Added gcc-10.2.0 patches #106

saidganim · 2020-08-29T16:26:29Z

No description provided.

richfelker · 2020-08-29T17:22:01Z

It looks like a good many of the patches from 9.x have been dropped. Are they all upstream? I notice the s390x arch support patch is still there but m68k was dropped, so was one of these upstreamed but not the other? And in particular, omitting the the invalid tls model and gth-weak patches would produce silent breakage if they're not upstream.

saidganim · 2020-08-29T17:43:24Z

@richfelker hey, yes, I didn't include only those patches, which are in upstream. So it seems to be safe :)

saidganim · 2020-09-19T10:56:35Z

@richfelker Are we merging this PR ?

TOCK-Chiu

Just a minor question regarding 0009-static-pie.diff.

TOCK-Chiu · 2020-09-25T10:43:48Z

patches/gcc-10.1.0/0009-static-pie.diff

+   "%{!static:%{fvtable-verify=none:%s; \
+      fvtable-verify=preinit:vtv_end_preinit.o%s; \
+      fvtable-verify=std:vtv_end.o%s}} \


According to the commit that changed this line between GCC 9 and 10, I think we should exclude static-pie as well, no?

For example (borrowed from Szabolcs Nagy):

Suggested change

"%{!static:%{fvtable-verify=none:%s; \

fvtable-verify=preinit:vtv_end_preinit.o%s; \

fvtable-verify=std:vtv_end.o%s}} \

"%{static|static-pie:; \

fvtable-verify=none:%s; \

fvtable-verify=preinit:vtv_end_preinit.o%s; \

fvtable-verify=std:vtv_end.o%s} \

Good point. Thank you. Seems, you are right I think we should ~~include~~ exclude it, otherwise compiling static-pie binary should give 'multiple definition of ...' linking error. To be honest, I am not sure whether it was a good idea in gcc10 to make tentative variables go into bss .. (I want to be wrong).

@richfelker any thoughts?

TOCK-Chiu

Just some more questions.

TOCK-Chiu · 2020-09-29T07:31:17Z

patches/gcc-10.1.0/0003-libatomic-test-fix.diff

+diff --git a/libatomic/testsuite/lib/libatomic.exp b/libatomic/testsuite/lib/libatomic.exp
+index 38f3e5673e2..02ec2e8b48a 100644
+--- a/libatomic/testsuite/lib/libatomic.exp
+++ b/libatomic/testsuite/lib/libatomic.exp
+@@ -77,6 +77,7 @@ proc libatomic_init { args } {
+     global ALWAYS_CFLAGS
+     global CFLAGS
+     global TOOL_EXECUTABLE TOOL_OPTIONS
+    global BUILD_CC
+     global GCC_UNDER_TEST
+     global TESTING_IN_BUILD_TREE
+     global target_triplet
+@@ -92,6 +93,8 @@ proc libatomic_init { args } {
+     if ![info exists GCC_UNDER_TEST] then {
+ 	if [info exists TOOL_EXECUTABLE] {
+ 	    set GCC_UNDER_TEST $TOOL_EXECUTABLE
+	} elseif [info exists BUILD_CC] {
+	    set GCC_UNDER_TEST $BUILD_CC
+ 	} else {
+ 	    set GCC_UNDER_TEST "[find_gcc]"
+ 	}


This might have been fixed upstream in another way?

TOCK-Chiu · 2020-09-29T07:35:48Z

patches/gcc-10.1.0/0004-libgomp-test-fix.diff

+diff --git a/libgomp/testsuite/Makefile.am b/libgomp/testsuite/Makefile.am
+index 655a413c160..916e9d30f15 100644
+--- a/libgomp/testsuite/Makefile.am
+++ b/libgomp/testsuite/Makefile.am
+@@ -14,6 +14,13 @@ RUNTESTDEFAULTFLAGS = --tool $$tool --srcdir $$srcdir
+
+ EXTRA_DEJAGNU_SITE_CONFIG = libgomp-site-extra.exp
+
+EXTRA_DEJAGNU_SITE_CONFIG = extra.exp
+
+extra.exp:
+	echo 'set BUILD_CC "$(CC)"' > [email protected]
+	mv [email protected] $@
+
+
+ # Instead of directly in ../testsuite/libgomp-test-support.exp.in, the
+ # following variables have to be "routed through" this Makefile, for expansion
+ # of the several (Makefile) variables used therein.
+diff --git a/libgomp/testsuite/Makefile.in b/libgomp/testsuite/Makefile.in
+index 52aa6c5fbc9..617da8ddbf2 100644
+--- a/libgomp/testsuite/Makefile.in
+++ b/libgomp/testsuite/Makefile.in
+@@ -311,6 +311,7 @@ _RUNTEST = $(shell if test -f $(top_srcdir)/../dejagnu/runtest; then \
+ 	     echo $(top_srcdir)/../dejagnu/runtest; else echo runtest; fi)
+
+ RUNTESTDEFAULTFLAGS = --tool $$tool --srcdir $$srcdir
+EXTRA_DEJAGNU_SITE_CONFIG = extra.exp
+ EXTRA_DEJAGNU_SITE_CONFIG = libgomp-site-extra.exp
+ all: all-am
+
+@@ -475,6 +476,10 @@ uninstall-am:
+ .PRECIOUS: Makefile
+
+
+extra.exp:
+	echo 'set BUILD_CC "$(CC)"' > [email protected]
+	mv [email protected] $@
+
+ # Instead of directly in ../testsuite/libgomp-test-support.exp.in, the
+ # following variables have to be "routed through" this Makefile, for expansion
+ # of the several (Makefile) variables used therein.
+diff --git a/libgomp/testsuite/lib/libgomp.exp b/libgomp/testsuite/lib/libgomp.exp
+index ee5f0e57b19..183081b5802 100644
+--- a/libgomp/testsuite/lib/libgomp.exp
+++ b/libgomp/testsuite/lib/libgomp.exp
+@@ -68,6 +68,7 @@ proc libgomp_init { args } {
+     global ALWAYS_CFLAGS
+     global CFLAGS
+     global TOOL_EXECUTABLE TOOL_OPTIONS
+    global BUILD_CC
+     global GCC_UNDER_TEST
+     global TESTING_IN_BUILD_TREE
+     global target_triplet
+@@ -90,6 +91,8 @@ proc libgomp_init { args } {
+     if ![info exists GCC_UNDER_TEST] then {
+ 	if [info exists TOOL_EXECUTABLE] {
+ 	    set GCC_UNDER_TEST $TOOL_EXECUTABLE
+	} elseif [info exists BUILD_CC] {
+	    set GCC_UNDER_TEST $BUILD_CC        
+ 	} else {
+ 	    set GCC_UNDER_TEST "[find_gcc]"
+ 	}


This might have been fixed upstream in another way, as well?

You are right. I am generating patches for gcc10.2.0 anyway, I will update this PR

Thank you for your revision

TOCK-Chiu · 2020-09-29T07:41:42Z

patches/gcc-10.1.0/0008-s390x-muslldso.diff

+#define MUSL_DYNAMIC_LINKER32 "/lib/ld-musl-s390.so.1"
+#define MUSL_DYNAMIC_LINKER64 "/lib/ld-musl-s390x.so.1"


This has been added upstream, along with another fix.

You are right.

rofl0r · 2020-09-29T09:22:03Z

i think it would be preferable to merge the patches done by szabolcs nagy, as he's a long-time contributor to both musl and gcc and certainly did a careful forward port of the missing patches.
note that he also provided most of the existing patch sets.

richfelker · 2020-10-01T19:16:46Z

@rofl0r Looking at https://port70.net/~nsz/musl/gcc-10.1.0/, it seems there are patches missing that are included in this set: s390x ldso, time64 futex support, and m68k sqrt, at least. It sounds like the test fixes might be fixed upstream and no longer needed though.

Also, the time64 futex patch needs a fix to be riscv32-ready. I've pushed some pending changes I had here that can be incorporated.

rofl0r · 2021-07-07T16:27:10Z

superceded by #124 , this can be closed.

TOCK-Chiu reviewed Sep 25, 2020

View reviewed changes

TOCK-Chiu reviewed Sep 29, 2020

View reviewed changes

saidganim force-pushed the master branch from b3d6497 to 3d8d726 Compare October 1, 2020 20:04

saidganim changed the title ~~Added gcc-10.1.0 patches~~ Added gcc-10.2.0 patches Oct 1, 2020

saidganim force-pushed the master branch from 3d8d726 to bd450bc Compare October 1, 2020 20:06

Added gcc-10.2.0 patches

bd450bc

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Added gcc-10.2.0 patches #106

Added gcc-10.2.0 patches #106

saidganim commented Aug 29, 2020

richfelker commented Aug 29, 2020

saidganim commented Aug 29, 2020

saidganim commented Sep 19, 2020

TOCK-Chiu left a comment

TOCK-Chiu Sep 25, 2020

saidganim Sep 25, 2020 •

edited

Loading

saidganim Sep 25, 2020

TOCK-Chiu left a comment

TOCK-Chiu Sep 29, 2020

TOCK-Chiu Sep 29, 2020

saidganim Oct 1, 2020

saidganim Oct 1, 2020

TOCK-Chiu Sep 29, 2020

saidganim Oct 1, 2020

rofl0r commented Sep 29, 2020

richfelker commented Oct 1, 2020

rofl0r commented Jul 7, 2021

		+#define MUSL_DYNAMIC_LINKER32 "/lib/ld-musl-s390.so.1"
		+#define MUSL_DYNAMIC_LINKER64 "/lib/ld-musl-s390x.so.1"

Added gcc-10.2.0 patches #106

Are you sure you want to change the base?

Added gcc-10.2.0 patches #106

Conversation

saidganim commented Aug 29, 2020

richfelker commented Aug 29, 2020

saidganim commented Aug 29, 2020

saidganim commented Sep 19, 2020

TOCK-Chiu left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

saidganim Sep 25, 2020 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

TOCK-Chiu left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rofl0r commented Sep 29, 2020

richfelker commented Oct 1, 2020

rofl0r commented Jul 7, 2021

saidganim Sep 25, 2020 •

edited

Loading