Andrew Burgess [Tue, 22 Apr 2025 20:31:02 +0000 (21:31 +0100)]
gdb/python: stop using PyObject_IsInstance in py-disasm.c
The PyObject_IsInstance function can return -1 for errors, 0 to
indicate false, and 1 to indicate true.
I noticed in python/py-disasm.c that we treat the result of
PyObject_IsInstance as a bool. This means that if PyObject_IsInstance
returns -1, then this will be treated as true. The consequence of
this is that we will invoke undefined behaviour by treating the result
from the _print_insn call as if it was a DisassemblerResult object,
even though PyObject_IsInstance raised an error, and the result might
not be of the required type.
I could fix this by taking the -1 result into account, however,
gdb.DisassemblerResult cannot be sub-classed, the type doesn't have
the Py_TPFLAGS_BASETYPE flag. As such, we can switch to using
PyObject_TypeCheck instead, which only return 0 or 1, with no error
case.
I have also taken the opportunity to improve the error message emitted
if the result has the wrong type. Better error message make debugging
issues easier.
I've added a test which exposes the problem when using
PyObject_IsInstance, and I've updated the existing test for the
improved error message.
Commit b9c7eed0c2409fc640129a38d80a2bf1212b464a recently introduced
a build failure, because the file gdb/riscv-canonicalize-syscall-gen.c
hasn't been added to the ALL_64_TARGET_OBS variable in the makefile,
leading to a linker issue. This commit fixes that.
Also, turns out, the new file was slightly outdated, as the gdb_old_mmap
syscall has been renamed to gdb_sys_old_mmap in commit 432eca4113d5748ad284a068873455f9962b44fe. This commit also fixes that
on the generated file itself, to quickly fix the build. A followup
commit will fix the python file responsible for generating the .c file.
Guinevere Larsen [Wed, 26 Mar 2025 14:14:52 +0000 (11:14 -0300)]
GDB: Introduce "info linker-namespaces" command
Continuing to improve GDB's ability to debug linker namespaces, this
commit adds the command "info linker- namespaces". The command is
similar to "info sharedlibrary" but focused on improved readability
when the inferior has multiple linker namespaces active. This command
can be used in 2 different ways, with or without an argument.
When called without argument, the command will print the number of
namespaces, and for each active namespace, it's identifier, how many
libraries are loaded in it, and all the libraries (in a similar table to
what "info sharedlibrary" shows). As an example, this is what GDB's
output could look like:
(gdb) info linker-namespaces
There are 2 linker namespaces loaded
There are 3 libraries loaded in linker namespace [[0]]
Displaying libraries for linker namespace [[0]]:
From To Syms Read Shared Object Library
0x00007ffff7fc6000 0x00007ffff7fff000 Yes /lib64/ld-linux-x86-64.so.2
0x00007ffff7ebc000 0x00007ffff7fa2000 Yes (*) /lib64/libm.so.6
0x00007ffff7cc9000 0x00007ffff7ebc000 Yes (*) /lib64/libc.so.6
(*): Shared library is missing debugging information.
There are 4 libraries loaded in linker namespace [[1]]
Displaying libraries for linker namespace [[1]]:
From To Syms Read Shared Object Library
0x00007ffff7fc6000 0x00007ffff7fff000 Yes /lib64/ld-linux-x86-64.so.2
0x00007ffff7fb9000 0x00007ffff7fbe000 Yes gdb.base/dlmopen-ns-ids/dlmopen-lib.so
0x00007ffff7bc4000 0x00007ffff7caa000 Yes (*) /lib64/libm.so.6
0x00007ffff79d1000 0x00007ffff7bc4000 Yes (*) /lib64/libc.so.6
(*): Shared library is missing debugging information.
When called with an argument, the argument must be a namespace
identifier (either with or without the square brackets decorators). In
this situation, the command will truncate the output to only show the
relevant information for the requested namespace. For example:
(gdb) info linker-namespaces 0
There are 3 libraries loaded in linker namespace [[0]]
Displaying libraries for linker namespace [[0]]:
From To Syms Read Shared Object Library
0x00007ffff7fc6000 0x00007ffff7fff000 Yes /lib64/ld-linux-x86-64.so.2
0x00007ffff7ebc000 0x00007ffff7fa2000 Yes (*) /lib64/libm.so.6
0x00007ffff7cc9000 0x00007ffff7ebc000 Yes (*) /lib64/libc.so.6
(*): Shared library is missing debugging information.
The test gdb.base/dlmopen-ns-id.exp has been extended to test this new
command. Because some gcc and glibc defaults can change between
systems, we are not guaranteed to always have libc and libm loaded in a
namespace, so we can't guarantee the number of libraries, but the range
of the result is 2, so we can still check for glaring issues.
Reviewed-By: Eli Zaretskii <eliz@gnu.org> Approved-by: Kevin Buettner <kevinb@redhat.com>
Guinevere Larsen [Tue, 25 Mar 2025 20:27:49 +0000 (17:27 -0300)]
gdb: factor out printing a table of solibs for info sharedlibrary
The next patch will add a new command that will print libraries in a
manner very similar to the existing "info sharedlibrary" command. To
make that patch simpler to review, this commit does the bulk of
refactoring work, since it ends up being a non-trivial diff to review.
No functional changes are expected after this commit.
Guinevere Larsen [Fri, 21 Mar 2025 19:35:07 +0000 (16:35 -0300)]
gdb: add convenience variables around linker namespace debugging
This commit adds 2 simple built-in convenience variables to help users
debug an inferior with multiple linker namespaces. The first is
$_active_linker_namespaces, which just counts how many namespaces have SOs
loaded onto them. The second is $_current_linker_namespace, and it tracks
which namespace the current location in the inferior belongs to.
This commit also introduces a test ensuring that we track namespaces
correctly, and that a user can use the $_current_linker_namespace
variable to set a conditional breakpoint, while linespec changes aren't
finalized to make it more convenient.
Reviewed-By: Eli Zaretskii <eliz@gnu.org> Approved-by: Kevin Buettner <kevinb@redhat.com>
Timur [Tue, 18 Mar 2025 18:49:44 +0000 (21:49 +0300)]
This commit adds record full support for rv64gc instruction set
It includes changes to the following files:
- gdb/riscv-linux-tdep.c, gdb/riscv-linux-tdep.h: adds facilities to record
syscalls.
- gdb/riscv-tdep.c, gdb/riscv-tdep.h: adds facilities to record execution of
rv64gc instructions.
- gdb/configure.tgt: adds new files for compilation.
- gdb/testsuite/lib/gdb.exp: enables testing of full record mode for RISC-V
targets.
- gdb/syscalls/riscv-canonicalize-syscall-gen.py: a script to generate
function that canonicalizes RISC-V syscall. This script can simplify support
for syscalls on rv32 and rv64 system (currently support only for rv64). To
use this script you need to pass a path to a file with syscalls description
from riscv-glibc (example is in the help message). The script produces a
mapping from syscall names to gdb_syscall enum.
- gdb/riscv-canonicalize-syscall.c: the file generated by the previous script.
- gdb/doc/gdb.texinfo: notification that record mode is enabled in RISC-V.
- gdb/NEWS: notification of new functionality.
Approved-By: Guinevere Larsen <guinevere@redhat.com> Approved-By: Andrew Burgess <aburgess@redhat.com>
Tom de Vries [Wed, 23 Apr 2025 09:02:53 +0000 (11:02 +0200)]
[gdb/testsuite] Add selftest disassemble-s390x
In commit a98a6fa2d8e ("s390: Add arch15 instructions"), support for
new instructions was added to libopcodes, but the added tests only exercise
this for gas.
Add a unit test disassemble-s390x that checks gdb's ability to
disassemble one of these instructions:
...
$ gdb -q -batch -ex "maint selftest -v disassemble-s390x"
Running selftest disassemble-s390x.
0xb9 0x68 0x00 0x03 -> clzg %r0,%r3
Ran 1 unit tests, 0 failed
...
Alan Modra [Wed, 23 Apr 2025 01:20:18 +0000 (10:50 +0930)]
string merge section map output
This fixes an inconsistency in the linker map file, where string merge
sections (other than the first) kept their sizes. String merge
sections of like entsize all are accounted in the fisrt string merge
section size.
* ldlang.c (print_input_section): Print SEC_EXCLUDE section size
as zero.
Tom Tromey [Wed, 19 Mar 2025 20:47:29 +0000 (14:47 -0600)]
Handle DWARF 5 separate debug sections
DWARF 5 standardized the .gnu_debugaltlink section that dwz emits in
multi-file mode. This is handled via some new forms, and a new
.debug_sup section.
This patch adds support for this to gdb. It is largely
straightforward, I think, though one oddity is that I chose not to
have this code search the system build-id directories for the
supplementary file. My feeling was that, while it makes sense for a
distro to unify the build-id concept with the hash stored in the
.debug_sup section, there's no intrinsic need to do so.
This in turn means that a few tests -- for example those that test the
index cache -- will not work in this mode.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=32808 Acked-By: Simon Marchi <simon.marchi@efficios.com>
Tom Tromey [Sat, 22 Mar 2025 16:39:43 +0000 (10:39 -0600)]
Remove 'read' call from dwz_file::read_string
dwz_file::read_string calls 'read' on the section, but this isn't
needed as the sections have all been pre-read.
This patch makes this change, and refactors dwz_file a bit to make
this more obvious -- by making it clear that only the "static
constructor" can create a dwz_file.
Approved-By: Simon Marchi <simon.marchi@efficios.com> Tested-By: Alexandra Petlanova Hajkova <ahajkova@redhat.com>
Andrew Burgess [Tue, 22 Apr 2025 18:45:57 +0000 (19:45 +0100)]
gdb/testsuite: fix incorrect comment in py-color.exp
There was a comment in gdb.python/py-color.exp that was probably left
over from a copy & paste, it incorrectly described what the test
script was testing.
Fixed in this commit.
There's no change in what is tested with this commit.
Andrew Burgess [Sat, 19 Apr 2025 21:33:32 +0000 (22:33 +0100)]
gdb/python: address some coding style issues in py-color.c
A few minor GNU/GDB coding style issues in py-color.c:
- Space after '&' reference operator in one place.
- Some excessive indentation on a couple of lines.
- Spaces after '!' logical negation operator.
- Using a pointer as a bool in a couple of places.
There should be no functional changes after this commit.
it was pointed out that I should use @samp{} around some text I was
adding to the documentation. However, the offending snippet of
documentation was something I copied from elsewhere in python.texi.
This commit fixes the original to use @samp{}.
has an unnecessary `Py_INCREF (self);` in gdb.Color.__init__. This
means that the reference count on all gdb.Color objects (that pass
through __init__) will be +1 from where they should normally be, and
this will stop the gdb.Color objects from being deallocated.
Andrew Burgess [Sat, 19 Apr 2025 13:41:17 +0000 (14:41 +0100)]
gdb/python: restructure the existing memory leak tests
We currently have two memory leak tests in gdb.python/ and there's a
lot of duplication between these two.
In the next commit I'd like to add yet another memory leak test, which
would mean a third set of scripts which duplicate the existing two.
And three is where I draw the line.
This commit factors out the core of the memory leak tests into a new
module gdb_leak_detector.py, which can then be imported by each
tests's Python file in order to make writing the memory leak tests
easier.
I've also added a helper function to lib/gdb-python.exp which captures
some of the common steps needed in the TCL file in order to run a
memory leak test.
Finally, I use this new infrastructure to rewrite the two existing
memory leak tests.
What I considered, but ultimately didn't do, is merge the two memory
leak tests into a single TCL script. I did consider this, and for the
existing tests this would be possible, but future tests might require
different enough setup that this might not be possible for all future
tests, and now that we have helper functions in a central location,
the each individual test is actually pretty small now, so leaving them
separate seemed OK.
There should be no change in what is actually being tested after this
commit.
This fixes a crash on Windows NT 4.0, where windows-nat failed dynamic loading
some Win32 functions and print a warning message with styled string, which
depends on ui-style regex. By using `compiled_regex` constructor, the regex is
guaranteed to be initialized before `_initialize_xxx` functions.
Alan Modra [Sun, 20 Apr 2025 23:33:52 +0000 (09:03 +0930)]
avoid bogus format-overflow error
Seen on x86_64-linux Ubuntu 24.04.2 using gcc-13.3.0 with
CFLAGS="-m32 -g -O2 -fsanitize=address,undefined"
In function ‘sprintf’,
inlined from ‘s_mri_for’ at gas/config/tc-m68k.c:6941:5:
/usr/include/bits/stdio2.h:30:10: error: null destination pointer [-Werror=format-overflow=]
30 | return __builtin___sprintf_chk (__s, __USE_FORTIFY_LEVEL - 1,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
31 | __glibc_objsize (__s), __fmt,
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
32 | __va_arg_pack ());
| ~~~~~~~~~~~~~~~~~
Rewrite the code without sprintf, as in other parts of s_mri_for.
See also commit 760fb390fd4c and following commits.
Note that adding -D_FORTIFY_SOURCE=0 to CFLAGS (which is a good idea
when building with sanitizers) merely transforms the sprintf_chk error
here into one regarding plain sprintf.
Alan Modra [Sat, 19 Apr 2025 05:12:23 +0000 (14:42 +0930)]
bfd_check_format_matches error paths
Tidy early out errors which didn't free matching_vector. Don't
bfd_preserve_restore if we get to err_ret from the first
bfd_preserve_save, which might fail from a memory allocation leaving
preserve.marker NULL. Also take bfd_lock a little earlier before
modifying abfd->format to simplify error return path from a lock
failure.
Simon Marchi [Thu, 17 Apr 2025 19:37:26 +0000 (15:37 -0400)]
gdb/dwarf: make some more functions methods of cutu_reader
These are only used by cutu_reader, so make them methods of cutu_reader.
This makes it a bit more obvious in which context this code is called.
lookup_dwo_unit_in_dwp can't be made a method of cutu_reader, as it is
used in another context (lookup_dwp_signatured_type /
lookup_signatured_type), which happens during CU expansion.
Change-Id: Ic62c3119dd6ec198411768323aaf640ed165f51b Approved-By: Tom Tromey <tom@tromey.com>
Simon Marchi [Thu, 17 Apr 2025 19:37:25 +0000 (15:37 -0400)]
gdb/dwarf: look up .dwp file ahead of time
get_dwp_file lazily looks for a .dwp file for the given objfile. It is
called by indexing workers, when a cutu_reader object looks for a DWO
file. It is called with the "dwo_lock" held, meaning that the first
worker to get there will do the work, while the others will wait at the
lock.
I'm trying to reduce the time where this lock is taken and do other
refactorings to make it easier to reason about the DWARF reader code.
Moving the lookup of the .dwp file ahead, before we start parallelizing
work, helps makes things simpler, because we can then assume everywhere
else that we have already checked for a .dwp file.
Put the call to open_and_init_dwp_file in dwarf2_has_info, right next to
where we look up .dwz files. I used the same try-catch pattern as for
the .dwz file lookup.
Change-Id: I615da85f62a66d752607f0dbe9f0372dfa04b86b Approved-By: Tom Tromey <tom@tromey.com>
Simon Marchi [Thu, 17 Apr 2025 19:37:20 +0000 (15:37 -0400)]
gdb/dwarf: remove dwarf2_section_info::get_size
The comment over dwarf2_section_info::get_size says:
In other cases, you must call this function, because for compressed
sections the size field is not set correctly until the section has
been read
From what I can see (while debugging a test case compiled with -gz on
Linux), that's not true. For compressed sections, bfd_section_size
returns the uncompressed size. asection::size contains the uncompressed
size while asection::compressed_size contains the compressed size:
(top-gdb) p sec
$13 = (asection *) 0x521000119778
(top-gdb) p sec.compressed_size
$14 = 6191
(top-gdb) p sec.size
$15 = 12116
I therefore propose to remove dwarf2_section_info::get_size, as it
appears that reading in the section is orthogonal to knowing its size.
If the assumption above is false, it would be nice to document in which
case it's false.
I checked the callers, and I don't think that we need to add any
dwarf2_section_info::read calls to compensate for the fact that get_size
used to do it.
Change-Id: I428571e532301d49f1d8242d687e1fcb819b75c1 Approved-By: Tom Tromey <tom@tromey.com>
Tom Tromey [Sat, 19 Apr 2025 17:33:12 +0000 (11:33 -0600)]
Remove some obsolete comments from ada-varobj.c
I noticed a few spots in ada-varobj.c that refer to calling xfree,
where the type in question has changed to std::string. This patch
removes these obsolete comments.
Alan Modra [Tue, 15 Apr 2025 21:42:46 +0000 (07:12 +0930)]
loongarch gas resolving constant expressions
The test added in commit 4fe96ddaf614 results in an asan complaint:
loongarch-parse.y:225:16: runtime error: left shift of negative value -1
To avoid the complaint, perform left shifts as unsigned (which gives
the same result on 2's complement machines). Do the same for
addition, subtraction and multiplication. Furthermore, warn on
divide/modulus by zero.
Tom de Vries [Thu, 17 Apr 2025 15:57:29 +0000 (17:57 +0200)]
[gdb/testsuite] Don't run to main in gdb.cp/cplusfuncs.exp
After building gdb with -fsanitize=threads, and running test-case
gdb.cp/cplusfuncs.exp, I run into a single timeout:
...
FAIL: gdb.cp/cplusfuncs.exp: info function operator=( (timeout)
...
and the test-case takes 2m33s to finish.
This is due to expanding CUs from libstdc++.
After de-installing package libstdc++6-debuginfo, the timeout disappears and
testing time goes down to 9 seconds.
Fix this by not running to main, which brings testing time down to 3 seconds.
With a gdb built without -fsanitize=threads, testing time goes down from 11
seconds to less than 1 second.
Tom Tromey [Tue, 15 Apr 2025 18:29:50 +0000 (12:29 -0600)]
Clean up value_struct_elt_bitpos
value_struct_elt_bitpos is weird: it takes an in/out value parameter,
and it takes an error string parameter. However, it only has a single
caller, which never uses the "out" value.
I think it was done this way to mimic value_struct_elt. However,
value_struct_elt is pretty ugly and I don't think it's worth
imitating.
This patch cleans up value_struct_elt_bitpos a bit.
Approved-By: Simon Marchi <simon.marchi@efficios.com>
Yury Khrustalev [Thu, 27 Mar 2025 16:48:24 +0000 (16:48 +0000)]
aarch64: ld: Fix scanning of GNU properties for AARCH64_FEATURE_1_AND
Fixes [1]. Previously iteration over GNU properties of an input file
could stop before reaching GNU_PROPERTY_AARCH64_FEATURE_1_AND which
would result in incorrect inference of properties of the output file.
In the particular use case described in [1], the memory seal property
GNU_PROPERTY_MEMORY_SEAL with number 3, if present in the input file,
prevented reading information from GNU_PROPERTY_AARCH64_FEATURE_1_AND
property due to filtering by property number.
Simon Marchi [Fri, 11 Apr 2025 15:08:06 +0000 (11:08 -0400)]
gdb: fix bugs in gdb/copyright.py, make it use glob patterns
gdb/copyright.py currently changes some files that it shouldn't:
- despite having a `gnulib/import` entry in EXCLUDE_LIST, it does
change the files under that directory
- it is missing `sim/Makefile.in`
Change the exclude list logic to use glob patterns. This makes it
easier to specify exclusions of full directories or files by basename,
while simplifying the code.
Merge EXCLUDE_LIST and NOT_FSF_LIST, since there's no fundamental reason
to keep them separate (they are treated identically). I kept the
comment that explains that some files are excluded due to not being
FSF-licensed.
Merge EXCLUDE_ALL_LIST in EXCLUDE_LIST, converting the entries to glob
patterns that match everywhere in the tree (e.g. `**/configure`).
Tested by running the script on the parent commit of d01e823438c7
("Update copyright dates to include 2025") and diff'ing the result with d01e823438c7. The only differences are:
- the files that we don't want to modify (gnulib/import and
sim/Makefile.in)
- the files that need to be modified by hand
Running the script on latest master produces no diff.
Simon Marchi [Fri, 11 Apr 2025 15:08:05 +0000 (11:08 -0400)]
gdb: make typing strict in gdb/copyright.py
Add `pyright: strict` at the top of the file, then adjust the fallouts.
This annotation is understood by pyright, and thus any IDE using pyright
behind the scenes (VSCode and probably others).
I presume that any GDB developer running this script is using a recent
enough version of Python, so specify the type annotations using the
actual types when possible (e.g. `list[str]` instead of
`typing.List[str]`). I believe this required Python 3.9.
Tom de Vries [Wed, 16 Apr 2025 15:39:41 +0000 (17:39 +0200)]
[gdb/testsuite] Fix another timeout in gdb.base/bg-execution-repeat.exp
With test-case gdb.base/bg-execution-repeat.exp, occasionally I run into a
timeout:
...
(gdb) c 1&
Will stop next time breakpoint 1 is reached. Continuing.
(gdb) PASS: $exp: c 1&: c 1&
Breakpoint 2, foo () at bg-execution-repeat.c:23
23 return 0; /* set break here */
PASS: $exp: c 1&: breakpoint hit 1
Will stop next time breakpoint 2 is reached. Continuing.
(gdb) PASS: $exp: c 1&: repeat bg command
print 1
$1 = 1
(gdb) PASS: $exp: c 1&: input still accepted
interrupt
(gdb) PASS: $exp: c 1&: interrupt
Program received signal SIGINT, Interrupt.
foo () at bg-execution-repeat.c:24
24 }
PASS: $exp: c 1&: interrupt received
set var do_wait=0
(gdb) PASS: $exp: c 1&: set var do_wait=0
continue&
Continuing.
(gdb) PASS: $exp: c 1&: continue&
FAIL: $exp: c 1&: breakpoint hit 2 (timeout)
...
I can reproduce it reliably by adding a "sleep (1)" before the "do_wait = 1"
in the .c file.
The timeout happens as follows:
- with the inferior stopped at main, gdb continues (command c 1&)
- the inferior hits the breakpoint at foo
- gdb continues (using the repeat command functionality)
- the inferior is interrupted
- inferior variable do_wait gets set to 0. The assumption here is that the
inferior has progressed enough that do_wait is set to 1 at that point, but
that happens not to be the case. Consequently, this has no effect.
- gdb continues
- the inferior sets do_wait to 1
- the inferior enters the wait function, and wait for do_wait to become 0,
which never happens.
Fix this by moving the "do_wait = 1" to before the first call to foo.
Tested on x86_64-linux.
Reviewed-By: Alexandra Petlanova Hajkova <ahajkova@redhat.com>
Tom de Vries [Tue, 15 Apr 2025 14:59:32 +0000 (16:59 +0200)]
[gdb/ada] Fix gdb.ada/overloads.exp on s390x
On s390x-linux, with test-case gdb.ada/overloads.exp and gcc 7.5.0 I run into:
...
(gdb) print Oload(CA)^M
Could not find a match for oload^M
(gdb) FAIL: $exp: print Oload(CA)
...
The mismatch happens here in ada_type_match:
...
return ftype->code () == atype->code ();
...
with:
...
(gdb) p ftype->code ()
$3 = TYPE_CODE_TYPEDEF
(gdb) p atype->code ()
$4 = TYPE_CODE_ARRAY
...
At the start of ada_type_match, typedefs are skipped:
...
ftype = ada_check_typedef (ftype);
atype = ada_check_typedef (atype);
...
but immediately after this, refs are skipped:
...
if (ftype->code () == TYPE_CODE_REF)
ftype = ftype->target_type ();
if (atype->code () == TYPE_CODE_REF)
atype = atype->target_type ();
...
which in this case makes ftype a typedef.
Fix this by using ada_check_typedef after skipping the refs as well.
Tested on x86_64-linux and s390x-linux.
Approved-By: Tom Tromey <tom@tromey.com>
PR ada/32409
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=32409
Simon Marchi [Fri, 11 Apr 2025 20:38:04 +0000 (16:38 -0400)]
gdb/dwarf: skip type units in create_dwo_cus_hash_table
When compiling with -gsplit-dwarf -fdebug-types-section, DWARF 5
.debug_info.dwo sections may contain some type units:
$ llvm-dwarfdump -F -color a-test.dwo | head -n 5
a-test.dwo: file format elf64-x86-64
.debug_info.dwo contents:
0x00000000: Type Unit: length = 0x000008a0, format = DWARF32, version = 0x0005, unit_type = DW_UT_split_type, abbr_offset = 0x0000, addr_size = 0x08, name = 'vector<int, std::allocator<int> >', type_signature = 0xb499dcf29e2928c4, type_offset = 0x0023 (next unit at 0x000008a4)
In this case, create_dwo_cus_hash_table wrongly creates a dwo_unit for
it and adds it to dwo_file::cus. create_dwo_debug_type_hash_table later
correctly creates a dwo_unit that it puts in dwo_file::tus.
Fix it by skipping anything that isn't a compile unit in
create_dwo_cus_hash_table. After this patch, the debug output of
create_dwo_cus_hash_table only shows one created dwo_unit, as we expect.
I couldn't find any user-visible problem related to this, I just noticed
it while debugging.
Change-Id: I7dddf766fe1164123b6702027b1beb56114f25b1 Reviewed-By: Tom de Vries <tdevries@suse.de>
Alan Modra [Mon, 14 Apr 2025 09:41:53 +0000 (19:11 +0930)]
windres: don't exit so much on errors in read_coff_rsrc
windres code has the habit of exiting on any error. That's not so
bad, but it does make oss-fuzz ineffective when testing windres. Fix
many places that print errors and exit to instead print the error and
pass status up the call chain. In the process of doing this, I
noticed write_res_file was calling bfd_close without checking return
status. Fixing that resulted in lots of testsuite failures. The
problem was a lack of bfd_set_format in windres_open_as_binary, which
leaves the output file as bfd_unknown format. As it happens this
doesn't make any difference in writing the output binary file, except
for the bfd_close return status.
Alan Modra [Mon, 14 Apr 2025 01:01:32 +0000 (10:31 +0930)]
windres: buffer overflow in bin_to_res_toolbar
oss-fuzz testcase manages to hit a buffer overflow. Sanity check
by passing the buffer length to bin_to_res_toolbar and ensuring reads
don't go off the end of the buffer.
Jan Beulich [Mon, 14 Apr 2025 12:24:28 +0000 (14:24 +0200)]
ld/PE: restrict non-zero default DLL characteristics to MinGW
While commit ef6379e16dd1 ("Set the default DLL chracteristics to 0 for
Cygwin based targets") tried to undo the too broad earlier 514b4e191d5f
("Change the default characteristics of DLLs built by the linker to more
secure settings"), it didn't go quite far enough. Apparently the
assumption was that if it's not MinGW, it must be Cygwin. Whether it
really is okay to default three of the flags to non-zero on MinGW also
remains unclear - sadly neither of the commits came with any description
whatsoever. (Documentation also wasn't updated to indicate the restored
default.)
Setting effectively any of the DLL characteristics flags depends on
properties of the binary being linked. While defaulting to "more secure"
is a fair goal, it's only the programmer who can know whether their code
is actually compatible with the respective settings. On the assumption
that the change of defaults was indeed deliberate (and justifiable) for
MinGW, limit them to just that. In particular, don't default any of the
flags to set also for non-MinGW, non-Cygwin targets, like e.g. UEFI. At
least the mere applicability of the high-entropy-VA bit is pretty
questionable there in the first place - UEFI applications, after all,
run in "physical mode", i.e. either unpaged or (where paging is a
requirement, like for x86-64) direct-mapped.
The situation is particularly problematic with NX-compat: Many UEFI
implementations respect the "physical mode" property, where permissions
can't be enforced anyway. Some, like reportedly OVMF, even have a build
option to behave either way. Hence successfully testing a UEFI binary on
any number of systems does not guarantee it won't crash elsewhere if the
flag is wrongly set.
Jan Beulich [Mon, 14 Apr 2025 12:23:53 +0000 (14:23 +0200)]
bfd/ELF/x86: avoid layering violation in link hash table entry init
There's no reason not to do as the comment says, just like all other
architectures do when they need custom field: Call the allocation method
of the "superclass". Which is the ELF one, of which in turn the BFD one
is the "superclass", dealt with accordingly by
_bfd_elf_link_hash_newfunc().
Jan Beulich [Mon, 14 Apr 2025 12:23:29 +0000 (14:23 +0200)]
bfd/aout: drop add_one_symbol() hook
The need for this has disappeared with c65c21e1ffd1 ("various i386-aout
and i386-coff target removal"), with a few other users having got
removed just a few days earlier; avoid the unnecessary indirection.
Jan Beulich [Mon, 14 Apr 2025 12:22:49 +0000 (14:22 +0200)]
bfd/COFF: propagate function size when copying/linking ELF objects
While COFF, unlike ELF, doesn't have a generic way to express symbol
size, there is a means to do so for functions. When inputs are ELF,
propagate function sizes, including the fact that a symbol denotes a
function, to the output's symbol table.
Note that this requires hackery (cross-object-format processing) in two
places - when linking, global symbols are entered into a global hash
table, and hence relevant information needs to be updated there in that
case, while otherwise the original symbol structures can be consulted.
For the setting of ->u.syment.n_type the later writing of the field to
literal 0 needs to be dropped from coff_write_alien_symbol(). It was
redundant anyway with an earlier write of the field using C_NUL.
Andrew Burgess [Sun, 13 Apr 2025 13:01:59 +0000 (14:01 +0100)]
gdb: add an assert to cmd_list_element constructor
The cmd_list_element::doc variable must be non-nullptr, otherwise, in
`help_cmd` (cli/cli-decode.c), we will trigger an assert when we run
one of these lines:
gdb_puts (c->doc, stream);
or,
gdb_puts (alias->doc, stream);
as gdb_puts requires that the first argument (the doc string) be
non-nullptr.
Better, I think, to assert when the cmd_list_element is created,
rather than catching an assert later when 'help CMD' is used.
I only ran into this case when messing with the Python API command
creation code, I accidentally created a command with a nullptr doc
string, and only found out when I ran 'help CMD' and got an
assertion.
While I'm adding this assertion, I figure I should also assert that
the command name is not nullptr too. Looking through cli-decode.c,
there seems to be plenty of places where we assume a non-nullptr name.
Built and tested on x86-64 GNU/Linux with an all-targets build; I
don't see any regressions, so (I hope) there are no commands that
currently violate this assertion.
Approved-By: Simon Marchi <simon.marchi@efficios.com>
WANG Xuerui [Sun, 23 Mar 2025 09:30:38 +0000 (17:30 +0800)]
LoongArch: Support LA32R aliases rdcnt{vl,vh,id}.w
These LA32R instructions are in fact special cases of the LA32S/LA64
rdtime{l,h}.w (with only one output operand instead of two, the other
one being forced to $zero), but are named differently in the LA32R
ISA manual nevertheless.
As the LA32R names are more memorable to a degree (especially for those
having difficulties remembering which operand corresponds to the node
ID), support them by making them aliases of the corresponding LA32S/LA64
instruction respectively, and make them render as such in disassembly.
Andrew Burgess [Wed, 12 Mar 2025 11:16:42 +0000 (11:16 +0000)]
gdb: silence some 'Can't open file' warnings from core file loading
But PR gdb/20126 highlights a case where GDB emits a large number of
warnings like:
warning: Can't open file /anon_hugepage (deleted) during file-backed mapping note processing
warning: Can't open file /dev/shm/PostgreSQL.1150234652 during file-backed mapping note processing
warning: Can't open file /dev/shm/PostgreSQL.535700290 during file-backed mapping note processing
warning: Can't open file /SYSV604b7d00 (deleted) during file-backed mapping note processing
... etc ...
when opening a core file. This commit aims to avoid at least some of
these warnings.
What we know is that, for at least some of these cases, (e.g. the
'(deleted)' mappings), the content of the mapping will have been
written into the core file itself. As such, the fact that the file
isn't available ('/SYSV604b7d00' at least is a shared memory mapping),
isn't really relevant, GDB can still provide access to the mapping, by
reading the content from the core file itself.
What I propose is that, when processing the file backed mappings, if
all of the mappings for a file are covered by segments within the core
file itself, then there is no need to warn the user that the file
can't be opened again. The debug experience should be unchanged, as
GDB would have read from the in-core mapping anyway.
Andrew Burgess [Sun, 6 Apr 2025 20:52:50 +0000 (21:52 +0100)]
bfd: fix missing warnings from bfd_check_format_matches
In PR gdb/31846 the user reported an issue where GDB is unable to find
the build-id within an ELF, despite the build-id being
present (confirmed using readelf).
The user was able to try several different builds of GDB, and in one
build they observed the warning:
warning: BFD: FILENAME: unable to decompress section .debug_info
But in may other builds of GDB this warning was missing.
There are, I think, a couple of issues that the user is running into,
but this patch is about why the above warning is often missing from
GDB's output.
I wasn't able to reproduce a corrupted .debug_info section such that
the above warning would be triggered, but it is pretty easy to patch
the _bfd_elf_make_section_from_shdr function (in bfd/elf.c) such that
the call to bfd_init_section_decompress_status is reported as a
failure, thus triggering the warning. There is a patch that achieves
this in the bug report.
I did this, and can confirm that on my build of GDB, I don't see the
above warning, even though I can confirm that the _bfd_error_handler
call (in _bfd_elf_make_section_from_shdr) is being reached.
The problem is back in format.c, in bfd_check_format_matches. This
function intercepts all the warnings and places them into a
per_xvec_messages structure. These warnings are then printed with a
call to print_and_clear_messages.
If bfd_check_format_matches finds a single matching format, then
print_and_clear_messages, will print all warnings associated with that
single format.
But if no format matches, print_and_clear_messages will print all the
warnings, so long as all targets have emitted the same set of
warnings, and unfortunately, that is not the case for me.
The warnings are collected by iterating over bfd_target_vector and
trying each target. My target happens to be x86_64_elf64_vec, and, as
expected this target appears in bfd_target_vector.
However, bfd_target_vector also includes DEFAULT_VECTOR near the top.
And in my build, DEFAULT_VECTOR is x86_64_elf64_vec. Thus, for me,
the x86_64_elf64_vec entry appears twice in bfd_target_vector, this
means that x86_64_elf64_vec ends up being tried twice, and, as each
try generates one warning, the x86_64_elf64_vec entry in the
per_xvec_messages structure, has two warnings, while the other
per_xvec_messages entries only have one copy of the warning.
Because of this difference, print_and_clear_messages decides not to
print any of the warnings, which is not very helpful.
I considered a few different approaches to fix this issue:
We could de-duplicate warnings in the per_xvec_messages structure as
new entries are added. So for any particular xvec, each time a new
warning arrives, if the new warning is identical to an existing
warning, then don't record it. This might be an interesting change in
the future, but for now I rejected this solution as it felt like a
bodge, the duplicate warnings aren't really from a single attempt at
an xvec, but are from two distinct attempts at the same xvec. And so:
I wondered if we could remove the duplicate entries from
bfd_target_vector. Or if we could avoid processing the same xvec
twice maybe? For the single DEFAULT_VECTOR this wouldn't be too hard
to do, but bfd_target_vector also includes SELECT_VECS, which I think
could contain more duplicates. Changing bfd_check_format_matches to
avoid attempting any duplicate vectors would now require more
complexity than a single flag, and I felt there was an easier
solution, which was:
I propose that within bfd_check_format_matches, within the loop that
tries each entry from bfd_target_vector, as we switch to each vector
in turn, we should delete any existing warnings within the
per_xvec_messages structure for the target vector we are about to try.
This means that, if we repeat a target, only the last set of warnings
will survive.
With this change in place, print_and_clear_messages now sees the same
set of warnings for each target, and so prints out the warning
message.
Additionally, while I was investigating this issue I managed to call
print_and_clear_messages twice. This caused a crash because the first
call to print_and_clear_messages frees all the associated memory, but
leaves the per_xvec_messages::next field pointing to the now
deallocated object. I'd like to propose that we set the next field to
NULL in print_and_clear_messages. This clearly isn't needed so long
as print_and_clear_messages is only called once, but (personally) I
like to set pointers back to NULL if the object they are pointing to
is free and the parent object is going to live for some additional
time. I can drop this extra change if people don't like it.
This change doesn't really "fix" PR gdb/31846, but it does mean that
the warning about being unable to decompress .debug_info should now be
printed consistently, which is a good thing.
gdb/testsuite: fix gdb.base/dlmopen-ns-ids.exp racy test
The recently included gdb.base/dlmopen-ns-ids.exp test can sometimes
fail the call to get_integer_valueof when trying to check the namespace
ID of the fourth dlopened SO, for apparently no reason.
What's happening is that the call to get_first_so_ns doesn't necessarily
consume the GDB prompt, and so get_integer_valueof will see the prompt
immediately and not find the value the test is looking for.
To fix this, the test was changed so that we consume all of the output
of the command "info sharedlibrary", but only set the namespace ID for
the first occurrence of the SO we're looking for. The command now also
gets the solib name as a parameter, to reduce the amount of output.
Co-Authored-By: Tom de Vries <tdevries@suse.de> Approved-By: Tom de Vries <tdevries@suse.de>
skipped the LTO archive member even when the earlier item is also an
archive. Instead, skip the LTO archive member only if the earlier item
is a shared library.
bfd/
PR ld/32846
PR ld/32854
* elflink.c (elf_link_add_archive_symbols): Skip the LTO archive
member only if the earlier item is a shared library.
Tom de Vries [Thu, 10 Apr 2025 16:42:06 +0000 (18:42 +0200)]
[gdb/unittests] Ignore spellcheck warning in rsp-low-selftests.c
Ignore the following spellcheck warning:
...
$ codespell --config gdb/contrib/setup.cfg gdb/unittests
gdb/unittests/rsp-low-selftests.c:54: fo ==> of, for, to, do, go
...
and add gdb/unittests to the pre-commit codespell configuration.
Tom de Vries [Thu, 10 Apr 2025 02:50:26 +0000 (04:50 +0200)]
[gdb/testsuite] Fix gdb.dwarf2/fission-with-type-unit.exp with remote host
When running test-case gdb.dwarf2/fission-with-type-unit.exp with a remote
host configuration, say host board local-remote-host and target board
remote-gdbserver-on-localhost, I run into:
...
(gdb) maint expand-symtabs^M
During symbol reading: Could not find DWO CU \
fission-with-type-unit.dwo(0xf00d) referenced by CU at offset 0x2d7 \
[in module /home/remote-host/fission-with-type-unit]^M
warning: Could not find DWO CU fission-with-type-unit.dwo(0xf00d) referenced \
by CU at offset 0x2d7 [in module /home/remote-host/fission-with-type-unit]^M
(gdb) FAIL: gdb.dwarf2/fission-with-type-unit.exp: maint expand-symtabs
...
Fix this by adding the missing download to remote host of the .dwo file.
Tested by running make-check-all.sh on x86_64-linux.
projects / thirdparty / binutils-gdb.git / log
