Break i386-dis-evex.h into small files such that each file is included
just once.
* i386-dis-evex.h: Break into ...
* i386-dis-evex-len.h: New file.
* i386-dis-evex-mod.h: Likewise.
* i386-dis-evex-prefix.h: Likewise.
* i386-dis-evex-reg.h: Likewise.
* i386-dis-evex-w.h: Likewise.
* i386-dis.c: Include i386-dis-evex-reg.h, i386-dis-evex-prefix.h,
i386-dis-evex.h, i386-dis-evex-len.h, i386-dis-evex-w.h and
i386-dis-evex-mod.h.
This patch removes uses of VEC (dwarf2_section_info_def) in favor of
std::vector<dwarf2_section_info>. The conversion is relatively
straightforward, no function changes are intended.
gdb/ChangeLog:
* dwarf2read.h (dwarf2_section_info_def): Remove.
(DEF_VEC_O (dwarf2_section_info_def)): Remove.
* dwarf2read.c (struct dwo_sections) <types>: Change type to
std::vector<dwarf2_section_info>.
(struct dwo_file) <~dwo_file>: Remove.
(dwarf2_per_objfile::~dwarf2_per_objfile): Don't manually free
types field.
(dwarf2_per_objfile::locate_sections): Adjust to std::vector.
(dwarf2_read_debug_names): Likewise.
(create_debug_types_hash_table): Change parameter type to
array_view, adjust code accordingly.
(dwarf2_locate_dwo_sections): Adjust to std::vector.
(partial_die_info::fixup): Likewise.
(determine_prefix): Likewise.
* dwarf-index-write.c (write_psymtabs_to_index): Adjust.
This removes the manual call to gdb_bfd_ref in favor of gdb_bfd_ref_ptr.
gdb/ChangeLog:
* dwarf2read.c (struct dwo_file) <dbfd>: Change type to
gdb_bfd_ref_ptr.
<~dwo_file>: Remove call to gdb_bfd_unref.
(open_and_init_dwo_file): Move gdb_bfd_ref_ptr into dbfd field. Call
gdb_bfd_ref_ptr::get.
This patch changes dwo_file to be allocated/deallocated with new/delete,
so that we can start using C++ features in it, and in struct
dwo_sections.
The free_dwo_file function becomes the destructor of struct dwo_file
(and will disappear in upcoming patches, which will use gdb_bfd_ref_ptr
for dbfd and an std::vector for sections.types).
gdb/ChangeLog:
* dwarf2read.h (struct dwarf2_per_objfile) <dwo_files>: Change
type to htab_up.
* dwarf2read.c (struct dwo_file): Initialize fields.
<~dwo_file>: New.
(free_dwo_file): Remove, move content to ~dwo_file.
(struct dwo_file_deleter): Remove.
(dwo_file_up>: Remove custom deleter.
(free_dwo_files): Remove.
(dwarf2_per_objfile::~dwarf2_per_objfile): Don't explicitly free
dwo_files.
(process_skeletonless_type_units): Call unique_ptr::get.
(allocate_dwo_file_hash_table): Add deleter to created hash
table. Change return type to htab_up.
(lookup_dwo_file_slot): Don't memset dwo_file, call
unique_ptr::get.
(create_dwo_unit_in_dwp_v1): Allocate dwo_file with new.
(create_dwo_unit_in_dwp_v2): Likewise.
(open_and_init_dwo_file): Likewise.
(free_dwo_file_from_slot): Remove.
Use bool instead of char where applicable in dwarf2_section_info.
No functional changes intended.
gdb/ChangeLog:
* dwarf2read.h (struct dwarf2_section_info) <readin,
is_virtual>: Change type to bool.
* dwarf2read.c (dwarf2_read_section, create_dwp_v2_section): Use
true instead of 1.
When testing gdb with board cc-with-gdb-index, we run into:
...
FAIL: gdb.ada/ptype_union.exp: ptype global
FAIL: gdb.ada/ptype_union.exp: print global
...
The index is not supported for Ada (PR24713), and cc-with-gdb-index does not
add an index for Ada test-cases. However, this test-case compiles C sources,
for which cc-with-gdb-index does add an index. In gdb we load the executable
containing the index and set the language to Ada, resulting in gdb trying to
handle something that is not supported.
Fix the fail by marking this unsupported.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2019-06-21 Tom de Vries <tdevries@suse.de>
PR testsuite/24518
PR ada/24713
* gdb.ada/ptype_union.exp: Mark as unsupported if executable contains
index.
This commit fixes two paths in update-gnulib.sh that weren't updated
when gnulib was moved to toplevel.
gnulib/ChangeLog:
* update-gnulib.sh: Adjust paths.
ctf_open (or, rather, ctf_fdopen, which underlies it) has several
endianness problems, even though it was written after the
endian-swapping code was implemented, so should have been endian-aware.
Even though the comment right above the relevant check says that it wil
check for CTF magic in any endianness, it only checks in the native
endianness, so opening raw LE CTF files on BE, or vice-versa, will fail.
It also checks the CTF version by hand, without ever endianness-swapping
the header, so that too will fail, and is entirely redundant because
ctf_simple_open does the job properly in any case. We have a similar
problem in the next if block, which checks for raw CTF archives: we are
checking in the native endianness while we should be doing a le64toh()
on it to check in little-endian form only: so opening CTF archives
created on the local machine will fail if the local machine is
big-endian.
Adding insult to injury, if ctf_simple_open then fails, we go on and try
to turn it into a single-element CTF archive regardless, throwing the
error away. Since this involves dereferencing null pointers it is not
likely to work very well.
libctf/
* ctf-open-bfd.c: Add swap.h and ctf-endian.h.
(ctf_fdopen): Check for endian-swapped raw CTF magic, and
little-endian CTF archive magic. Do not check the CTF version:
ctf_simple_open does that in endian-safe ways. Do not dereference
null pointers on open failure.
Testing of the first code to generate CTF_K_SLICEs on big-endian
revealed a bunch of new problems in this area. Most importantly, the
trick we did earlier to avoid wasting two bytes on padding in the
ctf_slice_t is best avoided: because it leads to the whole file after
that point no longer being naturally aligned, all multibyte accesses
from then on must use memmove() to avoid unaligned access on platforms
where that is fatal. In future, this is planned, but for now we are
still doing direct access in many places, so we must revert to making
ctf_slice_t properly aligned for storage in an array.
Rather than wasting bytes on padding, we boost the size of cts_offset
and cts_bits. This is still a waste of space (we cannot have offsets or
bits in bitfields > 256) but it cannot be avoided for now, and slices
are not so common that this will be a serious problem.
A possibly-worse endianness problem fixed at the same time involves
a codepath used only for foreign-endian, uncompressed CTF files, where
we were not copying the actual CTF data into the buffer, leading to
libctf reading only zeroes (or, possibly, uninitialized garbage).
Finally, when we read in a CTF file, we copy the header and work from
the copy. We were flipping the endianness of the header copy, and of
the body of the file buffer, but not of the header in the file buffer
itself: so if we write the file back out again we end up with an
unreadable frankenfile with header and body of different endiannesses.
Fix by flipping both copies of the header.
include/
* ctf.h (ctf_slice_t): Make cts_offset and cts_bits unsigned
short, so following structures are properly aligned.
libctf/
* ctf-open.c (get_vbytes_common): Return the new slice size.
(ctf_bufopen): Flip the endianness of the CTF-section header copy.
Remember to copy in the CTF data when opening an uncompressed
foreign-endian CTF file. Prune useless variable manipulation.
If we see a CTF type with a kind we do not recognize in its ctt_info
during opening, we cannot skip it and continue opening the file: if the
type kind is unknown, we do not know how long its vlen is, and we cannot
have skipped past it: so if we continue reading we will almost certainly
read in part of the vlen as if it were a new ctf_type_t.
Avoid this trouble by considering unknown type kinds to be a reason to
return ECTF_CORRUPT, just like everything else that reads in type kinds
does.
libctf/
* ctf-open.c (ctf_types): Fail when unidentified type kinds are
seen.
This is an essential first piece of info needed to debug both libctf
writing and reading problems, and we weren't recording it anywhere!
(This is a short-term fix: fairly soon, we will record all of this in a
form that outlives ctf_bufopen, and then ctf_dump() will be able to dump
it like it can everything else.)
libctf/
* ctf-open.c (ctf_bufopen): Dump header offsets into the debugging
output.
This allocator has the ostensible benefit that it lets us mprotect() the
memory used for CTF storage: but in exchange for this it adds
considerable complexity, since we have to track allocation sizes
ourselves for use at freeing time, note whether the data we are storing
was ctf_data_alloc()ed or not so we know if we can safely mprotect()
it... and while the mprotect()ing has found few bugs, it *has* been the
cause of more than one due to errors in all this tracking leading to us
mprotect()ing bits of the heap and stuff like that.
We are about to start composing CTF buffers from pieces so that we can
do usage-based optimizations on the strtab. This means we need
realloc(), which needs nonportable mremap() and *more* tracking of the
*original* allocation size, and the complexity and bureaucracy of all of
this is just too high for its negligible benefits.
Drop the whole thing and just use malloc() like everyone else. It knows
better than we do when it is safe to use mmap() under the covers,
anyway.
While we're at it, don't leak the entire buffer if ctf_compress_write()
fails to compress it.
libctf/
* ctf-subr.c (_PAGESIZE): Remove.
(ctf_data_alloc): Likewise.
(ctf_data_free): Likewise.
(ctf_data_protect): Likewise.
* ctf-impl.h: Remove declarations.
* ctf-create.c (ctf_update): No longer call ctf_data_protect: use
ctf_free, not ctf_data_free.
(ctf_compress_write): Use ctf_data_alloc, not ctf_alloc. Free
the buffer again on compression error.
* ctf-open.c (ctf_set_base): No longer track the size: call
ctf_free, not ctf_data_free.
(upgrade_types): Likewise. Call ctf_alloc, not ctf_data_alloc.
(ctf_bufopen): Likewise. No longer call ctf_data_protect.
We were missing several cases where dynhash insertion might fail, likely
due to OOM but possibly for other reasons. Pass the errors on.
libctf/
* ctf-create.c (ctf_dtd_insert): Pass on error returns from
ctf_dynhash_insert.
(ctf_dvd_insert): Likewise.
(ctf_add_generic): Likewise.
(ctf_add_variable): Likewise.
* ctf-impl.h: Adjust declarations.
The testcase in the PR had a e_shstrndx section of type SHT_GROUP.
hdr->contents were initialized by setup_group rather than being read
from the file, thus last byte was not zero and string dereference ran
off the end of the buffer.
PR 24689
* elfcode.h (elf_object_p): Check type of e_shstrndx section.
When building gdb on s390x with -m31, we run into this error:
...
gdb/gdbserver/linux-s390-ipa.c: \
In function 'const target_desc* get_ipa_tdesc(int)':
gdb/gdbserver/linux-s390-ipa.c:371:18: error: 's390_te_ft_collect_regmap' \
was not declared in this scope
SET_REGMAP(s390_te_ft_collect_regmap, 0);
The offending line is part of this code snippet:
...
case S390_TDESC_GS:
SET_REGMAP(s390_te_ft_collect_regmap, 0);
return tdesc_s390_gs_linux64;
...
introduced in commit ce29f8439f "S390: Make IPA recognize tdescs with guarded
storage".
The snippet is part of an #ifdef __s390x__ construct, in the false branch, and
in the true branch we find a snippet introduced by the same commit:
...
case S390_TDESC_GS:
SET_REGMAP(s390x_te_ft_collect_regmap, 0);
return tdesc_s390x_gs_linux64;
...
which is paired with a comment update for s390x_te_ft_collect_regmap:
...
-/* Used for s390x-te-linux64, s390x-tevx-linux64. */
+/* Used for s390x-te-linux64, s390x-tevx-linux64, and
+ s390x-gs-linux64. */
static const int s390x_te_ft_collect_regmap[] = {
...
A similar comment update is added in the same commit for
s390_te_linux64_ft_collect_regmap:
...
-/* Used for s390-te-linux64, s390-tevx-linux64. */
+/* Used for s390-te-linux64, s390-tevx-linux64, and s390-gs-linux64. */
static const int s390_te_linux64_ft_collect_regmap[] = {
...
but not paired with any update.
Fix the build breaker by making the offending SET_REGMAP use the regmap
indicated by the comment.
...
- SET_REGMAP(s390_te_ft_collect_regmap, 0);
+ SET_REGMAP(s390_te_linux64_ft_collect_regmap, 0);
...
Build on s390x-linux with -m31.
gdb/gdbserver/ChangeLog:
2019-06-20 Tom de Vries <tdevries@suse.de>
* linux-s390-ipa.c (get_ipa_tdesc)[!__s390x__]: Use
s390_te_linux64_ft_collect_regmap for S390_TDESC_GS.
I noticed that tui_init_content_element is declared but never defined.
This removes the declaration. Tested by rebuilding. (I should have
merged this with the previous patch but I had forgotten that I found
two of these.)
gdb/ChangeLog
2019-06-19 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (tui_init_content_element): Don't declare.
I noticed that tui_init_win_info is declared but never defined. This
removes the declaration. Tested by rebuilding.
gdb/ChangeLog
2019-06-19 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (tui_init_win_info): Don't declare.
Building gdb with clang, I run into:
...
src/gdb/gdbserver/linux-low.c:6190:41: error: comparison of unsigned \
expression < 0 is always false [-Werror,-Wtautological-compare]
if (debug_write ("sigchld_handler\n",
sizeof ("sigchld_handler\n") - 1) < 0)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~
...
This regression is introduced by commit a7e559cc08 "gdbserver: Ensure all
debug output uses debug functions", which replaces calls to write with result
type ssize_t with calls to debug_write with result type size_t.
Fix this by making debug_write return ssize_t.
Build and reg-tested on x86_64-linux.
gdb/gdbserver/ChangeLog:
2019-06-19 Tom de Vries <tdevries@suse.de>
* debug.h (debug_write): Change return type to ssize_t.
* debug.c (debug_write): Same.
When compiling with gcc 4.8, we run into:
...
/usr/include/c++/4.8/bits/unordered_map.h💯18: required from \
‘class std::unordered_map<sect_offset, std::vector<sect_offset> >’
src/gdb/dwarf2read.h:260:5: required from here
/usr/include/c++/4.8/bits/hashtable_policy.h:1070:12: error: invalid use of \
incomplete type ‘struct std::hash<sect_offset>’
...
Fix this by setting the Hash template parameter of the unordered_map to
gdb::hash_enum<sect_offset>, rather than using the default
std::hash<sect_offset>.
Build and reg-tested on x86_64-linux with gcc 4.8.
gdb/ChangeLog:
2019-06-19 Tom de Vries <tdevries@suse.de>
* dwarf2read.h (abstract_to_concrete): Change type to
std::unordered_map<sect_offset, std::vector<sect_offset>,
gdb::hash_enum<sect_offset>>.
Pedro pointed out that the new length_cond.exp test did not have an
intro comment. This adds one.
gdb/testsuite/ChangeLog
2019-06-19 Tom Tromey <tromey@adacore.com>
* gdb.ada/length_cond.exp: Add intro comment.
gdb could crash when setting a breakpoint condition on a breakpoint
when using the Ada language. The problem occurred because the
ada_evaluate_subexp would try to evaluate the array to compute its
attributes, but evaluating can't really be done at this time.
This patch fixes the problem by arranging not to try to evaluate in
EVAL_AVOID_SIDE_EFFECTS mode when computing an attribute.
Tested on x86-64 Fedora 29. Because this is Ada-specific, and because
Joel approved it internally, I am checking it in.
gdb/ChangeLog
2019-06-19 Tom Tromey <tromey@adacore.com>
* ada-lang.c (ada_evaluate_subexp) <case OP_ATR_FIRST>: Handle
EVAL_AVOID_SIDE_EFFECTS specially.
gdb/testsuite/ChangeLog
2019-06-19 Tom Tromey <tromey@adacore.com>
* gdb.ada/length_cond.exp: New file.
* gdb.ada/length_cond/length_cond.adb: New file.
* gdb.ada/length_cond/pck.adb: New file.
* gdb.ada/length_cond/pck.ads: New file.
It occurred to me that there's no reason to make a new source
highlighter each time gdb needs to highlight some source code.
Instead, a single one can be created and then simply reused each time.
This patch implements this idea. Tested on x86-64 Fedora 29.
gdb/ChangeLog
2019-06-19 Tom Tromey <tromey@adacore.com>
* source-cache.c (highlighter): New global.
(source_cache::get_source_lines): Create a highlighter on demand.
Calls from functions that don't have a valid toc pointer in r2 (these
calls are marked with _NOTOC relocs) to functions that require r2
valid must go via the callee global entry point. This patch corrects
the condition the linker was using to detect functions that require r2
to be valid. Values of both zero and one in st_other local entry bits
mean a function doesn't care about r2.
* elf64-ppc.c (ppc64_elf_inline_plt): Correct st_other test for
functions that require r2 valid to use local entry.
(ppc64_elf_size_stubs, ppc64_elf_relocate_section): Likewise.
The deprecated_interactive_hook is not used within GDB. It is used in
gdbtk, however this patch removes that use:
https://sourceware.org/ml/insight/2019-q2/msg00001.html
So I think there is no longer a reason to keep this hook around.
This patch removes it. There should be no user visible changes after
this commit.
gdb/ChangeLog:
* defs.h (deprecated_interactive_hook): Delete declaration.
* interps.c (clear_interpreter_hooks): Remove use of
deprecated_interactive_hook.
* top.c (deprecated_interactive_hook): Delete definition.
* utils.c (maybe_quit): Remove use of deprecated_interactive_hook.
The target board fission.exp requires the gold linker (because it supports
--gdb-index).
When running the target board on a system where the default linker is not
gold, most tests will fail to compile.
Fix this by adding "-fuse-ld=gold" ( supported in gcc since version 4.8).
gdb/testsuite/ChangeLog:
2019-06-18 Tom de Vries <tdevries@suse.de>
* boards/fission.exp (debug_flags): Add "-fuse-ld=gold".
The test-case varval.exp fails here:
...
FAIL: gdb.dwarf2/varval.exp: print varval2
...
with boards readnow/cc-with-gdb-index/cc-with-debug-names, as well as if gdb
is build with -fsanitize=address -lasan.
The problem is that the abstract_to_concrete map in which we track the
association of abstract to concrete DIEs (for DW_OP_GNU_variable_value
support) has type std::unordered_map<die_info_ptr, std::vector<die_info_ptr>>,
and the die_info_ptrs that we register in the map may be invalid by the time
that we start to lookup DIEs in the map.
Fix this by using the sect_offset instead to identify the DIEs in the map.
Build and tested on x86_64-linux.
gdb/ChangeLog:
2019-06-18 Tom de Vries <tdevries@suse.de>
PR gdb/24515
* dwarf2read.h (abstract_to_concrete): Change type from
std::unordered_map<die_info_ptr, std::vector<die_info_ptr>> to
std::unordered_map<sect_offset, std::vector<sect_offset>>.
* dwarf2read.c (read_variable): Update.
(dwarf2_fetch_die_loc_sect_off): Update.
Section "Invoking GDB" of the manual states that if you try to launch
gdb with:
gdb program 1234
it will try to attach to the process with id 1234, unless there is a
file named 1234 in the current working directory, in which case it will
try to open that file as a core. In fact, when the second argument
starts with a digit, GDB tries to attach to process 1234 first, before
trying to open file 1234 as a core. So that last remark is not true and
therefore this patch removes it.
The same remark is present in the man page, so it is removed there too.
Section "Choosing Files" correctly states:
If the second argument begins with a decimal digit, GDB will first
attempt to attach to it as a process, and if that fails, attempt to
open it as a corefile.
so it is unchanged.
Finally, the man page has an additional detail compared to section
"Invoking GDB", regarding the use of the -p switch, so I added the same
detail to the "Invoking GDB" section.
gdb/doc/ChangeLog:
* gdb.texinfo (Invoking GDB): Remove sentence about how GDB
deals with a file that has the same name as the specified pid to
attach to. Add example using -p option.
(gdb man): Remove same sentence as in previous item.
Pedro suggested adding a comment to list0.h to explain the control
character.
Tested on x86-64 Fedora 29.
gdb/testsuite/ChangeLog
2019-06-18 Tom Tromey <tromey@adacore.com>
* gdb.base/list0.h: Add comment explaining control character.
When compiling gdb with '-lasan -fsanitizer=address' and running tests with:
- export ASAN_OPTIONS="detect_leaks=0:alloc_dealloc_mismatch=0", and
- a target board using local-board.exp, which sets sysroot to ""
we run into a heap-buffer-overflow in child_path for f.i. gdb.arch/amd64-byte:
...
==3997==ERROR: AddressSanitizer: heap-buffer-overflow on address \
0x60200002abcf at pc 0x5602acdf6872 bp 0x7ffe5237a090 sp 0x7ffe5237a080
READ of size 1 at 0x60200002abcf thread T0
#0 0x5602acdf6871 in child_path(char const*, char const*) \
gdb/common/pathstuff.c:161
#1 0x5602adb06587 in find_separate_debug_file gdb/symfile.c:1483
#2 0x5602adb06f2f in find_separate_debug_file_by_debuglink[abi:cxx11](...) \
gdb/symfile.c:1563
#3 0x5602ad13b743 in elf_symfile_read gdb/elfread.c:1293
#4 0x5602adb01cfa in read_symbols gdb/symfile.c:798
#5 0x5602adb03769 in syms_from_objfile_1 gdb/symfile.c:1000
#6 0x5602adb039d0 in syms_from_objfile gdb/symfile.c:1017
#7 0x5602adb04551 in symbol_file_add_with_addrs gdb/symfile.c:1124
#8 0x5602adb04ebf in symbol_file_add_from_bfd(...) gdb/symfile.c:1204
#9 0x5602ada5a78d in solib_read_symbols(...) gdb/solib.c:695
#10 0x5602ada5bdae in solib_add(char const*, int, int) gdb/solib.c:1004
#11 0x5602ada49bcd in enable_break gdb/solib-svr4.c:2394
#12 0x5602ada4dae9 in svr4_solib_create_inferior_hook gdb/solib-svr4.c:3028
#13 0x5602ada5d4f1 in solib_create_inferior_hook(int) gdb/solib.c:1215
#14 0x5602ad347f66 in post_create_inferior(target_ops*, int) \
gdb/infcmd.c:467
#15 0x5602ad348b3c in run_command_1 gdb/infcmd.c:663
#16 0x5602ad348e55 in run_command gdb/infcmd.c:686
#17 0x5602acd7d32b in do_const_cfunc gdb/cli/cli-decode.c:106
#18 0x5602acd84bfe in cmd_func(cmd_list_element*, char const*, int) \
gdb/cli/cli-decode.c:1892
#19 0x5602adc62a90 in execute_command(char const*, int) gdb/top.c:630
#20 0x5602ad5053e6 in catch_command_errors gdb/main.c:372
#21 0x5602ad507eb1 in captured_main_1 gdb/main.c:1138
#22 0x5602ad5081ec in captured_main gdb/main.c:1163
#23 0x5602ad508281 in gdb_main(captured_main_args*) gdb/main.c:1188
#24 0x5602ac9ddc3a in main gdb/gdb.c:32
#25 0x7f582b56eb96 in __libc_start_main \
(/lib/x86_64-linux-gnu/libc.so.6+0x21b96)
#26 0x5602ac9dda09 in _start \
(/home/smarchi/build/binutils-gdb/gdb/gdb+0x19a2a09)
0x60200002abcf is located 1 bytes to the left of 1-byte region \
[0x60200002abd0,0x60200002abd1)
allocated by thread T0 here:
#0 0x7f582e0e4b50 in __interceptor_malloc \
(/usr/lib/x86_64-linux-gnu/libasan.so.4+0xdeb50)
#1 0x5602acdd3656 in xmalloc gdb/common/common-utils.c:44
#2 0x5602aefe17d1 in xstrdup libiberty/xstrdup.c:34
#3 0x5602acdf61f6 in gdb_realpath(char const*) gdb/common/pathstuff.c:80
#4 0x5602adb06278 in find_separate_debug_file gdb/symfile.c:1444
#5 0x5602adb06f2f in find_separate_debug_file_by_debuglink[abi:cxx11](...) \
gdb/symfile.c:1563
#6 0x5602ad13b743 in elf_symfile_read gdb/elfread.c:1293
#7 0x5602adb01cfa in read_symbols gdb/symfile.c:798
#8 0x5602adb03769 in syms_from_objfile_1 gdb/symfile.c:1000
#9 0x5602adb039d0 in syms_from_objfile gdb/symfile.c:1017
#10 0x5602adb04551 in symbol_file_add_with_addrs gdb/symfile.c:1124
#11 0x5602adb04ebf in symbol_file_add_from_bfd(...) gdb/solib.c:695
#13 0x5602ada5bdae in solib_add(char const*, int, int) gdb/solib.c:1004
#14 0x5602ada49bcd in enable_break gdb/solib-svr4.c:2394
#15 0x5602ada4dae9 in svr4_solib_create_inferior_hook gdb/solib-svr4.c:3028
#16 0x5602ada5d4f1 in solib_create_inferior_hook(int) gdb/solib.c:1215
#17 0x5602ad347f66 in post_create_inferior(target_ops*, int) \
gdb/infcmd.c:467
#18 0x5602ad348b3c in run_command_1 gdb/infcmd.c:663
#19 0x5602ad348e55 in run_command gdb/infcmd.c:686
#20 0x5602acd7d32b in do_const_cfunc gdb/cli/cli-decode.c:106
#21 0x5602acd84bfe in cmd_func(cmd_list_element*, char const*, int) \
gdb/cli/cli-decode.c:1892
#22 0x5602adc62a90 in execute_command(char const*, int) gdb/top.c:630
#23 0x5602ad5053e6 in catch_command_errors gdb/main.c:372
#24 0x5602ad507eb1 in captured_main_1 gdb/main.c:1138
#25 0x5602ad5081ec in captured_main gdb/main.c:1163
#26 0x5602ad508281 in gdb_main(captured_main_args*) gdb/main.c:1188
#27 0x5602ac9ddc3a in main gdb/gdb.c:32
#28 0x7f582b56eb96 in __libc_start_main \
(/lib/x86_64-linux-gnu/libc.so.6+0x21b96)
SUMMARY: AddressSanitizer: heap-buffer-overflow gdb/common/pathstuff.c:161 \
in child_path(char const*, char const*)
Shadow bytes around the buggy address:
0x0c047fffd520: fa fa fd fd fa fa fd fd fa fa fd fa fa fa fd fa
0x0c047fffd530: fa fa fd fa fa fa fd fa fa fa fd fa fa fa fd fa
0x0c047fffd540: fa fa fd fa fa fa fd fa fa fa fd fa fa fa fd fa
0x0c047fffd550: fa fa fd fd fa fa fd fd fa fa fd fd fa fa fd fa
0x0c047fffd560: fa fa fd fa fa fa fd fa fa fa fd fa fa fa 00 00
=>0x0c047fffd570: fa fa 07 fa fa fa 00 fa fa[fa]01 fa fa fa fa fa
0x0c047fffd580: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
0x0c047fffd590: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
0x0c047fffd5a0: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
0x0c047fffd5b0: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
0x0c047fffd5c0: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa
Shadow byte legend (one shadow byte represents 8 application bytes):
Addressable: 00
Partially addressable: 01 02 03 04 05 06 07
Heap left redzone: fa
Freed heap region: fd
Stack left redzone: f1
Stack mid redzone: f2
Stack right redzone: f3
Stack after return: f5
Stack use after scope: f8
Global redzone: f9
Global init order: f6
Poisoned by user: f7
Container overflow: fc
Array cookie: ac
Intra object redzone: bb
ASan internal: fe
Left alloca redzone: ca
Right alloca redzone: cb
==3997==ABORTING
...
The direct cause is that child_path gets called with parent == "", so this
test:
...
if (IS_DIR_SEPARATOR (parent[parent_len - 1]))
...
accesses parent[-1].
[ There is an open discussion (1) about whether an empty sysroot should indeed
be represented internally as "". But this patch focuses on fixing the
heap-buffer-overflow without any redesign. ]
Fix this by guarding the test with 'parent_len > 0'.
Note that the fix makes child_path behave the same for:
- parent == "/" && child == "/foo" (returns "foo")
- parent == "" and child == "/foo" (returns "foo").
Build and reg-tested on x86_64-linux.
(1) https://sourceware.org/ml/gdb-patches/2019-05/msg00193.html
gdb/ChangeLog:
2019-06-17 Tom de Vries <tdevries@suse.de>
PR gdb/24617
* common/pathstuff.c (child_path): Make sure parent_len > 0 before
accessing parent[parent_len - 1].
The loc_hash_table should only contain local ifunc symbols. The current
code already aborts if there is anything else and for defined ifunc
symbols elfNN_aarch64_allocate_dynrelocs is a no-op.
bfd/ChangeLog:
* elfnn-aarch64.c (elfNN_aarch64_allocate_local_dynrelocs): Remove.
(elfNN_aarch64_size_dynamic_sections): Remove loc_hash_table traversal
with elfNN_aarch64_allocate_local_dynrelocs.
Atm gdb-add-index.exp fails with target board cc-with-dwz-m.
Fix this by updating gdb/contrib/gdb-add-index.sh to handle a dwz-m-ed
executable.
Tested on x86_64-linux.
gdb/ChangeLog:
2019-06-16 Tom de Vries <tdevries@suse.de>
PR gdb/24445
* contrib/gdb-add-index.sh: Update to handle dwz-m-ed executable.
I found a few spots that directly cast a tui_win_info to a
tui_gen_win_info. However, I think it's a bit better here to take the
address of the "generic" member. As far as I know, nothing relies on
being able to downcast here, so this gives us the freedom to rearrange
the structure.
gdb/ChangeLog
2019-06-16 Tom Tromey <tom@tromey.com>
* tui/tui-wingeneral.c (tui_unhighlight_win, tui_highlight_win)
(make_all_visible): Use address of member.
I found a few more spots that unnecessarily cast NULL to a pointer
type. My earlier search included a "*" in the cast, but these use a
typedef to a pointer type instead. This patch removes these casts.
gdb/ChangeLog
2019-06-16 Tom Tromey <tom@tromey.com>
* tui/tui-data.c (tui_clear_win_detail, init_win_info)
(tui_free_window, free_content, free_content_elements): Remove
unnecessary cast.
* tui/tui-windata.c (tui_display_all_data): Remove unnecessary
cast.
* tui/tui-regs.c (tui_show_register_group)
(tui_display_registers_from, tui_display_reg_element_at_line):
Remove unnecessary cast.
In the following commit:
commit 7feb7d068a
Date: Mon May 11 12:08:03 2009 +0000
The last useful uses of normal_mask in linux-nat.c were removed, since
then this variable has sat around being initialised, but never used.
There should be no user visible changes after this commit.
gdb/ChangeLog:
* linux-nat.c (normal_mask): Delete.
(_initialize_linux_nat): Don't initialise normal_mask.
PR 24445 ("dwz multifile index not written to index cache") exposed the
fact that we are not doing things right when we generate an index for an
object file that has is linked to a dwz file. The same happens whether
the index is generated with the intent of populating the index cache or
using the save gdb-index command.
The problem can be observed when running these tests with the
cc-with-dwz-m board:
FAIL: gdb.base/index-cache.exp: test_cache_enabled_hit: check index-cache stats
FAIL: gdb.dwarf2/gdb-index.exp: index used
FAIL: gdb.dwarf2/gdb-index.exp: index used after symbol reloading
When generating the index for such file and inspecting the CU list of the
resulting index (with readelf --debug-dump=gdb_index), we can see something
like:
CU table:
[ 0] 0x0 - 0xb9
[ 1] 0x0 - 0x44
This is supposed to be a sorted list of the ranges of all CUs in the
file this index represents, so already having some overlap is a red
flag. It turns out that we save the ranges of CUs coming from both the
main file and the dwz file in the same index.
After digging a little bit, it became quite obvious that the index in
the main file should only list the CUs present in the main file, and a
separate index should be generated for the dwz file, listing the CUs
present in that file.
First, that's what happens if you run dwz on a file that already has a
GDB index embedded. Second, dwarf2read.c has code to read an index from
a dwz file. The index in the dwz file is actually required to be
present, if the main file has an index.
So this patch changes write_psymtabs_to_index to generate an index for
the dwz file, if present. That index only contains a CU list, just like
what the dwz tool does when processing a file that already contains an
index.
Some notes about the implementation:
- The file management (creating a temp file, make sure it's
close/removed on error - in the right order) is a bit heavy in
write_psymtabs_to_index, and I needed to add a third file. I factored
this pattern in a separate class, index_wip_file.
- It became a bit tedious to keep the call to assert_file_size in
write_psymtabs_to_index, write_gdbindex would have had to return two
sizes. Instead, I moved the calls to assert_file_size where the file
is written. The downside is that we lose the filename at this point,
but it was only used for the very improbable case of ftell failing, so
I think it's not a problem.
- The actual writing of the index file is factored out to
write_gdbindex_1, so it can be re-used for both index files.
- While the "save gdb-index" command will now write two .gdb-index
files, this patch does not update the gdb-add-index.sh script, this
will come in a later patch.
gdb/ChangeLog:
YYYY-MM-DD Simon Marchi <simon.marchi@efficios.com>
PR gdb/24445
* dwarf-index-write.h (write_psymtabs_to_index): Add
dwz_basename parameter.
* dwarf-index-write.c (write_gdbindex): Move file writing to
write_gdbindex_1. Change return type void.
(assert_file_size): Move up, remove filename parameter.
(write_gdbindex_1): New function.
(write_debug_names): Change return type to void, call
assert_file_size.
(struct index_wip_file): New struct.
(write_psymtabs_to_index): Add dwz_basename parameter. Move
file logic to index_wip_file. Write index for dwz file if
needed.
(save_gdb_index_command): Pass basename of dwz file, if present.
* dwarf-index-cache.c (index_cache::store): Obtain and pass
build-id of dwz file, if present.
* dwarf2read.c (struct dwz_file): Move to dwarf2read.h.
(dwarf2_get_dwz_file): Likewise.
* dwarf2read.h (struct dwz_file): Move from dwarf2read.c.
(dwarf2_get_dwz_file): Likewise.
gdb/testsuite/ChangeLog:
YYYY-MM-DD Tom de Vries <tdevries@suse.de>
PR gdb/24445
* gdb.dwarf2/gdb-index.exp (add_gdb_index): Update dwz file with
generated index.
I noticed a couple of spots using concat that could use xstrdup
instead. This patch fixes these.
gdb/ChangeLog
2019-06-16 Tom Tromey <tom@tromey.com>
* coffread.c (process_coff_symbol): Use xstrdup.
* value.c (create_internalvar): Use xstrdup.
In bug gdb/24686 a testsuite failure was reported, this failure was
actually just another instance of bug gdb/24541, however, due to the
non-deterministic nature of bug gdb/24541 the testsuite pattern that
was intended to catch this bug failed.
This commit adds a second pattern to help detect gdb/24541, which
should change the FAIL reported in gdb/24686 into a KFAIL.
gdb/testsuite/ChangeLog:
PR gdb/24686
* gdb.mi/mi-catch-cpp-exceptions.exp: Add an extra pattern to
improve detection of bug gdb/24541.
I found a few spots in the TUI that were NULL-checking the result of
XNEW. This cannot return NULL, so this patch removes the checks.
gdb/ChangeLog
2019-06-16 Tom Tromey <tom@tromey.com>
* tui/tui-data.c (tui_alloc_generic_win_info)
(tui_alloc_win_info, tui_add_content_elements): Remove NULL
checks.
