dmalloc directly calls into $(LD) to generate a shared library our of
the static one.
To detect what command it should run, ./configure tries various
incantations of ld with various command line options until one does not
fail. One of those is (basically):
$(LD) --whole-archive -o contest.o.t contest.a
This makes ./configure conclude what the command to link a shared
library in the Makefile should be, and thus stores that in a variable:
shlinkargs='$(LD) --whole-archive -o $@'
... which is then AC_SUBST()ed into Makefile.in with a rule like:
$(SHLIB): $(LIBRARY)
@shlinkargs@ $(LIRARY)
which once substiuted, gives:
$(SHLIB): $(LIBRARY)
$(LD) --whole-archive -o $@ $(LIRARY)
However, when SSP is enabled, the __stack_chk_fail_local and co symbols
are provided by additional libraries or object files, and that is the
responsibility of gcc to pass those when linking. But as dmalloc
directly calls ld, it misses those.
Changing dmalloc to use $(CC) is not trivial, however.
First, we can't pass LD=$(TARGET_CC), otherwise the whole package
explodes [0]: indeed --whole-archive is unknown to gcc, so it must be
passed as -Wl,--whole archive instead. So we'd need to add a new test
that uses $(CC), like so:
$(CC) -Wl,--whole-archive -o contest.o.t contest.a
However, in that case, gcc does pass additional libs/objs (like, for
eample, the SSP ones) to the linker. But then those are also included
in the whole-archive section. This causes the linker to add all the
symbols form those libs/objs, even those not needed for SSP; on some
archs, like PPC, that may require floating point symbols (__muldiv3 et
al.) which are in another library, and thus the linker can't find them.
The proper solution wouild be to add -Wl,--no-whole-archive, but that
would have to be added _after_ the library we want to link, i.e.e we
should be able to evntually run:
$(CC) -Wl,--whole-archive -o $@ $(LIRARY) -Wl,--no-whole-archive
That would require that we introduce a new variable that is added
_after_ the $(LIBRARY), e.g. @shlinkargs_post@ or so...
This is a bigger endeavour than we want to pursue...
Since dmalloc is a debugging utility, it is not supposed to go into
production builds, and during debugging, it would not be surprising that
it needs to poke around arrays to debug them.
So, we go the easier route: disable SSP altogether.
[0] with lots of nice colors, but that's not the point, is it?
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Our dmalloc packaging is very old and carries historical baggage and
idiosyncracies that we have long stopped doing in the rest ofthe code
base.
Drop our post-patch hooks that seds the configure and Makefile.in files,
and add patches (that could be upstreamed one day).
We provide the results in the environment, like would be done with
actual autoconf cache variables (ac_cv_*).
Note: those are the result of cleaning up for further patches that did
not manifest because it was too complex to add proper SSP support to
dmalloc (instead, we're going to forcibly disable it in the following
commit).
Note-2: those patches have not been submitted upstream, as it's mostly
dead: even though there's been some commit activities recently-ish, there
has been no review or comments or the many PR pending for many years
now.
Note-3: we patch both configure and configure.ac, rather than
autoreconf, for two reasons: 1. the both are in the upstream git tree,
so submitting these patches would require patching both, and 2. dmalloc
does not autoreconf nicely out of the box, and it was deemed too much
hassle to fix that in addition.
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Fix strndup detection as already done for strdup in commit
f9287b263f to avoid the following build
failure raised since bump to version 5.6.5 in commit
736932c715:
In file included from /nvmedata/autobuild/instance-0/output-1/host/mipsel-buildroot-linux-gnu/sysroot/usr/include/string.h:630:0,
from append.c:34:
dmalloc.h:396:7: error: expected identifier or '(' before '__extension__'
char *strndup(const char *string, const DMALLOC_SIZE max_len);
^
Fixes:
- http://autobuild.buildroot.org/results/2164dac43e2607a3570e8f98957a70f579d06cd3
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Fix the following build failure raised since bump to version 5.6.5 in
commit 736932c715:
In file included from /home/buildroot/autobuild/instance-0/output-1/host/arm-buildroot-linux-gnueabi/sysroot/usr/include/string.h:631:0,
from arg_check.c:31:
dmalloc.h:377:7: error: expected identifier or '(' before '__extension__'
char *strdup(const char *string);
^
This build failure is raised because 0004-fix-strdup.patch was wrongly
removed and because configure.ac wrongly assumes that strdup macro is
not available when cross-compiling
As configure.ac can't be patched because autoreconf is not working,
patch configure as already done in DMALLOC_POST_PATCH. An upstream patch
on configure.ac will be sent as soon as upstream merges this old PR:
https://github.com/j256/dmalloc/pull/44
Fixes:
- http://autobuild.buildroot.org/results/00565c4b0618ea5f553d8c22284637b6574e4a93
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
- Drop second to fifth patches (already in version)
- License is ISC since version 5.6.2 and
eda734a8ba
- Update indentation in hash file (two spaces)
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Drop first patch and pass -fPIC to configure to fix the following build
failure on ARC:
ERROR: architecture for "/usr/lib/libdmalloc.so" is "ARCompact", should be "ARCv2"
ERROR: architecture for "/usr/lib/libdmallocth.so" is "ARCompact", should be "ARCv2"
ERROR: architecture for "/usr/lib/libdmallocthcxx.so" is "ARCompact", should be "ARCv2"
ERROR: architecture for "/usr/lib/libdmallocxx.so" is "ARCompact", should be "ARCv2"
This build failure is due to the following configure error:
checking shared library link args... ./configure: line 4467: 10229 Segmentation fault ( ${LD-ld} -shared --whole-archive -soname conftest.so -o conftest.so.t conftest.a ) 2>&5
/home/buildroot/autobuild/run/instance-0/output-1/host/bin/arc-buildroot-linux-gnu-ld -G -o $@.t
This configure error is due to missing -fPIC:
configure:4392: checking shared library link args
configure:4398: /home/buildroot/autobuild/run/instance-0/output-1/host/bin/arc-buildroot-linux-gnu-gcc -c -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -Os -g2 -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 conftest.c >&5
configure:4404: $? = 0
configure:4408: test -z
|| test ! -s conftest.err
configure:4411: $? = 0
configure:4414: test -s conftest.o
configure:4417: $? = 0
/home/buildroot/autobuild/run/instance-0/output-1/host/bin/arc-buildroot-linux-gnu-ld: conftest.a(conftest.o): relocation R_ARC_32_ME against `__stack_chk_guard' can not be used when making a shared object; recompile with -fPIC
/home/buildroot/autobuild/run/instance-0/output-1/host/bin/arc-buildroot-linux-gnu-ld: BFD (GNU Binutils) 2.33.50.20191002 assertion fail elf32-arc.c:1805
/home/buildroot/autobuild/run/instance-0/output-1/host/bin/arc-buildroot-linux-gnu-ld: unrecognized option '-all'
/home/buildroot/autobuild/run/instance-0/output-1/host/bin/arc-buildroot-linux-gnu-ld: use the --help option for usage information
configure:4475: result: /home/buildroot/autobuild/run/instance-0/output-1/host/bin/arc-buildroot-linux-gnu-ld -G -o $@.t
Fixes:
- http://autobuild.buildroot.org/results/65677d889c27649e1f3ca1f3b6c70df7c89779f6
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Building dmalloc on Microblaze with optimization enabled and gcc < 8.x
fails with:
Error: operation combines symbols in different segments
This is due to gcc bug 63261 (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63261).
To avoid this, the dmalloc package has a !BR2_microblaze
dependency. However, gcc bug 63261 only triggers when optimization is
enabled, so we can work around the issue by passing -O0, which is what
we do in other Buildroot packages to work around this bug.
So, this commit passes -O0 when BR2_TOOLCHAIN_HAS_GCC_BUG_63261, and
re-enables dmalloc on Microblaze.
Signed-off-by: Giulio Benetti <giulio.benetti@micronovasrl.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
To be consistent with the recent change of FOO_MAKE_OPT into FOO_MAKE_OPTS,
make the same change for FOO_CONF_OPT.
Sed command used:
find * -type f | xargs sed -i 's#_CONF_OPT\>#&S#g'
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Fix libdmalloc*.so files being deleted from the staging directory as the
package is installed into the target.
Signed-off-by: Reuben Dowle <reuben.dowle at 4rf.com>
Finally get rid of all := used for variable definitions in packages,
as we suggest in our manual and during the review of new packages.
While I was at it, I also sometimes added a few missing new lines
between the header and the first variable definition.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Thanks to the pkgparentdir and pkgname functions, we can rewrite the
AUTOTARGETS macro in a way that avoids the need for each package to
repeat its name and the directory in which it is present.
[Peter: pkgdir->pkgparentdir]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
The selection of linuxthreads, linuxthreads old or NPTL doesn't make a
lot of sense for external toolchains. So, instead, we :
* Introduce an hidden BR2_TOOLCHAIN_HAS_THREADS option, which must be
selected by toolchain specific options when thread support is
available. Package needing to test thread support should use this
option.
* Move the none/linuxthreads/linuxthreads old/NPTL selection to
Buildroot internal toolchain configuration.
* Add an option in external toolchain to tell if thread support is
available or not in the external toolchain. We assume that glibc
without threads is not possible, as Ulrich Drepper said in
http://sourceware.org/ml/libc-alpha/2005-08/msg00091.html
ffmpeg, dmalloc and openvpn are fixed to use the new
BR2_TOOLCHAIN_HAS_THREADS option. For openvpn, --enable-threads=posix
is no longer used, as the configure script doesn't even understand
this option.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Changes by Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
* Target installation is the default behaviour
* No installation to staging is the default behaviour
* Remove DMALLOC_TARGET_BINARY variable
Signed-off-by: cmchao <cmchao@gmail.com>
We have been passing -q to ./configure when using 'make -s' for
packages using Makefile.autotools.in for some time. Do the same
for packages using autotools, but not using the
Makefile.autotools.in infrastructure, taking care to not do it
for packages with hand written configure scripts.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
A C library will have been built by the toolchain makefiles, so there is no
need for packages to explicitly depend on uclibc.
Signed-off-by: Will Newton <will.newton@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
they should be configured with --prefix=/usr and we then need to use
make DESTDIR=$(STAGING_DIR) install to get things installed into the
staging directory. The current situation for many packages, which use
--prefix=$(STAGING_DIR) results in the staging_dir paths getting compiled
into the binary itself.
This also adds in a pile of libtool fixups. Between broken pkgconfig,
broken libtool handling, and broken --prefix settings, its a wonder
things have worked as well as they have up till now.
-Erik
