buildroot/package/valgrind, branch 2016.11

valgrind: do not use host mpicc

2016-11-09T21:15:06+00:00

Valgrind may use 'mpicc' (from openmi project) to compile libmpiwrap-*.so. Without any option, it will detect and use 'mpicc' from host and generate libmpiwrap-*.so for host: $ file target/usr/lib/valgrind/libmpiwrap-arm-linux.so target/usr/lib/valgrind/libmpiwrap-arm-linux.so: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, not stripped We simply disable openmi support for valgrind. Signed-off-by: Jérôme Pouiller [Thomas: refactor with other unconditional CONF_OPTS.] Signed-off-by: Thomas Petazzoni

valgrind: bump version to 3.12.0

2016-10-26T20:39:02+00:00

See [1] for release notes. - removed 0003-mips-replace-addi-with-addiu.patch (applied upstream, see [2]) [1] http://valgrind.org/docs/manual/dist.news.html [2] https://bugs.kde.org/show_bug.cgi?id=356112 Signed-off-by: Peter Seiderer Signed-off-by: Thomas Petazzoni

valgrind: let the valgrind configure script detect TLS availability

2016-08-31T19:45:33+00:00

Back in 2005, in commit a2c326396a43ecbc8d02c3d815d4010a7ba2e004 ("update valgrind to the latest and greatest"), an explicit --disable-tls option was added. More recently, in commit 31a3f4bd54e12b8d6de286ab8fb6d9651990e2f5 ("valgrind: enable tls support") changed this to be conditional on BR2_GCC_ENABLE_TLS. However, the configure script of valgrind is perfectly capable of detecting TLS support, even in a cross-compilation case: it tries to compile a program that uses __thread and sees if it works. Since we're about to modify how BR2_GCC_ENABLE_TLS is handled, we'd better remove its usage from packages, and valgrind is the only package using this config option. Signed-off-by: Thomas Petazzoni Reviewed-by: Arnout Vandecappelle (Essensium/Mind)

valgrind: disable for MIPS soft-float

2016-05-24T11:16:08+00:00

As stated here [1], recent changes on the MIPS binutils sources have made it necessary for GCC to pass the -msoft-float to the assembler. Due to that, valgrind fails to build for MIPS soft-float when using a version of binutils >= 2.25 because its using some hard-float instructions. However, we cannot just disable it for the soft-float and binutils >= 2.25 combination since external toolchains don't provide information about the binutils version they use. So, instead, we simply disable it for soft-float. 1: https://gcc.gnu.org/ml/gcc-patches/2014-08/msg00905.html Fixes: http://autobuild.buildroot.net/results/5f5/5f576c7f8d56058a19ed0e7ff4b1ec620bcafb65/ Signed-off-by: Vicente Olivert Riera Signed-off-by: Peter Korsgaard

valgrind: fix musl compile

2016-02-13T17:26:45+00:00

Add musl libc detection (based on suggestions by Romain Naour and Arnout Vandecappelle) and add some tweaks for musl support (based on OpenWRT patch [1]). Fixes ([2]): checking the glibc version... unsupported version configure: error: Valgrind requires glibc version 2.2 or later Patch suggested upstream (see [3]). [1] https://dev.openwrt.org/browser/trunk/package/devel/valgrind/patches/200-musl_fix.patch?rev=46302 [2] http://autobuild.buildroot.org/results/7b0/7b048ba58918f0a08498c61327fcf35a85a84837 [3] https://bugs.kde.org/show_bug.cgi?id=359202 Signed-off-by: Peter Seiderer Reviewed-by: Arnout Vandecappelle (Essensium/Mind) Signed-off-by: Thomas Petazzoni

valgrind: update 0003-mips-replace-addi-with-addiu.patch

2015-12-17T21:01:12+00:00

This patch has been updated in the upstream bug report, so let's update it in Buildroot as well. It will fix the build issue for mips64r6: m_dispatch/dispatch-mips64-linux.S:199: Error: opcode not supported on this processor: mips64r6 (mips64r6) `daddi $13,$13,8' Fixes: http://autobuild.buildroot.net/results/6ca/6ca3a31d1542fbbb44238e296565b40d6afcd5fa/ Signed-off-by: Vicente Olivert Riera Signed-off-by: Thomas Petazzoni

valgrind: add a patch to fix MIPS R6 by replacing ADDI with ADDIU

2015-11-30T16:08:10+00:00

The ADDI instruction has been removed for MIPS R6, so valgrind fails to build for this revision level because it uses that assembly instruction. Using ADDIU instead fixes the problem. The patch has been sent upstream: https://bugs.kde.org/show_bug.cgi?id=356112 Fixes: http://autobuild.buildroot.net/results/488/488937704b3b6bae8daed7b24da8c740b56f4d84/ Signed-off-by: Vicente Olivert Riera Signed-off-by: Peter Korsgaard

valgrind: enable tls support

2015-11-02T14:24:57+00:00

Tested with example program from [1] with qemu_x86_64. [1] http://valgrind.10908.n7.nabble.com/Thread-local-storage-TLS-support-td40815.html Signed-off-by: Peter Seiderer Signed-off-by: Thomas Petazzoni

valgrind: bump version to 3.11.0

2015-10-31T14:42:36+00:00

- rebase patch 0001-workaround-SIGSEGV-on-PPC.patch - remove upstream applied patches 0002-Fix-configure-for-Linux-kernel-4.0-rc1.patch 0004-configure.ac-Generalize-glibc-version-check.patch - rebase 0003-add-missing-ptrace-getsiginfo-on-powerpc.patch (convert to git patch format, rename to 0002-Define-PTRACE_GETSIGINFO-on-PowerPC-when-not-availab.patch) - remove legacy VALGRIND_AUTORECONF=YES (no more patch touching confgure.ac) - add '--disable-ubsan' for new undefined behaviour sanitiser option Signed-off-by: Peter Seiderer Signed-off-by: Thomas Petazzoni

valgrind: build for the right MIPS ISA revision level

2015-10-03T13:26:21+00:00

When Valgrind detects a 32-bit MIPS architecture, it forcibly adds -march=mips32 to CFLAGS; when it detects a 64-bit MIPS architecture, it forcibly adds -march=mips64. This causes Valgrind to be built always for the first ISA revision level (R1), even when the user has configured Buildroot for the second ISA revision level (R2). Since R2 is backwards compatible with R1, you can run a Valgrind built for R1 in an R2 core. This is why nobody noticed about this problem, or at least nobody complained. But, since (I hope) we will support R6 in Buildroot in the near future, this problem will become very important because R6 is not backwards compatible with R1 or R2, so building Valgrind for R1 when your target is R6 will result in a non-working Valgrind. Override the CFLAGS variable (which Valgrind appends to its CFLAGS) and pass the right -march option, so they take precedence over Valgrind's wrongfully detected value. Signed-off-by: Vicente Olivert Riera Reviewed-by: Arnout Vandecappelle (Essensium/Mind) Reviewed-by: "Yann E. MORIN" Signed-off-by: Peter Korsgaard