build_address_symbolic contains some code which causes it to
prefer the minsym over the the function symbol in certain cases.
The cases where this occurs are the same as the "certain pathological
cases" that used to exist in find_frame_funname().
This commit largely disables that code; it will only prefer the
minsym when the address of minsym is identical to that of the address
under consideration AND the function address for the symbtab sym is
not the same as the address under consideration.
So, without this change, when using the dw2-ranges-func-lo-cold
executable from the gdb.dwarf2/dw2-ranges-func.exp test, GDB exhibits
the following behavior:
(gdb) x/5i foo_cold
0x40110d <foo+4294967277>: push %rbp
0x40110e <foo+4294967278>: mov %rsp,%rbp
0x401111 <foo+4294967281>: callq 0x401106 <baz>
0x401116 <foo+4294967286>: nop
0x401117 <foo+4294967287>: pop %rbp
On the other hand, still without this change, using the
dw2-ranges-func-hi-cold executable from the same test, GDB
does this instead:
(gdb) x/5i foo_cold
0x401128 <foo_cold>: push %rbp
0x401129 <foo_cold+1>: mov %rsp,%rbp
0x40112c <foo_cold+4>: callq 0x401134 <baz>
0x401131 <foo_cold+9>: nop
0x401132 <foo_cold+10>: pop %rbp
This is inconsistent behavior. When foo_cold is at a lower
address than the function's entry point, the symtab symbol (foo)
is displayed along with a large positive offset which would wrap
around the address space if the address space were only 32 bits wide.
(A later patch fixes this problem by displaying negative offsets.)
This commit makes the behavior uniform for both the "lo-cold" and
"hi-cold" cases:
lo-cold:
(gdb) x/5i foo_cold
0x40110d <foo_cold>: push %rbp
0x40110e <foo-18>: mov %rsp,%rbp
0x401111 <foo-15>: callq 0x401106 <baz>
0x401116 <foo-10>: nop
0x401117 <foo-9>: pop %rbp
hi-cold:
(gdb) x/5i foo_cold
0x401128 <foo_cold>: push %rbp
0x401129 <foo+35>: mov %rsp,%rbp
0x40112c <foo+38>: callq 0x401134 <baz>
0x401131 <foo+43>: nop
0x401132 <foo+44>: pop %rbp
In both cases, the symbol shown for the address at which foo_cold
resides is shown as <foo_cold>. Subsequent offsets are shown as
either negative or positive offsets from the entry pc for foo.
When disassembling a function, care must be taken to NOT display
<+0> as the offset for the second range. For this reason, I found
it necessary to add the "prefer_sym_over_minsym" parameter to
build_address_symbolic. The type of this flag is a bool; do_demangle
ought to be a bool also, so I made this change at the same time.
gdb/ChangeLog:
* valprint.h (build_address_symbolic): Add "prefer_sym_over_minsym"
parameter. Change type of "do_demangle" to bool.
* disasm.c (gdb_pretty_print_disassembler::pretty_print_insn):
Pass suitable "prefer_sym_over_minsym" flag to
build_address_symbolic(). Don't output "+" for negative offsets.
* printcmd.c (print_address_symbolic): Update invocation of
build_address_symbolic to include a "prefer_sym_over_minsym"
flag.
(build_address_symbolic): Add "prefer_sym_over_minsym" parameter.
Restrict cases in which use of minimal symbol is preferred to that
of a found symbol. Update comments.
The discussion on gdb-patches which led to this patch may be found
here:
https://www.sourceware.org/ml/gdb-patches/2019-05/msg00018.html
Here's a brief synopsis/analysis:
Eli Zaretskii, while debugging a Windows emacs executable, found
that functions comprised of more than one (non-contiguous)
address range were not being displayed correctly in a backtrace. This
is the example that Eli provided:
(gdb) bt
#0 0x76a63227 in KERNELBASE!DebugBreak ()
from C:\Windows\syswow64\KernelBase.dll
#1 0x012e7b89 in emacs_abort () at w32fns.c:10768
#2 0x012e1f3b in print_vectorlike.cold () at print.c:1824
#3 0x011d2dec in print_object (obj=<optimized out>, printcharfun=XIL(0),
escapeflag=true) at print.c:2150
The function print_vectorlike consists of two address ranges, one of
which contains "cold" code which is expected to not execute very often.
There is a minimal symbol, print_vectorlike.cold.65, which is the address
of the "cold" range.
GDB is prefering this minsym over the the name provided by the
DWARF info due to some really old code in GDB which handles
"certain pathological cases". This comment reads as follows:
/* In certain pathological cases, the symtabs give the wrong
function (when we are in the first function in a file which
is compiled without debugging symbols, the previous function
is compiled with debugging symbols, and the "foo.o" symbol
that is supposed to tell us where the file with debugging
symbols ends has been truncated by ar because it is longer
than 15 characters). This also occurs if the user uses asm()
to create a function but not stabs for it (in a file compiled
with -g).
So look in the minimal symbol tables as well, and if it comes
up with a larger address for the function use that instead.
I don't think this can ever cause any problems; there
shouldn't be any minimal symbols in the middle of a function;
if this is ever changed many parts of GDB will need to be
changed (and we'll create a find_pc_minimal_function or some
such). */
In an earlier version of this patch, I had left the code for the
pathological case intact, but those who reviwed that patch recommended
removing it. So that's what I've done - I've removed it.
gdb/ChangeLog:
* stack.c (find_frame_funname): Remove code which preferred
minsym over symtab sym in "certain pathological cases".
To recap the bug report:
Commit a068643 introduced a small typo that breaks the gdb build on OpenBSD.
Line 38 of obsd-nat.c needs to be changed from std::sring to std::string.
gdb/ChangeLog
2019-07-26 Brian Callahan <bcallah@openbsd.org>
PR gdb/24839:
* gdb/obsd-nat.c (obsd_nat_target::pid_to_str): Fix typo in return
type.
I ran into this error:
...
ERROR: tcl error sourcing gdb/testsuite/gdb.arch/i386-pkru.exp.
ERROR: missing "
while executing
"untested ""
invoked from within
"if { [prepare_for_testing "failed to prepare" ${testfile} ${srcfile} \
[list debug additional_flags=${comp_flags}]] } {
untested "failed to c..."
(file "gdb/testsuite/gdb.arch/i386-pkru.exp" line 25)
invoked from within
...
caused by:
...
untested "failed to compile x86 PKEYS test.
...
Fix the unterminated string.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2019-07-26 Tom de Vries <tdevries@suse.de>
* gdb.arch/i386-pkru.exp: Fix unterminated string.
PR 24798
* dwarf.c (process_cu_tu_index): Avoid integer overflow on 64-bit
systems by casting ncols and nslots expressions to size_t. Display
number of columns and slots before giving up due to buffer overflow.
Use %u to display unsigned ints. Perform more pointer wrap tests.
This field effectively became usused a long time ago, perhaps as early
as 1994.
* elf-bfd.h (struct output_elf_obj_tdata): Delete "linker" field.
(elf_linker): Don't define.
* elflink.c (bfd_elf_final_link): Don't set elf_linker.
Also fixes the date in the changelog of my last commit.
gdb/ChangeLog:
2019-07-25 Christian Biesinger <cbiesinger@google.com>
* python/py-objfile.c (add_separate_debug_file): Fix comment about
this function's Python signature.
On a system without SDT probes in libstdc++, we run into:
...
FAIL: gdb.mi/mi-catch-cpp-exceptions.exp: all with invalid regexp: run until \
breakpoint in main (unknown output after running)
...
The test-case uses a regexp argument for the catch throw/rethrow/catch
command, which is only supported on systems with SDT probes in libstdc++.
Fix this by marking the portions of the test-case that use a regexp argument
as unsupported on a system without SDT probes.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2019-07-25 Tom de Vries <tdevries@suse.de>
PR testsuite/24830
* gdb.mi/mi-catch-cpp-exceptions.exp: Call
mi_skip_libstdcxx_probe_tests, and skip unsupported tests.
* lib/gdb.exp (skip_libstdcxx_probe_tests_prompt): Factor out of ...
(skip_libstdcxx_probe_tests): ... here.
* lib/mi-support.exp (mi_skip_libstdcxx_probe_tests): New proc.
This has no behavior change in itself, but allows a future patch
to add a function to the Python API to look up symbols in the
static block.
gdb/ChangeLog:
2019-07-24 Christian Biesinger <cbiesinger@google.com>
* compile/compile-object-load.c (compile_object_load): Pass GLOBAL_SCOPE.
* solib-spu.c (spu_lookup_lib_symbol): Pass GLOBAL_SCOPE.
* solib-svr4.c (elf_lookup_lib_symbol): Pass GLOBAL_SCOPE.
* symtab.c (lookup_global_symbol_from_objfile): Add a scope parameter.
* symtab.h (lookup_global_symbol_from_objfile): Likewise.
When running gdb.objc/objcdecode.exp we get:
...
objcdecode.m: In function '-[Decode multipleDef]':
objcdecode.m:14:3: warning: incompatible implicit declaration of built-in \
function 'printf'
printf("method multipleDef\n");
^~~~~~
objcdecode.m:14:3: note: include '<stdio.h>' or provide a declaration of \
'printf'
...
Fix this in the three gdb.objc/*.m test-cases by including stdio.h.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2019-07-24 Tom de Vries <tdevries@suse.de>
PR testsuite/24807
* gdb.objc/basicclass.m: Include stdio.h.
* gdb.objc/nondebug.m: Same.
* gdb.objc/objcdecode.m: Same.
Revert
commit 8c728a9d93
Author: Martin Liska <mliska@suse.cz>
Date: Mon Jul 22 14:23:32 2019 +0200
Remove tests that test __gnu_lto_v1 symbol.
since outputs of these tests are used by later tests. Check the normal
symbol, foo, instead of __gnu_lto_v.*, which GCC stopped emitting after
r273662.
* testsuite/ld-plugin/lto-3r.d: Restored. Check foo instead
of __gnu_lto_v.*.
* testsuite/ld-plugin/lto-5r.d: Likewise.
* testsuite/ld-plugin/lto.exp: Run lto-3r and lto-5r tests.
When running gdb.base/infoline-reloc-main-from-zero.exp, I see:
...
Running gdb/testsuite/gdb.base/infoline-reloc-main-from-zero.exp ...
gdb compile failed, ld: infoline-reloc-main-from-zero: \
not enough room for program headers, try linking with -N
ld: final link failed: bad value
collect2: error: ld returned 1 exit status
UNTESTED: gdb.base/infoline-reloc-main-from-zero.exp: infoline-reloc-main-from-zero.exp
UNTESTED: gdb.base/infoline-reloc-main-from-zero.exp: failed to compile
...
Fix this by following the suggestion:
...
-set opts {debug "additional_flags=-nostdlib -emain -Wl,-Ttext=0x00"}
+set opts {debug "additional_flags=-nostdlib -emain -Wl,-Ttext=0x00 -Wl,-N"}
...
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2019-07-24 Tom de Vries <tdevries@suse.de>
PR testsuite/24612
* gdb.base/infoline-reloc-main-from-zero.exp: Add -Wl,-N to
additional_flags.
New instruction are added, and some of them are overlapping. Update
disassembler to correctly recognize them. Introduce nps400 option.
opcodes/
xxxx-xx-xx Claudiu Zissulescu <claziss@synopsys.com>
* arc-dis.c (skip_this_opcode): Check also for 0x07 major opcodes,
and MPY class instructions.
(parse_option): Add nps400 option.
(print_arc_disassembler_options): Add nps400 info.
gas/
xxxx-xx-xx Claudiu Zissulescu <claziss@synopsys.com>
* testsuite/gas/arc/nps400-6.d: Update test.
Add linker relaxation. The first relaxation added is converting
GOTPC32 to PCREL relocations. This relaxation doesn't change the size of
the binary.
bfd/
xxxx-xx-xx Claudiu Zissulescu <claziss@synopsys.com>
* elf32-arc.c (bfd_get_32_me): New function.
(bfd_put_32_me): Likewise.
(arc_elf_relax_section): Likewise.
(bfd_elf32_bfd_relax_section): Define.
ld/testsuite/
xxxx-xx-xx Claudiu Zissulescu <claziss@synopsys.com>
* ld-arc/relax-local-pic.d: New test.
* ld-arc/relax-local-pic.s: New file.
H8/300H general register name "ER0" - "ER7".
But gdb using "R0" - "R7".
This changes register name "ER0" - "ER7" in h8300h machine mode.
gdb/ChangeLog:
* h8300-tdep.c (h8300_register_name_common): New.
h8300_register_name): Use h8300_register_name_common.
(h8300s_register_name): Likewise.
(h8300sx_register_name): Likewise.
(h8300h_register_nam): New.
(h8300_gdbarch_init): Use h8300h_register_name in h8300h machine.
I missed some early exits from final_write_processing that mean
_bfd_elf_final_write_processing could be missed.
* elf-vxworks.c (elf_vxworks_final_write_processing): Don't return
early.
* elf32-arc.c (arc_elf_final_write_processing): Likewise.
* elf32-xtensa.c (elf_xtensa_final_write_processing): Likewise.
We currently have these FAILs:
...
FAIL: gdb.multi/tids.exp: two inferiors: info threads -1
FAIL: gdb.multi/tids.exp: two inferiors: info threads -$one
...
because we're expecting:
...
Invalid thread ID: -1
...
but instead we have:
...
Unrecognized option at: -1
...
This error message for info threads has changed since commit 54d6600669
'Make "info threads" use the gdb::option framework'.
Update the test accordingly.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2019-07-24 Tom de Vries <tdevries@suse.de>
PR testsuite/24831
* gdb.multi/tids.exp: Update error messages for info threads.
On openSUSE Leap 15.0, I get:
...
FAIL: gdb.base/info-types.exp: l=c: info types
FAIL: gdb.base/info-types.exp: l=c++: info types
...
because the info type command prints info for files info-types.c, stddef.h,
elf-init.c and init.c, while the regexp in the test-case expect only info for
info-types.c.
Fix this by extending the regexp.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2019-07-24 Tom de Vries <tdevries@suse.de>
* gdb.base/info-types.exp: Allow info types to print info for more than
one file.
PR 24818
* objdump.c (is_relocateable): Delete.
(load_specific_debug_section): Test the abfd for relocations
directly, rather than relying upon is_relocateable.
(dump_dwarf): Delete initlialization of is_relocateable.
When running multidictionary.exp in conjunction with:
...
$ stress -c $(($(cat /proc/cpuinfo | grep -c "^processor") + 1))
...
we get:
...
Running gdb/testsuite/gdb.dwarf2/multidictionary.exp ...
ERROR: Couldn't load multidictionary into gdb.
=== gdb Summary ===
nr of unresolved testcases 1
...
The multidictionary test-case needs -readnow, and achieves this using:
...
gdb_spawn_with_cmdline_opts "-readnow"
gdb_load
...
but the initial gdb prompt is not read. Usually, the following gdb_load
command accidentally consumes that initial prompt (at the gdb_expect for the
kill command in gdb_file_cmd). But under high load, that doesn't happen and
we run into the error.
Fix this by consuming the initial gdb prompt after spawning gdb.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2019-07-23 Tom de Vries <tdevries@suse.de>
PR testsuite/24842
* gdb.dwarf2/multidictionary.exp: Consume initial prompt after
gdb_spawn_with_cmdline_opts.
GDB is not able to execute "step" command on function calls of Armv8-M cmse secure entry functions.
Everytime GNU linker come across definition of any cmse secure entry function in object file(s),
it creates two new instructions secure gateway (sg) and original branch destination (b.w),
place those two instructions in ".gnu.sgstubs" section of executable.
Any function calls to these cmse secure entry functions is re-directed through secure gateway (sg)
present in ".gnu.sgstubs" section.
Example:
Following is a function call to cmse secure entry function "foo":
...
bl xxxx <foo> --->(a)
...
<foo>
xxxx: push {r7, lr}
GNU linker on finding out "foo" is a cmse secure entry function, created sg and b.w instructions and
place them in ".gnu.sgstubs" section (marked by c).
The "bl" instruction (marked by a) which is a call to cmse secure entry function is modified by GNU linker
(as marked by b) and call flow is re-directly through secure gateway (sg) in ".gnu.sgstubs" section.
...
bl yyyy <foo> ---> (b)
...
section .gnu.sgstubs: ---> (c)
yyyy <foo>
yyyy: sg // secure gateway
b.w xxxx <__acle_se_foo> // original_branch_dest
...
0000xxxx <__acle_se_foo>
xxxx: push {r7, lr} ---> (d)
On invoking GDB, when the control is at "b" and we pass "step" command, the pc returns "yyyy"
(sg address) which is a trampoline and which does not exist in source code. So GDB jumps
to next line without jumping to "__acle_se_foo" (marked by d).
The above details are published on the Arm website [1], please refer to section 5.4 (Entry functions)
and section 3.4.4 (C level development flow of secure code).
[1] https://developer.arm.com/architectures/cpu-architecture/m-profile/docs/ecm0359818/latest/armv8-m-security-extensions-requirements-on-development-tools-engineering-specification
This patch fixes above problem by returning target pc "xxxx" to GDB on executing "step"
command at "b", so that the control jumps to "__acle_se_foo" (marked by d).
gdb/ChangeLog:
* arm-tdep.c (arm_skip_cmse_entry): New function.
(arm_is_sgstubs_section): New function.
(arm_skip_stub): Add call to arm_skip_cmse_entry function.
gdb/testsuite/ChangeLog:
* gdb.arch/arm-cmse-sgstubs.c: New test.
* gdb.arch/arm-cmse-sgstubs.exp: New file.
ld/ChangeLog:
2019-07-22 Martin Liska <mliska@suse.cz>
* testsuite/ld-plugin/lto-3r.d: Remove.
* testsuite/ld-plugin/lto-5r.d: Remove.
* testsuite/ld-plugin/lto.exp: Do not run lto-3r and lto-5r
tests.
When SHF_GNU_MBIND was added in the SHF_LOOS to SHF_HIOS range, it
should have required ELFOSABI_GNU since these flags are already in use
by other OSes. HPUX SHF_HP_TLS in fact has the same value. That
means no place in binutils should test SHF_GNU_MBIND without first
checking OSABI, and SHF_GNU_MBIND should not be set without also
setting OSABI. At least, that's the ideal, but the patch accepts
SHF_GNU_MBIND on ELFOSABI_NONE object files since gas didn't always
set OSABI. However, to reinforce the fact that SHF_GNU_MBIND isn't
proper without a non-zero OSABI, readelf will display the flag as
LOOS+0 if OSABI isn't set.
The clash with SHF_HP_TLS means that hppa64-linux either has that flag
on .tbss sections or supports GNU_MBIND, not both. (hppa64-linux
users, if there are any, may have noticed that GNU ld since 2017
mysteriously aligned their .tbss sections to a 4k boundary. That was
one consequence of SHF_HP_TLS being blindly interpreted as
SHF_GNU_MBIND.) Since it seems that binutils, gdb, gcc, glibc, and
the linux kernel don't care about SHF_HP_TLS I took that flag out of
.tbss for hppa64-linux.
bfd/
* elf-bfd.h (enum elf_gnu_osabi): Add elf_gnu_osabi_mbind.
* elf.c (_bfd_elf_make_section_from_shdr): Set elf_gnu_osabi_mbind.
(get_program_header_size): Formatting. Only test SH_GNU_MBIND
when elf_gnu_osabi_mbind is set.
(_bfd_elf_map_sections_to_segments): Likewise.
(_bfd_elf_init_private_section_data): Likewise.
(_bfd_elf_final_write_processing): Update comment.
* elf64-hppa.c (elf64_hppa_special_sections): Move .tbss entry.
(elf_backend_special_sections): Define without .tbss for linux.
binutils/
* readelf.c (get_parisc_segment_type): Split off hpux entries..
(get_ia64_segment_type): ..and these..
(get_hpux_segment_type): ..to here.
(get_segment_type): Condition GNU_MBIND on osabi. Use
get_hpux_segment_type.
(get_symbol_binding): Do not print UNIQUE for ELFOSABI_NONE.
(get_symbol_type): Do not print IFUNC for ELFOSABI_NONE.
gas/
* config/obj-elf.c (obj_elf_change_section): Don't emit a fatal
error for non-SHF_ALLOC SHF_GNU_MBIND here.
(obj_elf_parse_section_letters): Return SHF_GNU_MBIND in new
gnu_attr param.
(obj_elf_section): Adjust obj_elf_parse_section_letters call.
Formatting. Set SHF_GNU_MBIND and elf_osabi from gnu_attr.
Emit normal error for non-SHF_ALLOC SHF_GNU_MBIND and wrong osabi.
(obj_elf_type): Set elf_osabi for ifunc.
* testsuite/gas/elf/section12a.d: xfail msp430 and hpux.
* testsuite/gas/elf/section12b.d: Likewise.
* testsuite/gas/elf/section13.d: Likewise.
* testsuite/gas/elf/section13.l: Adjust expected error.
ld/
* emultempl/elf32.em (gld${EMULATION_NAME}_place_orphan): Condition
SHF_GNU_MBIND on osabi. Set output elf_gnu_osabi_mbind.
Also restore them somewhat closer to the original. They originally
failed on many targets, and a month later I "simplified" them as part
of a larger patch fixing other failing tests. That unfortunately lost
their main purpose, which was to test TLS layout.
* testsuite/ld-elf/size-1.d,
* testsuite/ld-elf/size-1.s,
* testsuite/ld-elf/size-1.t: New test.
* testsuite/ld-elf/size-2.d,
* testsuite/ld-elf/size-2.s,
* testsuite/ld-elf/size-2.t: New test.
* testsuite/ld-scripts/size-1.d,
* testsuite/ld-scripts/size-1.s,
* testsuite/ld-scripts/size-1.t,
* testsuite/ld-scripts/size-2.s,
* testsuite/ld-scripts/size-2.d,
* testsuite/ld-scripts/size-2.t,
* testsuite/ld-scripts/size.exp: Delete.
When running gdb.mi/mi-complete.exp in conjunction with:
...
$ stress -c $(($(cat /proc/cpuinfo | grep -c "^processor") + 1))
...
we get less than 50% full passes:
...
$ for n in $(seq 1 100); do \
make V=1 -O check \
'RUNTESTFLAGS=gdb.mi/mi-complete.exp --target_board=unix'; \
done 2>&1 \
| grep "expected passes" | sort | uniq -c
45 # of expected passes 7
9 # of expected passes 8
46 # of expected passes 9
...
A diff between a passing and failing gdb.log shows this difference:
...
-&"set max-completions 1\n"
2-complete br
+&"set max-completions 1\n"
...
The problem is that the test-case issues the "set max-completion <n>" command,
and without waiting for the output issues a next command, and tries to parse
the results of both commands, expecting a specific interleaving of the various
output streams.
Fix the FAIL by waiting for the result of the "set max-completion <n>" command
before issuing another command.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2019-07-23 Tom de Vries <tdevries@suse.de>
PR testsuite/24711
* gdb.mi/mi-complete.exp: Wait for "set max-completions" result before
issuing next command.
In https://sourceware.org/ml/gdb-patches/2019-07/msg00509.html, Jan
pointed out that clang points out that
make_invisible_and_set_new_height self-assigns "height".
This patch fixes the bug by renaming the formal parameter.
gdb/ChangeLog
2019-07-22 Tom Tromey <tom@tromey.com>
* tui/tui-win.c (tui_win_info::make_invisible_and_set_new_height):
Don't self-assign.
