Sandra Loosemore [Fri, 31 Jan 2020 18:32:48 +0000 (10:32 -0800)]
nios2: recognize %gotoff relocation in assembler
The nios2 ABI documentation lists %gotoff as assembler syntax for the
R_NIOS2_GOTOFF relocation, used to represent a 32-bit GOT-relative offset
in data sections. This was previously unimplemented in GAS.
Andre Vieira [Fri, 31 Jan 2020 17:09:27 +0000 (17:09 +0000)]
arm: PR gas/25472 Enable DSP instructions with +mve
We noticed +mve was not enabling DSP instructions as it should, reported in PR
25472.
The MVE architecture extension for Armv8.1-M Mainline implies DSP extensions.
This patch reflects that in the '+mve' command line option.
gas/ChangeLog:
2020-01-31 Andre Vieira <andre.simoesdiasvieira@arm.com>
PR gas/25472
* config/tc-arm.c (armv8m_main_ext_table): Refactored +dsp adding.
(armv8_1m_main_ext_table): Refactored +dsp adding and enabled dsp for
+mve.
* testsuite/gas/arm/mve_dsp.d: New test.
[ARM]: Add support for vldmia/vldmdb/vstmia/vstmdb instructions in MVE.
This patch adds support for assembly instructions vldmia, vldmdb, vstmia
and vstmdb in MVE. This instructions are already supported for Armv8-M
Floating-point Extension.
* config/tc-arm.c (fldmias): Moved inside "THUMB_VARIANT & arm_ext_v6t2"
to support VLDMIA instruction for MVE.
(fldmdbs): Moved inside "THUMB_VARIANT & arm_ext_v6t2" to support VLDMDB
instruction for MVE.
(fstmias): Moved inside "THUMB_VARIANT & arm_ext_v6t2" to support VSTMIA
instruction for MVE.
(fstmdbs): Moved inside "THUMB_VARIANT & arm_ext_v6t2" to support VSTMDB
instruction for MVE.
* testsuite/gas/arm/mve-ldst.d: New test.
* testsuite/gas/arm/mve-ldst.s: Likewise.
bfcvt converts a .S input to a .H output, so any predicated movprfx
needs to operate on .S rather than .H. In common with SVE2 narrowing
top operations, bfcvtnt doesn't accept movprfx.
2020-01-31 Richard Sandiford <richard.sandiford@arm.com>
opcodes/
* aarch64-tbl.h (aarch64_opcode): Set C_MAX_ELEM for SVE bfcvt.
Remove C_SCAN_MOVPRFX for SVE bfcvtnt.
gas/
* testsuite/gas/aarch64/sve-bfloat-movprfx.s: Use .h rather than
.s for the movprfx.
* testsuite/gas/aarch64/sve-bfloat-movprfx.d: Update accordingly.
* testsuite/gas/aarch64/sve-movprfx_28.d,
* testsuite/gas/aarch64/sve-movprfx_28.l,
* testsuite/gas/aarch64/sve-movprfx_28.s: New test.
Tom Tromey [Wed, 22 Jan 2020 19:30:40 +0000 (12:30 -0700)]
Fix ravenscar-thread.c for multi-target
ravenscar-thread.c needed a change to adapt to multi-target:
ravenscar_thread_target::mourn_inferior called the mourn_inferior
method on the target beneat -- but when the target beneath was the
remote target, this resulted in the ravenscar target being deleted.
Switching the order of the calls to unpush_target and the beneath's
mourn_inferior fixes this problem.
gdb/ChangeLog
2020-01-31 Tom Tromey <tromey@adacore.com>
* ravenscar-thread.c (ravenscar_thread_target::mourn_inferior):
Call beneath target's mourn_inferior after unpushing.
Andrew Burgess [Fri, 24 Jan 2020 12:46:56 +0000 (12:46 +0000)]
gdb/tui: Disassembler scrolling of very small programs
In TUI mode, if the disassembly output for the program is less than
one screen long, then currently if the user scrolls down until on the
last assembly instruction is displayed and then tries to scroll up
using Page-Up, the display doesn't update - they are stuck viewing the
last line.
If the user tries to scroll up using the Up-Arrow, then the display
scrolls normally.
What is happening is on the Page-Up we ask GDB to scroll backward the
same number of lines as the height of the TUI ASM window. The back
scanner, which looks for a good place to start disassembling, fails to
find a starting address which will provide the requested number of new
lines before we get back to the original starting address (which is
not surprising, our whole program contains less than a screen height
of instructions), as a result the back scanner gives up and returns
the original starting address.
When we scroll with Up-Arrow we only ask the back scanner to find 1
new instruction, which it manages to do, so this scroll works.
The solution here is, when we fail to find enough instructions, to
return the lowest address we did manage to find. This will ensure we
jump to the lowest possible address in the disassembly output.
gdb/ChangeLog:
PR tui/9765
* tui/tui-disasm.c (tui_find_disassembly_address): If we don't
have enough lines to fill the screen, still return the lowest
address we found.
gdb/testsuite/ChangeLog:
PR tui/9765
* gdb.tui/tui-layout-asm-short-prog.S: New file.
* gdb.tui/tui-layout-asm-short-prog.exp: New file.
Andrew Burgess [Fri, 24 Jan 2020 14:18:56 +0000 (14:18 +0000)]
gdb/tui: Update help text for scroll commands
GDB has some commands ('+', '-', '<', and '>') for scrolling the SRC
and ASM TUI windows from the CMD window, however the help text for
these commands lists the arguments in the wrong order.
This commit updates the help text to match how GDB actually works, and
also extends the text to describe what the arguments mean, and what
the defaults are.
There should be no change in GDBs functionality after this commit.
gdb/ChangeLog:
* tui/tui-win.c (_initialize_tui_win): Update help text for '+',
'-', '<', and '>' commands.
Alan Modra [Thu, 30 Jan 2020 23:51:02 +0000 (10:21 +1030)]
Tidy bfd.pot
This patch removes the leak of Nick's source directory into bfd.pot,
and emits #line for some generated files so that those files aren't
referenced by comments in the .pot file. You can see both of these
effects in the following diff. I've also removed use of an
unnecessary temp file in the make rules.
Alan Modra [Thu, 30 Jan 2020 14:23:59 +0000 (00:53 +1030)]
OOM in setup_group
We alloc, seek and read using section sizes in object files. Fuzzed
objects can have silly sizes, but that's OK if the system supports
memory over-commit. The read fails because we hit EOF and that
usually results in a graceful exit.
But if we memset before the read then the invalid size results in
attempting to write to a huge number of memory pages, and an eventual
Out Of Memory after probably swapping like crazy. So don't memset.
There really isn't a need to clear the section contents anyway. All
bytes are written with a good object file by the read and following
loop converting section index in target order to ELF section header
pointer, and the only untidy bytes are the 4 bytes past the group
flags when pointers are 8 bytes. Those don't matter but the patch
clears them for anyone poking around in a debugger. On error paths
it's as good to free section contents as it is to clear them.
Noticed when looking at PR4110 fourth test case.
PR 4110
* elf.c (setup_group): Don't clear entire section contents,
just the padding after group flags. Release alloc'd memory
after a seek or read failure.
Jan Beulich [Thu, 30 Jan 2020 16:03:22 +0000 (17:03 +0100)]
x86: prevent undue use of GOT32X and alike relocations
Comparison of i.tm.base_opcode against particular but not sufficiently
specific values needs to be accompanied by other qualification. Exclude
VEX and alike encodings here, and also exclude all forms of prefixes
explicitly specified in the opcodes table. While using @GOT with such
insns may not be very useful, it also isn't with e.g. ADC and SBB, yet
these get explicitly listed in comments as supported.
Jan Beulich [Thu, 30 Jan 2020 16:02:33 +0000 (17:02 +0100)]
ld/doc: drop blank between @option and brace
Commit 9e7028aa1e78 ("PowerPC64 __tls_get_addr_desc") introduced a use
of @option which apparently newer makeinfo tolerates, but older ones
reject. Drop the unnecessary (a per all other uses of @option) blank.
Alan Modra [Thu, 30 Jan 2020 11:29:20 +0000 (21:59 +1030)]
ubsan: m32c: left shift of negative value
More nonsense fixing "bugs" with left shifts of signed values. Yes,
the C standard does say this is undefined (and right shifts of signed
values are implementation defined BTW) but in practice there is no
problem with current machines. 1's complement is a thing of the past.
cpu/
* m32c.cpu (f-src32-rn-unprefixed-QI): Shift before inverting.
(f-src32-rn-prefixed-QI, f-dst32-rn-unprefixed-QI): Likewise.
(f-dst32-rn-prefixed-QI): Likewise.
(f-dsp-32-s32): Mask before shifting left.
(f-dsp-48-u32, f-dsp-48-s32): Likewise.
(f-bitbase32-16-s11-unprefixed): Multiply signed field rather than
shifting left.
(f-bitbase32-24-s11-prefixed, f-bitbase32-24-s19-prefixed): Likewise.
(h-gr-SI): Mask before shifting.
opcodes/
* m32c-ibld.c: Regenerate.
Jon Turney [Wed, 15 Jan 2020 17:03:48 +0000 (17:03 +0000)]
Identify reproducible builds in 'objdump -p' output for PE files
These are produced by MSVC when the '/Brepro' flag is used.
To quote from the PE specification [1]:
"The presence of an entry of type IMAGE_DEBUG_TYPE_REPRO indicates the
PE file is built in a way to achieve determinism or reproducibility. If
the input does not change, the output PE file is guaranteed to be
bit-for-bit identical no matter when or where the PE is produced.
Various date/time stamp fields in the PE file are filled with part or
all the bits from a calculated hash value that uses PE file content as
input, and therefore no longer represent the actual date and time when a
PE file or related specific data within the PE is produced. The raw data
of this debug entry may be empty, or may contain a calculated hash value
preceded by a four-byte value that represents the hash value length."
2020-01-16 Jon Turney <jon.turney@dronecode.org.uk>
* peXXigen.c (pe_is_repro): New function.
(_bfd_XX_print_private_bfd_data_common): Note timestamp is
actually a build hash if PE_IMAGE_DEBUG_TYPE_REPRO is present.
Jon Turney [Wed, 15 Jan 2020 18:50:31 +0000 (18:50 +0000)]
Add some new PE_IMAGE_DEBUG_TYPE values
IMAGE_DEBUG_TYPE_REPRO is defined in the latest version of the PE
specification [1]. The others are defined in Windows SDK headers and/or
reported by DUMPBIN.
Jon Turney [Wed, 15 Jan 2020 18:48:13 +0000 (18:48 +0000)]
Bugfixes for pe_print_debugdata()
Use a separate iteration variable for inner loop (:blush:). This
generally prevented any debug directory entries after a
IMAGE_DEBUG_TYPE_CODEVIEW entry from being reported.
Don't leak the memory allocated for the section containing the debug
directory.
bfd/ChangeLog:
2020-01-16 Jon Turney <jon.turney@dronecode.org.uk>
* peXXigen.c (pe_print_debugdata): Fix the iteration variable for
inner loop. Fix a memory leak.
Jan Beulich [Thu, 30 Jan 2020 10:36:33 +0000 (11:36 +0100)]
x86-64: honor vendor specifics for near RET
While vendors agree about default operand size (64 bits) and hence
unavilability of a 32-bit form, AMD honors a 16-bit operand size
override (0x66) while Intel doesn't.
Jan Beulich [Thu, 30 Jan 2020 10:35:20 +0000 (11:35 +0100)]
x86-64: also diagnose far returns / IRET with ambiguous operand size
Other than near returns these default to 32-bit operand size, and hence
it isn't really unlikely that 64-bit forms are meant. Hence these should
have disambiguating suffixes. In Intel mode, however, don't error in
these cases unconditionally - MASM accepts these without suffix _and_
without warning.
Pedro Alves [Wed, 29 Jan 2020 17:53:55 +0000 (12:53 -0500)]
Fix -Werror-stringop error on infcmd.c:construct_inferior_arguments
While testing a GCC 10 build of our git HEAD, Sergio noticed an error
triggered by -Werror-stringop on
infcmd.c:construct_inferior_arguments. One of the things the function
does is calculate the length of the string that will hold the
inferior's arguments. GCC warns us that 'length' can be 0, which can
lead to undesired behaviour:
../../gdb/infcmd.c: In function 'char* construct_inferior_arguments(int, char**)':
../../gdb/infcmd.c:369:17: error: writing 1 byte into a region of size 0 [-Werror=stringop-overflow=]
369 | result[0] = '\0';
| ~~~~~~~~~~^~~~~~
../../gdb/infcmd.c:368:33: note: at offset 0 to an object with size 0 allocated by 'xmalloc' here
368 | result = (char *) xmalloc (length);
| ~~~~~~~~^~~~~~~~
The solution here is to assert that 'argc' is greater than 0 on entry,
which makes GCC understand that the loops always run at least once,
and thus 'length' is always > 0.
Tested by rebuilding.
gdb/ChangeLog:
2020-01-29 Pedro Alves <palves@redhat.com>
Sergio Durigan Junior <sergiodj@redhat.com>
* infcmd.c (construct_inferior_arguments): Assert that
'argc' is greater than 0.
Adjust src-release.sh's getver due to gdbsupport's move to toplevel
The move of gdbsupport to the top level directory requires a small
change to src-release.sh's "getver" function, which is responsible for
determining the version string that will be appended to the release
tarball: now the create-version.sh script lives under ./gdbsupport,
and not under gdb/gdbsupport anymore.
This patch unbreaks the snapshot generation, which hasn't been working
since January 14th.
An error in commit 42cd72aa0279520384327a1f6d0ebd2eb2200645 caused
srv_tgtobj to be overwritten and linux-ppc-low.o to be missed when
linking gdbserver for Linux on PowerPC. This patch fixes the error.
gdb/gdbserver/ChangeLog:
2020-01-29 Pedro Franco de Carvalho <pedromfc@linux.ibm.com>
* configure.srv (powerpc*-*-linux*): Use srv_tgtobj in second
assignment instead of srv_linux_obj.
Luis Machado [Mon, 23 Dec 2019 15:04:26 +0000 (12:04 -0300)]
Recognize more program breakpoint patterns
New in v3:
- Code cleanups based on reviews.
New in v2:
- Fixed misc problems based on reviews.
- Switched to using gdbarch_program_breakpoint_here_p as opposed to
gdbarch_insn_is_breakpoint.
- Fixed matching of brk instructions. Previously the mask was incorrect, which
was showing up as a few failures in the testsuite. Now it is clean.
- New testcase (separate patch).
- Moved program_breakpoint_here () to arch-utils.c and made it the default
implementation of gdbarch_program_breakpoint_here_p.
--
It was reported to me that program breakpoints (permanent ones inserted into
the code itself) other than the one GDB uses for AArch64 (0xd4200000) do not
generate visible stops when continuing, and GDB will continue spinning
infinitely.
This happens because GDB, upon hitting one of those program breakpoints, thinks
the SIGTRAP came from a delayed breakpoint hit...
If the program breakpoint is one GDB recognizes, then it will stop when it
hits it.
(gdb) x/i $pc
=> 0x4005c0 <problem_function>: brk #0x0
(gdb) c
Continuing.
infrun: clear_proceed_status_thread (process 14193)
infrun: proceed (addr=0xffffffffffffffff, signal=GDB_SIGNAL_DEFAULT)
infrun: proceed: resuming process 14193
infrun: resume (step=0, signal=GDB_SIGNAL_0), trap_expected=0, current thread [process 14193] at 0x4005c0
infrun: infrun_async(1)
infrun: prepare_to_wait
infrun: target_wait (-1.0.0, status) =
infrun: 14193.14193.0 [process 14193],
infrun: status->kind = stopped, signal = GDB_SIGNAL_TRAP
infrun: handle_inferior_event status->kind = stopped, signal = GDB_SIGNAL_TRAP
infrun: stop_pc = 0x4005c0
infrun: random signal (GDB_SIGNAL_TRAP)
infrun: stop_waiting
infrun: stop_all_threads
infrun: stop_all_threads, pass=0, iterations=0
infrun: process 14193 not executing
infrun: stop_all_threads, pass=1, iterations=1
infrun: process 14193 not executing
infrun: stop_all_threads done
Program received signal SIGTRAP, Trace/breakpoint trap.
problem_function () at brk_0.c:7
7 asm("brk %0\n\t" ::"n"(0x0));
infrun: infrun_async(0)
Otherwise GDB will keep trying to resume the inferior and will keep
seeing the SIGTRAP's, without stopping.
To the user it appears GDB has gone into an infinite loop, interruptible only
by Ctrl-C.
Also, windbg seems to use a different variation of AArch64 breakpoint compared
to GDB. This causes problems when debugging Windows on ARM binaries, when
program breakpoints are being used.
The proposed patch creates a new gdbarch method (gdbarch_program_breakpoint_here_p)
that tells GDB whether the underlying instruction is a breakpoint instruction
or not.
This is more general than only checking for the instruction GDB uses as
breakpoint.
The existing logic is still preserved for targets that do not implement this
new gdbarch method.
The end result is like so:
(gdb) x/i $pc
=> 0x4005c0 <problem_function>: brk #0x90f
(gdb) c
Continuing.
infrun: clear_proceed_status_thread (process 16417)
infrun: proceed (addr=0xffffffffffffffff, signal=GDB_SIGNAL_DEFAULT)
infrun: proceed: resuming process 16417
infrun: resume (step=0, signal=GDB_SIGNAL_0), trap_expected=0, current thread [process 16417] at 0x4005c0
infrun: infrun_async(1)
infrun: prepare_to_wait
infrun: target_wait (-1.0.0, status) =
infrun: 16417.16417.0 [process 16417],
infrun: status->kind = stopped, signal = GDB_SIGNAL_TRAP
infrun: handle_inferior_event status->kind = stopped, signal = GDB_SIGNAL_TRAP
infrun: stop_pc = 0x4005c0
infrun: random signal (GDB_SIGNAL_TRAP)
infrun: stop_waiting
infrun: stop_all_threads
infrun: stop_all_threads, pass=0, iterations=0
infrun: process 16417 not executing
infrun: stop_all_threads, pass=1, iterations=1
infrun: process 16417 not executing
infrun: stop_all_threads done
Program received signal SIGTRAP, Trace/breakpoint trap.
problem_function () at brk.c:7
7 asm("brk %0\n\t" ::"n"(0x900 + 0xf));
infrun: infrun_async(0)
gdb/ChangeLog:
2020-01-29 Luis Machado <luis.machado@linaro.org>
* aarch64-tdep.c (BRK_INSN_MASK): Define to 0xffe0001f.
(BRK_INSN_MASK): Define to 0xd4200000.
(aarch64_program_breakpoint_here_p): New function.
(aarch64_gdbarch_init): Set gdbarch_program_breakpoint_here_p hook.
* arch-utils.c (default_program_breakpoint_here_p): Moved from
breakpoint.c.
* arch-utils.h (default_program_breakpoint_here_p): Moved from
breakpoint.h
* breakpoint.c (bp_loc_is_permanent): Changed return type to bool and
call gdbarch_program_breakpoint_here_p.
(program_breakpoint_here): Moved to arch-utils.c, renamed to
default_program_breakpoint_here_p, changed return type to bool and
simplified.
* breakpoint.h (program_breakpoint_here): Moved prototype to
arch-utils.h, renamed to default_program_breakpoint_here_p and changed
return type to bool.
* gdbarch.c: Regenerate.
* gdbarch.h: Regenerate.
* gdbarch.sh (program_breakpoint_here_p): New method.
* infrun.c (handle_signal_stop): Call
gdbarch_program_breakpoint_here_p.
testsuite, cp: add expected failures to pass-by-ref tests for certain compilers
There exist expected failures in the pass-by-ref.exp and
pass-by-ref-2.exp tests based on the GCC and Clang version.
* GCC version <= 6 and Clang do not emit DW_AT_deleted and
DW_AT_defaulted.
* Clang version >= 7 emits DW_AT_calling_convention, which helps the
debugger make the right calling convention decision in some cases
despite lacking the 'defaulted' and 'deleted' attributes.
Mark the related tests as XFAIL based on the compiler version.
Tested on X86_64 using GCC 5.5.0, 6.5.0, 7.4.0, 8.3.0, 9.2.1;
and Clang 5.0.1, 6.0.0, 7.0.0, 8.0.0, 9.0.1, 10.0.0.
gdb/testsuite/ChangeLog:
2020-01-29 Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
* gdb.cp/pass-by-ref-2.exp: Mark some tests as XFAIL based on the
GCC/Clang version.
* gdb.cp/pass-by-ref.exp: Ditto.
I ran into:
...
Thread 3.1 "watchpoint-fork" hit Breakpoint 3, marker () at \
watchpoint-fork-mt.c:42^M
42 }^M
(gdb) parent2: 1945^M
FAIL: gdb.threads/watchpoint-fork.exp: child: multithreaded: breakpoint (A) \
after the second fork (timeout)
...
The problem is that the FAILing gdb_test expects '(gdb) ' to be the last thing
printed, but the inferior prints something after that.
A similar FAIL is described in the sources in watchpoint-fork-parent.c:
...
printf ("child%d: %d\n", nr, (int) getpid ());
/* Delay to get both the "child%d" and "parent%d" message printed
without a race breaking expect by its endless wait on `$gdb_prompt$':
Breakpoint 3, marker () at watchpoint-fork.c:33
33 }
(gdb) parent2: 14223 */
i = sleep (1);
...
I noticed that while the executables print output, the output is not verified in
the test-case, so it's merely debug output.
Fix this by:
- guarding the prints in the executables (as well as related
sleep and setbuf calls) with #if DEBUG, and
- compiling by default with DEBUG=0.
gdb/testsuite/ChangeLog:
2020-01-29 Tom de Vries <tdevries@suse.de>
* gdb.threads/watchpoint-fork-child.c: Guard prints with #if DEBUG.
* gdb.threads/watchpoint-fork-mt.c: Same.
* gdb.threads/watchpoint-fork-parent.c: Same.
* gdb.threads/watchpoint-fork-st.c: Same.
* gdb.threads/watchpoint-fork.exp: Compile with DEBUG=0.
Hannes Domani [Tue, 28 Jan 2020 17:24:31 +0000 (18:24 +0100)]
Fix library segment-address for 64bit values
The address was written as a long value, but long is always a 32bit value
on Windows, which lead to truncated addresses.
The solution was to use paddress instead.
gdb/gdbserver/ChangeLog:
2020-01-28 Hannes Domani <ssbssa@yahoo.de>
* server.c (handle_qxfer_libraries): Write segment-address with
paddress.
Jim Wilson [Mon, 27 Jan 2020 23:19:30 +0000 (15:19 -0800)]
RISC-V: Fix gdbserver problem with handling arch strings.
Maciej reported a problem found by his RISC-V gdbserver port.
warning: while parsing target description (at line 4): Target description specified unknown architecture "riscv:rv64id"
warning: Could not load XML target description; ignoring
We only have two arches defined, riscv:rv32 and riscv:rv64. Both bfd and
gdb are creating arch strings that have extension letters added to the base
architecture. The bfd_default_scan function requires an exact match, so
these strings fail to map to a bfd_arch. I think we should ignore the
extension letters in a RISC-V specific scan function.
bfd/
* cpu-riscv.c (riscv_scan): New.
(N): Change bfd_default_scan to riscv_scan.
Luis Machado [Tue, 14 Jan 2020 16:46:07 +0000 (13:46 -0300)]
Harden gdb.base/step-over-syscall.exp
New in v3:
- Verify if the syscall number matches what is expected for the target.
- Used gdb_assert for one more check.
New in v2:
- Set initial values to -1 instead of 0.
- Rewrote RE to prevent unexpected matching when parsing one character at a
time.
- Used gdb_assert for an additional check.
- Validated with check-read1
There are a couple problems with this test.
First
--
gdb.base/step-over-syscall.exp records the address of a syscall instruction
within fork/vfork/clone functions and also the address of the instruction
after that syscall instruction.
It uses these couples addresses to make sure we stepped over a syscall
instruction (fork/vfork/clone events) correctly.
The way the test fetches the addresses of the instructions is by stepi-ing
its way through the fork/vfork/clone functions until it finds a match for
a syscall. Then it stepi's once again to get the address of the next
instruction.
This assumes that stepi-ing over a syscall is working correctly and landing
in the right PC. This is not the case for AArch64/Linux, where we're
landing a couple instructions after the syscall in some cases.
The following patch lets the test execute as before, but adds a new instruction
address check using the x command as opposed to stepi.
I didn't want to change how the test works since we may also be
interested in checking if stepi-ing over the syscall under different
conditions (displaced stepping on/off) yields the same results. I don't
feel strongly about this, so i'm OK with changing how we compare PC's for
the entire test if folks decide it is reasonable.
Second
--
FAIL: gdb.base/step-over-syscall.exp: vfork: displaced=off: continue to vfork (3rd time) (the program exited)
FAIL: gdb.base/step-over-syscall.exp: vfork: displaced=off: continue to syscall insn vfork (the program is no longer running)
FAIL: gdb.base/step-over-syscall.exp: vfork: displaced=off: single step over vfork (the program is no longer running)
Depending on the glibc version we may have different code generated for the
fork/vfork/clone functions.
I ran into the situation where vfork for newer glibc's on AArch64/Linux is
very short, so "break vfork" will put a breakpoint right at the syscall
instruction, which is something the testcase isn't expecting (a off-by-1
of sorts).
The patch adds extra code to handle this case. If the test detects we're
already sitting at a syscall instruction, it records the address and moves
on to record the address after that particular instruction.
Another measure is to "break *$syscall" instead of "break $syscall". That
guarantees we're stopping at the first instruction of the syscall function,
if it ever happens that the syscall instruction is the first instruction of
those functions.
With these changes i can fix some failures for aarch64-linux-gnu and also
expose the problems i've reported here:
These tests now fail for aarch64-linux-gnu (patch for this is going through
reviews):
FAIL: gdb.base/step-over-syscall.exp: vfork: displaced=off: pc after stepi matches insn addr after syscall
FAIL: gdb.base/step-over-syscall.exp: vfork: displaced=on: pc after stepi matches insn addr after syscall
gdb/testsuite/ChangeLog:
2020-01-27 Luis Machado <luis.machado@linaro.org>
* gdb.base/step-over-syscall.exp (setup): Check if we're already
sitting at a syscall instruction when we hit the syscall function's
breakpoint.
Check PC against one obtained with the x command.
Validate syscall number.
(step_over_syscall): Don't continue to the syscall instruction if
we're already there.
H.J. Lu [Mon, 27 Jan 2020 12:38:10 +0000 (04:38 -0800)]
x86-64: Properly encode and decode movsxd
movsxd is a 64-bit only instruction. It supports both 16-bit and 32-bit
destination registers. Its AT&T mnemonic is movslq which only supports
64-bit destination register. There is also a discrepancy between AMD64
and Intel64 on movsxd with 16-bit destination register. AMD64 supports
32-bit source operand and Intel64 supports 16-bit source operand.
This patch updates movsxd encoding and decoding to alow 16-bit and 32-bit
destination registers. It also handles movsxd with 16-bit destination
register for AMD64 and Intel 64.
gas/
PR binutils/25445
* config/tc-i386.c (check_long_reg): Also convert to QWORD for
movsxd.
* doc/c-i386.texi: Add a node for AMD64 vs. Intel64 ISA
differences. Document movslq and movsxd.
* testsuite/gas/i386/i386.exp: Run PR binutils/25445 tests.
* testsuite/gas/i386/x86-64-movsxd-intel.d: New file.
* testsuite/gas/i386/x86-64-movsxd-intel64-intel.d: Likewise.
* testsuite/gas/i386/x86-64-movsxd-intel64-inval.l: Likewise.
* testsuite/gas/i386/x86-64-movsxd-intel64-inval.s: Likewise.
* testsuite/gas/i386/x86-64-movsxd-intel64.d: Likewise.
* testsuite/gas/i386/x86-64-movsxd-intel64.s: Likewise.
* testsuite/gas/i386/x86-64-movsxd-inval.l: Likewise.
* testsuite/gas/i386/x86-64-movsxd-inval.s: Likewise.
* testsuite/gas/i386/x86-64-movsxd.d: Likewise.
* testsuite/gas/i386/x86-64-movsxd.s: Likewise.
opcodes/
PR binutils/25445
* i386-dis.c (MOVSXD_Fixup): New function.
(movsxd_mode): New enum.
(x86_64_table): Use MOVSXD_Fixup and movsxd_mode on movsxd.
(intel_operand_size): Handle movsxd_mode.
(OP_E_register): Likewise.
(OP_G): Likewise.
* i386-opc.tbl: Remove Rex64 and allow 32-bit destination
register on movsxd. Add movsxd with 16-bit destination register
for AMD64 and Intel64 ISAs.
* i386-tbl.h: Regenerated.
Tamar Christina [Mon, 27 Jan 2020 10:40:02 +0000 (10:40 +0000)]
AArch64: Fix cfinv disassembly issues
This fixes the preferred disassembly for cfinv. The Armv8.4-a instruction
overlaps with the possible encoding space for msr. This because msr allows you
to use unallocated encoding space using the general sA_B_cC_cD_E form.
However when an encoding does become allocated then we need to ensure that it's
used as the preferred disassembly. The problem with cfinv is that its mask has
all bits sets because it has no arguments.
This causes issues for the Alias resolver in gas as it uses the mask to build
alias graph. In this case it can't do it since it thinks almost everything
would alias with cfinv. So instead we can only fix this by moving cfinv before
msr.
Tom Tromey [Wed, 23 Oct 2019 15:50:58 +0000 (09:50 -0600)]
Virtualize "readin" and "compunit_symtab"
This patch removes the "readin" and "compunit_symtab" members from
partial_symtab, replacing them with methods. Then it introduces a new
"standard_psymtab" class, which restores these members; and changes
the symbol readers to use this intermediate class as the base class of
their partial symtab subclasses.
The reason for this is to make it possible for a symbol reader to
implement an alternate mapping between partial and full symbol tables.
This is important in order to be able to share psymtabs across
objfiles -- whether a psymtab has been "readin" is objfile-dependent,
as are the pointers to the full symbol tables.
gdb/ChangeLog
2020-01-26 Tom Tromey <tom@tromey.com>
* psymtab.c (partial_map_expand_apply)
(psym_find_pc_sect_compunit_symtab, psym_lookup_symbol)
(psymtab_to_symtab, psym_find_last_source_symtab, dump_psymtab)
(psym_print_stats, psym_expand_symtabs_for_function)
(psym_map_symbol_filenames, psym_map_matching_symbols)
(psym_expand_symtabs_matching)
(partial_symtab::read_dependencies, maintenance_info_psymtabs)
(maintenance_check_psymtabs): Use new methods.
* psympriv.h (struct partial_symtab) <readin_p,
get_compunit_symtab>: New methods.
<readin, compunit_symtab>: Remove members.
(struct standard_psymtab): New.
(struct legacy_psymtab): Derive from standard_psymtab.
* dwarf2read.h (struct dwarf2_psymtab): Derive from
standard_psymtab.
* ctfread.c (struct ctf_psymtab): Derive from standard_psymtab.
Tom Tromey [Wed, 23 Oct 2019 15:46:25 +0000 (09:46 -0600)]
Consolidate partial symtab dependency reading
Most of the symbol readers have code to iterate over a partial symtabs
dependencies, expanding each one and optionally printing a message.
Now that the "second-stage" psymtab expansion is available as a
method, these implementations can all be merged.
This patch also changes a couple more warnings into assertions.
gdb/ChangeLog
2020-01-26 Tom Tromey <tom@tromey.com>
Tom Tromey [Wed, 23 Oct 2019 15:40:54 +0000 (09:40 -0600)]
Introduce partial_symtab::expand_psymtab method
The symbol readers generally used two functions to expand a partial
symtab: an outer function (now the "read_symtab" method), and an inner
function, typically named something like "psymtab_to_symtab".
This patch changes this second step to be a method on partial_symtab,
and updates all the callers. For legacy_psymtab, a new function
pointer member is introduced.
This patch enables a subsequent cleanup.
gdb/ChangeLog
2020-01-26 Tom Tromey <tom@tromey.com>
Tom Tromey [Tue, 22 Oct 2019 23:28:37 +0000 (17:28 -0600)]
Introduce partial_symtab::read_symtab method
This introduces a new partial_symtab::read_symtab method, and updates
the symbol readers to subclass partial_symtab and implement this
method. The old read_symtab and read_symtab_private members are
removed.
In practice, only DWARF and CTF are truly updated to take advantage of
the new setup. The other symbol readers are less actively maintained,
and so this patch also introduces a "legacy_psymtab", which
essentially works the same way as the old partial_symtab.
(Note that, without more knowledge of the interaction between these
symbol readers, fixing this to remove the new (small) overhead is not
trivial, because these readers copy the read_symtab pointer between
partial symtabs.)
gdb/ChangeLog
2020-01-26 Tom Tromey <tom@tromey.com>
* xcoffread.c (this_symtab_psymtab, read_xcoff_symtab)
(xcoff_psymtab_to_symtab_1, xcoff_read_symtab)
(xcoff_start_psymtab, xcoff_end_psymtab, scan_xcoff_symtab): Use
legacy_symtab.
* stabsread.h (dbx_end_psymtab): Use legacy_symtab.
* psymtab.c (psymtab_to_symtab): Call method.
(dump_psymtab): Update.
* psympriv.h (struct partial_symtab): Add virtual destructor.
<read_symtab>: New method.
(struct legacy_symtab): New.
* mdebugread.c (mdebug_read_symtab): Use legacy_psymtab.
(struct pst_map) <pst>: Now a legacy_psymtab.
(parse_procedure, parse_partial_symbols, psymtab_to_symtab_1)
(new_psymtab): Use legacy_psymtab.
* dwarf2read.h (struct dwarf2_psymtab): New.
(struct dwarf2_per_cu_data) <psymtab>: Use it.
* dwarf2read.c (dwarf2_create_include_psymtab)
(dwarf2_build_include_psymtabs, create_type_unit_group)
(create_partial_symtab, process_psymtab_comp_unit_reader)
(build_type_psymtabs_reader, build_type_psymtab_dependencies)
(set_partial_user): Use dwarf2_psymtab.
(dwarf2_psymtab::read_symtab): Rename from dwarf2_read_symtab.
(psymtab_to_symtab_1, process_full_comp_unit)
(process_full_type_unit, dwarf2_ranges_read)
(dwarf2_get_pc_bounds, psymtab_include_file_name)
(dwarf_decode_lines): Use dwarf2_psymtab.
* dwarf-index-write.c (psym_index_map): Use dwarf2_psymtab.
(add_address_entry_worker, write_one_signatured_type)
(recursively_count_psymbols, recursively_write_psymbols)
(write_one_signatured_type, psyms_seen_size, write_gdbindex)
(write_debug_names): Likewise.
* dbxread.c (struct header_file_location): Take a legacy_psymtab.
<pst>: Now a legacy_psymtab.
(find_corresponding_bincl_psymtab): Return a legacy_psymtab.
(read_dbx_symtab, start_psymtab, dbx_end_psymtab)
(dbx_psymtab_to_symtab_1, read_ofile_symtab): Use legacy_psymtab.
* ctfread.c (struct ctf_psymtab): New.
(ctf_start_symtab, ctf_end_symtab, psymtab_to_symtab): Take a
ctf_psymtab.
(ctf_psymtab::read_symtab): Rename from ctf_read_symtab.
(create_partial_symtab): Return a ctf_psymtab.
(scan_partial_symbols): Update.
Tom Tromey [Tue, 22 Oct 2019 23:08:16 +0000 (17:08 -0600)]
Turn start_psymtab_common into a constructor
This turns start_psymtab_common into a constructor, and then changes
the callers to use "new" directly. This completes the psymtab
allocation transition -- now it is possible for symbol readers to
subclass struct partial_symtab.
gdb/ChangeLog
2020-01-26 Tom Tromey <tom@tromey.com>
* xcoffread.c (xcoff_start_psymtab): Use new.
* psymtab.c (partial_symtab::partial_symtab): New constructor,
renamed from start_psymtab_common.
* psympriv.h (struct partial_symtab): Add new constructor.
(start_psymtab_common): Don't declare.
* mdebugread.c (parse_partial_symbols): Use new.
* dwarf2read.c (create_partial_symtab): Use new.
* dbxread.c (start_psymtab): Use new.
* ctfread.c (create_partial_symtab): Use new.
Tom Tromey [Tue, 22 Oct 2019 22:57:35 +0000 (16:57 -0600)]
Change allocate_psymtab to be a constructor
This is the next step in getting the symbol readers to allocate
psymtabs themselves: change allocate_psymtab to be an ordinary
constructor, and then use "new" at the previous call sites. Note that
this doesn't get us all the way -- start_psymtab_common is still
allocating a partial symtab.
gdb/ChangeLog
2020-01-26 Tom Tromey <tom@tromey.com>
* xcoffread.c (xcoff_end_psymtab): Use new.
* psymtab.c (start_psymtab_common): Use new.
(partial_symtab::partial_symtab): Rename from allocate_psymtab.
Update.
* psympriv.h (struct partial_symtab): Add parameters to
constructor. Don't inline.
(allocate_psymtab): Don't declare.
* mdebugread.c (new_psymtab): Use new.
* dwarf2read.c (dwarf2_create_include_psymtab): Use new.
* dbxread.c (dbx_end_psymtab): Use new.
Tom Tromey [Tue, 22 Oct 2019 22:51:55 +0000 (16:51 -0600)]
Do not allocate psymtabs via psymtab_storage
Currently, partial symbol tables are allocated by a method in
psymtab_storage. However, eventually we want to subclass partial
symtabs in the symbol readers, so the calls to "new" will have to
happen there. This patch is a first step, moving the allocation from
psymtab_storage and into allocate_psymtab.
gdb/ChangeLog
2020-01-26 Tom Tromey <tom@tromey.com>
* psymtab.h (class psymtab_storage) <install_psymtab>: Rename from
allocate_psymtab. Update documentation.
* psymtab.c (psymtab_storage::install_psymtab): Rename from
allocate_psymtab. Do not use new.
(allocate_psymtab): Use new. Update.
Tom Tromey [Thu, 24 Oct 2019 16:30:13 +0000 (10:30 -0600)]
Document m68k floating point feature correspondence
From what I can tell, The m68k floating point target feature should
apparently always be called "org.gnu.gdb.coldfire.fp" -- even when the
primary feature is not "coldfire", because m68k_gdbarch_init only
checks for this feature when assigning register numbers.
However, the floating point registers are expected to match what gdb
thinks are the register sizes for the primary feature. For example,
if the main feature is "coldfire", then the floating point registers
should be 64 bits.
See this note for some an instance of this confusion:
Let me know what you think. An alternate approach here might be to
make gdb adapt to the register sizes as actually reported. I'm not
sure if this makes sense or not.
gdb/doc/ChangeLog
2020-01-26 Tom Tromey <tromey@adacore.com>
Modify gdb.base/attach.exp to test the behaviour of the option
exec-file-mismatch. Note that this test can also be run using/
make check RUNTESTFLAGS="--target_board=native-extended-gdbserver" TESTS=gdb.base/attach.exp
to test the behaviour of attaching to running program using a gdb server.
Note: when running the test with a gdbserver, the tests in
test_command_line_attach_run fail because the command "run" is not supported.
I tried to extend the condition
if ![isnative] then {
unsupported "commandline attach run test"
return 0
}
but unclear to me how to best do that. The below trials all failed
to work properly:
if { ![isnative] || [target_is_gdbserver] } then {
if { ![isnative] || [use_gdb_stub] } then {
if { ![isnative] || [is_remote target] } then {
=> could never obtain a condition that was true with gdbserver.
2020-01-25 Philippe Waroquiers <philippe.waroquiers@skynet.be>
* gdb.base/attach.exp: Test 'set exec-file-mismatch'.
This option allows to tell GDB to detect and possibly handle mismatched exec-files.
A recurrent problem with GDB is that GDB uses the wrong exec-file
when using the attach/detach commands successively.
Also, in case the user specifies a file on the command line but attaches
to the wrong PID, this error is not made visible and gives a not user
understandable behaviour.
For example:
$ gdb
...
(gdb) atta 2682 ############################################ PID running 'sleepers' executable
Attaching to process 2682
[New LWP 2683]
[New LWP 2684]
[New LWP 2685]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
0x00007f5ff829f603 in select () at ../sysdeps/unix/syscall-template.S:84
84 ../sysdeps/unix/syscall-template.S: No such file or directory.
(gdb) det
Detaching from program: /home/philippe/valgrind/git/trunk_untouched/gdbserver_tests/sleepers, process 2682
[Inferior 1 (process 2682) detached]
(gdb) atta 31069 ############################################ PID running 'gdb' executable
Attaching to program: /home/philippe/valgrind/git/trunk_untouched/gdbserver_tests/sleepers, process 31069
Reading symbols from /lib64/ld-linux-x86-64.so.2...
Reading symbols from /usr/lib/debug/.build-id/60/6df9c355103e82140d513bc7a25a635591c153.debug...
0x00007f43c23478a0 in ?? ()
(gdb) bt
#0 0x00007f43c23478a0 in ?? ()
#1 0x0000558909e3ad91 in ?? ()
#2 0x0000202962646700 in ?? ()
#3 0x00007ffc69c74e70 in ?? ()
#4 0x000055890c1d2350 in ?? ()
#5 0x0000000000000000 in ?? ()
(gdb)
The second attach has kept the executable of the first attach.
(in this case, 31069 is the PID of a GDB, that has nothing to do
with the first determined 'sleepers' executable).
Similarly, if specifying an executable, but attaching to a wrong pid,
we get:
gdb /home/philippe/valgrind/git/trunk_untouched/gdbserver_tests/sleepers
...
Reading symbols from /home/philippe/valgrind/git/trunk_untouched/gdbserver_tests/sleepers...
(gdb) atta 31069 ############################################ PID running 'gdb' executable
Attaching to program: /home/philippe/valgrind/git/trunk_untouched/gdbserver_tests/sleepers, process 31069
Reading symbols from /lib64/ld-linux-x86-64.so.2...
Reading symbols from /usr/lib/debug/.build-id/60/6df9c355103e82140d513bc7a25a635591c153.debug...
0x00007f43c23478a0 in ?? ()
(gdb) bt
#0 0x00007f43c23478a0 in ?? ()
#1 0x0000558909e3ad91 in ?? ()
#2 0x0000202962646700 in ?? ()
#3 0x00007ffc69c74e70 in ?? ()
#4 0x000055890c1d2350 in ?? ()
#5 0x0000000000000000 in ?? ()
(gdb)
And it is unclear to the user what has happened/what is going wrong.
This patch series implements a new option:
(gdb) apropos exec-file-mismatch
set exec-file-mismatch -- Set exec-file-mismatch handling (ask|warn|off).
show exec-file-mismatch -- Show exec-file-mismatch handling (ask|warn|off).
(gdb) help set exec-file-mismatch
Set exec-file-mismatch handling (ask|warn|off).
Specifies how to handle a mismatch between the current exec-file name
loaded by GDB and the exec-file name automatically determined when attaching
to a process:
ask - warn the user and ask whether to load the determined exec-file.
warn - warn the user, but do not change the exec-file.
off - do not check for mismatch.
"ask" means: in case of mismatch between the current exec-file name
and the automatically determined exec-file name of the PID we are attaching to,
give a warning to the user and ask whether to load the automatically determined
exec-file.
"warn" means: in case of mismatch, just give a warning to the user.
"off" means: do not check for mismatch.
This fixes PR gdb/17626.
There was a previous trial to fix this PR.
See https://sourceware.org/ml/gdb-patches/2015-07/msg00118.html
This trial was however only fixing the problem for the automatically
determined executable files when doing attach.
It was differentiating the 'user specified executable files' ("sticky")
from the executable files automatically found by GDB.
But such user specified sticky executables are in most cases due
to a wrong manipulation by the user, giving unexpected results
such as backtrace showing no function like in the above example.
This patch ensures that whenever a process executable can be
determined, that the user is warned if there is a mismatch.
The same tests as above then give:
(gdb) atta 2682
Attaching to process 2682
[New LWP 2683]
[New LWP 2684]
[New LWP 2685]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
0x00007f5ff829f603 in select () at ../sysdeps/unix/syscall-template.S:84
84 ../sysdeps/unix/syscall-template.S: No such file or directory.
(gdb) det
Detaching from program: /home/philippe/valgrind/git/trunk_untouched/gdbserver_tests/sleepers, process 2682
[Inferior 1 (process 2682) detached]
(gdb) atta 31069
Attaching to program: /home/philippe/valgrind/git/trunk_untouched/gdbserver_tests/sleepers, process 31069
warning: Mismatch between current exec-file /home/philippe/valgrind/git/trunk_untouched/gdbserver_tests/sleepers
and automatically determined exec-file /bd/home/philippe/gdb/git/build_fixes/gdb/gdb
exec-file-mismatch handling is currently "ask"
Load new symbol table from "/bd/home/philippe/gdb/git/build_fixes/gdb/gdb"? (y or n) y
Reading symbols from /bd/home/philippe/gdb/git/build_fixes/gdb/gdb...
Setting up the environment for debugging gdb.
...
Reading symbols from /usr/lib/debug/.build-id/60/6df9c355103e82140d513bc7a25a635591c153.debug...
0x00007f43c23478a0 in __poll_nocancel () at ../sysdeps/unix/syscall-template.S:84
84 ../sysdeps/unix/syscall-template.S: No such file or directory.
(top-gdb) bt
During symbol reading: incomplete CFI data; unspecified registers (e.g., rax) at 0x7f43c23478ad
During symbol reading: unsupported tag: 'DW_TAG_unspecified_type'
During symbol reading: cannot get low and high bounds for subprogram DIE at 0x12282a7
During symbol reading: Child DIE 0x12288ba and its abstract origin 0x1228b26 have different parents
During symbol reading: DW_AT_call_target target DIE has invalid low pc, for referencing DIE 0x1229540 [in module /bd/home/philippe/gdb/git/build_fixes/gdb/gdb]
#0 0x00007f43c23478a0 in __poll_nocancel () at ../sysdeps/unix/syscall-template.S:84
#1 0x0000558909e3ad91 in poll (__timeout=-1, __nfds=<optimized out>, __fds=<optimized out>) at /usr/include/x86_64-linux-gnu/bits/poll2.h:46
#2 gdb_wait_for_event (block=block@entry=1) at ../../fixes/gdb/event-loop.c:772
#3 0x0000558909e3aef4 in gdb_do_one_event () at ../../fixes/gdb/event-loop.c:347
#4 0x0000558909e3b085 in gdb_do_one_event () at ../../fixes/gdb/common/common-exceptions.h:219
#5 start_event_loop () at ../../fixes/gdb/event-loop.c:371
During symbol reading: Member function "~_Sp_counted_base" (offset 0x1c69bf7) is virtual but the vtable offset is not specified
During symbol reading: Multiple children of DIE 0x1c8f5a0 refer to DIE 0x1c8f0ee as their abstract origin
#6 0x0000558909ed3b78 in captured_command_loop () at ../../fixes/gdb/main.c:331
#7 0x0000558909ed4b6d in captured_main (data=<optimized out>) at ../../fixes/gdb/main.c:1174
#8 gdb_main (args=<optimized out>) at ../../fixes/gdb/main.c:1190
#9 0x0000558909c1e9a8 in main (argc=<optimized out>, argv=<optimized out>) at ../../fixes/gdb/gdb.c:32
(top-gdb)
gdb /home/philippe/valgrind/git/trunk_untouched/gdbserver_tests/sleepers
...
Reading symbols from /home/philippe/valgrind/git/trunk_untouched/gdbserver_tests/sleepers...
(gdb) atta 31069
Attaching to program: /home/philippe/valgrind/git/trunk_untouched/gdbserver_tests/sleepers, process 31069
warning: Mismatch between current exec-file /home/philippe/valgrind/git/trunk_untouched/gdbserver_tests/sleepers
and automatically determined exec-file /bd/home/philippe/gdb/git/build_fixes/gdb/gdb
exec-file-mismatch handling is currently "ask"
Load new symbol table from "/bd/home/philippe/gdb/git/build_fixes/gdb/gdb"? (y or n) y
Reading symbols from /bd/home/philippe/gdb/git/build_fixes/gdb/gdb...
Setting up the environment for debugging gdb.
....
In other words, it now works as intuitively expected by the user.
If ever the user gave the correct executable on the command line,
then attached to the wrong pid, then confirmed loading the wrong executable,
the user can simply fix this by detaching, and attaching to the correct pid,
GDB will then tell again to the user that the exec-file might better
be loaded.
The default value of "ask" is chosen instead of e.g. "warn" as in most
cases, switching of executable will be the correct action,
and in any case, the user can decide to not load the executable,
as GDB asks a confirmation to the user to load the new executable.
For settings "ask" and "warn", the new function validate_exec_file ()
tries to get the inferior pid exec file and compares it with the current
exec file. In case of mismatch, it warns the user and optionally load
the executable.
This function is called in the attach_command implementation to cover
most cases of attaching to a running process.
It must also be called in remote.c, as the attach command is not supported
for all types of remote gdbserver.
gdb/ChangeLog
2020-01-25 Philippe Waroquiers <philippe.waroquiers@skynet.be>
Andrew Burgess [Mon, 11 Nov 2019 22:41:13 +0000 (22:41 +0000)]
gdb: Better frame tracking for inline frames
This commit improves GDB's handling of inline functions when there are
more than one inline function in a stack, so for example if we have a
stack like:
main -> aaa -> bbb -> ccc -> ddd
And aaa, bbb, and ccc are all inline within main GDB should (when
given sufficient debug information) be able to step from main through
aaa, bbb, and ccc. Unfortunately, this currently doesn't work, here's
an example session:
(gdb) start
Temporary breakpoint 1 at 0x4003b0: file test.c, line 38.
Starting program: /project/gdb/tests/inline/test
Temporary breakpoint 1, main () at test.c:38
38 global_var = 0;
(gdb) step
39 return aaa () + 1;
(gdb) step
aaa () at test.c:39
39 return aaa () + 1;
(gdb) step
bbb () at test.c:39
39 return aaa () + 1;
(gdb) step
ccc () at test.c:39
39 return aaa () + 1;
(gdb) step
ddd () at test.c:32
32 return global_var;
(gdb) bt
#0 ddd () at test.c:32
#1 0x00000000004003c1 in ccc () at test.c:39
#2 bbb () at test.c:26
#3 aaa () at test.c:14
#4 main () at test.c:39
Notice that once we get to line 39 in main, GDB keeps reporting line
39 in main as the location despite understanding that the inferior is
stepping through the nested inline functions with each use of step.
The problem is that as soon as the inferior stops we call
skip_inline_frames (from inline-frame.c) which calculates the
inferiors current state in relation to inline functions - it figures
out if we're in an inline function, and if we are counts how many
inline frames there are at the current location.
So, in our example above, when we step from line 38 in main to line 39
we stop at a location that is simultaneously in all of main, aaa, bbb,
and ccc. The block structure reflects the order in which the
functions would be called, with ccc being the most inner block and
main being the most outer block. When we stop GDB naturally finds the
block for ccc, however within skip_inline_frames we spot that bbb,
aaa, and main are super-blocks of the current location and that each
layer represents an inline function. The skip_inline_frames then
records the depth of inline functions (3 in this case for aaa, bbb,
and ccc) and also the symbol of the outermost inline function (in this
case 'aaa' as main isn't an inline function, it just has things inline
within it).
Now GDB understands the stack to be main -> aaa -> bbb -> ccc,
however, the state initialised in skip_inline_frames starts off
indicating that we should hide 3 frames from the user, so we report
that we're in main at line 39. The location of main, line 39 is
derived by asking the inline function state for the last symbol in the
stack (aaa in this case), and then asking for it's location - the
location of an inlined function symbol is its call site, so main, line
39 in this case.
If the user then asks GDB to step we don't actually move the inferior
at all, instead we spot that we are in an inline function stack,
lookup the inline state data, and reduce the skip depth by 1. We then
report to the user that GDB has stopped. GDB now understands that we
are in 'aaa'. In order to get the precise location we again ask GDB
for the last symbol from the inline data structure, and we are again
told 'aaa', we then get the location from 'aaa', and report that we
are in main, line 39.
Hopefully it's clear what the mistake here is, once we've reduced the
inline skip depth we should not be using 'aaa' to compute the precise
location, instead we should be using 'bbb'. That is what this patch
does.
Now, when we call skip_inline_frames instead of just recording the
last skipped symbol we now record all symbols in the inline frame
stack. When we ask GDB for the last skipped symbol we return a symbol
based on how many frames we are skipping, not just the last know
symbol.
With this fix in place, the same session as above now looks much
better:
(gdb) start
Temporary breakpoint 1 at 0x4003b0: file test.c, line 38.
Starting program: /project/gdb/tests/inline/test
Temporary breakpoint 1, main () at test.c:38
38 global_var = 0;
(gdb) s
39 return aaa () + 1;
(gdb) s
aaa () at test.c:14
14 return bbb () + 1;
(gdb) s
bbb () at test.c:26
26 return ccc () + 1;
(gdb) s
ccc () at test.c:20
20 return ddd () + 1;
(gdb) s
ddd () at test.c:32
32 return global_var;
(gdb) bt
#0 ddd () at test.c:32
#1 0x00000000004003c1 in ccc () at test.c:20
#2 bbb () at test.c:26
#3 aaa () at test.c:14
#4 main () at test.c:39
gdb/ChangeLog:
* frame.c (find_frame_sal): Move call to get_next_frame into more
inner scope.
* inline-frame.c (inilne_state) <inline_state>: Update argument
types.
(inilne_state) <skipped_symbol>: Rename to...
(inilne_state) <skipped_symbols>: ...this, and change to a vector.
(skip_inline_frames): Build vector of skipped symbols and use this
to reate the inline_state.
(inline_skipped_symbol): Add a comment and some assertions, fetch
skipped symbol from the list.
gdb/testsuite/ChangeLog:
* gdb.dwarf2/dw2-inline-many-frames.c: New file.
* gdb.dwarf2/dw2-inline-many-frames.exp: New file.
Andrew Burgess [Wed, 6 Nov 2019 10:17:05 +0000 (10:17 +0000)]
gdb: Don't reorder line table entries too much when sorting.
Don't reorder line table entries for the same address when sorting the
line table, maintain the compiler given line order. Usually this will
reflect the order in which lines are conceptually encountered at a
given address.
The interesting entries are those for lines 16 and 10 at address
0x4003ba, these represent the call to foo and the inlined body of
foo.
With the current line table sorting GDB builds the line table like
this (as shown by 'maintenance info line-table'):
INDEX LINE ADDRESS
0 14 0x00000000004003b0
1 15 0x00000000004003b0
2 10 0x00000000004003ba
3 16 0x00000000004003ba
4 END 0x00000000004003c1
5 4 0x00000000004004b0
6 5 0x00000000004004b0
7 END 0x00000000004004b7
Notice that entries 2 and 3 for lines 10 and 16 are now in a different
order to the line table as given by the compiler. With this patch
applied the order is now:
INDEX LINE ADDRESS
0 14 0x00000000004003b0
1 15 0x00000000004003b0
2 16 0x00000000004003ba
3 10 0x00000000004003ba
4 END 0x00000000004003c1
5 4 0x00000000004004b0
6 5 0x00000000004004b0
7 END 0x00000000004004b7
Notice that entries 2 and 3 are now in their original order again.
The consequence of the incorrect ordering is that when stepping
through inlined functions GDB will display the wrong line for the
inner most frame. Here's a GDB session before this patch is applied:
Starting program: /home/andrew/tmp/inline/test
Temporary breakpoint 1, main () at test.c:15
15 /* 15 */ global_var = 0;
(gdb) step
16 /* 16 */ return foo ();
(gdb) step
foo () at test.c:16
16 /* 16 */ return foo ();
(gdb) step
bar () at test.c:5
5 /* 5 */ return global_var;
The step from line 15 to 16 was fine, but the next step should have
taken us to line 10, instead we are left at line 16. The final step
to line 5 is as expected.
With this patch applied the session goes better:
Starting program: /home/andrew/tmp/inline/test
Temporary breakpoint 1, main () at test.c:15
15 /* 15 */ global_var = 0;
(gdb) step
16 /* 16 */ return foo ();
(gdb) step
foo () at test.c:10
10 /* 10 */ return bar ();
(gdb) step
bar () at test.c:5
5 /* 5 */ return global_var;
We now visit the lines as 15, 16, 10, 5 as we would like.
The reason for this issue is that the inline frame unwinder is
detecting that foo is inlined in main. When we stop at the shared
address 0x4003ba the inline frame unwinder first shows us the outer
frame, this information is extracted from the DWARF's
DW_TAG_inlined_subroutine entries and passed via GDB's block data.
When we step again the inlined frame unwinder moves us up the call
stack to the inner most frame at which point the frame is displayed as
normal, with the location for the address being looked up in the line
table.
As GDB uses the last line table entry for an address as "the" line to
report for that address it is critical that GDB maintain the order of
the line table entries. In the first case, by reordering the line
table we report the wrong location.
I had to make a small adjustment in find_pc_sect_line in order to
correctly find the previous line in the line table. In some line
tables I was seeing an actual line entry and an end of sequence marker
at the same address, before this commit these would reorder to move
the end of sequence marker before the line entry (end of sequence has
line number 0). Now the end of sequence marker remains in its correct
location, and in order to find a previous line we should step backward
over any end of sequence markers.
As an example, the binary:
gdb/testsuite/outputs/gdb.dwarf2/dw2-ranges-func/dw2-ranges-func-lo-cold
Has this line table before the patch:
INDEX LINE ADDRESS
0 48 0x0000000000400487
1 END 0x000000000040048e
2 52 0x000000000040048e
3 54 0x0000000000400492
4 56 0x0000000000400497
5 END 0x000000000040049a
6 62 0x000000000040049a
7 END 0x00000000004004a1
8 66 0x00000000004004a1
9 68 0x00000000004004a5
10 70 0x00000000004004aa
11 72 0x00000000004004b9
12 END 0x00000000004004bc
13 76 0x00000000004004bc
14 78 0x00000000004004c0
15 80 0x00000000004004c5
16 END 0x00000000004004cc
And after this patch:
INDEX LINE ADDRESS
0 48 0x0000000000400487
1 52 0x000000000040048e
2 END 0x000000000040048e
3 54 0x0000000000400492
4 56 0x0000000000400497
5 END 0x000000000040049a
6 62 0x000000000040049a
7 66 0x00000000004004a1
8 END 0x00000000004004a1
9 68 0x00000000004004a5
10 70 0x00000000004004aa
11 72 0x00000000004004b9
12 END 0x00000000004004bc
13 76 0x00000000004004bc
14 78 0x00000000004004c0
15 80 0x00000000004004c5
16 END 0x00000000004004cc
When calling find_pc_sect_line with the address 0x000000000040048e, in
both cases we find entry #3, we then try to find the previous entry,
which originally found this entry '2 52 0x000000000040048e',
after the patch it finds '2 END 0x000000000040048e', which
cases the lookup to fail.
By skipping the END marker after this patch we get back to the correct
entry, which is now #1: '1 52 0x000000000040048e', and
everything works again.
gdb/ChangeLog:
* buildsym.c (lte_is_less_than): Delete.
(buildsym_compunit::end_symtab_with_blockvector): Create local
lambda function to sort line table entries, and use
std::stable_sort instead of std::sort.
* symtab.c (find_pc_sect_line): Skip backward over end of sequence
markers when looking for a previous line.
gdb/testsuite/ChangeLog:
* gdb.dwarf2/dw2-inline-stepping.c: New file.
* gdb.dwarf2/dw2-inline-stepping.exp: New file.
Add struct to record dwarf line number state machine.
I believe an unintended change was made to how we store the DWARF line
table, the end of sequence markers between sequences of lines were
lost from the line table.
This commit fixes this small oversight and restores the end of
sequence markers.
Given that we've survived this long without noticing is clearly an
indication that this isn't that serious, however, a later patch that I
am developing would benefit from having the markers in place, so I'd
like to restore them.
Having the markers also means that the output of 'maintenance info
line-table' now more closely reflects the DWARF line table.
I've taken this opportunity to improve how 'maintenance info
line-table' displays the end of sequence markers - it now uses the END
keyword, rather than just printing an entry with line number 0. So we
see this:
INDEX LINE ADDRESS
0 12 0x00000000004003b0
1 17 0x00000000004003b0
2 18 0x00000000004003b0
3 END 0x00000000004003b7
4 5 0x00000000004004a0
5 6 0x00000000004004a0
6 END 0x00000000004004a7
Instead of what we would have seen, which was this:
I've added a small test that uses 'maintenance info line-table' to
ensure that we don't regress this again.
gdb/ChangeLog:
* dwarf2read.c (lnp_state_machine::record_line): Include
end_sequence parameter in debug print out. Record the line if we
are at an end_sequence marker even if it's not the start of a
statement.
* symmisc.c (maintenance_print_one_line_table): Print end of
sequence markers with 'END' not '0'.
gdb/testsuite/ChangeLog:
* gdb.base/maint.exp: Update line table parsing test.
* gdb.dwarf2/dw2-ranges-base.exp: Add new line table parsing test.
Jim Wilson [Fri, 24 Jan 2020 22:24:09 +0000 (14:24 -0800)]
RISC-V: Minor cleanup for s extension support.
Looking at older versions of the patch, I confirmed that the odd comment
I referred to earlier was indeed from the removal of the sx support. It
also explains an oddly formatted switch statement. This patch fixes both
minor problems.
bfd/
* elfxx-riscv.c (riscv_get_prefix_class): Format s case like others.
(riscv_parse_prefixed_ext): Fix s extension comment and reword to
avoid over long line.
Pedro Alves [Fri, 24 Jan 2020 18:46:20 +0000 (18:46 +0000)]
Fix re-runs of a second inferior (PR gdb/25410)
This fixes a latent bug exposed by the multi-target patch (5b6d1e4fa
"Multi-target support), and then fixes two other latent bugs exposed
by fixing that first latent bug.
The symptom described in the bug report is that starting a first
inferior, then trying to run a second (multi-threaded) inferior twice,
causes libthread_db to fail to load, along with other erratic
behavior:
(gdb) run
Starting program: /tmp/foo
warning: td_ta_new failed: generic error
Going a bit deeply, I found that if the two inferiors have different
symbols, we can see that just after inferior 2 exits, we are left with
inferior 2 selected, which is correct, but the symbols in scope belong
to inferior 1, which is obviously incorrect...
This problem is that there's a path in
scoped_restore_current_thread::restore() that switches to no thread
selected, and switches the current inferior, but leaves the current
program space as is, resulting in leaving the program space pointing
to the wrong program space (the one of the other inferior). This was
happening after handling TARGET_WAITKIND_NO_RESUMED, which is an event
that triggers after TARGET_WAITKIND_EXITED for the previous inferior
exit. Subsequent symbol lookups find the symbols of the wrong
inferior.
The fix is to use switch_to_inferior_no_thread in that problem spot.
This function was recently added along with the multi-target work
exactly for these situations.
As for testing, this patch adds a new testcase that tests symbol
printing just after inferior exit, which exercises the root cause of
the problem more directly. And then, to cover the use case described
in the bug too, it also exercises the lithread_db.so mis-loading, by
using TLS printing as a proxy for being sure that threaded debugging
was activated sucessfully. The testcase fails without the fix like
this, for the "print symbol just after exit" bits:
...
[Inferior 1 (process 8719) exited normally]
(gdb) PASS: gdb.multi/multi-re-run.exp: re_run_inf=1: iter=1: continue until exit
print re_run_var_1
No symbol "re_run_var_1" in current context.
(gdb) FAIL: gdb.multi/multi-re-run.exp: re_run_inf=1: iter=1: print re_run_var_1
...
And like this for the "libthread_db.so loading" bits:
Thread 1.1 "multi-re-run" hit Breakpoint 3, all_started () at /home/pedro/gdb/binutils-gdb/build/../src/gdb/testsuite/gdb.multi/multi-re-run.c:44
44 }
(gdb) PASS: gdb.multi/multi-re-run.exp: re_run_inf=1: iter=2: running to all_started in runto
print tls_var
Cannot find thread-local storage for LWP 27000, executable file /home/pedro/gdb/binutils-gdb/build/gdb/testsuite/outputs/gdb.multi/multi-re-run/multi-re-run:
Cannot find thread-local variables on this target
(gdb) FAIL: gdb.multi/multi-re-run.exp: re_run_inf=1: iter=2: print tls_var
As mentioned, that fix above goes on to expose a couple other latent
bugs. This commit fixes those as well.
The first latent bug exposed is in
infrun.c:handle_vfork_child_exec_or_exit. The current code is leaving
inf->pspace == NULL while calling clone_program_space. The idea was
to make it so that the breakpoints module doesn't use this inferior's
pspace to set breakpoints. With that, any
scoped_restore_current_thread use from within clone_program_space
tries to restore a NULL program space, which hits an assertion:
Attaching after Thread 0x7ffff74b8700 (LWP 27276) vfork to child process 27277]
[New inferior 2 (process 27277)]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
/home/pedro/gdb/binutils-gdb/build/../src/gdb/progspace.c:243: internal-error: void set_current_program_space(program_space*): Assertion `pspace != NULL' faile
d.
A problem internal to GDB has been detected,
further debugging may prove unreliable.
Quit this debugging session? (y or n) FAIL: gdb.threads/vfork-follow-child-exit.exp: detach-on-fork=off: continue (GDB internal error)
That NULL pspace idea was legitimate, but it's no longer necessary,
since commit b2e586e850db ("Defer breakpoint reset when cloning
progspace for fork child"). So the fix is to just set the inferior's
program space earlier.
The other latent bug exposed is in exec.c. When exec_close is called
from the program_space destructor, it is purposedly called with a
current program space that is not the current inferior's program
space. The problem is that the multi-target work added some code to
remove_target_sections that loops over all inferiors, and uses
scoped_restore_current_thread to save/restore the previous
thread/inferior/frame state. This makes it so that exec_close returns
with the current program space set to the current inferior's program
space, which is exactly what we did not want. Then the program_space
destructor continues into free_all_objfiles, but it is now running
that method on the wrong program space, resulting in:
Reading symbols from /home/pedro/gdb/binutils-gdb/build/gdb/testsuite/outputs/gdb.threads/fork-plus-threads/fork-plus-threads...
Reading symbols from /usr/lib/debug/usr/lib64/libpthread-2.26.so.debug...
Reading symbols from /usr/lib/debug/usr/lib64/libm-2.26.so.debug...
Reading symbols from /usr/lib/debug/usr/lib64/libc-2.26.so.debug...
Reading symbols from /usr/lib/debug/usr/lib64/ld-2.26.so.debug...
[Inferior 3 (process 9583) exited normally]
/home/pedro/gdb/binutils-gdb/build/../src/gdb/progspace.c:170: internal-error: void program_space::free_all_objfiles(): Assertion `so->objfile == NULL' failed.
A problem internal to GDB has been detected,
further debugging may prove unreliable.
Quit this debugging session? (y or n) FAIL: gdb.threads/fork-plus-threads.exp: detach-on-fork=off: inferior 1 exited (GDB internal error)
The fix is to use scoped_restore_current_pspace_and_thread instead of
scoped_restore_current_thread.
gdb/ChangeLog:
2020-01-24 Pedro Alves <palves@redhat.com>
PR gdb/25410
* thread.c (scoped_restore_current_thread::restore): Use
switch_to_inferior_no_thread.
* exec.c: Include "progspace-and-thread.h".
(add_target_sections, remove_target_sections):
scoped_restore_current_pspace_and_thread instead of
scoped_restore_current_thread.
* infrun.c (handle_vfork_child_exec_or_exit): Assign the pspace
and aspace to the inferior before calling clone_program_space.
Remove stale comment.
gdb/testsuite/ChangeLog:
2020-01-24 Pedro Alves <palves@redhat.com>
Fixes the below compile error on ARM NetBSD 9.0_RC1 (the only version I
tested). types.h does not define register_t by default.
We already use this define elsewhere, notably in bsd-kvm.c.
In file included from ../../gdb/arm-nbsd-nat.c:28:
/usr/include/machine/frame.h:54:2: error: unknown type name 'register_t'; did you mean '__register_t'?
register_t tf_spsr;
^
/usr/include/machine/types.h:77:14: note: '__register_t' declared here
typedef int __register_t;
^
There are other compile errors that this does not fix.
gdb/ChangeLog:
2020-01-24 Christian Biesinger <cbiesinger@google.com>
* arm-nbsd-nat.c: Define _KERNTYPES to get the declaration of
register_t.
On NetBSD, pthread_setname_np takes a printf-style format string plus
one argument:
https://netbsd.gw.com/cgi-bin/man-cgi?pthread_setname_np++NetBSD-current
This patch makes thread-pool.c handle that.
gdbsupport/ChangeLog:
2020-01-24 Christian Biesinger <cbiesinger@google.com>
* thread-pool.c (set_thread_name): Add an overload for the NetBSD
version of pthread_setname_np.
Nick Clifton [Fri, 24 Jan 2020 13:19:48 +0000 (13:19 +0000)]
Fix an illegal call to free() when copying a PE format file.
PR 25447
* coffgen.c (_bfd_coff_close_and_cleanup): Do not clear the keep
syms and keep strings flags as these may have been set in order to
prevent a bogus call to free.
make: *** No rule to make target '.../gdb/gdbserver/arch/arm.c', needed by 'TAGS'. Stop.
error produced by `make TAGS' by making the list of sources processed
match actual file locations and by moving host-specific object files
listed in DEPFILES to nat/ or target/ subdirectories as appropriate so
that the location of the corresponding source file can be mechanically
determined.
gdb/gdbserver/
* Makefile.in (SFILES): Adjust paths to point to real files.
(OBS): Move waitstatus.o to target/waitstatus.o.
(TAGS): Transform paths appropriately.
(%.o): Rename to...
(nat/%.o): ... this pattern rule.
(%.o): Rename to...
(target/%.o): ... this pattern rule.
* configure.srv: Adjust paths throughout to include nat/ prefix
with the revant files.
* configure.ac: Add `nat' and `target' to CONFIG_SRC_SUBDIR.
* configure: Regenerate.
gdbserver: Remove a stale TAGS recipe for config files
Complement commit 7ea814144a31 ("Fully disentangle gdb and gdbserver"),
<https://sourceware.org/ml/gdb-patches/2002-02/msg00692.html> (from
2002!), and remove a recipe to include config files in `make TAGS',
which are no longer used by `gdbserver' as from that commit.
gdb/gdbserver/
* Makefile.in (TAGS): Remove config files from the recipe.
regset_from_core_section doesn't exist anymore; it has been replaced
by the iterate_over_regset_sections gdbarch method. Update comments
accordingly to not confuse readers.
gdb/ChangeLog:
2020-01-24 Christian Biesinger <cbiesinger@google.com>
* aarch64-fbsd-tdep.c (aarch64_fbsd_iterate_over_regset_sections):
Update comment.
* aarch64-linux-tdep.c (aarch64_linux_iterate_over_regset_sections):
Likewise.
* arm-fbsd-tdep.c (arm_fbsd_iterate_over_regset_sections): Likewise.
* gdbcore.h (deprecated_add_core_fns): Update comment to point to
the correct replacement (iterate_over_regset_sections).
* riscv-fbsd-tdep.c (riscv_fbsd_iterate_over_regset_sections):
Update comment.
This patch attempts to address PR gdb/23718 by re-enabling stdin
whenever an exception is caught during gdb.execute().
When Python gdb.execute() is called, an exception could occur (e.g. the
target disappearing), which is then converted into a Python exception. If
stdin was disabled before the exception is caught, it is not re-enabled,
because the exception doesn't propagate to the top level of the event loop,
whose catch block would otherwise enable it.
The result is that when execution of a Python script completes, GDB does
not prompt or accept input, and is effectively hung.
This change rectifies the issue by re-enabling stdin in the catch block of
execute_gdb_command, prior to converting the exception to a Python
exception.
Since this patch was originally posted I've added a test, and also I
converted the code to re-enable stdin from this:
SWITCH_THRU_ALL_UIS ()
{
async_enable_stdin ();
}
to simply this:
async_enable_stdin ();
My reasoning is that we only need the SWITCH_THRU_ALL_UIS if, at the time
the exception is caught, the current_ui might be different than at the time
we called async_disable_stdin. Within python's execute_gdb_command I think
it should be impossible to switch current_ui, so the SWITCH_THRU_ALL_UIS
isn't needed.
gdb/ChangeLog:
PR gdb/23718
* gdb/python/python.c (execute_gdb_command): Call
async_enable_stdin in catch block.
gdb/testsuite/ChangeLog:
PR gdb/23718
* gdb.server/server-kill-python.exp: New file.
Andrew Burgess [Tue, 19 Nov 2019 18:09:43 +0000 (18:09 +0000)]
gdb: Re-enable stdin for all UIs from start_event_loop
If we catch an exception in start_event_loop's call to
gdb_do_one_event, then it is possible that the current_ui has changed
since we called async_disable_stdin. If that's the case then calling
async_enable_stdin will be called on the wrong UI.
To solve this problem we wrap the call to async_enable_stdin with
SWITCH_THRU_ALL_UIS, this causes us to try and re-enable stdin for all
UIs, which will catch any for which we called async_disable_stdin.
gdb/ChangeLog:
* event-loop.c (start_event_loop): Wrap async_enable_stdin with
SWITCH_THRU_ALL_UIS.
gdb/testsuite/ChangeLog:
* gdb.server/multi-ui-errors.c: New file.
* gdb.server/multi-ui-errors.exp: New file.
Andrew Burgess [Sat, 11 Jan 2020 01:38:28 +0000 (01:38 +0000)]
gdb/tui: asm window handles invalid memory and scrolls better
This started as a patch to enable the asm window to handle attempts to
disassemble invalid memory, but it ended up expanding into a
significant rewrite of how the asm window handles scrolling. These
two things ended up being tied together as it was impossible to
correctly test scrolling into invalid memory when the asm window would
randomly behave weirdly while scrolling.
Things that should work nicely now; scrolling to the bottom or top of
the listing with PageUp, PageDown, Up Arrow, Down Arrow and we should
be able to scroll past small areas of memory that don't have symbols
associated with them. It should also be possible to scroll to the
start of a section even if there's no symbol at the start of the
section.
Adding tests for this scrolling was a little bit of a problem. First
I would have liked to add tests for PageUp / PageDown, but the tuiterm
library we use doesn't support these commands right now due to only
emulating a basic ascii terminal. Changing this to emulate a more
complex terminal would require adding support for more escape sequence
control codes, so I've not tried to tackle that in this patch.
Next, I would have liked to test scrolling to the start or end of the
assembler listing and then trying to scroll even more, however, this
is a problem because in a well behaving GDB a scroll at the start/end
has no effect. What we need to do is:
- Move to start of assembler listing,
- Send scroll up command,
- Wait for all curses output,
- Ensure the assembler listing is unchanged, we're still at the
start of the listing.
The problem is that there is no curses output, so how long do we wait
at step 3? The same problem exists for scrolling to the bottom of the
assembler listing. However, when scrolling down you can at least see
the end coming, so I added a test for this case, however, this feels
like an area of code that is massively under tested.
gdb/ChangeLog:
PR tui/9765
* minsyms.c (lookup_minimal_symbol_by_pc_section): Update header
comment, add extra parameter, and update to store previous symbol
when appropriate.
* minsyms.h (lookup_minimal_symbol_by_pc_section): Update comment,
add extra parameter.
* tui/tui-disasm.c (tui_disassemble): Update header comment,
remove unneeded parameter, add try/catch around gdb_print_insn,
rewrite to add items to asm_lines vector.
(tui_find_backward_disassembly_start_address): New function.
(tui_find_disassembly_address): Updated throughout.
(tui_disasm_window::set_contents): Update for changes to
tui_disassemble.
(tui_disasm_window::do_scroll_vertical): No need to adjust the
number of lines to scroll.
gdb/testsuite/ChangeLog:
PR tui/9765
* gdb.tui/tui-layout-asm.exp: Add scrolling test for asm window.
Pedro Alves [Sat, 11 Jan 2020 01:37:26 +0000 (01:37 +0000)]
gdb/tui: Prevent exceptions from trying to cross readline
This is triggered by simply scrolling off the end of the dissasembly
window. This commit doesn't fix the actual exception that is being
thrown, which will still need to be fixed, but makes sure that we
don't ever throw an exception out to readline.
gdb/ChangeLog:
yyyy-mm-dd Pedro Alves <palves@redhat.com>
PR tui/9765
* tui/tui-io.c (tui_getc): Rename to ...
(tui_getc_1): ... this.
(tui_get): New, reimplent as try/catch wrapper around tui_getc_1.
Simon Marchi [Thu, 23 Jan 2020 22:44:22 +0000 (17:44 -0500)]
gdb: fix variable shadowing error in darwin-nat.c
We encounter this error when building on macOS with GCC.
CXX darwin-nat.o
/src-local/binutils-gdb/gdb/darwin-nat.c: In member function 'ptid_t darwin_nat_target::wait_1(ptid_t, target_waitstatus*)':
/src-local/binutils-gdb/gdb/darwin-nat.c:1264:18: error: declaration of 'inf' shadows a previous local [-Werror=shadow=compatible-local]
for (inferior *inf : all_inferiors (this))
^~~
/src-local/binutils-gdb/gdb/darwin-nat.c:1205:20: note: shadowed declaration is here
struct inferior *inf;
^~~
Fix it by moving the declaration of `inf` in the specific scopes that
need it. I think it's clearer this way anyway, as it shows that it's
not the same `inf` that is used in these different scopes.
Thanks to Iain Sandoe for reporting this. I did not see this error at
first, because I compile with the default system compiler on macOS,
which is clang. The compiler flag we try to enable for this is
`-Wshadow=local`, which is not one recognized by clang. I checked to
see if there would a version of the -Wshadow* warnings [1] we could
enable for clang, that would catch this, but the only one that would is
`-Wshadow` itself, and this is too invasive for us (which is why we
enabled just -Wshadow=local in the first place).
Simon Marchi [Thu, 23 Jan 2020 19:55:50 +0000 (14:55 -0500)]
gdb: fix darwin-nat.c build / adapt to multi-target
The darwin-nat.c file doesn't build since the multi-target changes
(5b6d1e4f, "Multi-target support"). This patch makes it build. I have
access to a macOS vm, so I am able to build it, but I wasn't able to
successfully codesign it and try to actually debug something, so I don't
know if it works. I don't have much more time to put on this to figure
it out, so I thought I'd sent the patch anyway, as it's at least a step
in the right direction.
The bulk of the patch is to change a bunch of functions to be methods of
the darwin_nat_target object, so that this can pass `this` to
find_inferior_ptid and other functions that now require a
process_stratum_target pointer.
The darwin_ptrace_him function (renamed to darwin_nat_target::ptrace_him
in this patch) is passed to fork_inferior as the `init_trace_fun`
parameter. Since the method can't be passed as a plain function pointer
(we need the `this` pointer), I changed the `init_trace_fun` parameter
of fork_inferior to be a gdb::function_view, so we can pass a lambda and
capture `this`.
The changes in darwin-nat.h are only to move definition higher in the
file, so that forward declarations are not needed.
Hannes Domani [Sat, 21 Dec 2019 16:08:14 +0000 (17:08 +0100)]
Cache the text section offset of shared libraries
Each time a dll is loaded, update_solib_list is called.
This in turn calls deep down xfer_partial -> windows_xfer_shared_libraries,
which calls windows_xfer_shared_library for each loaded dll,
and pe_text_section_offset reads the dll for the text section offset.
Also if the data provided by xfer_partial is bigger than 4K,
then all of this is done for each 4K chunk (see target_read_alloc_1).
Caching of the text section offset improves the startup time of
an application with >300 dynamically loaded plugins from 2m10s to 10s.
And the shutdown time improves from 2m to 2s.
Jim Wilson [Thu, 23 Jan 2020 00:45:04 +0000 (16:45 -0800)]
RISC-V: Change -march parsing.
bfd/
2020-01-22 Maxim Blinov <maxim.blinov@embecosm.com>
* bfd/elfnn-riscv.c (riscv_skip_prefix): New.
(riscv_prefix_cmp): Likewise.
(riscv_non_std_ext_p): Deleted.
(riscv_std_sv_ext_p): Likewise.
(riscv_non_std_sv_ext_p): Likewise.
(riscv_merge_non_std_and_sv_ext): Rename to...
(riscv_merge_multi_letter_ext): and modified to use riscv_prefix_cmp.
(riscv_merge_arch_attr_info): Replace 3 calls to
riscv_merge_non_std_and_sv_ext with single call to
riscv_merge_multi_letter_ext.
* bfd/elfxx-riscv.c (riscv_parse_std_ext): Break if we
encounter a 'z' prefix.
(riscv_get_prefix_class): New function, return prefix class based
on first few characters of input string.
(riscv_parse_config): New structure to factor out minor differences
in extension class parsing behaviour.
(riscv_parse_sv_or_non_std_ext): Rename to...
(riscv_parse_prefixed_ext): and parameterise with
riscv_parse_config.
(riscv_std_z_ext_strtab, riscv_std_s_ext_strtab): New.
(riscv_multi_letter_ext_valid_p): New.
(riscv_ext_x_valid_p, riscv_ext_z_valid_p, riscv_ext_s_valid_p): New.
(riscv_parse_subset): Delegate all non-single-letter parsing work
to riscv_parse_prefixed_ext.
* bfd/elfxx-riscv.h (riscv_isa_ext_class): New type.
(riscv_get_prefix_class): Declare.
gas/
2020-01-22 Maxim Blinov <maxim.blinov@embecosm.com>
* testsuite/gas/riscv/march-ok-s.d: sx is no longer valid and
s exts must be known, so rename *ok* to *fail*.
* testsuite/gas/riscv/march-ok-sx.d: Likewise.
* testsuite/gas/riscv/march-ok-s-with-version: Likewise.
* testsuite/gas/riscv/march-fail-s.l: Expected error messages for
above change.
* testsuite/gas/riscv/march-fail-sx.l: Likewise.
* testsuite/gas/riscv/march-fail-sx-with-version.l: Likewise.
MSP430: Fix simulator execution of RRUX instruction
The MSP430X RRUX instruction (unsigned right shift) is synthesized as
the RRC (rotate right through carry) instruction, but with the ZC
(zero carry) bit of the opcode extention word set.
Ensure the carry flag is ignored when the ZC bit is set.
sim/msp430/ChangeLog:
2020-01-22 Jozef Lawrynowicz <jozef.l@mittosystems.com>
* msp430-sim.c (msp430_step_once): Ignore the carry flag when executing
an RRC instruction, if the ZC bit of the extension word is set.
sim/testsuite/sim/msp430/ChangeLog:
2020-01-22 Jozef Lawrynowicz <jozef.l@mittosystems.com>
projects / thirdparty / binutils-gdb.git / log
