Sourceware infrastructure community updates for Q1 2025
Sourceware has provided the infrastructure for core toolchain
and developer tools projects for more than 25 years.
https://sourceware.org/sourceware-25-roadmap.html
Over the last couple of years, Sourceware has transformed from a
purely volu
Hi hackers,
TLDR; When using https://patchwork.sourceware.org or Bunsen
https://builder.sourceware.org/testruns/ you might now have to enable
javascript. This should not impact any scripts, just browsers (or bots
pretending to be browsers). If it does cause trouble, please let us
know. If this wor
Hi Aaron, Hi Constantine,
On Thu, 2025-04-10 at 10:52 -0400, Aaron Merey wrote:
> handle_dynamic_symtab can attempt to read symbol and version data from
> file offset 0 if the associated DT_ tags aren't found.
>
> Fix this by only reading symbol and version data when non-zero file
> offsets have
Hi Aaron,
On Tue, Apr 08, 2025 at 01:34:28PM -0400, Aaron Merey wrote:
> LGTM. The optional main function approach opens up a lot of
> possibilities for testing static functions throughout all elfutils
> libraries.
I hope to use the same mechanism to write some fuzzing wrappers so you
don't have
Hi Dmitry,
On Fri, Apr 04, 2025 at 04:07:26PM +0300, Dmitry V. Levin wrote:
> On Fri, Apr 04, 2025 at 01:55:48PM +0200, Mark Wielaard wrote:
> > So to be (pedantically) correct should we include the attached?
>
> Yes, this should be fine, along with
>
> --- a/lib/crc32.
Hi Dmitry,
On Fri, Apr 04, 2025 at 04:04:21PM +0300, Dmitry V. Levin wrote:
> On Fri, Apr 04, 2025 at 01:38:01PM +0200, Mark Wielaard wrote:
> > But maybe we should just use AC_CHECK_HEADERS([stdatomic.h]) and drop
> > the AC_COMPILE_IFELSE trick?
>
> Sure. And if we want
Hi Michael,
On Thu, 2025-03-13 at 12:05 +0100, Michael Trapp wrote:
> Use MHD_OPTION_SOCK_ADDR to bind the http listen socket to a single address.
> The address can be any IPv4 or IPv6 address configured on the system:
> --http-addr=127.0.0.1
> --http-addr=::1
> --http-addr='ANY_ACTIVE
Hi Dmitry,
On Thu, 2025-04-03 at 19:09 +0300, Dmitry V. Levin wrote:
> On Thu, Aug 29, 2019 at 03:16:12PM +0200, Mark Wielaard wrote:
> > From: Jonathon Anderson
> >
> > Uses the stdatomic.h provided by FreeBSD when GCC doesn't (ie. GCC < 4.9)
> >
&
Hi,
On Fri, 2025-04-04 at 12:39 +, buil...@sourceware.org wrote:
> A new failure has been detected on builder elfutils-debian-armhf while
> building elfutils.
>
> Full details are available at:
> https://builder.sourceware.org/buildbot/#/builders/6/builds/403
>
> Build state: failed tes
> of including config.h twice for all files that include both
> and "system.h".
So to be (pedantically) correct should we include the attached?
Thanks,
Mark
From f14c00adb24bf15059821ca972e5d20a58b893fa Mon Sep 17 00:00:00 2001
From: Mark Wielaard
Date: Fri, 4 Apr 2025 13:50:04 +02
Hi Samuel,
On Thu, 2025-04-03 at 17:24 +1000, Samuel Zeter wrote:
> When printing section headers, also include a key to what each flag
> is at the end of the section header output.
>
> Signed-off-by: Samuel Zeter
Thanks, looks good. Committed with a slightly tweaked commit message:
commit ef1
CHECK.
(main): Likewise.
Signed-off-by: Mark Wielaard
---
libdw/Makefile.am | 11
libdw/dwarf.h | 2 +
libdw/dwarf_default_lower_bound.c | 2 +
libdw/dwarf_srclang.c | 87 ++-
4 files changed, 101 insertions(
Hi Michael,
On Thu, 2025-03-27 at 17:06 +0100, Michael Trapp wrote:
> Use MHD_OPTION_SOCK_ADDR to bind the http listen socket to a single address.
> The address should be an IPv4 or IPv6 address configured on the system:
> --listen-address=127.0.0.1
> --listen-address=::1
> --listen-ad
On Fri, Mar 14, 2025 at 01:40:42AM +0100, Mark Wielaard wrote:
> The Sourceware Project Leadership Committee would like to know who our
> users are, which hosted projects they feel part of, what services they
> rely on and what the priorities should be for new initiatives.
&
Hi Michael,
On Thu, Mar 27, 2025 at 04:43:49PM +0100, Michael Trapp wrote:
> Use MHD_OPTION_SOCK_ADDR to bind the http listen socket to a single address.
> The address should be an IPv4 or IPv6 address configured on the system:
> --listen-address=127.0.0.1
> --listen-address=::1
> --li
Hi Aaron,
On Thu, Mar 27, 2025 at 12:07:18AM -0400, Aaron Merey wrote:
> Ensure that dwarf_lock is held before accessing next_tu_offset and
> next_cu_offset.
>
> This fixes a TOCTOU bug in __libdw_findcu that causes NULL to be
> incorrectly returned.
Could you explain what the issue is in a few
Using [[ is a bashism. We can just use [ since this is a simple
comparision.
* tests/thread-safety-subr.sh: Use if [ instead of if [[.
Signed-off-by: Mark Wielaard
---
tests/thread-safety-subr.sh | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tests/thread-safety
Hi Michael,
On Thu, Mar 20, 2025 at 02:58:34PM +, Trapp, Michael wrote:
> > Am 19.03.2025 um 13:35 schrieb Mark Wielaard :
> > The code itself does look ok, although I think it could be simplified a
> > little if we go for something like --listen-local only (assuming th
Hi Aaron,
On Sun, 2025-03-16 at 21:51 -0400, Aaron Merey wrote:
> * libdw/dwarf_end.c (cu_free): Free str_off_base_lock.
> * libdw/libdwP.h (struct Dwarf_CU): Add str_off_base_lock member.
> (str_offsets_base_off): Add locking.
> * libdw/libdw_findcu.c (__libdw_intern_next_
Hi Aaron,
On Sun, 2025-03-16 at 21:51 -0400, Aaron Merey wrote:
> * libdw/dwarf_begin_elf.c (dwarf_begin_elf): Init macro_lock.
> * libdw/dwarf_end.c (cu_free): Free src_lock.
> (dwarf_end): Free macro_lock.
> * libdw/dwarf_getsrcfiles.c (dwarf_getsrcfiles): Use src_lock.
>
Hi Serhei,
On Sun, 2025-03-16 at 19:14 -0400, Serhei Makarov wrote:
> New data structure to coordinate caching Elf data among multiple Dwfl
> structs attached to different processes. Meant to reduce the overhead
> for profilers that use elfutils for unwinding.
>
> The caching is well-justified, a
Hi Serhei,
On Sun, Mar 16, 2025 at 07:13:00PM -0400, Serhei Makarov wrote:
> Renaming to dwfl_set_initial_registers_thread.
>
> This callback was private to one file, but now that other
> tools (eu-stacktrace, sysprof via the dwfl_perf_sample_getframes) are
> invoking the ebl set_initial_register
Hi Serhei,
On Sun, Mar 16, 2025 at 07:14:11PM -0400, Serhei Makarov wrote:
> Since libebl is a private interface, subsequent patches in the series
> introduce another api wrapping the libebl perf register handling. In
> this patch, add an interface to access the preferred set of registers
> that
Hi Serhei,
On Sun, Mar 16, 2025 at 07:13:41PM -0400, Serhei Makarov wrote:
> Change the sample_set_initial_registers callback in eu-stacktrace to
> use the proper libebl ebl_set_initial_registers_sample function.
>
> * src/Makefile.am (stacktrace_LDADD): Add libebl.
> * src/stacktrace.c (sample_r
Hi Serhei,
On Sun, 2025-03-16 at 19:12 -0400, Serhei Makarov wrote:
> First patch of a series that reworks eu-stacktrace functionality
> into a library interface for other profiling tools.
>
> * * *
>
> As it happens, Linux perf_events and DWARF can prescribe completely
> different layouts for t
Hi Hengqi,
On Tue, 2025-03-11 at 13:27 +0800, Hengqi Chen wrote:
> I want to ask you a question regarding elf internals.
> How to calculate a symbol's file offset (which is kernel uprobe expects)
> in an elf (executable or shared object)?
Could you point me to a description of what uprobe expects
Hi Samuel,
On Mon, 2025-03-17 at 17:39 +1000, Samuel Zeter wrote:
> When printing section headers, also include a key to what each flag
> is at the end of the section header output.
Did you run make check after your patch? It seems various tests fail
because they aren't expecting the new flag inf
Hi Serhei,
On Mon, 2025-03-17 at 10:22 -0400, Serhei Makarov wrote:
> On Sun, Mar 16, 2025, at 7:12 PM, Serhei Makarov wrote:
> > +bool
> > +x86_64_set_initial_registers_sample (const Dwarf_Word *regs, uint32_t
> > n_regs,
> > +uint64_t regs_mask, uint32_t abi,
> >
The Sourceware Project Leadership Committee would like to know who our
users are, which hosted projects they feel part of, what services they
rely on and what the priorities should be for new initiatives.
https://nextcloud.sfconservancy.org/apps/forms/s/xmGgmJFzSb2FZNd58cXMtAZp
The survey will ru
.
* tests/Makefile.am (check_PROGRAMS): Add dwarf_language_lower_bound.
(TESTS): Likewise.
(dwarf_language_lower_bound_LDADD): New variable.
Signed-off-by: Mark Wielaard
---
NEWS | 2 +
libdw/dwarf_aggregate_size.c | 8 +-
libdw
.
(dwarf_srclang): If there is no DW_AT_language, try
DW_AT_language_name and DW_AT_language_version, use
language_to_srclang.
Signed-off-by: Mark Wielaard
---
libdw/dwarf_srclang.c | 221 +-
1 file changed, 218 insertions(+), 3 deletions
dwarf_lname_name.
* run-readelf-lnames.sh: New test.
* testfile-lnames.bz2: New testfile.
* tests/Makefile.am (TESTS): Add run-readelf-lnames.sh.
(EXTRA_DIST): Add run-readelf-lnames.sh and
testfile-lnames.bz2.
Signed-off-by: Mark Wielaard
---
libdw/dwarf.h
-dwarf.awk: Handle DW_LNAME.
* libdw/dwarf.h: Add DW_AT_language_name and DW_AT_language_version
plus all currently defined DW_LNAME constants. Add DW_LANG_V and
DW_LANG_Algol68.
* libdw/dwarf_default_lower_bound.c: Add DW_LANG_V and
DW_LANG_Algol68.
Signed-off-by: Mark Wielaard
Hi Aaron,
On Wed, Feb 19, 2025 at 11:36:39PM -0500, Aaron Merey wrote:
> Like --enable-valgrind but uses helgrind instead of memcheck.
>
> If both --enable-valgrind and --enable-helgrind are given then
> helgrind takes priority.
>
> --enable-helgrind requires --enable-valgrind-annotations.
>
>
Hi Frank,
On Thu, 2025-02-27 at 18:47 -0500, Frank Ch. Eigler wrote:
> > We should make dwarf_getsrcfiles and dwarf_getsrclines independent so
> > callers that onle need the file table don't need to "pay" for parsing
> > the whole line table. (I thought we had a bug for that, but cannot
> > find i
Hi Aaron,
On Wed, Feb 19, 2025 at 11:36:38PM -0500, Aaron Merey wrote:
> * libdw/dwarf_end.c (cu_free): Free str_off_base_lock.
> * libdw/libdwP.h (struct Dwarf_CU): Add str_off_base_lock member.
> (str_offsets_base_off): Add locking.
> * libdw/libdw_findcu.c (__libdw_inter
e dwarf_macro_getsrclines locking added in this patch.
That makes sense.
> On Wed, Feb 12, 2025 at 8:17 AM Mark Wielaard wrote:
> >
> > Hi Aaron,
> >
> > On Tue, 2025-02-04 at 16:50 -0500, Aaron Merey wrote:
> > > * libdw/dwarf_getsrcfiles.
Hi Aaron,
On Wed, Feb 19, 2025 at 11:36:36PM -0500, Aaron Merey wrote:
> Change type of dwarf_lock to mutex in order to take advantage of
> built-in support for recursive locking.
>
> * lib/locks.h: Add macros for locking, unlocking, initializing
> and destroying mutexes.
> * li
Hi Anton,
On Thu, Feb 13, 2025 at 08:19:44PM +0300, Anton Moryakov wrote:
> Static analyzer reported:
> Return value of a function 'elf_getdata' is dereferenced at unstrip.c:1977
> without checking for NULL, but it is usually checked for this function
> (97/101).
>
> Corrections explained:
> - A
Hi Anton,
On Thu, Feb 13, 2025 at 07:52:00PM +0300, Anton Moryakov wrote:
> Static analyzer reported:
> Return value of a function 'gelf_getehdr' is dereferenced at readelf.c:12443
> without checking for NULL, but it is usually checked for this function
> (53/54).
I can see how a static analyzer
rk
>From dfa7b2c23ddabcba2a4972fa67d3c670ae31f1ee Mon Sep 17 00:00:00 2001
From: Mark Wielaard
Date: Thu, 27 Feb 2025 21:22:49 +0100
Subject: [PATCH] readelf: Use section_name instead of elf_strptr in
print_debug_frame_section
All other print_debug_* functions use section_name (ebl, shdr) t
Hi Anton,
On Thu, 2025-02-13 at 18:57 +0300, Anton Moryakov wrote:
> Report of the static analyzer:
> DEREF_OF_NULL.RET Pointer, returned from function 'elf_getarhdr' at
> readelf.c:13551,
> may be NULL and is dereferenced at readelf.c:13553.
>
> Corrections explained:
> - Added a NULL check fo
Hi Anton,
The subject isn't super helpful unless you know the specific
terminology of the statuc analyzer you are using. It would be better to
say something like:
ar: check whether elf_getarhdr returns NULL
On Thu, 2025-02-13 at 18:00 +0300, Anton Moryakov wrote:
> Report of the static analyze
st-manyfuncs.sh: Likewise.
* tests/test-manyfuncs.c: New test.
Signed-off-by: Mark Wielaard
---
libelf/elf32_getshdr.c | 14 --
libelf/elf_begin.c | 26 ---
libelf/elf_scnshndx.c| 52 -
libelf/libelf.h | 9 +-
libelf/libelfP.h
Hi Markus,
On Fri, Feb 21, 2025 at 11:19:34AM -0800, Markus Mayer wrote:
> On Aarch64, Linux is using Pointer Authentication Code (PAC) for pointer
> authentication.[1] The struct "user_pac_mask" has been part of the Linux
> kernel since version 5.0 as part of this feature. However, older kernels
Hi,
On Thu, Feb 20, 2025 at 08:18:15PM +, buil...@sourceware.org wrote:
> A new failure has been detected on builder elfutils-snapshots-coverage while
> building elfutils.
>
> Full details are available at:
> https://builder.sourceware.org/buildbot/#/builders/250/builds/210
>
> Build st
Hi Frank,
On Thu, Feb 20, 2025 at 02:18:04PM -0500, Frank Ch. Eigler wrote:
> > [...]
> > This does sounds like a bug in glibc sscanf. I cannot find a
> > description of what exactly happens with 'm' modifier allocated
> > buffers on error. So I can imagine a double free if sscanf frees the
> > bu
One of the (libcurl) libraries tries to close invalid file descriptors
on fedora s390. Skip running the testsuite under valgrind for now.
---
builder/master.cfg | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/builder/master.cfg b/builder/master.cfg
index 0ccd23df9800..7a55e
Hi Frank,
On Tue, Feb 18, 2025 at 10:30:44PM -0500, Frank Ch. Eigler wrote:
> Planning to commit this shortly:
>
> commit a71bac67f4705b84368b71f5ece54deedaa1abf1 (HEAD -> master1)
> Author: Frank Ch. Eigler
> Date: Tue Feb 18 22:09:12 2025 -0500
>
> debuginfod-client: correct invalid fre
Hi Aaron,
On Thu, Feb 13, 2025 at 03:35:03PM -0500, Aaron Merey wrote:
> On Mon, Feb 10, 2025 at 1:27 PM Mark Wielaard wrote:
> > When dynsym/str was read through eu-readelf --dynamic by readelf
> > process_symtab the string data was not validated, possibly printing
> > un
kes a
message string to display.
(INTERNAL_ERROR): Use INTERNAL_ERROR_MSG with elf_errmsg (-1).
(remove_debug_relocations): Check the sh_link referenced
section is real and isn't a SHT_NOBITS section.
https://sourceware.org/bugzilla/show_bug.cgi?id=32673
Signed-of
d yet.
* libelf/elf_strptr.c (elf_strptr): Check strscn->rawdata_base
is not NULL.
https://sourceware.org/bugzilla/show_bug.cgi?id=32672
Signed-off-by: Mark Wielaard
---
libelf/elf_strptr.c | 10 +++---
1 file changed, 7 insertions(+), 3 deletions(-)
diff --git a/libelf/elf_
Hi Aaron,
On Tue, 2025-02-04 at 16:50 -0500, Aaron Merey wrote:
> thread-safety-subr.sh contains subroutine check_thread_safety_enabled
> which skips further testing when USE_LOCKS is false.
>
> * tests/Makefile.am (EXTRA_DIST): Add thread-safety-subr.sh.
> * tests/thread-safety-subr.
Hi Aaron,
On Tue, 2025-02-04 at 16:50 -0500, Aaron Merey wrote:
> Like --enable-valgrind but uses helgrind instead of memcheck.
>
> If both --enable-valgrind and --enable-helgrind are given then
> helgrind takes priority.
>
> --enable-helgrind requires --enable-valgrind-annotations.
>
> *
Hi Aaron,
On Tue, 2025-02-04 at 16:50 -0500, Aaron Merey wrote:
> * libdw/dwarf_offdie.c (__libdw_offdie): Use dwarf_lock.
>
> Signed-off-by: Aaron Merey
> ---
> v2 changes: Lock only for the duration of __libdw_findcu.
__libdw_findcu already has internal locking.
Isn't that enough?
> l
Hi Aaron,
On Tue, 2025-02-04 at 16:50 -0500, Aaron Merey wrote:
> * libdw/dwarf_getsrcfiles.c (dwarf_getsrcfiles): Use dwarf_lock.
> * libdw/dwarf_getsrclines.c (dwarf_getsrclines): Use dwarf_lock.
>
> Signed-off-by: Aaron Merey
> ---
> v2 changes: Combined from v1 patches 04/15 and
Hi Aaron,
On Tue, 2025-02-04 at 16:50 -0500, Aaron Merey wrote:
> * libdw/dwarf_getsrclines.c (read_srcfiles): Initialize Dwarf
> member.
> * libdw/dwarf_filesrc.c (dwarf_filesrc): Use dwarf_lock.
> * libdw/libdwP.h (struct Dwarf_Files_s): Add Dwarf member.
>
> Signed-off-
On Tue, 2025-02-04 at 16:50 -0500, Aaron Merey wrote:
> Change type of dwarf_lock to mutex in order to take advantage of
> built-in support for recursive locking.
>
> * lib/locks.h: Add macros for locking, unlocking, initializing
> and destroying mutexes.
> * libdw/dwarf_begin_el
Hi Anton,
On Mon, 2025-02-03 at 14:45 +0100, Mark Wielaard wrote:
> So the code path where prefix != NULL also implies suffix != NULL.
>
> Maybe the code needs a comment, check or assert to verify this?
I added an assert.
Pushed,
Mark
From 316077dd4b34c90f7e4c4228395a580e99239347 M
x27;t try to gnu decompress a
section without a name.
(print_string_section): Likewise.
https://sourceware.org/bugzilla/show_bug.cgi?id=32656
Signed-off-by: Mark Wielaard
---
src/readelf.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/readelf.c b/src/readelf.c
gelf_getphdr returns NULL.
https://sourceware.org/bugzilla/show_bug.cgi?id=32655
Signed-off-by: Mark Wielaard
---
src/readelf.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/readelf.c b/src/readelf.c
index c43fda35075a..21b92a08a6b6 100644
--- a/src/readelf.c
+++ b
://sourceware.org/bugzilla/show_bug.cgi?id=32654
Signed-off-by: Mark Wielaard
---
lib/system.h| 27 +++
libelf/elf_strptr.c | 18 --
src/readelf.c | 18 +++---
3 files changed, 42 insertions(+), 21 deletions(-)
diff --git a/lib/system.h b/lib
abbrev into abbrevp on success.
* libdw/libdw.h (dwarf_offabbrev): Document return values.
* libdw/libdwP.h (__libdw_getabbrev): Don't take Dwarf_Abbrev
result argument.
https://sourceware.org/bugzilla/show_bug.cgi?id=32650
Signed-off-by: Mark Wielaard
---
libdw/dw
Hi Anton,
I changed the indention from using spaces to tabs and rewrote the
subject line to better explain the change.
Pushed as attached.
Thanks,
Mark
>From 6c76ce7331e82a6fc6c2004e8e6ae51ca91f51d7 Mon Sep 17 00:00:00 2001
From: Anton Moryakov
Date: Wed, 5 Feb 2025 00:50:23 +0300
Subject: [P
Hi Aaron,
On Tue, Feb 04, 2025 at 05:00:21PM -0500, Aaron Merey wrote:
> On Mon, Feb 3, 2025 at 8:36 AM Mark Wielaard wrote:
> > So I don't think close should be called unconditionally. Maybe the
> > correct solution is to do if (!close_fd) close (fd); after elf_end?
Hi Anton,
On Sat, Feb 01, 2025 at 02:21:24AM +0300, Anton Moryakov wrote:
> Report of the static analyzer:
> Pointer, returned from function 'elf_getarhdr' at
> objdump.c:314, may be NULL and is dereferenced at
> objdump.c:317. (CWE476, CWE690)
Nice catch.
> Corrections explained:
> When process
Hi Anton,
On Sat, Feb 01, 2025 at 02:10:03AM +0300, Anton Moryakov wrote:
> Report of the static analyzer:
> After having been assigned to a NULL value at
> dwfl_segment_report_module.c:187, pointer 'retval' is
> dereferenced at dwfl_segment_report_module.c:195 by
> calling function 'strcmp'. (CWE
Hi Anton,
On Sat, Feb 01, 2025 at 01:56:34AM +0300, Anton Moryakov wrote:
> Report of the static analyzer:
> After having been compared to a NULL value at
> dwarf_ranges.c:492, pointer 'd' is dereferenced at
> dwarf_ranges.c:531. (CWE476)
But there is a lot of code between those points. On line
Hi Aaron,
On Thu, Jan 30, 2025 at 09:35:54PM -0500, Aaron Merey wrote:
> Add calls to close for all test programs that leak file descriptors
> in order to prevent test failures when run under valgrind
> --track-fds=yes.
These all look correct to me.
Could you commit this before enabling valgrind
Hi Anton,
On Sat, Feb 01, 2025 at 01:43:44AM +0300, Anton Moryakov wrote:
> Report of the static analyzer:
> After having been compared to a NULL value at
> elflint.c:252, pointer 'suffix' is dereferenced at elflint.c:260
> by calling function 'stpcpy'
>
> Corrections explained:
> When processing
Hi Aaron,
On Thu, Jan 30, 2025 at 09:35:53PM -0500, Aaron Merey wrote:
> test-elf_cntl_gelf_getshdr conditionally closes a file descriptor
> depending on a command line argument. This causes an error when run
> under valgrind --track-fds=yes.
>
> Fix this by unconditionally closing the fd.
I thi
Hi Aaron,
On Thu, Jan 30, 2025 at 09:35:52PM -0500, Aaron Merey wrote:
> valgrind --track-fds=yes might incorrectly report an error due to the use
> of inherited file descriptors in check_native_core.
Note that this fixed upstream by
commit 9f0e4107c140b47ea2a9c097afcac73a8454e17f
https://bugs.kd
Hi Aaron,
On Thu, Jan 30, 2025 at 09:35:51PM -0500, Aaron Merey wrote:
> process_archive may be called with an fd argument of -1, which
> libelf interprets as "no file opened". However when closing
> the fd process_archive does not check whether the fd is valid
> and may attempt to close an fd of
Hi Aaron,
On Thu, Jan 30, 2025 at 09:35:50PM -0500, Aaron Merey wrote:
> `valgrind --track-fds=yes` will report errors for file descriptor leaks
> and attempts at closing invalid file descriptors.
This looks good to me, but could we apply the patches in "reverse"
order so we don't get test failur
.c: New file.
> * tests/run-eu-search-macros.sh: New file.
>
> Signed-off-by: Heather S. McIntyre
> Signed-off-by: Aaron Merey
> Signed-off-by: Mark Wielaard
> ---
> v5 changes: pthread added to eu_search_macros_LDFLAGS unconditionally.
> Add missing error exit to thread_work.
Looks good.
Thanks,
Mark
d-off-by: Heather S. McIntyre
> Signed-off-by: Aaron Merey
> Signed-off-by: Mark Wielaard
> ---
> v5 changes: pthread added to eu_search_cfi_LDFLAGS unconditionally.
This looks good.
Thanks,
Mark
-off-by: Heather S. McIntyre
> Signed-off-by: Aaron Merey
> Signed-off-by: Mark Wielaard
> ---
> v5 changes: pthread added to eu_search_die_LDFLAGS unconditionally.
Looks good.
Thanks,
Mark
* tests/Makefile.am: Add eu_search_lines,
> run-eu-search-lines.sh.
> * tests/eu_search_lines.c: New file.
> tests/run-eu-search-lines.sh: New file.
>
> Signed-off-by: Heather S. McIntyre
> Signed-off-by: Aaron Merey
> Signed-off-by: Mark Wielaard
>
On Thu, Jan 30, 2025 at 11:05:20AM +, parona wrote:
> > Could you show the configure/build flags and the error the compiler
> > gives? I think the patch is correct, but I have trouble replicating
> > the issue.
>
> export CFLAGS="-O0 -pipe"
> autoreconf -fi
> ./configure --enable-stacktrace --
Hi,
On Thu, Jan 30, 2025 at 12:05:59AM +, parona wrote:
> While building elfutils on musl with different features enabled I
> hit this.
>
> Curiously this also affects glibc if you explicitly build without
> optimizations.
Could you show the configure/build flags and the error the compiler
gi
Hi,
On Wed, Jan 29, 2025 at 07:57:37AM +, parona wrote:
> On Monday, January 27th, 2025 at 10:40 PM, Mark Wielaard
> wrote:
> > So maybe our configure check is wrong and it should just test whether
> > there is an error.h but also if you can link the result?
>
> In
Hi,
On Mon, Jan 27, 2025 at 05:23:15AM +, parona wrote:
> Background for this patch is https://bugs.gentoo.org/948878 where
> are user had installed a standalone error implementation for a
> different package leading to a build failure.
There is also some discussion in this elfutils bug:
http
Hi Aaron,
On Fri, Jan 24, 2025 at 08:32:58PM -0500, Aaron Merey wrote:
> skel_fd is passed to create_dwfl, which calls dup() on skel_fd.
> create_dwfl handles closing the dup'ed fd but not the original.
>
> Ensure the original skel_fd is closed after it's passed to create_dwfl.
Nice find.
We sh
Hi Aaron,
On Fri, Jan 24, 2025 at 08:32:48PM -0500, Aaron Merey wrote:
> debuginfod_validate_imasig might call free on an uninitialized sig_buf
> due to a goto that can occur before sig_buf is set to NULL.
>
> Fix this by setting sig_buf to NULL before the goto.
The first thing after exit_valida
Hi Aaron,
On Thu, Jan 23, 2025 at 09:15:10PM -0500, Aaron Merey wrote:
> > On Sun, Jan 19, 2025 at 10:20:30PM -0500, Aaron Merey wrote:
> > > res = __libdw_getsrcfiles (cu->dbg, debug_line_offset,
> > >__libdw_getcompdir (cudie),
> >
> > So is the lock t
Hi Aaron,
On Sun, 2025-01-19 at 22:20 -0500, Aaron Merey wrote:
> From: Heather McIntyre
>
> * tests/.gitignore: Add eu_search_die.
> * tests/Makefile.am: Add eu_search_die, run-eu-search-die.sh.
> * tests/eu_search_die.c: New file.
> * tests/run-eu-search-die.sh:
Hi Aaron,
On Thu, 2025-01-23 at 21:13 -0500, Aaron Merey wrote:
> > > If performance suffers too much from serializing all accesses to Dwarf
> > > objects, we could instead consider using an rwlock with a custom wrapper
> > > that implements recursive locking.
> >
> > We could also use multiple l
Hi Aaron,
On Sun, 2025-01-19 at 22:20 -0500, Aaron Merey wrote:
> From: Heather McIntyre
>
> * tests/.gitignore: Add eu_search-lines.
> * tests/Makefile.am: Add eu_search_lines,
> run-eu-search-lines.sh.
> * tests/eu_search_lines.c: New file.
> * tests/run-eu-search
Hi Aaron,
On Sun, 2025-01-19 at 22:20 -0500, Aaron Merey wrote:
> From: Heather McIntyre
>
> * tests/.gitignore: Add eu_search_macros.sh
> * tests/Makefile.am: Add eu_search_macros,
> run-eu-search-macros.sh.
> * tests/eu_search_macros.c: New file.
> * tests/run-eu-
Hi Aaron,
On Sun, 2025-01-19 at 22:20 -0500, Aaron Merey wrote:
> From: Heather McIntyre
>
> * tests/.gitignore: Add eu_search_cfi.
> * tests/Makefile.am: Add eu_search_cfi, run-eu-search-cfi.sh.
> * tests/eu_search_cfi.c: New file.
> * tests/run-eu-search-cfi.sh: New f
Hi Aaron,
On Sun, 2025-01-19 at 22:20 -0500, Aaron Merey wrote:
> thread-safety-subr.sh contains subroutine check_thread_safety_enabled
> which skips further testing when either USE_LOCKS is false or when
> attempting to use valgrind with address or memory sanitizer enabled.
>
> * tests/Mak
Hi Aaron,
On Sun, 2025-01-19 at 22:20 -0500, Aaron Merey wrote:
> Like --enable-valgrind but uses helgrind instead of memcheck. If both
> --enable-valgrind and --enable-helgrind are given then helgrind takes
> priority.
>
> * configure.ac: Add --enable-helgrind option.
> * tests/Make
Hi Aaron,
On Sun, Jan 19, 2025 at 10:20:35PM -0500, Aaron Merey wrote:
> * libdw/dwarf_onesrcline.c (dwarf_onesrcline): Use dwarf_lock.
Is this really necessary? When the user calls dwarf_onesrcline
wouldn't they already have a fully setup Dwarf_Lines? Can it still
change concurrently?
> S
Hi Aaron,
On Sun, Jan 19, 2025 at 10:20:34PM -0500, Aaron Merey wrote:
> * libdw/dwarf_offdie.c (__libdw_offdie): Use dwarf_lock.
What exactly is "protected" here? If it is just the __libdw_findcu
call, then it seems the lock is too wide/early. It could just be taken
just before that call,
Hi Aaron,
On Sun, Jan 19, 2025 at 10:20:33PM -0500, Aaron Merey wrote:
> * libdw/dwarf_lineaddr.c (dwarf_lineaddr): Use dwarf_lock.
Why is this necessary? It seems this function can only be called with a
known/setup Dwarf_Line. Can it be concurrently changes at this point?
> Signed-off-by
Hi Aaron,
On Sun, Jan 19, 2025 at 10:20:32PM -0500, Aaron Merey wrote:
> * libdw/dwarf_getsrclines.c (read_srcfiles): Initialize dbg
> member.
> * libdw/libdwP.h (struct Dwarf_Files_s): Add Dwarf member.
> (struct Dwarf_Lines_s): Ditto.
Right, this is what you use in earli
Hi Aaron,
On Sun, Jan 19, 2025 at 10:20:30PM -0500, Aaron Merey wrote:
> * libdw/dwarf_getsrcfiles.c (dwarf_getsrcfiles): Use dwarf_lock.
>
> ---
> libdw/dwarf_getsrcfiles.c | 11 +++
> 1 file changed, 7 insertions(+), 4 deletions(-)
>
> diff --git a/libdw/dwarf_getsrcfiles.c b/lib
On Sun, Jan 19, 2025 at 10:20:31PM -0500, Aaron Merey wrote:
> * libdw/dwarf_getsrclines.c (dwarf_getsrclines): Use dwarf_lock.
So same lock used for getting/setting Dwarf_Lines. It is a little
unfortunate that lines and files table reading is so intertwined.
I understand why you use the sa
On Sun, Jan 19, 2025 at 10:20:29PM -0500, Aaron Merey wrote:
> * libdw/dwarf_filesrc.c (dwarf_filesrc): Use dwarf_lock.
I didn't got this patch till I saw 06/15 Add Dwarf member to
Dwarf_Lines and Dwarf_Files. So this patch should at least come after
that.
In general I find it a little diff
Hi Aaron,
On Sun, Jan 19, 2025 at 10:20:28PM -0500, Aaron Merey wrote:
> Change type of dwarf_lock to mutex in order to take advantage of
> built-in support for recursive locking.
When/where do we need recursive locking?
> * lib/locks.h: Add macros for locking, unlocking, initializing
>
1 - 100 of 1174 matches
Mail list logo
