libgomp.$shlib_ext]
if {$libgompdir != ""} {
set v3-libgomp 1
set libgompdir [file dirname $libgompdir]
@@ -156,7 +159,7 @@

# Locate libstdc++ shared library. (ie libstdc++.so.)
set v3-sharedlib 0
- set sharedlibdir [lookfor_file $blddir src/.libs/libstdc++.so]
+ set sharedlibdir [lookfor_file $blddir src/.libs/libstdc++.$shlib_ext]
if {$sharedlibdir != ""} {
if { [string match "*-*-linux*" $target_triplet] && [isnative] } {
set v3-sharedlib 1
@@ -165,6 +168,15 @@
}
v3track v3-sharedlib 3

+ set STATIC_LIBCXXFLAGS ""
+ set staticlibdir [lookfor_file $blddir src/.libs/libstdc++.a]
+ if {$staticlibdir != ""} {
+ set staticlibdir [file dirname $staticlibdir]
+ # Some targets use libstdc++.a%s in their specs, so they need a -B option
+ # for uninstalled testing.
+ set STATIC_LIBCXXFLAGS " -B${staticlibdir} "
+ }
+
# Compute what needs to be added to the existing LD_LIBRARY_PATH.
if {$gccdir != ""} {
set compiler ${gccdir}/g++
@@ -403,6 +415,7 @@
global cxxflags
global cxxldflags
global includes
+ global STATIC_LIBCXXFLAGS

if { [target_info needs_status_wrapper] != "" && [info exists gluefile] } {
lappend options "libs=${gluefile}"
@@ -412,6 +425,7 @@
set cxx_final $cxx
set cxxlibglossflags [libgloss_link_flags]
set cxx_final [concat $cxx_final $cxxlibglossflags]
+ set cxx_final [concat $cxx_final $STATIC_LIBCXXFLAGS]
set cxx_final [concat $cxx_final $cxxflags]
set cxx_final [concat $cxx_final $includes]

@@ -443,6 +457,7 @@
global blddir
global cc
global cxxflags
+ global STATIC_LIBCXXFLAGS

if { [target_info needs_status_wrapper] != "" && [info exists gluefile] } {
lappend options "libs=${gluefile}"
@@ -453,6 +468,7 @@
set cc_final $cc
set cxxlibglossflags [libgloss_link_flags]
set cc_final [concat $cc_final $cxxlibglossflags]
+ set cc_final [concat $cc_final $STATIC_LIBCXXFLAGS]
set cc_final [concat $cc_final $cxxflags]
set cc_final [concat $cc_final $includes]
regsub -all {\s[-]nostdinc[+][+]} $cc_final "" cc_final
Index: libgomp/testsuite/lib/libgomp.exp
===================================================================
--- libgomp/testsuite/lib/libgomp.exp (revision 155320)
+++ libgomp/testsuite/lib/libgomp.exp (working copy)
@@ -130,6 +130,9 @@
set ALWAYS_CFLAGS ""
if { $blddir != "" } {
lappend ALWAYS_CFLAGS "additional_flags=-B${blddir}/"
+ # targets that use libgomp.a%s in their specs need a -B option
+ # for uninstalled testing.
+ lappend ALWAYS_CFLAGS "additional_flags=-B${blddir}/.libs"
lappend ALWAYS_CFLAGS "additional_flags=-I${blddir}"
lappend ALWAYS_CFLAGS "ldflags=-L${blddir}/.libs"
}

[patch, testsuite, libgomp.fortran] darwin no longer requires static linkage to access TLS.

2009-12-17T12:33:25Z

Hi All,

As discussed on IRC last night, Darwin targets no longer need -static-
libgcc to provide TLS. Consequently, the two tests that force this
(libgomp.fortran/crayptr2.f90 and libgomp.fortran/pr32550.f90) can be
tidied.

in each case:
-! { dg-options "-static-libgcc" { target *-*-darwin* } }
+! { dg-require-effective-target tls_runtime }

this is (IMO) the correct situation in any event - it seems misleading
to force non-standard options to make a target pass a few tests.
checked on darwin9 ppc/i686 and darwin10 (thanks to Dominique).

cheers,
Iain

====

2009-12-17 Iain Sandoe <iain.sandoe@...>

*libgomp/testsuite/libgomp.fortran/crayptr2.f90 remove forced static
linkage for darwin, protect the test with require-effective-target
tls_runtime
*libgomp/testsuite/libgomp.fortran/pr32550.f90 remove forced static
linkage for darwin, protect the test with require-effective-target
tls_runtime

====

Index: libgomp/testsuite/libgomp.fortran/crayptr2.f90
===================================================================
--- libgomp/testsuite/libgomp.fortran/crayptr2.f90 (revision 155320)
+++ libgomp/testsuite/libgomp.fortran/crayptr2.f90 (working copy)
@@ -1,6 +1,6 @@
! { dg-do run }
! { dg-options "-fopenmp -fcray-pointer" }
-! { dg-options "-fopenmp -fcray-pointer -static-libgcc" { target *-*-darwin* } }
+! { dg-require-effective-target tls_runtime }

use omp_lib
integer :: a, b, c, d, p
Index: libgomp/testsuite/libgomp.fortran/pr32550.f90
===================================================================
--- libgomp/testsuite/libgomp.fortran/pr32550.f90 (revision 155320)
+++ libgomp/testsuite/libgomp.fortran/pr32550.f90 (working copy)
@@ -1,6 +1,6 @@
! PR fortran/32550
! { dg-do run }
-! { dg-options "-static-libgcc" { target *-*-darwin* } }
+! { dg-require-effective-target tls_runtime }

integer, pointer, save :: ptr
integer, target :: targ

[ping, Patch, driver gcc/g++] updated patch for PR41594, 41596

2009-12-17T12:20:21Z

following discussion on this, I've updated the patch that deals with
suppression of library commands when making PCH.
(whitespace issues kindly noted by Dave, have been mopped up)

The only change is that "-static-libstdc++" is now re-written for all
targets, including those that have -Bstatic/dynamic - thus allowing
all the option of spec substitution.

Bootstrapped on powerpc-apple-darwin9.

====

2009-10-05 Iain Sandoe <iain.sandoe@...>

PR driver/41594

*gcc/gcc.c: Add -static-libstdc++ to list of recognized options.

Index: gcc/gcc.c
===================================================================
--- gcc/gcc.c (revision 155320)
+++ gcc/gcc.c (working copy)
@@ -4551,12 +4551,14 @@
switches[n_switches].validated = 0;
switches[n_switches].ordering = 0;
/* These are always valid, since gcc.c itself understands the
- first four and gfortranspec.c understands -static-libgfortran. */
+ first four, gfortranspec.c understands -static-libgfortran
+ and g++spec.c understands -static-libstdc++ */
if (!strcmp (p, "save-temps")
|| !strcmp (p, "static-libgcc")
|| !strcmp (p, "shared-libgcc")
|| !strcmp (p, "pipe")
- || !strcmp (p, "static-libgfortran"))
+ || !strcmp (p, "static-libgfortran")
+ || !strcmp (p, "static-libstdc++"))
switches[n_switches].validated = 1;
else
{

2009-12-17 Iain Sandoe <iain.sandoe@...>

PR driver/41596

*gcc/cp/g++spec.c(lang_specific_driver): Do not insert -lstdc++
unless we really need it.
re-write '-static-libstdc++' to the CL to allow for spec substitution.

Index: gcc/cp/g++spec.c
===================================================================
--- gcc/cp/g++spec.c (revision 155320)
+++ gcc/cp/g++spec.c (working copy)
@@ -60,10 +60,12 @@
int saw_profile_flag = 0;

/* What do with libstdc++:
- -1 means we should not link in libstdc++
- 0 means we should link in libstdc++ if it is needed
- 1 means libstdc++ is needed and should be linked in.
- 2 means libstdc++ is needed and should be linked statically. */
+ -1 means a forced "don't link" libstdc++,
+ 0 means we don't need libstdc++,
+ bit 0 = 1 means libstdc++ is needed and should be linked in.
+ bit 1 = 1 means when libstdc++ is needed it should be linked statically.
+ during re-writing the args the -1 case is set to 0.
+ */
int library = 0;

/* The number of arguments being added to what's in argv, other than
@@ -160,19 +162,19 @@
quote = arg = argv[i+1];
else /* Error condition, message will be printed later. */
arg = "";
- if (library == 0
+ if (library >= 0
&& (strcmp (arg, "c++") == 0
|| strcmp (arg, "c++-cpp-output") == 0
|| strcmp (arg, "objective-c++") == 0
|| strcmp (arg, "objective-c++-cpp-output") == 0))
- library = 1;
+ library |= 1;

saw_speclang = 1;
}
else if (strcmp (argv[i], "-ObjC++") == 0)
{
- if (library == 0)
- library = 1;
+ if (library >= 0)
+ library |= 1;
saw_speclang = 1;
}
/* Arguments that go directly to the linker might be .o files,
@@ -180,14 +182,14 @@
else if (strcmp (argv[i], "-Xlinker") == 0)
{
quote = argv[i];
- if (library == 0)
- library = 1;
+ if (library >= 0)
+ library |= 1;
}
else if (strncmp (argv[i], "-Wl,", 4) == 0)
- library = (library == 0) ? 1 : library;
+ library |= (library >= 0) ? 1 : library;
/* Unrecognized libraries (e.g. -lfoo) may require libstdc++. */
else if (strncmp (argv[i], "-l", 2) == 0)
- library = (library == 0) ? 1 : library;
+ library |= (library >= 0) ? 1 : library;
else if (((argv[i][2] == '\0'
&& strchr ("bBVDUoeTuIYmLiA", argv[i][1]) != NULL)
|| strcmp (argv[i], "-Tdata") == 0))
@@ -207,7 +209,7 @@
shared_libgcc = 0;
else if (strcmp (argv[i], "-static-libstdc++") == 0)
{
- library = library >= 0 ? 2 : library;
+ library |= library >= 0 ? 2 : library;
args[i] |= SKIPOPT;
}
else if (DEFAULT_WORD_SWITCH_TAKES_ARG (&argv[i][1]))
@@ -241,7 +243,7 @@

/* If we don't know that this is a header file, we might
need to be linking in the libraries. */
- if (library == 0)
+ if (library >= 0 && (library & 1) == 0)
{
if ((len <= 2 || strcmp (argv[i] + (len - 2), ".H") != 0)
&& (len <= 2 || strcmp (argv[i] + (len - 2), ".h") != 0)
@@ -252,7 +254,7 @@
&& (len <= 4 || strcmp (argv[i] + (len - 4), ".HPP") != 0)
&& (len <= 4 || strcmp (argv[i] + (len - 4), ".tcc") != 0)
&& (len <= 3 || strcmp (argv[i] + (len - 3), ".hh") != 0))
- library = 1;
+ library |= 1;
}
}
}
@@ -260,6 +262,10 @@
if (quote)
fatal ("argument to '%s' missing\n", quote);

+ /* if library < 0 it means it was forced off, so don't include it */
+ if (library < 0)
+ library = 0;
+
/* There's no point adding -shared-libgcc if we don't have a shared
libgcc. */
#ifndef ENABLE_SHARED_LIBGCC
@@ -284,13 +290,13 @@

/* Make sure -lstdc++ is before the math library, since libstdc++
itself uses those math routines. */
- if (!saw_math && (args[i] & MATHLIB) && library > 0)
+ if (!saw_math && (args[i] & MATHLIB) && (library & 1) != 0)
{
--j;
saw_math = argv[i];
}

- if (!saw_libc && (args[i] & WITHLIBC) && library > 0)
+ if (!saw_libc && (args[i] & WITHLIBC) && (library & 1) != 0)
{
--j;
saw_libc = argv[i];
@@ -327,21 +333,23 @@
}

/* Add `-lstdc++' if we haven't already done so. */
- if (library > 0)
+ if ((library & 1) != 0)
{
+ if ((library & 2) && !static_link)
+ {
+ arglist[j] = "-static-libstdc++"; /* allow for link spec substitution */
+ j++;
#ifdef HAVE_LD_STATIC_DYNAMIC
- if (library > 1 && !static_link)
- {
arglist[j] = "-Wl,-Bstatic";
j++;
+#endif
}
-#endif
arglist[j] = saw_profile_flag ? LIBSTDCXX_PROFILE : LIBSTDCXX;
if (arglist[j][0] != '-' || arglist[j][1] == 'l')
added_libraries++;
j++;
/* Add target-dependent static library, if necessary. */
- if ((static_link || library > 1) && LIBSTDCXX_STATIC != NULL)
+ if ((static_link || (library & 2)) && LIBSTDCXX_STATIC != NULL)
{
arglist[j] = LIBSTDCXX_STATIC;
if (arglist[j][0] != '-' || arglist[j][1] == 'l')
@@ -349,7 +357,7 @@
j++;
}
#ifdef HAVE_LD_STATIC_DYNAMIC
- if (library > 1 && !static_link)
+ if ((library & 2) && !static_link)
{
arglist[j] = "-Wl,-Bdynamic";
j++;
@@ -358,7 +366,7 @@
}
if (saw_math)
arglist[j++] = saw_math;
- else if (library > 0 && need_math)
+ else if ((library & 1) && need_math)
{
arglist[j] = saw_profile_flag ? MATH_LIBRARY_PROFILE : MATH_LIBRARY;
if (arglist[j][0] != '-' || arglist[j][1] == 'l')

===

[PATCH] Fix same_comdat_group handling in cgraph_remove_unreachable_nodes (PR c++/42386, take 2)

2009-12-17T11:34:50Z

On Wed, Dec 16, 2009 at 04:59:03PM -0500, Jakub Jelinek wrote:

> On Wed, Dec 16, 2009 at 10:46:43PM +0100, Jan Hubicka wrote:
> > > We shouldn't consider same_comdat_group links to make nodes reachable for
> > > DECL_EXTERNAL nodes - for those whether they are in the same comdat group or
> > > not doesn't matter.
> >
> > What is supposed semnatics of DECL_EXTERNAL nodes in comdat groups and
> > how they differ from normal DECL_EXTERNAL functions? It seems that we
> > might not want to put them into the linkist at all.
>
> They don't differ, but whether some decl is emitted in the current TU or is
> DECL_EXTERNAL is not necessarily finalized at the beginning. While in this
> PR's testcase already maybe_clone_body can know it, in some other cases it
> doesn't know that and the FE only determines it much later (so
> same_comdat_group links have to be set up). And I wonder if
> IPA/-fwhole-program can change that as well even later.
> So having the links around, just not considering them for DECL_EXTERNAL
> seems more robust to me, otherwise the FE during
> cp_write_global_declarations will need to unlink the links if they stayed
> really DECL_EXTERNAL, and possibly other places that might change
> non-external into DECL_EXTERNAL as well.

This is a patch variant that Honza preapproved on IRC and I've committed to
trunk after bootstrapping/regtesting on x86_64-linux.

2009-12-17 Jakub Jelinek <jakub@...>

PR c++/42386
* ipa.c (function_and_variable_visibility): Clear same_comdat_group
links of DECL_EXTERNAL nodes.

* g++.dg/opt/dtor3.C: New test.

--- gcc/ipa.c.jj 2009-12-17 15:05:58.000000000 +0100
+++ gcc/ipa.c 2009-12-17 15:54:33.000000000 +0100
@@ -370,6 +370,23 @@ function_and_variable_visibility (bool w
happy. Clear the flag here to avoid confusion in middle-end. */
if (DECL_COMDAT (node->decl) && !TREE_PUBLIC (node->decl))
DECL_COMDAT (node->decl) = 0;
+ /* For external decls stop tracking same_comdat_group, it doesn't matter
+ what comdat group they are in when they won't be emitted in this TU,
+ and simplifies later passes. */
+ if (node->same_comdat_group && DECL_EXTERNAL (node->decl))
+ {
+ struct cgraph_node *n = node, *next;
+ do
+ {
+ /* If at least one of same comdat group functions is external,
+ all of them have to be, otherwise it is a front-end bug. */
+ gcc_assert (DECL_EXTERNAL (n->decl));
+ next = n->same_comdat_group;
+ n->same_comdat_group = NULL;
+ n = next;
+ }
+ while (n != node);
+ }
gcc_assert ((!DECL_WEAK (node->decl) && !DECL_COMDAT (node->decl))
|| TREE_PUBLIC (node->decl) || DECL_EXTERNAL (node->decl));
if (cgraph_externally_visible_p (node, whole_program))
--- gcc/testsuite/g++.dg/opt/dtor3.C.jj 2009-12-16 18:08:11.000000000 +0100
+++ gcc/testsuite/g++.dg/opt/dtor3.C 2009-12-16 18:06:36.000000000 +0100
@@ -0,0 +1,11 @@
+// PR c++/42386
+// { dg-do compile }
+// { dg-options "-O2" }
+# 1 "A.h" 1
+#pragma interface
+struct D { virtual bool d () const; };
+struct E { virtual ~E (); virtual void *e () const = 0; };
+struct A : public D, public E { ~A () {} };
+# 5 "dtor3.C" 1
+struct F : public A { void *f () const; void *e () const; };
+void *F::e () const { return __null; }

Jakub

Re: [PATCH] PR c++/42225

2009-12-17T09:46:56Z

On 12/17/2009 09:21 AM, Dodji Seketeli wrote:
> Right now that DECL_CONTEXT is not set for template type parms. So the
> patch below sets it.

Once we do that, perhaps structural_comptypes can compare the template
parms from the context of template type parms, so we don't need special
handling for typedefs at all?

Jason

Re: [PATCH c++/31665] Move the fragments of conditional expressions to full sentences

2009-12-17T09:39:11Z

On Wed, Dec 16, 2009 at 8:31 PM, Shujing Zhao <pearly.zhao@...> wrote:

> On 12/16/2009 11:15 PM, Gabriel Dos Reis wrote:
>>
>> On Wed, Dec 16, 2009 at 5:23 AM, Paolo Bonzini <bonzini@...> wrote:
>>>
>>> On 12/16/2009 10:55 AM, Shujing Zhao wrote:
>>>>
>>>> @@ -3934,9 +3936,9 @@ check_tag_decl (cp_decl_specifier_seq *d
>>>> {
>>>> if (declspecs->specs[(int)ds_inline]
>>>> || declspecs->specs[(int)ds_virtual])
>>>> - error ("%qs can only be specified for functions",
>>>> - declspecs->specs[(int)ds_inline]
>>>> - ? "inline" : "virtual");
>>>> + error (declspecs->specs[(int)ds_inline]
>>>> + ? G_("%<inline%> can only be specified for functions")
>>>> + : G_("%<virtual%> can only be specified for
>>>> functions"));
>>>> else if (saw_friend
>>>> && (!current_class_type
>>>> || current_scope () != current_class_type))
>>>>
>>>> @@ -19808,10 +19809,13 @@ static void
>>>> cp_parser_check_class_key (enum tag_types class_key, tree type)
>>>> {
>>>> if ((TREE_CODE (type) == UNION_TYPE) != (class_key == union_type))
>>>> - permerror (input_location, "%qs tag used in naming %q#T",
>>>> - class_key == union_type ? "union"
>>>> - : class_key == record_type ? "struct" : "class",
>>>> - type);
>>>> + permerror (input_location,
>>>> + class_key == union_type
>>>> + ? G_("%<union%> tag used in naming %q#T")
>>>> + : class_key == record_type
>>>> + ? G_("%<struct%> tag used in naming %q#T")
>>>> + : G_("%<class%> tag used in naming %q#T"),
>>>> + type);
>>>> }
>>>
>>> Pedantically these are not needed, because keywords are not translated
>>> anyway (and it gives 3 more strings to translate). However, I don't
>>> think
>>> this is a problem.
>>>
>>> Paolo
>>>
>>
>> yes, keyword should be left as such.
>
> Ok, the keyword conditional expressions are left.
>
> Tested again on i686.
> Is the revised patch ok for trunk?

Yes.
Thanks,

-- Gaby

[PATCH, ARM] Fix problem with regrename on ARM

2009-12-17T09:16:51Z

The pic_load_tls_plus_eight pattern on ARM was somehow mistakenly added
with a +r constraint when =r should have been used. This caused
reg-rename to think it was modifying a non-live value and thus that the
result couldn't really matter either. An unfortunate chain of events
then lead to a critical insn being deleted that then caused assembly
failure because we still had a reference to a label that was no-longer
being generated.

Fixed by modifying the constraint. Now everything is happy again.

R.
2009-12-17 Richard Earnshaw <rearnsha@...>

PR target/42372
* arm.md (pic_add_dot_plus_eight): Change output constraint from
update to write.

[tls.patch]

*** gcc/config/arm/arm.md (revision 155323) --- gcc/config/arm/arm.md (local) *************** (define_insn "pic_add_dot_plus_eight" *** 5292,5298 **** ) (define_insn "tls_load_dot_plus_eight" ! [(set (match_operand:SI 0 "register_operand" "+r") (mem:SI (unspec:SI [(match_operand:SI 1 "register_operand" "r") (const_int 8) (match_operand 2 "" "")] --- 5292,5298 ---- ) (define_insn "tls_load_dot_plus_eight" ! [(set (match_operand:SI 0 "register_operand" "=r") (mem:SI (unspec:SI [(match_operand:SI 1 "register_operand" "r") (const_int 8) (match_operand 2 "" "")]

Re: unwind info for epilogues

2009-12-17T08:43:31Z

On Mon, Jun 29, 2009 at 1:54 PM, H.J. Lu <hjl.tools@...> wrote:

> On Sat, May 30, 2009 at 6:21 AM, H.J. Lu<hjl.tools@...> wrote:
>> On Fri, May 29, 2009 at 5:42 PM, Richard Henderson <rth@...> wrote:
>>> Joseph S. Myers wrote:
>>>>
>>>> On Wed, 6 May 2009, Richard Henderson wrote:
>>>>
>>>>> Since the beginning, -fasynchronous-unwind-tables has not held correct
>>>>> information for function epilogues. This is an attempt at adding that.
>>>>
>>>> See also Nathan Froyd's patch (doing this for x86_64) described in his
>>>> 2006 Summit paper. I haven't compared the approaches (but fully support
>>>> adding this feature).
>>>>
>>>> http://gcc.gnu.org/ml/gcc-patches/2006-03/msg00426.html
>>>> http://gcc.gnu.org/ml/gcc-patches/2006-02/msg01091.html
>>>>
>>>
>>> Thanks for the pointers. I did incorporate a few of Nathan's ideas into
>>> this version of the patch -- primarily, when duplicating the epilogue,
>>> make sure the duplicated insns are also recorded as epilogue insns.
>>>
>>> The other large change from the previous patch is the ability to have
>>> the eh_return epilogue from _Unwind_Resume (and friends) marked
>>> properly. This required the addition of an EH_RETURN rtx, so that
>>> the middle-end could recognize when epilogue expansion should happen,
>>> rather than the add-hoc unspecs that ports had been using. As it
>>> happens, only i386 and bfin implement eh_return via special epilogues;
>>> most ports only need to overwrite one or more registers before using
>>> a normal epilogue.
>>>
>>> Tested on x86_64, i686; committed.
>>>
>>> r~
>>>
>>> * cfgcleanup.c (try_crossjump_to_edge): Only skip past
>>> NOTE_INSN_BASIC_BLOCK.
>>> * cfglayout.c (duplicate_insn_chain): Copy epilogue insn marks.
>>> Duplicate NOTE_INSN_EPILOGUE_BEG notes.
>>> * cfgrtl.c (can_delete_note_p): Allow NOTE_INSN_EPILOGUE_BEG
>>> to be deleted.
>>> * dwarf2out.c (struct cfa_loc): Change indirect field to bitfield,
>>> add in_use field.
>>> (add_cfi): Disable check redefining cfa away from drap.
>>> (lookup_cfa_1): Add remember argument; handle remember/restore.
>>> (lookup_cfa): Pass remember argument.
>>> (cfa_remember): New.
>>> (compute_barrier_args_size_1): Remove sibcall check.
>>> (dwarf2out_frame_debug_def_cfa): New.
>>> (dwarf2out_frame_debug_adjust_cfa): New.
>>> (dwarf2out_frame_debug_cfa_offset): New.
>>> (dwarf2out_frame_debug_cfa_register): New.
>>> (dwarf2out_frame_debug_cfa_restore): New.
>>> (dwarf2out_frame_debug): Handle REG_CFA_* notes.
>>> (dwarf2out_begin_epilogue): New.
>>> (dwarf2out_frame_debug_restore_state): New.
>>> (dw_cfi_oprnd1_desc): Handle DW_CFA_remember_state,
>>> DW_CFA_restore_state.
>>> (output_cfi_directive): Likewise.
>>> (convert_cfa_to_fb_loc_list): Likewise.
>>> (dw_cfi_oprnd1_desc): Handle DW_CFA_restore.
>>> * dwarf2out.h: Update.
>>> * emit-rtl.c (try_split): Don't split RTX_FRAME_RELATED_P.
>>> (copy_insn_1): Early out for null.
>>> * final.c (final_scan_insn): Call dwarf2out_begin_epilogue
>>> and dwarf2out_frame_debug_restore_state.
>>> * function.c (prologue, epilogue, sibcall_epilogue): Remove.
>>> (prologue_insn_hash, epilogue_insn_hash): New.
>>> (free_after_compilation): Adjust freeing accordingly.
>>> (record_insns): Create hash table if needed; push insns into
>>> hash instead of array.
>>> (maybe_copy_epilogue_insn): New.
>>> (contains): Search hash table instead of array.
>>> (sibcall_epilogue_contains): Remove.
>>> (thread_prologue_and_epilogue_insns): Split eh_return insns
>>> and mark them as epilogues.
>>> (reposition_prologue_and_epilogue_notes): Rewrite epilogue
>>> scanning in terms of basic blocks.
>>> * insn-notes.def (CFA_RESTORE_STATE): New.
>>> * jump.c (returnjump_p_1): Accept EH_RETURN.
>>> (eh_returnjump_p_1, eh_returnjump_p): New.
>>> * reg-notes.def (CFA_DEF_CFA, CFA_ADJUST_CFA, CFA_OFFSET,
>>> CFA_REGISTER, CFA_RESTORE): New.
>>> * rtl.def (EH_RETURN): New.
>>> * rtl.h (eh_returnjump_p, maybe_copy_epilogue_insn): Declare.
>>>
>>> * config/bfin/bfin.md (UNSPEC_VOLATILE_EH_RETURN): Remove.
>>> (eh_return_internal): Use eh_return rtx; split w/ epilogue.
>>>
>>> * config/i386/i386.c (gen_push): Update cfa state.
>>> (pro_epilogue_adjust_stack): Add set_cfa argument. When true,
>>> add a CFA_ADJUST_CFA note.
>>> (ix86_dwarf_handle_frame_unspec): Remove.
>>> (ix86_expand_prologue): Update cfa state.
>>> (ix86_emit_restore_reg_using_pop): New.
>>> (ix86_emit_restore_regs_using_pop): New.
>>> (ix86_emit_leave): New.
>>> (ix86_emit_restore_regs_using_mov): Add CFA_RESTORE notes.
>>> (ix86_expand_epilogue): Add notes for unwinding the epilogue.
>>> * config/i386/i386.h (struct machine_cfa_state): New.
>>> (ix86_cfa_state): New.
>>> * config/i386/i386.md (UNSPEC_EH_RETURN): Remove.
>>> (eh_return_internal): Merge from eh_return_<mode>,
>>> use eh_return rtx, split w/ epilogue.
>>>
>>
>> This breaks stack unwind on Linux/ia32:
>>
>> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40304
>>
>>
>
> This also breaks debug info for local variable on Linux/ia32:
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40596
>

This also caused:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42396

--
H.J.

Re: [www] Document GDB 7.0+ is needed to debug GCC 4.5 compiled code (PR other/41255)

2009-12-17T08:32:05Z

On Thu, Dec 17, 2009 at 4:57 PM, Jakub Jelinek <jakub@...> wrote:
> Hi!
>
> Is this ok, or do you prefer some other wording?

Sounds good to me. There is also
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41255
and we should recommend a gdb version in the Debuging Options section, maybe
right at the beginning.

Richard.

> --- htdocs/gcc-4.5/changes.html.jj 2009-12-17 16:43:11.000000000 +0100
> +++ htdocs/gcc-4.5/changes.html 2009-12-17 16:53:27.000000000 +0100
> @@ -30,6 +30,15 @@
> <li>Support has been removed for tuning for Itanium1 (Merced) variants.
> Note that code tuned for Itanium2 should also run correctly on Itanium1.</li>
>
> + <li>GCC now generates unwind info also for epilogues. DWARF debuginfo
> + generated by GCC now uses more features of DWARF3 than it used to
> + do and also some DWARF4 features. GDB older than 7.0 is not able to
> + handle either of these, so to debug GCC 4.5 generated binaries or
> + libraries GDB 7.0 or later is needed. You can disable use of DWARF4
> + features with <code>-gdwarf-3 -gstrict-dwarf</code> options, or with
> + <code>-gdwarf-2 -gstrict-dwarf</code> restrict GCC to just DWARF2
> + standard, but epilogue unwind info is emitted unconditionally whenever
> + unwind info is emitted. </li>
> </ul>
>
> <h2>General Optimizer Improvements</h2>
>
> Jakub
>

[www] Document GDB 7.0+ is needed to debug GCC 4.5 compiled code (PR other/41255)

2009-12-17T07:57:36Z

Hi!

Is this ok, or do you prefer some other wording?

--- htdocs/gcc-4.5/changes.html.jj 2009-12-17 16:43:11.000000000 +0100
+++ htdocs/gcc-4.5/changes.html 2009-12-17 16:53:27.000000000 +0100
@@ -30,6 +30,15 @@
<li>Support has been removed for tuning for Itanium1 (Merced) variants.
Note that code tuned for Itanium2 should also run correctly on Itanium1.</li>

+ <li>GCC now generates unwind info also for epilogues. DWARF debuginfo
+ generated by GCC now uses more features of DWARF3 than it used to
+ do and also some DWARF4 features. GDB older than 7.0 is not able to
+ handle either of these, so to debug GCC 4.5 generated binaries or
+ libraries GDB 7.0 or later is needed. You can disable use of DWARF4
+ features with <code>-gdwarf-3 -gstrict-dwarf</code> options, or with
+ <code>-gdwarf-2 -gstrict-dwarf</code> restrict GCC to just DWARF2
+ standard, but epilogue unwind info is emitted unconditionally whenever
+ unwind info is emitted. </li>
</ul>

<h2>General Optimizer Improvements</h2>

Jakub

[Docs patch, committed] clarify vec_select behavior

2009-12-17T07:53:36Z

I was confused by the description of vec_select in the internals manual and had
to look at various examples to understand what it returns. Having figured it
out, it's now obvious :-) so I have checked in this patch to the docs.

-Sandra

2009-12-17 Sandra Loosemore <sandra@...>

gcc/
* doc/rtl.texi (Vector Operations): Clarify vec_select result mode.

Index: gcc/doc/rtl.texi
===================================================================
--- gcc/doc/rtl.texi (revision 155316)
+++ gcc/doc/rtl.texi (working copy)
@@ -2614,9 +2614,12 @@ a set bit indicates it is taken from @va
@findex vec_select
@item (vec_select:@var{m} @var{vec1} @var{selection})
This describes an operation that selects parts of a vector. @var{vec1} is
-the source vector, @var{selection} is a @code{parallel} that contains a
+the source vector, and @var{selection} is a @code{parallel} that contains a
@code{const_int} for each of the subparts of the result vector, giving the
-number of the source subpart that should be stored into it.
+number of the source subpart that should be stored into it.
+The result mode @var{m} is either the submode for a single element of
+@var{vec1} (if only one subpart is selected), or another vector mode
+with that element submode (if multiple subparts are selected).

@findex vec_concat
@item (vec_concat:@var{m} @var{vec1} @var{vec2})

Re: [Patch, Fortran] PR 42144: [OOP] deferred TBPs do not work

2009-12-17T07:43:37Z

Hi Tobias,

> I have now merged the trunk into the fortran-dev branch

thanks a bunch.

> (a) I ignored the trans-expr.c change as it does not apply
> to the vtable.

Yes, that's correct.

> (b) I have included the test case. Thus, you will
> probably get a FAIL when regtesting the branch.

I don't think so. The vtable implementation should be able to handle
deferred TBPs already.

Cheers,
Janus

Re: Version script vs. mangled names on systems where size_type != unsigned int

2009-12-17T07:39:18Z

Paolo Carlini wrote:
> On 12/17/2009 04:32 PM, Dave Korn wrote:
>> Righto, I'll add the general comment as well as the two shorter ones. Did
>> your abi_check complete OK?
>>
> Sure, the entire regression suite actually. You are cleared to commit
> and close the PR.

Thanks, all done :)

cheers,
DaveK

Re: [Patch, Fortran] PR 42353: [OOP] Bogus Error: Name 'vtype$...' at (1) is an ambiguous reference ...

2009-12-17T07:38:35Z

> the attached patch fixes a small OOP problem, where under certain
> circumstances one would get 'ambiguous reference' errors for the
> vtypes. The easiest thing I could think of to fix this is to make the
> vtypes private, which means that each module which uses a vtype will
> have its own private copy (which in principle should not be a problem,
> should it?).

In contrast to the patch I sent earlier, the version attached here
also fixes the second error in the PR, which I almost forgot about
("Error: The element in the derived type constructor at (1), for
pointer component '$extends', is DERIVED but should be DERIVED").

Although the patch obviously works, I'm not sure if it may have any
negative consequences and if it's the best thing to do. At least it
fixes the test case and regtests cleanly.

Comments? Suggestions? Ok for trunk?

Cheers,
Janus

pr42353_2.diff (670 bytes) Download Attachment

Re: [Patch, Fortran] PR 42144: [OOP] deferred TBPs do not work

2009-12-17T07:37:41Z

On 12/17/2009 10:41 AM, Janus Weil wrote:
>>> The patch was regtested on x86_64-unknown-linux-gnu. Ok for trunk?
>>>
>> OK for trunk.
>>
> Thanks. Committed as r155305.
>

I have now merged the trunk into the fortran-dev branch; that means for
this patch: (a) I ignored the trans-expr.c change as it does not apply
to the vtable. (b) I have included the test case. Thus, you will
probably get a FAIL when regtesting the branch. (And I did some
renaming; as this patch's testfile name was the same as an existing one
on the branch, I have renamed the branch one and updated
ChangeLog.fortran-dev.)

Transmitting file data
..........................................................................................................
Committed revision 155317.

Tobias

Re: Version script vs. mangled names on systems where size_type != unsigned int

2009-12-17T07:18:41Z

On 12/17/2009 04:32 PM, Dave Korn wrote:
> Righto, I'll add the general comment as well as the two shorter ones. Did
> your abi_check complete OK?
>
Sure, the entire regression suite actually. You are cleared to commit
and close the PR.

Thanks,
Paolo.

Re: Version script vs. mangled names on systems where size_type != unsigned int

2009-12-17T07:16:25Z

Paolo Carlini wrote:
> On 12/17/2009 04:23 PM, Dave Korn wrote:
>> Do you mean something like this ...
> Sure. A general comment like the one you posted before cannot hurt, but
> really next time I have to write one of those patterns I want to be
> reminded that 'x' and 'y' are also necessary in some cases (in the past,
> if x86-64-linux multilib was fine I could consider a change more or less
> safe from the linker script point of view, now isn't true anymore, and I
> don't want to make mistakes)

Righto, I'll add the general comment as well as the two shorter ones. Did
your abi_check complete OK?

cheers,
DaveK

Re: PING [Patch, driver] PR41594 recognize -static-libstdc++ as a valid option.

2009-12-17T07:14:57Z

IainS wrote:

>
> On 17 Dec 2009, at 14:23, Dave Korn wrote:
>
>> IainS wrote:
>>
>>> Although your patch resolves the issue of spec substitution - it will
>>> not resolve PR41596.
>>> I believe my patch resolves both (and agreed is more complex because of
>>> that).
>>
>> Yep, agreed; my patch only addresses the 41594 side of things, but I
>> think
>> if we used it rather than your one for that issue, you would only then
>> need a
>> much simpler patch on top to solve 41596, no?
>
> I don't think I can make the solution to PR41596 simpler...
>
> .. in fact, re-emitting the -static-libstdc++ flag without checking that
> we really need to link will actually make the situation worse for cases
> where we're compiling PCH.
>
> In order to solve PR41596 - one needs to detect that PCH generation is
> happening and make sure that no library commands are emitted - since the
> latter will cause a (redundant) invocation of collect2.

Ahh, tricky. So I guess you really can't get away without removing it. Hm.

> as I said, I'm happy to unwrap the re-emission so that it occurs for all
> targets (if that's what people want).

I certainly would like to see it appear on the command-line whenever a link
is taking place (assuming it was specified by the user in the first place of
course!).

cheers,
DaveK

Re: Version script vs. mangled names on systems where size_type != unsigned int

2009-12-17T07:13:13Z

On 12/17/2009 04:23 PM, Dave Korn wrote:
> Do you mean something like this ...
Sure. A general comment like the one you posted before cannot hurt, but
really next time I have to write one of those patterns I want to be
reminded that 'x' and 'y' are also necessary in some cases (in the past,
if x86-64-linux multilib was fine I could consider a change more or less
safe from the linker script point of view, now isn't true anymore, and I
don't want to make mistakes)

Thanks,
Paolo.

Re: Version script vs. mangled names on systems where size_type != unsigned int

2009-12-17T07:07:32Z

Paolo Carlini wrote:
> On 12/17/2009 03:28 PM, Dave Korn wrote:
>> I'm not sure what first occurrence you mean or what goes before "...and
>> below" in that sentence, but how about this instead:
>>
> The first occurrence of 'x' and 'y', what else? Just a one-line comment
> before the first time you use 'x' and 'y', describing those, reminding
> about LLP64 and finishing with something like "... and below", to mean
> that there are more 'x' and 'y' in the patterns below. I don't find that
> so strange...

Do you mean something like this ...

> @@ -167,20 +167,22 @@ GLIBCXX_3.4 {
> # Names not in an 'extern' block are mangled names.
>
> # std::string
> - _ZNSsC[12][EI][PRjmvN]*;
> + # 'y' here and below represents 'unsigned long long'
> + # where it is used for size_type on LLP64 platforms.
> + _ZNSsC[12][EI][PRjmvyN]*;

... and this?

> @@ -342,7 +344,9 @@ GLIBCXX_3.4 {
> _ZNKSi[0-9][a-z]*;
> _ZNSi[0-9][a-h]*;
> _ZNSi[0-9][j-z]*;
> - _ZNSi6ignoreE[il][il];
> + # 'x' here and below matches 'long long' where it
> + # is used for signed sizetypes on LLP64 platforms.
> + _ZNSi6ignoreE[ilx][ilx];
> _ZNSirsE*[^g];

cheers,
DaveK

Re: Version script vs. mangled names on systems where size_type != unsigned int

2009-12-17T06:33:59Z

On 12/17/2009 03:28 PM, Dave Korn wrote:
> I'm not sure what first occurrence you mean or what goes before "...and
> below" in that sentence, but how about this instead:
>
The first occurrence of 'x' and 'y', what else? Just a one-line comment
before the first time you use 'x' and 'y', describing those, reminding
about LLP64 and finishing with something like "... and below", to mean
that there are more 'x' and 'y' in the patterns below. I don't find that
so strange...

Paolo.

[PATCH] Fix PR42397

2009-12-17T06:29:28Z

This fixes an ICE with an LTO testcase that uncovers a middle-end
problem. get_object_alignment does not properly handle &CONST_DECL.

Fixed thus, bootstrapped and tested on x86_64-unknown-linux-gnu,
committed to trunk.

Richard.

2009-12-17 Richard Guenther <rguenther@...>

PR middle-end/42397
* builtins.c (get_object_alignment): Properly deal with
a CONST_DECL base.
* expr.c (emit_block_move_hints): Assert the alignment makes
sense.

Index: gcc/builtins.c
===================================================================
*** gcc/builtins.c (revision 155297)
--- gcc/builtins.c (working copy)
*************** get_object_alignment (tree exp, unsigned
*** 323,328 ****
--- 323,330 ----
offset = next_offset;
}
}
+ if (TREE_CODE (exp) == CONST_DECL)
+ exp = DECL_INITIAL (exp);
if (DECL_P (exp))
align = MIN (inner, DECL_ALIGN (exp));
#ifdef CONSTANT_ALIGNMENT
Index: gcc/expr.c
===================================================================
*** gcc/expr.c (revision 155297)
--- gcc/expr.c (working copy)
*************** emit_block_move_hints (rtx x, rtx y, rtx
*** 1194,1199 ****
--- 1194,1200 ----
}

align = MIN (MEM_ALIGN (x), MEM_ALIGN (y));
+ gcc_assert (align >= BITS_PER_UNIT);

gcc_assert (MEM_P (x));
gcc_assert (MEM_P (y));