Package: libunwind
Version: 1.0.1-4
Severity: normal
Tags: patch
User: ubuntu-de...@lists.ubuntu.com
Usertags: origin-ubuntu raring ubuntu-patch
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256
Dear Maintainer,
In Ubuntu, the attached patch was applied to achieve the following:
* Enable test suite for amd64 architecture:
+ d/rules: Execute test suite when building for amd64 only.
+ d/patches/20122307_fix_setjmp.patch: Cherry picked fix from upstream
VCS to fixup setjmp test failure.
+ d/patches/20123007_disable_known_test_failures.patch: Disable *-dyn1
tests as they use unsupported features on amd64.
+ d/patches/20123007_workaround_ltest-nocalloc.patch: Make dummy call to
do_backtrace to ensure that malloc is not called post thread init.
+ d/patches/20123007_workaround_test-ptrace.patch: Walk default test
arguments to ensure child process has visibility.
The libunwind test suite is only really usable on amd64 but it serves as a
useful canary for any core integration issues.
Thanks for considering the patch.
- -- System Information:
Debian Release: wheezy/sid
APT prefers quantal-updates
APT policy: (500, 'quantal-updates'), (500, 'quantal-security'), (500,
'quantal')
Architecture: amd64 (x86_64)
Foreign Architectures: i386
Kernel: Linux 3.5.0-19-generic (SMP w/8 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)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=eeth
-----END PGP SIGNATURE-----
diff -Nru libunwind-1.0.1/debian/changelog libunwind-1.0.1/debian/changelog
diff -Nru libunwind-1.0.1/debian/patches/20122307_fix_setjmp.patch libunwind-1.0.1/debian/patches/20122307_fix_setjmp.patch
--- libunwind-1.0.1/debian/patches/20122307_fix_setjmp.patch 1970-01-01 01:00:00.000000000 +0100
+++ libunwind-1.0.1/debian/patches/20122307_fix_setjmp.patch 2012-07-30 10:10:02.000000000 +0100
@@ -0,0 +1,109 @@
+From 297d9cd07d7ea9b541fb13bffe418c8b2a8c3aa6 Mon Sep 17 00:00:00 2001
+From: Paul Pluzhnikov <ppluzhni...@google.com>
+Date: Tue, 15 Nov 2011 07:45:09 +0000
+Subject: Fix for failing test-setjmp
+
+On Mon, Nov 14, 2011 at 5:02 PM, Paul Pluzhnikov <ppluzhni...@google.com> wrote:
+
+> P.S. test-setjmp is failing for me (before or after the patch).
+> When I enable assertions (to confirm my new assertions are correct), I see:
+>
+> lt-test-setjmp: ../../src/dwarf/Gparser.c:754: apply_reg_state: \
+> Assertion `rs->reg[17].where == DWARF_WHERE_EXPR' failed.
+>
+> which likely explains that failure.
+
+The problem is actually two-fold:
+
+First, the loops in {sig,}longjmp.c are "do { ... } while (unw_step() >= 0);"
+
+But unw_step() returns 0 on reaching the end of the chain (_start),
+and the loop should stop there.
+
+The second problem is that with this commit:
+ http://repo.or.cz/w/glibc.git/commitdiff/c67da0b50e3d20f89d7bb352cd67dcf66d808e50
+glibc obfuscates value of SP in jmp_buf, so we might as well just give up.
+
+Patch attached.
+
+Thanks,
+--
+Paul Pluzhnikov
+---
+Index: libunwind/src/setjmp/longjmp.c
+===================================================================
+--- libunwind.orig/src/setjmp/longjmp.c 2012-07-23 10:59:17.449688000 +0100
++++ libunwind/src/setjmp/longjmp.c 2012-07-23 11:00:31.632589776 +0100
+@@ -34,6 +34,20 @@
+ #include "jmpbuf.h"
+ #include "setjmp_i.h"
+
++#if defined(__GLIBC__) && __GLIBC_PREREQ(2, 4)
++
++/* Starting with glibc-2.4, {sig,}setjmp in GLIBC obfuscates the
++ register values in jmp_buf by XORing them with a "random"
++ canary value.
++
++ This makes it impossible to implement longjmp, as we
++ can never match wp[JB_SP], unless we decode the canary first.
++
++ Doing so is possible, but doesn't appear to be worth the trouble,
++ so we simply defer to glibc longjmp here. */
++
++#else
++
+ void
+ _longjmp (jmp_buf env, int val)
+ {
+@@ -74,7 +88,7 @@
+
+ abort ();
+ }
+- while (unw_step (&c) >= 0);
++ while (unw_step (&c) > 0);
+
+ abort ();
+ }
+@@ -89,4 +103,6 @@
+ _longjmp (env, val);
+ }
+
++#endif /* __GLIBC__ */
++
+ #endif
+Index: libunwind/src/setjmp/siglongjmp.c
+===================================================================
+--- libunwind.orig/src/setjmp/siglongjmp.c 2012-07-23 10:59:17.449688000 +0100
++++ libunwind/src/setjmp/siglongjmp.c 2012-07-23 11:00:31.632589776 +0100
+@@ -35,6 +35,20 @@
+ # define _NSIG (_SIG_MAXSIG - 1)
+ #endif
+
++#if defined(__GLIBC__) && __GLIBC_PREREQ(2, 4)
++
++/* Starting with glibc-2.4, {sig,}setjmp in GLIBC obfuscates the
++ register values in jmp_buf by XORing them with a "random"
++ canary value.
++
++ This makes it impossible to implement longjmp, as we
++ can never match wp[JB_SP], unless we decode the canary first.
++
++ Doing so is possible, but doesn't appear to be worth the trouble,
++ so we simply defer to glibc siglongjmp here. */
++
++#else
++
+ void
+ siglongjmp (sigjmp_buf env, int val)
+ {
+@@ -96,7 +110,9 @@
+
+ abort ();
+ }
+- while (unw_step (&c) >= 0);
++ while (unw_step (&c) > 0);
+
+ abort ();
+ }
++
++#endif /* __GLIBC__ */
diff -Nru libunwind-1.0.1/debian/patches/20123007_disable_known_test_failures.patch libunwind-1.0.1/debian/patches/20123007_disable_known_test_failures.patch
--- libunwind-1.0.1/debian/patches/20123007_disable_known_test_failures.patch 1970-01-01 01:00:00.000000000 +0100
+++ libunwind-1.0.1/debian/patches/20123007_disable_known_test_failures.patch 2012-07-30 12:09:41.000000000 +0100
@@ -0,0 +1,19 @@
+Description: Some tests within the libunwind test suite
+ are known failures (see Origin and README for verification of this).
+ .
+ The package currently only executes tests for amd64:
+ - *-dyn1 tests use feature which are not avaliable on this arch
+Origin: http://lists.nongnu.org/archive/html/libunwind-devel/2012-07/msg00005.html
+
+Index: libunwind/tests/Makefile.am
+===================================================================
+--- libunwind.orig/tests/Makefile.am 2012-07-30 10:05:35.271168000 +0100
++++ libunwind/tests/Makefile.am 2012-07-30 11:57:02.332769610 +0100
+@@ -41,7 +41,6 @@
+ Gtest-init Ltest-init \
+ Gtest-concurrent Ltest-concurrent \
+ Gtest-resume-sig Ltest-resume-sig \
+- Gtest-dyn1 Ltest-dyn1 \
+ Gtest-trace Ltest-trace \
+ test-async-sig test-flush-cache test-init-remote \
+ test-mem test-setjmp test-ptrace \
diff -Nru libunwind-1.0.1/debian/patches/20123007_workaround_ltest-nocalloc.patch libunwind-1.0.1/debian/patches/20123007_workaround_ltest-nocalloc.patch
--- libunwind-1.0.1/debian/patches/20123007_workaround_ltest-nocalloc.patch 1970-01-01 01:00:00.000000000 +0100
+++ libunwind-1.0.1/debian/patches/20123007_workaround_ltest-nocalloc.patch 2012-07-30 10:21:11.000000000 +0100
@@ -0,0 +1,22 @@
+Description: Make dummy call to backtrace from thread creation to ensure
+ that libunwind per-thread state only gets initialized once resulting
+ in no futher calls to calloc.
+ .
+ This can create issues if libunwind is being used within a signal handler
+ (such as provided by google-perftools).
+Origin: https://groups.google.com/forum/#!msg/google-perftools/Df10Uy4Djrg/XzbESvLZhiYJ
+Forwarded: not-needed
+
+Index: libunwind/tests/Gtest-nocalloc.c
+===================================================================
+--- libunwind.orig/tests/Gtest-nocalloc.c 2012-07-30 10:05:35.271168000 +0100
++++ libunwind/tests/Gtest-nocalloc.c 2012-07-30 10:17:34.794107721 +0100
+@@ -123,6 +123,8 @@
+ if (pthread_key_create (&key, NULL))
+ panic ("FAILURE: unable to create key %d\n", i);
+ }
++ do_backtrace();
++ num_mallocs = num_callocs = 0;
+ foo1 ();
+ num_errors = num_mallocs + num_callocs;
+ if (num_errors > 0)
diff -Nru libunwind-1.0.1/debian/patches/20123007_workaround_test-ptrace.patch libunwind-1.0.1/debian/patches/20123007_workaround_test-ptrace.patch
--- libunwind-1.0.1/debian/patches/20123007_workaround_test-ptrace.patch 1970-01-01 01:00:00.000000000 +0100
+++ libunwind-1.0.1/debian/patches/20123007_workaround_test-ptrace.patch 2012-07-30 11:39:41.000000000 +0100
@@ -0,0 +1,16 @@
+Index: libunwind/tests/test-ptrace.c
+===================================================================
+--- libunwind.orig/tests/test-ptrace.c 2012-07-30 10:05:35.271168000 +0100
++++ libunwind/tests/test-ptrace.c 2012-07-30 11:39:37.309158623 +0100
+@@ -182,6 +182,11 @@
+
+ /* automated test case */
+ argv = args;
++
++ /* Unless the args array is 'walked' the child
++ process is unable to access it and dies with a segfault */
++ fprintf(stderr, "Automated test (%s,%s,%s,%s)\n",
++ args[0],args[1],args[2],args[3]);
+ }
+ else if (argc > 1)
+ while (argv[optind][0] == '-')
diff -Nru libunwind-1.0.1/debian/patches/series libunwind-1.0.1/debian/patches/series
--- libunwind-1.0.1/debian/patches/series 2012-09-17 06:32:11.000000000 +0100
+++ libunwind-1.0.1/debian/patches/series 2012-12-06 18:31:40.000000000 +0000
@@ -2,3 +2,7 @@
20120422_tests_Makefile_am.patch
20120704_fix_arm_ftbfs.patch
20120707_ia64_FTBFS.patch
+20122307_fix_setjmp.patch
+20123007_workaround_ltest-nocalloc.patch
+20123007_disable_known_test_failures.patch
+20123007_workaround_test-ptrace.patch
diff -Nru libunwind-1.0.1/debian/rules libunwind-1.0.1/debian/rules
--- libunwind-1.0.1/debian/rules 2012-09-17 06:32:11.000000000 +0100
+++ libunwind-1.0.1/debian/rules 2012-11-14 11:22:58.000000000 +0000
@@ -5,7 +5,6 @@
#export DH_VERBOSE=1
export DEB_BUILD_MAINT_OPTIONS = hardening=-all,+format
-DEB_BUILD_OPTIONS=nocheck
%:
dh $@ --parallel --with autoreconf
@@ -22,6 +21,15 @@
override_dh_installchangelogs:
dh_installchangelogs ChangeLog
+override_dh_auto_test:
+ifeq (,$(findstring nocheck, $(DEB_BUILD_OPTIONS)))
+# Test suite is only useful on amd64 architecture;
+# other ARCH's have quite a few know test faulures.
+ifeq (amd64, $(DEB_BUILD_ARCH))
+ $(MAKE) check
+endif # amd64
+endif # nocheck
+
override_dh_strip:
dh_strip -plibunwind8 --dbg-package=libunwind8-dbg
dh_strip -plibunwind-setjmp0 --dbg-package=libunwind-setjmp0-dbg
