afl-fuzz or some other fuzzing
framework, or simply as extra testcase.
Please also see:
https://sourceware.org/git/?p=elfutils.git;f=CONTRIBUTING;hb=HEAD
Cheers,
Mark
and/or
-fsanitize-coverage=trace-cmp (which I believe is command line
compatible with what clang uses).
Cheers,
Mark
> On Mon, Dec 23, 2019 at 3:12 AM Mark Wielaard wrote:
>
> > Hi Berkeley,
> >
> > On Fri, 2019-12-20 at 17:21 +0200, Berkeley Churchill wrote:
> >
s
> flowing - at all - the rest will follow as rapidly as the network
> allows.
>
> That suggests one timeout could be sufficient - the progress timeout
> you the one you found - just not too short and not too fast.
How about the attached (untested) patch?
Cheers,
Mark>From 9b90be
Hi Dmitry,
On Thu, 2020-01-02 at 03:02 +0300, Dmitry V. Levin wrote:
> On Thu, Aug 29, 2019 at 11:43:58PM +0200, Mark Wielaard wrote:
> > All archive members from libebl.a are now in libdw.a. We don't generate
> > separate backend shared libraries anymore. So
on DEBUGINFOD.
That looks correct. Pushed to master.
Thanks,
Mark
quot; DEBUGINFOD_CLIENT_USERAGENT_2' >> $@
Eep. We really should pick up this info during runtime instead of
during build time. We do want a reproducible build. And this will most
likely produce wrong information if the package build server is on a
different OS or release than the OS/release it is producing packages
for. uname -sm might be "stable", but probably not when cross-
compiling. But where does lsb_release come from? I don't have that on
my systems.
Cheers,
Mark
>
rings on a couple of my
systems out of the box:
Linux/i686 4.19.0-6-686-pae
debian/10
Linux/x86_64 3.10.0-1062.9.1.el7.x86_64
rhel/7.7
Linux/x86_64 4.19.0-5-amd64
pureos/9.0
Maybe something like that is usable?
Cheers,
Mark#include
#include
#include
#include
int
main ()
{
struct uts
and make it 90s. That is more than twice the
worst delay I ever observed.
Pushed the attached.
Cheers,
Mark
From b8d85ed024a745cff05e56c6337d95d654d5294a Mon Sep 17 00:00:00 2001
From: Mark Wielaard
Date: Thu, 2 Jan 2020 17:02:42 +0100
Subject: [PATCH] debuginfod: Use DEBUGINFOD_TIMEOUT as
Hi Dmitry,
On Thu, 2020-01-09 at 02:25 +0300, Dmitry V. Levin wrote:
> On Wed, Jan 08, 2020 at 03:09:32PM +0100, Mark Wielaard wrote:
> > Even if you could use libasm.h, linking with libasm always was kind
> > of
> > useless since there was no supported way to create an Ebl h
An earlier variant of the default progress function could write to any
file. Which is still in the documentation. But the actual implementation
just uses stderr. Fix the documentation to match.
Signed-off-by: Mark Wielaard
---
doc/ChangeLog | 5 +
doc
On Fri, Jan 10, 2020 at 12:43:44AM +0100, Mark Wielaard wrote:
> On Thu, 2020-01-09 at 02:25 +0300, Dmitry V. Levin wrote:
> > On Wed, Jan 08, 2020 at 03:09:32PM +0100, Mark Wielaard wrote:
> > > Short term I propose the attached patch.
> >
> > Thanks, it works: th
On Fri, Jan 10, 2020 at 03:50:49PM +0100, Mark Wielaard wrote:
> An earlier variant of the default progress function could write to any
> file. Which is still in the documentation. But the actual implementation
> just uses stderr. Fix the documentation to match.
I pushed this to master.
h (exit status: 1)
Full test-suite.log is at:
https://builder.wildebeest.org/buildbot/#/builders/11/builds/412/steps/8/logs/test-suite_log
Anyone a guess what is going on in this case?
Thanks,
Mark
I guess the comment "release the hounds" gets me confused. I don't know
precisely what that means. Is this needed for correctness, or is it simply to
have an hint that the scanners have ran because you register as an idler and so
at this point you could only run if no scanner is running? But given you
immediately tell them you are done a scanner could in theory start their work
up again while you are doing your own traversal?
Thanks,
Mark
Yes, because that's ideal for concurrency: while one thread fts(3)'s
> and enumerates the results, the scanner threads can process the
> already enumerated files.
OK. And if any idlers (grooming) would wake up that is no problem,
because that simply means the work queue was empty and adding work to
the queue can be done whether either the grooming thread (x)or scanner
threads are running concurrently.
Please do push if you feel confident about it being correct. I think it
is now that I understand how it actually works.
Thanks,
Mark
GCC10 on some arches will warn about possible NULL derefences.
In the libdwfl linux-kernel-modules.c cases it might be caught already
by earlier calls to get_release (). But it is hard to see that will
really always happen. So do an explicit NULL check just in case.
Signed-off-by: Mark Wielaard
On Fri, 2020-01-24 at 23:28 +0100, Mark Wielaard wrote:
> GCC10 on some arches will warn about possible NULL derefences.
> In the libdwfl linux-kernel-modules.c cases it might be caught already
> by earlier calls to get_release (). But it is hard to see that will
> really always happ
no problem
> multi-distroreq. separate servers one server ok
One of the use cases of debuginfod is working with local/team builds
for binaries not packaged by a distro. How does clear-linux handle that?
Cheers,
Mark
which you could provide
through the Fuse file system. On Fedora those are symlinks to the
binaries installed, but you could provide the actual binaries
instead. That would be somewhat similar to the debuginfod
/buildid/BUILDID/executable request then.
Cheers,
Mark
just pass the file with the given extension to libarchive directly
(then you also don't need the zstdcat conversion). And letting
libarchive handle it directly might be more efficient.
Cheers,
Mark
in the code to get the FILE* to pass
to archive_read_open_FILE? Or even just open and archive_read_open_fd?
Cheers,
Mark
The sh_info field of the symtab says how many symbols are in the
section. Make sure at least that many symbols fit in the section.
Reported-by: Ulrich Drepper
Signed-off-by: Mark Wielaard
---
src/ChangeLog | 4
src/elflint.c | 4
2 files changed, 8 insertions(+)
diff --git a/src
// "e" O_CLOEXEC?
+ dfn = pclose;
+ if (fp == NULL)
+throw libc_exception (errno, string("popen ") + popen_cmd);
+}
+ else
+{
+ fp = fopen (b_source0.c_str(), "r");
+ dfn = fclose;
+ if (fp == NULL)
+throw libc_exception (errno, string("fopen ") + b_source0);
+}
+ defer_dtor fp_closer (fp, dfn);
struct archive *a;
a = archive_read_new();
Cheers,
Mark
Hi Arjan,
On Fri, 2020-02-07 at 06:47 -0800, Arjan van de Ven wrote:
> On 2/7/2020 1:13 AM, Mark Wielaard wrote:
> > On Wed, Feb 05, 2020 at 02:05:42PM -0800, Arjan van de Ven wrote:
> > > On 2/5/2020 11:14 AM, Frank Ch. Eigler wrote:
> > > > AIUI that works by us
On Fri, 2020-02-07 at 18:20 +0100, Mark Wielaard wrote:
> The sh_info field of the symtab says how many symbols are in the
> section. Make sure at least that many symbols fit in the section.
I pushed this to master.
On Mon, 2020-02-10 at 22:34 +, build...@builder.wildebeest.org
wrote:
> The Buildbot has detected a failed build on builder whole buildset
> while building elfutils.
> Full details are available at:
> https://builder.wildebeest.org/buildbot/#builders/15/builds/269
This is a somewhat older
infod server, unless they test (and set) one themselves.
Signed-off-by: Mark Wielaard
---
tests/ChangeLog | 5 +
tests/Makefile.am | 2 ++
2 files changed, 7 insertions(+)
diff --git a/tests/ChangeLog b/tests/ChangeLog
index 01e4a083..32ca1ce0 100644
--- a/tests/ChangeLog
+++ b/tests/Chan
eriodically as this
> -program is reexecuted. The default is $HOME/.debuginfod_client_cache.
> +program is reexecuted. If XDG_CACHE_HOME is set then
> +$XDG_CACHE_HOME/debuginfod_client is the default location, otherwise
> +$HOME/.cache/debuginfod_client is used.
> +
>
> .S
Hi Aaron,
On Fri, 2020-02-21 at 12:21 -0500, Aaron Merey wrote:
> On Fri, Feb 21, 2020 at 11:13 AM Mark Wielaard wrote:
> > On Wed, 2020-02-19 at 16:04 -0500, Aaron Merey wrote:
> > > + /* Create XDG cache directory if it doesn't exist. */
> > > +
On Fri, 2020-02-21 at 13:49 +0100, Mark Wielaard wrote:
> If DEBUGINFOD_URLS is set various tests will try to query the debuginfod
> server which can stall the tests a bit. If other evironment variables
> like DEBUGINFOD_PROGRESS are set it will make various tests fail because
>
t it to the user supplied value (if any). Is that
intentional?
Cheers,
Mark
27;t really see the point of
how they are used. Is the only use case to pass the string "Progress"?
If there are no real users for this then I think we should not add
these at this time. Or is some other client using them? I am not really
against it, but would prefer if we add them separately to keep the
patches concise.
Thanks,
Mark
s for this then I think we should not add
> > these at this time. Or is some other client using them? I am not
> > really
> > against it, but would prefer if we add them separately to keep the
> > patches concise.
>
> GDB would use them pretty much immediately, to be able to prepare a
> more informative progress notification (like the file name whose
> debuginfo is being sought, instead of its buildid).
OK. Yes, that seems like a good use case.
But can we add this as a separate feature in a separate commit?
Thanks,
Mark
id=25600
Perfect patch, nice testcase. Thanks for the ChangeLog entries. I did
tweak a few whitespace issues, but the file itself wasn't really
consistent to begin with.
Pushed to master.
Cheers,
Mark
gment(&nsegments[i])
--18155:0: aspacem at m_aspacemgr/aspacemgr-linux.c:728 (preen_nsegments)
--18155:0: aspacem Exiting now.
I'll try to replicate it and figure this out. But I think it doesn't
have to do anything with the elfutils code.
Cheers,
Mark
f
> builds. Would it make sense to take —disable-shared here ?
Yes, that might make sense. It isn't really tested much though. So
someone would have to carefully audit the build files to see if it
really works as intended.
Cheers,
Mark
Hi Aaron,
On Thu, 2020-02-27 at 19:00 -0500, Aaron Merey wrote:
> On Fri, Feb 21, 2020 at 3:09 PM Mark Wielaard wrote:
> > But I think that means we should only create the cachedir (either
> > $XDG_CACHE_HOME or $HOME/.cache). Which is what we are doing here. But
> >
We no longer require rpm2cpio to unpack rpms.
We do now require dpkg-deb to unpack debs.
https://sourceware.org/bugzilla/show_bug.cgi?id=25583
Signed-off-by: Mark Wielaard
---
config/ChangeLog| 5 +
config/elfutils.spec.in | 5 +++--
2 files changed, 8 insertions(+), 2 deletions
ath was updated only when
> creating the new default directory and would otherwise be set to the
> old default path.
Nice catch and nice testcase. Looks good. Pushed to master.
Cheers,
Mark
04)
==27438==by 0x100318B: process_file (elfcompress.c:778)
==27438==by 0x100318B: main (elfcompress.c:1359)
Luckily upstream valgrind already has a patch. But that will need to be
backported and then a new package should be installed on the s390x
buildbot worker.
Cheers,
Mark
er inflateEnd() failed, which
might cause it to call inflateEnd() twice (which might be fine, I
dunno). Should we maybe ignore the error if inflateEnd() and just call
it unconditionally before (ignoring its return code)?
So, replace:
if (... Z_OK) zrc = inflateEnd (&z);
with unconditionally ending the stream:
(void)inflateEnd(&z);
What do you think?
Cheers,
Mark
@ __libelf_decompress (void *buf_in, size_t size_in, size_t
size_out)
}
zrc = inflateReset (&z);
}
- if (likely (zrc == Z_OK))
-zrc = inflateEnd (&z);
+ inflateEnd (&z);
if (unlikely (zrc != Z_OK) || unlikely (z.avail_out != 0))
{
Thanks,
Mark
Hi,
On Mon, Mar 16, 2020 at 01:05:26PM +0100, Mark Wielaard wrote:
> Thanks for looking into this so closely. I think it is best to just do
> as the example code does. Always call inflateEnd (), ignoring the
> return code and not care whether (in an error case) it might be called
> t
return
> Z_STREAM_END to indicate we are done. Hence change the condition.
It was that last condition I had missed.
Added a ChangeLog entry and pushed to master.
Thanks,
Mark
PN_XNUM and for some reason the
scns cache wasn't initialized. Do you happen to have a testcase?
Thanks,
Mark
ter location for that note.
> +.SS "USER DATA POINTER"
> +
> +A single \fIvoid *\fP pointer associated with the connection handle
> +may be set any time via
> +.BR \%debuginfod_set_user_data () ,
> +and retrieved via
> +.BR \%debuginfod_get_user_data () .
> +The value may be NULL.
So, if we decide that the initial value is NULL then this would be
where we would document that or say that the value is undefined unless
debuginfod_set_user_data () has never been called.
Thanks,
Mark
have to check for the empty string, just
for NULL.
Cheers,
Mark
|| (((uintptr_t) ((char *) ehdr + e_shoff)
> & (__alignof__ (Elf64_Shdr) - 1)) == 0)))
I had missed that part. So it is actually common to not have the phdrs
cache.
Thanks,
Mark
From 28734adeb8cd51c7124330f88c1dc49c9c187865 Mon Sep 17 00:00:00 2001
From: Mark Wielaard
for
> > adding the attached to the testsuite? It fails before, and passes
> > after
> > your fix.
>
> Sure thing, feel free to add
>
> Signed-off-by: Omar Sandoval
Done and pushed your original fix, plus this testcase.
Thanks,
Mark
check for the empty string, just
> > for NULL.
>
> OK.
Looks good. Thanks for updating and pushing. But please do sent the
final patch also to the list when the code changed. So people can track
which version eventually gets in.
Thanks,
Mark
s-debian-i386/build/tests/getphdrnum
contains non-lfs symbols: open
FAIL run-lfs-symbols.sh (exit status: 1)
The backends shared libraries don't exist anymore and should be removed
from the testcase.
But I don't immediately see what/where I forgot which flags to make the
new getphdrnum check program picks up the wrong "open".
Investigating...
Mark
Hi,
On Mon, 2020-03-23 at 15:08 +0100, Mark Wielaard wrote:
> Well, that is odd. Both are 32bit arches. The failure is the same in
> run-lfs-symbols.sh
>
>FAIL: run-lfs-symbols.sh
>
>./testfile-nolfs contains non-lfs symbols: fopen
>
ode got
pushed and why. Then I also know whether I can expect some patch to
show up when I do a git pull and whether it is (subtly) different from
what was posted to the list earlier.
Cheers,
Mark
the missing CURLOPT_PATH_AS_IS from comment #4
probably should be resolved also.
Are there any other "urgent" bug fixes pending?
Cheers,
Mark
t not be that simple to get things to compile and
run correctly against it.
Thanks,
Mark
Hi,
I pushed the following 2 commits to the code and htdocs repositories.
sourceware updated the mailinglist software recently. Update our
documentation (README and CONTRIBUTING) and the webpages to point to
the new subscription form.
Cheers,
Mark
From a344b8ce0bbe4f50fa2ace5fdfdcaccdee81d0e5
remove 'test-5124': Directory not empty
FAIL run-backtrace-native-core-biarch.sh (exit status: 1)
But I cannot replicate it. And it was caused by a change to the README
file. Rebuilding on the same buildbot worker also made it PASS... Sigh.
Don't know what to do about it, except watch whether it happens again
or not.
Cheers,
Mark
/tests/debuginfod-rpms/hello3.spec.
> b/tests/debuginfod-rpms/hello3.spec.
> new file mode 100644
> index ..ffb95134e2b6
> --- /dev/null
> +++ b/tests/debuginfod-rpms/hello3.spec.
> @@ -0,0 +1,60 @@
> +Summary: hello3 -- double hello, world rpm
> +Name: hello3
&
_start.tv_sec) + (tv_end.tv_usec -
> tv_start.tv_usec)*0.01;
> + obatched(clog) << conninfo(connection)
> + << ' ' << method << ' ' << url
> + << ' ' << http_code << '
On Thu, 2020-03-26 at 11:24 -0400, Frank Ch. Eigler via Elfutils-devel
wrote:
> Ah but that would be too easy.
:)
Also make sure the new test rpms are added to EXTRA_DIST.
Pushed the attached.
Cheers,
Mark
From 12305ff528ad9365b4d4e5ebcd307373cdd5bf83 Mon Sep 17 00:00:00 2001
From: M
> .SH DESCRIPTION
>
> @@ -159,6 +161,21 @@ The resulting string is owned by the library, and must
> not be modified
> or freed. The caller should copy it if it is needed beyond the release
> of the client object.
>
> +.SS "HTTP HEADER"
> +
> +Before a lo
stalled on the s390x and
ppc64[le] workers.
I am not against this patch, but I do wonder if it should not try to
fall back on dpkg if bsdtar isn't installed.
Cheers,
Mark
;", &alt_fd);
> > > }
> >
> > 0 instead of NULL?
> > I believe we went over this before, 0 is more C++?
>
> Yes.
OK. I guess I got stuck in C language mindset.
> > > +/* Add an outgoing HTTP request "Header: Value". Copies string. */
> > > +int debuginfod_add_http_header (debuginfod_client *client, const char*
> > > header);
> >
> > So here at least add "optional header" or something. A warning that
> > this is a hint only and might be ignored depending on backend/transport
> > used.
>
> Adding a http header to a non-http transport URL is obviously
> meaningless and harmless. I don't see why we should belabour it or
> give people any concern about it.
I just like the documentation to be clear and precise.
Cheers,
Mark
s is just into debuginfo-find
and experiment a bit with it before adding it as a public API to the
debuginfod-client API.
In debuginfod-find the code would also be slightly simpler since it
doesn't have to convert to a hex-string first, it can just pass the
build-id bytes as is.
Cheers,
Mark
s and data amounts.
This looks nice. Do we have to document these anywhere, or are
prometheus metrics "self documenting"? I assume tools just take what
they can get from the json export and create graphs based on the name
count/sum patterns?
Thanks,
Mark
-by: Mark Wielaard
---
debuginfod/ChangeLog | 4
debuginfod/debuginfod.cxx | 3 ++-
2 files changed, 6 insertions(+), 1 deletion(-)
diff --git a/debuginfod/ChangeLog b/debuginfod/ChangeLog
index 58ba85cf..018d62f6 100644
--- a/debuginfod/ChangeLog
+++ b/debuginfod/ChangeLog
@@ -1,3 +1,7
er testers
some time to digest and do a release on Monday or Tuesday?
Cheers,
Mark
correct. I added a simple check to the function to see it is at least
somewhat plausible, so that a user won't be surprised if a wrongly
formatted string accidentally removes a header instead of adding one.
Cheers,
Mark
From 7ddceee2b6b0a3fe752a2e8cc5d5cfd0f45d6897 Mon Sep 17 00:00:00 2001
Fr
to disambiguate files by
using ./
Thanks,
Mark
Set version to 0.179.
Update NEWS and elfutils.spec.in.
Regenerate po/*.po files.
Signed-off-by: Mark Wielaard
---
ChangeLog | 5 +
NEWS| 35 +++
config/ChangeLog| 4 +
config/elfutils.spec.in | 25 ++
configure.ac| 2 +-
po
ine (3):
debuginfod.h: fix typo in comment: legnth -> length
debuginfod/doc: Fix typo debuginfo_progressfn_t ->
debuginfod_progressfn_t
debuginfod: file:// URLs: handle curl resp. code
Mark Wielaard (27):
config: versioned libdebuginfod libraries should not be in
Default file permissions on sourceware have changed. Make sure that
the upload-release.sh script sets the permissions right.
Signed-off-by: Mark Wielaard
---
config/ChangeLog | 5 +
config/upload-release.sh | 2 ++
2 files changed, 7 insertions(+)
diff --git a/config/ChangeLog b
Hi,
Thanks for the patch, but elf.h comes from glibc. So please get your
fix accepted upstream (sent it to libc-al...@sourceware.org) and we'll
pick it up next time we sync with the glibc code.
Thanks,
Mark
function names. And avoid defining symbols
in asm statements.
Signed-off-by: Mark Wielaard
---
ChangeLog | 4
configure.ac| 13 +
lib/ChangeLog | 5 +
lib/eu-config.h | 20 ++--
4 files changed, 40 insertions(+), 2 deletions(-)
diff --git a
Hi,
On Fri, 2020-04-10 at 21:42 +0200, Mark Wielaard wrote:
> GCC 10 introduces a function attribute to define symbol versioning.
> Add a configure check to see if __attribute__((symver)) is supported.
> If it is then define the OLD_VERSION, NEW_VERSION, COMPAT_VERS
On Tue, 2020-04-14 at 17:25 +0200, Mark Wielaard wrote:
> On Fri, 2020-04-10 at 21:42 +0200, Mark Wielaard wrote:
> > GCC 10 introduces a function attribute to define symbol versioning.
> > Add a configure check to see if __attribute__((symver)) is supported.
> > If
https://sourceware.org/bugzilla/show_bug.cgi?id=25831
Signed-off-by: Mark Wielaard
---
libcpu/ChangeLog | 5 +
libcpu/i386_disasm.c | 11 +++
2 files changed, 12 insertions(+), 4 deletions(-)
diff --git a/libcpu/ChangeLog b/libcpu/ChangeLog
index 70796514..a8b2b951 100644
d use.
If the callchain I am inlining is so deep that I cannot figure out
maybe-uninitialized variables anymore I should stop inlining.
For now just help GCC out and initialize bits to NULL.
Signed-off-by: Mark Wielaard
---
libdwfl/ChangeLog| 5 +
libdwfl/find-debuginfo.c | 2 +-
2
Strictly speaking SHF_EXCLUDE is a processor specific section flag,
but it is used generically in the GNU toolchain. For example when
adding .gnu.lto_ sections.
Signed-off-by: Mark Wielaard
---
src/ChangeLog | 5 +
src/elflint.c | 5 +
2 files changed, 10 insertions(+)
diff --git a/src
GCC puts (partial) DWARF debuginfo into sections prefixed with
.gnu.debuglto_. Handle those sections as if they are normal .debug
sections (which they are). This allows showing the DWARF that gcc
puts into ET_REL files compiled with -flto.
Signed-off-by: Mark Wielaard
---
libdw/ChangeLog
test_nlist checks its own symbol table, and expects various symbols
to be in the order as specified in the source file. Explicitly set
minimal CFLAGS.
Signed-off-by: Mark Wielaard
---
tests/ChangeLog | 5 +
tests/Makefile.am | 9 +
2 files changed, 14 insertions(+)
diff --git a
files with --exprlocs to test that output
too (and not require addresses).
Signed-off-by: Mark Wielaard
---
tests/ChangeLog | 7 +++
tests/run-varlocs-self.sh | 4 +++-
tests/test-subr.sh| 13 +
3 files changed, 23 insertions(+), 1 deletion(-)
diff --git a
On Thu, Apr 16, 2020 at 06:53:41PM +0200, Mark Wielaard wrote:
> Strictly speaking SHF_EXCLUDE is a processor specific section flag,
> but it is used generically in the GNU toolchain. For example when
> adding .gnu.lto_ sections.
I pushed this to master.
g root
path is almost no cost. So it seems fine to add.
The default sysconfig is somewhat rpm/fedora/rh specific. Hopefully
now that we support other archive formats we can trick some
contributor into adding some sysconfig that is is optimal for other
setups.
Cheers,
Mark
On Fri, Apr 17, 2020 at 11:35:37AM +0200, Mark Wielaard wrote:
> GCC puts (partial) DWARF debuginfo into sections prefixed with
> .gnu.debuglto_. Handle those sections as if they are normal .debug
> sections (which they are). This allows showing the DWARF that gcc
> puts into ET_REL fi
On Fri, Apr 17, 2020 at 12:22:39PM +0200, Mark Wielaard wrote:
> test_nlist checks its own symbol table, and expects various symbols
> to be in the order as specified in the source file. Explicitly set
> minimal CFLAGS.
Pushed to master.
On Fri, Apr 17, 2020 at 12:34:17PM +0200, Mark Wielaard wrote:
> The varlocs test relies on finding addresses for the CUs, even in
> object files. This might not be true (for example when building with
> gcc -flto, which only emits partial, type only, debuginfo). Split
> running the s
| | |
| | (8) ...to here
| | (9) first ‘free’ here
|
Fixed by removing the free (out_buf) on line 116 as attached.
Cheers,
Mark
>From 0b2fc95c46dabf85d053b2f0c6aab217b9c5a9b8 Mon Sep 17 00:00:00 2001
From: Mark Wielaard
Date: Sa
would if something went wrong writing
out the file.
Signed-off-by: Mark Wielaard
---
libasm/ChangeLog | 4
libasm/asm_end.c | 2 +-
2 files changed, 5 insertions(+), 1 deletion(-)
diff --git a/libasm/ChangeLog b/libasm/ChangeLog
index 7b0d3df3..2c092abe 100644
--- a/libasm/ChangeLog
+++ b/l
GCC10 -fanalyzer found a possibly-NULL dereference after a failed
malloc in __libdw_alloc_tail. In this case we should call the Dwarf
oom_handler as is done in other places where an essential malloc
call fails. The oom_handler cannot return and will likely just abort.
Signed-off-by: Mark Wielaard
ned.
Signed-off-by: Mark Wielaard
---
libdwfl/ChangeLog | 5 +
libdwfl/gzip.c| 2 +-
2 files changed, 6 insertions(+), 1 deletion(-)
diff --git a/libdwfl/ChangeLog b/libdwfl/ChangeLog
index 4ddc9ad4..daedaed8 100644
--- a/libdwfl/ChangeLog
+++ b/libdwfl/ChangeLog
@@ -1,3 +1,8 @@
+2020-04
...
Signed-off-by: Mark Wielaard
---
libdwfl/ChangeLog| 5 +
libdwfl/find-debuginfo.c | 23 +--
2 files changed, 26 insertions(+), 2 deletions(-)
diff --git a/libdwfl/ChangeLog b/libdwfl/ChangeLog
index daedaed8..3f9cd665 100644
--- a/libdwfl/ChangeLog
+++ b/li
Hi,
On Fri, 2020-05-01 at 00:07 +0200, Mark Wielaard wrote:
> When we fall back to the debuginfod client then we need to do the
> same trick we do for local lookups in dwfl_build_id_find_debuginfo.
> If the debug file (dw) is already set, then we must be looking for
> the altfile. B
to dwarf_attr_integrate explaining
that the attribute returned might come from a different CU (and even
different Dwarf).
Signed-off-by: Mark Wielaard
---
libdw/ChangeLog | 7 +
libdw/dwarf_decl_file.c | 2 +-
libdw/dwarf_ranges.c | 5
libdw/libdw.h
d-off-by: Mark Wielaard
---
libdw/ChangeLog | 5 +
libdw/libdw_visit_scopes.c | 6 +-
tests/ChangeLog | 9 +
tests/Makefile.am | 1 +
tests/run-allfcts.sh| 27 +++
tests/testfile-lto-gcc8.bz2 | Bin 0 -&
probably fairly hard
to trigger. All are somewhat trivial and I intent to check them in soon.
Cheers,
Mark
GCC10 -fanalyzer noticed that we allocate, but don't always cleanup the
dwfl->user_core if it wasn't set yet on error. In theory dwfl_module_end
should take care of it, but it is cleaner and less confusing to just do
it here.
Signed-off-by: Mark Wielaard
---
libdwfl/Cha
d->d_buf, orig_size) == 0)
orig_buf can only be NULL when orig_size is zero, but it might still
be undefined behaviour. So don't try to be too smart and just check
whether we actually have an buffer.
Signed-off-by: Mark Wielaard
---
tests/ChangeLog | 5 +
tests/el
GCC10 -fanalyzer with -flto notices __elf64_getshdr_rdlock can fail
and because the result isn't checked in elf_strptr it can cause a
dereference of NULL.
Signed-off-by: Mark Wielaard
---
libelf/ChangeLog| 4
libelf/elf_strptr.c | 2 +-
2 files changed, 5 insertions(+), 1 del
GCC10 -fanalyzer with -flto notices __gelf_getehdr_rdlock can fail
and that the result of the call in __libelf_set_rawdata_wrlock isn't
checked, which can cause a dereference of NULL.
Signed-off-by: Mark Wielaard
---
libelf/ChangeLog | 5 +
libelf/elf_getdata.c | 2 ++
2 files ch
1601 - 1700 of 3433 matches
Mail list logo
