strlen is an ifunc function, and what we see above is the result of
calling the ifunc resolver in the inferior. That returns a pointer to
the actual target function that implements strlen on my machine. GDB
should have turned around and called the resolver automatically
without the user noticing.
Handle function aliases better (PR gdb/19487, errno printing)
which added the find_function_alias_target call to c-exp.y, to try to
find an alias with debug info for a minsym. For ifunc symbols, that
finds the ifunc's resolver if it has debug info (in the example it's
called "strlen_ifunc"), with the result that GDB calls that as a
regular function.
After this commit, we get now get:
(top-gdb) p strlen ("hello")
'__strlen_avx2' has unknown return type; cast the call to its declared return type
Which is correct, because __strlen_avx2 is written in assembly.
That'll be improved in a following patch, though.
gdb/ChangeLog:
2018-04-26 Pedro Alves <palves@redhat.com>
* c-exp.y (variable production): Skip finding an alias for ifunc
symbols.
Pedro Alves [Thu, 26 Apr 2018 12:02:26 +0000 (13:02 +0100)]
Fix breakpoints in ifunc after inferior resolved it (@got.plt symbol creation)
Setting a breakpoint on an ifunc symbol after the ifunc has already
been resolved by the inferior should result in creating a breakpoint
location at the ifunc target. However, that's not what happens on
current Fedora:
(gdb) n
53 i = gnu_ifunc (1); /* break-at-call */
(gdb)
54 assert (i == 2);
(gdb) b gnu_ifunc
Breakpoint 2 at gnu-indirect-function resolver at 0x7ffff7bd36ee
(gdb) info breakpoints
Num Type Disp Enb Address What
2 STT_GNU_IFUNC resolver keep y 0x00007ffff7bd36ee <gnu_ifunc+4>
The problem is that elf_gnu_ifunc_resolve_by_got never manages to
resolve an ifunc target. The reason is that GDB never actually
creates the internal got.plt symbols:
(gdb) p 'gnu_ifunc@got.plt'
No symbol "gnu_ifunc@got.plt" in current context.
and this is because GDB expects that rela.plt has relocations for
.plt, while it actually has relocations for .got.plt:
Relocation section [10] '.rela.plt' for section [22] '.got.plt' at offset 0x570 contains 2 entries:
Offset Type Value Addend Name
0x0000000000601018 X86_64_JUMP_SLOT 000000000000000000 +0 __assert_fail
0x0000000000601020 X86_64_JUMP_SLOT 000000000000000000 +0 gnu_ifunc
Using an older system on the GCC compile farm (machine gcc15, an
x86-64 running Debian 6.0.8, with GNU ld 2.20.1), we see that it used
to be that we'd get a .rela.plt section for .plt:
Relocation section [ 9] '.rela.plt' for section [11] '.plt' at offset 0x578 contains 3 entries:
Offset Type Value Addend Name
0x0000000000600cc0 X86_64_JUMP_SLOT 000000000000000000 +0 __assert_fail
0x0000000000600cc8 X86_64_JUMP_SLOT 000000000000000000 +0 __libc_start_main
0x0000000000600cd0 X86_64_JUMP_SLOT 000000000000000000 +0 gnu_ifunc
Those offsets did point into .got.plt, as seen with objdump -h:
Jan Beulich [Thu, 26 Apr 2018 06:55:02 +0000 (08:55 +0200)]
x86: fold various non-memory operand AVX512VL templates
There's little point carrying up to three templates per insn flavor
when the sole difference is operand size and the dependency on AVX512VL
being enabled. Instead the need for AVX512VL can be derived from an
operand allowing for ZMMword as well as one or both or XMMword and
YMMword (irrespective of whether this is a register or memory operand).
Without further abstraction to deal with the different Disp8MemShift
values between the templates, only a limited set (mostly ones only
allowing for non-memory operands) can be folded, which is being done
here.
Also drop IgnoreSize wherever possible from anything that's being
touched anyway.
Jan Beulich [Thu, 26 Apr 2018 06:53:20 +0000 (08:53 +0200)]
x86: also optimize zeroing-masking variants of insns
When zeroing an element of a register it doesn't matter whether the zero
results from the actual operation (xor, sub, or nand) or from the
zeroing-masking taking effect due to a clear mask register bit.
Jan Beulich [Thu, 26 Apr 2018 06:49:41 +0000 (08:49 +0200)]
x86: properly force / avoid forcing EVEX encoding
Pseudo prefixes are supposed to be a hint only - when the specific
encoding can't be used to encode an insn, silently override it. But
this overriding must only happen after the respective check, to
avoid forcing EVEX encoding because of something that isn't a valid
register name in the given context.
Jan Beulich [Thu, 26 Apr 2018 06:48:56 +0000 (08:48 +0200)]
x86: CpuXSAVE is a prereq for various other features
All of AVX, LWP, MPX, and PKU require XSAVE, and hence it as well as
XRSTOR should be enabled when enabling these ISA extensions. Leverage
these implications to shorten some of the cpu_flag_init[] entries.
Jan Beulich [Thu, 26 Apr 2018 06:48:01 +0000 (08:48 +0200)]
x86: drop CpuRegMMX, CpuReg[XYZ]MM, and CpuRegMask
It's not clear to me why they had been introduced - the respective
comments in opcodes/i386-gen.c are certainly wrong: ymm<N> registers
are very well supported (and necessary) with just AVX512F.
Jan Beulich [Thu, 26 Apr 2018 06:45:35 +0000 (08:45 +0200)]
x86: x87-related adjustments
Neither 287 wrt 8087 nor 387 wrt 287 are proper supersets - in each case
some insns get removed from the ISA (they become NOPs, but code intended
for newer co-processors should not use them).
Furthermore with .no87, ST should not be recognized as a register name.
Jan Beulich [Thu, 26 Apr 2018 06:31:14 +0000 (08:31 +0200)]
x86: fix indentation in build_modrm_byte()
The VEX3SOURCES code was (originally) written with just space
indentation, which is not in line with general coding style as well as
the style later in the function.
Jan Beulich [Thu, 26 Apr 2018 06:28:38 +0000 (08:28 +0200)]
x86: drop dead code from build_modrm_byte()
There are no templates with VexImmExt and ImmExt set at the same time.
There are also no VEX3SOURCES templates with CpuFMA. I assume both are
left-overs from the implementation of an early specification which was
later revised.
Christophe Lyon [Tue, 20 Mar 2018 09:55:29 +0000 (10:55 +0100)]
[ARM] Add TLS relocations for FDPIC.
Define and handle TLS relocations for FDPIC in BFD and gas.
In gas, the new relocations are rejected if the --fdpic option was not
specified.
We also define the __tdata_start symbol to mark the start of the
.tdata section. This allows FDPIC static binaries to find the start of
.tdata section, since phdr->p_vaddr of TLS segment is not a valid
value for FDPIC.
Christophe Lyon [Tue, 20 Mar 2018 09:54:50 +0000 (10:54 +0100)]
[ARM] Add FDPIC OSABI flag support.
ELF files targetting ARM FDPIC use the ELFOSABI_ARM_FDPIC flag.
Set it appropriately in file generators (eg. gas), and handle it in
readers (eg. readelf).
bfd/
* elf32-arm.c (elf32_arm_print_private_bfd_data): Support
EF_ARM_PIC and ELFOSABI_ARM_FDPIC.
(elf32_arm_post_process_headers): Support ELFOSABI_ARM_FDPIC.
(ELF_OSABI): Define to ELFOSABI_ARM_FDPIC.
binutils/
* readelf.c (decode_ARM_machine_flags): Support EF_ARM_PIC.
(get_osabi_name): Support ELFOSABI_ARM_FDPIC.
gas/
* config/tc-arm.c (arm_fdpic): New.
(elf32_arm_target_format): Support FDPIC.
(OPTION_FDPIC): New.
(md_longopts): Support FDPIC.
(md_parse_option): Likewise.
(md_show_usage): Likewise.
Pedro Alves [Wed, 25 Apr 2018 16:28:25 +0000 (17:28 +0100)]
Fix new inferior events output
Since f67c0c917150 ("Enable 'set print inferior-events' and improve
detach/fork/kill/exit messages"), when detaching a remote process, we
get, for detach against a remote target:
(gdb) detach
Detaching from program: ...., process 5388
Ending remote debugging.
[Inferior 1 (Thread 5388.5388) detached]
^^^^^^^^^^^^^^^^
That is incorrect, for it is printing a thread id as string while we
should be printing the process id instead. I.e., either one of:
depending on remote stub support for the multi-process extensions.
Similarly, after killing a process, we're printing thread ids while we
should be printing process ids. E.g., on native GNU/Linux:
(gdb) k
Kill the program being debugged? (y or n) y
[Inferior 1 (Thread 0x7ffff7faa8c0 (LWP 30721)) has been killed]
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
while it should have been:
Kill the program being debugged? (y or n) y
[Inferior 1 (process 30721) has been killed]
^^^^^^^^^^^^^
There's a wording inconsistency between detach and kill:
[Inferior 1 (process 30721) has been killed]
[Inferior 1 (process 30721) detached]
Given we were already saying "detached" instead of "has been
detached", and we used to say just "exited", and given that the "has
been" doesn't really add any information, this commit changes the
message to just "killed":
[Inferior 1 (process 30721) killed]
gdb/ChangeLog:
2018-04-25 Pedro Alves <palves@redhat.com>
* infcmd.c (kill_command): Print the pid as string, not the whole
thread's ptid. Add comment. s/has been killed/killed/ in output
message.
* remote.c (remote_detach_1): Print the pid as string, not the
whole thread's ptid.
gdb/testsuite/ChangeLog:
2018-04-25 Pedro Alves <palves@redhat.com>
* gdb.base/hook-stop.exp: Expect "killed" instead of "has been
killed".
* gdb.base/kill-after-signal.exp: Likewise.
* gdb.threads/kill.exp: Likewise.
Fix bug with relocation addends and merge sections with --icf.
During --icf processing, gold was incorrectly processing the relocation
addend for references to items in a merge section. PC-relative references
and other forms of reference with a biased base address require a
non-section local symbol, where the addend is purely the bias.
gold/
PR gold/20642
PR gold/22820
* gc.h (gc_process_relocs): Flag STT_SECTION symbols in symvec.
* icf.cc (get_section_contents): For merge sections, ignore the
addend for relocations against non-section symbols.
Alan Modra [Mon, 16 Apr 2018 11:03:36 +0000 (20:33 +0930)]
Remove arm-aout and arm-coff support
This also removes arm-netbsd (not arm-netbsdelf!), arm-openbsd, and
arm-riscix. Those targets weren't on the obsolete list but they are
all aout, and it doesn't make all that much sense to remove arm-aout
without removing them too.
Fix internal error caused by conflicting default version definitions.
gold/
PR gold/16504
* dynobj.cc (Versions::symbol_section_contents): Don't set
VERSYM_HIDDEN flag for undefined symbols.
* symtab.cc (Symbol_table::add_from_object): Don't override default
version definition with a different default version.
* symtab.h (Symbol::from_dyn): New method.
* testsuite/plugin_test.c (struct sym_info): Add ver field.
(claim_file_hook): Pass symbol version to plugin API.
(parse_readelf_line): Parse symbol version.
* testsuite/Makefile.am (ver_test_pr16504): New test case.
* testsuite/Makefile.in: Regenerate.
* testsuite/ver_test_pr16504.sh: New test script.
* testsuite/ver_test_pr16504_a.c: New source file.
* testsuite/ver_test_pr16504_a.script: New version script.
* testsuite/ver_test_pr16504_b.c: New source file.
* testsuite/ver_test_pr16504_b.script: New version script.
Enable 'set print inferior-events' and improve detach/fork/kill/exit messages
This patch aims to turn 'set print inferior-events' always on, and do
some cleanup on the messages printed by GDB when various inferior
events happen (attach, detach, fork, kill, exit).
To make sure that the patch is correct, I've tested it with a handful
of combinations of 'set follow-fork-mode', 'set detach-on-fork' and
'set print inferior-events'. In the end, I decided to make my
hand-made test into an official testcase. More on that below.
Using the following program as an example:
#include <unistd.h>
int main ()
{
fork ();
return 0;
}
We see the following outputs from the patched GDB:
- With 'set print inferior-events on':
(gdb) r
Starting program: a.out
[Detaching after fork from child process 27749]
[Inferior 1 (process 27745) exited normally]
(gdb)
(unpatched-gdb) r
Starting program: a.out
[New process 5702]
[Inferior 2 (process 5702) exited normally]
(unpatched-gdb)
It is possible to notice that, in this scenario, the patched GDB
will lose the '[New process %d]' message.
- With 'set print inferior-events on', 'set follow-fork-mode child'
and 'set detach-on-fork on':
(gdb) r
Starting program: a.out
[Attaching after process 27905 fork to child process 27909]
[New inferior 2 (process 27909)]
[Detaching after fork from parent process 27905]
[Inferior 1 (process 27905) detached]
[Inferior 2 (process 27909) exited normally]
(gdb)
Compare this output with an unpatched GDB, using the same settings:
As can be seen above, I've also made a few modifications to messages
that are printed when 'set print inferior-events' is on. For example,
a few of the messages did not contain the '[' and ']' as
prefix/suffix, which led to a few inconsistencies like:
Attaching after process 22995 fork to child process 22999.
[New inferior 22999]
Detaching after fork from child process 22999.
[Inferior 22995 detached]
[Inferior 2 (process 22999) exited normally]
So I took the opportunity and included the square brackets where
applicable. I have also made the existing messages more uniform, by
always printing "Inferior %d (process %d)..." where applicable. This
makes it easier to identify the inferior number and the PID number
from the messages.
As suggested by Pedro, the "[Inferior %d exited]" message from
'exit_inferior' has been removed, because it got duplicated when
'inferior-events' is on. I'm also using the
'add_{thread,inferior}_silent' versions (instead of their verbose
counterparts) on some locations, also to avoid duplicated messages.
For example, a patched GDB with 'set print inferior-events on', 'set
detach-on-fork on' and 'set follow-fork-mode child', but using
'add_thread', would print:
(gdb) run
Starting program: a.out
[Attaching after process 25088 fork to child process 25092.]
[New inferior 25092] <--- duplicated
[Detaching after fork from child process 25092.]
[Inferior 25088 detached]
[New process 25092] <--- duplicated
[Inferior 2 (process 25092) exited normally]
But if we use 'add_thread_silent' (with the same configuration as
before):
(gdb) run
Starting program: a.out
[Attaching after process 31606 fork to child process 31610]
[New inferior 2 (process 31610)]
[Detaching after fork from parent process 31606]
[Inferior 1 (process 31606) detached]
[Inferior 2 (process 31610) exited normally]
As for the tests, the configuration options being exercised are:
It was also necessary to perform adjustments on several testcases,
because the expected messages changed considerably.
Built and regtested on BuildBot, without regressions.
gdb/ChangeLog:
2018-04-24 Jan Kratochvil <jan.kratochvil@redhat.com>
Sergio Durigan Junior <sergiodj@redhat.com>
Pedro Alves <palves@redhat.com>
* infcmd.c (kill_command): Print message when inferior has
been killed.
* inferior.c (print_inferior_events): Remove 'static'. Set as
'1'.
(add_inferior): Improve message printed when
'print_inferior_events' is on.
(exit_inferior): Remove message printed when
'print_inferior_events' is on.
(detach_inferior): Improve message printed when
'print_inferior_events' is on.
(initialize_inferiors): Use 'add_inferior_silent' to set
'current_inferior_'.
* inferior.h (print_inferior_events): Declare here as
'extern'.
* infrun.c (follow_fork_inferior): Print '[Attaching...]' or
'[Detaching...]' messages when 'print_inferior_events' is on.
Use 'add_thread_silent' instead of 'add_thread'. Add '[' and ']'
as prefix/suffix for messages. Remove periods. Fix erroneous
'Detaching after fork from child...', replace it by '... from
parent...'.
(handle_vfork_child_exec_or_exit): Add '[' and ']' as
prefix/suffix when printing 'Detaching...' messages. Print
them when 'print_inferior_events' is on.
* remote.c (remote_detach_1): Print message when detaching
from inferior and '!is_fork_parent'.
gdb/testsuite/ChangeLog:
2018-04-24 Jan Kratochvil <jan.kratochvil@redhat.com>
Sergio Durigan Junior <sergiodj@redhat.com>
Pedro Alves <palves@redhat.com>
* gdb.base/attach-non-pgrp-leader.exp: Adjust 'Detaching...'
regexps to expect for '[Inferior ... detached]' as well.
* gdb.base/attach.exp: Likewise.
* gdb.base/catch-syscall.exp (check_for_program_end): Adjust
"gdb_continue_to_end".
(test_catch_syscall_with_wrong_args): Likewise.
* gdb.base/foll-fork.exp: Adjust regexps to match '[' and
']'. Don't set 'verbose' on.
* gdb.base/foll-vfork.exp: Likewise.
* gdb.base/fork-print-inferior-events.c: New file.
* gdb.base/fork-print-inferior-events.exp: New file.
* gdb.base/hook-stop.exp: Adjust regexps to expect for new
'[Inferior ... has been killed]' message.
* gdb.base/kill-after-signal.exp: Likewise.
* gdb.base/solib-overlap.exp: Adjust regexps to expect for new
detach message.
* gdb.threads/kill.exp: Adjust regexps to expect for new kill
message.
* gdb.threads/clone-attach-detach.exp: Adjust 'Detaching...'
regexps to expect for '[Inferior ... detached]' as well.
* gdb.threads/process-dies-while-detaching.exp: Likewise.
Simon Marchi [Tue, 24 Apr 2018 14:14:14 +0000 (10:14 -0400)]
info-shared.exp: Replace libs=-ldl with shlib_load
As reported in PR 23104, -ldl doesn't work on FreeBSD. Replace it with
shlib_load, which adds the right flags for dynamic library loading based
on the current target platform.
The test still passes on Linux, and should now pass on FreeBSD, though I
did not test personally.
gdb/testsuite/ChangeLog:
PR gdb/23104
* gdb.base/info-shared.exp: Replace libs=-ldl with shlib_load.
Tom Tromey [Sat, 21 Apr 2018 22:38:33 +0000 (16:38 -0600)]
Remove a cleanup from scm-frame.c
This removes a cleanup from scm-frame.c, replacing it with
unique_xmalloc_ptr and a new scope. I believe this also fixes a
latent bug involving calling do_cleanups twice for a single cleanup.
Regression tested using the gdb.guile test suite on x86-64 Fedora 26.
ChangeLog
2018-04-23 Tom Tromey <tom@tromey.com>
* guile/scm-frame.c (gdbscm_frame_read_var): Use
gdb::unique_xmalloc_ptr.
Rajendra SY [Sun, 22 Apr 2018 22:19:26 +0000 (18:19 -0400)]
Fixed test case to compile & run on FreeBSD
Problems:
1. linking -dl lib on FreeBSD platform
2. backtrace from ld-elf shows r_debug_state() instead of _dl_debug_state()
Cause:
1. There is no dl library on FreeBSD platform test has to ignore linking "-ldl"
2. The stop due to a shared library event shows backtrace frame #0
function as r_debug_state()
gdb/ChangeLog:
PR gdb/23095
* gdb/testsuite/gdb.base/break-probes.exp: Pass shlib_load to
prepare_for_testing. Set normal_bp to r_debug_state if target
is bsd.
Pedro Alves [Sat, 21 Apr 2018 17:19:30 +0000 (18:19 +0100)]
FreeBSD: Fix 'Couldn't get registers: Device busy' error (PR gdb/23077)
As Rajendra SY reported at
<https://sourceware.org/ml/gdb-patches/2018-04/msg00399.html>, several
attach-related tests are failing on FreeBSD. The "attach" command
errors with "Couldn't get registers: Device busy".
When the "attach" command is executed, it calls target_attach ->
inf_ptrace_attach, which just does the ptrace(PT_ATTACH), it does not
wait for the child to stop with SIGSTOP. Afterwards, the command is
complete and we go back to the event loop. The event loop wakes up
and we end up in target_wait -> fbsd_wait, and handle the SIGSTOP
stop.
At the end of execute_command, though, before going back to the event
loop, we check if the frame language changed via
check_frame_language_change(). That reads the current PC, which is
what leads to the registers read that fails.
The problem is that we fail to mark the attached-to thread as
executing between the initial attach, and the subsequent target_wait.
Until we see the thread stop with SIGSTOP, we shouldn't try to read
registers off of it. I guess there may a timing issue here - if
you're "lucky", the thread may stop before gdb reads its registers,
masking the problem.
With that fixed, check_frame_language_change() becomes a nop until the
thread is marked not-executing again, after target_wait is called and
we go through handle_inferior_event -> normal_stop.
We haven't seen the problem on Linux because there, the target_attach
implementation waits for the thread to stop before returning. Still,
that's supposedly hidden from the core, since the Linux target, like
most targets, is a '!to_attach_no_wait' target.
This fixes:
FAIL: gdb.base/attach.exp: attach1, after setting file
FAIL: gdb.base/attach.exp: attach2, with no file
FAIL: gdb.base/attach.exp: load file manually, after attach2 (re-read) (got interactive prompt)
FAIL: gdb.base/attach.exp: attach when process' a.out not in cwd
FAIL: gdb.base/dprintf-detach.exp: bai=on ds=gdb dd=on: re-attach to inferior
FAIL: gdb.base/dprintf-detach.exp: bai=on ds=gdb dd=off: re-attach to inferior
FAIL: gdb.base/dprintf-detach.exp: bai=on ds=call dd=on: re-attach to inferior
FAIL: gdb.base/dprintf-detach.exp: bai=on ds=call dd=off: re-attach to inferior
FAIL: gdb.base/dprintf-detach.exp: bai=on ds=agent dd=on: re-attach to inferior
FAIL: gdb.base/dprintf-detach.exp: bai=on ds=agent dd=off: re-attach to inferior
FAIL: gdb.base/dprintf-detach.exp: bai=off ds=gdb dd=on: re-attach to inferior
FAIL: gdb.base/dprintf-detach.exp: bai=off ds=gdb dd=off: re-attach to inferior
FAIL: gdb.base/dprintf-detach.exp: bai=off ds=call dd=on: re-attach to inferior
FAIL: gdb.base/dprintf-detach.exp: bai=off ds=call dd=off: re-attach to inferior
FAIL: gdb.base/dprintf-detach.exp: bai=off ds=agent dd=on: re-attach to inferior
FAIL: gdb.base/dprintf-detach.exp: bai=off ds=agent dd=off: re-attach to inferior
gdb/ChangeLog:
2018-04-21 Pedro Alves <palves@redhat.com>
Rajendra SY <rajendra.sy@gmail.com>
* inf-ptrace.c (inf_ptrace_attach): Mark the thread as executing.
* remote.c (extended_remote_attach): In all-stop mode, mark the
thread as executing.
Alan Modra [Sat, 21 Apr 2018 02:36:57 +0000 (12:06 +0930)]
Test that gcc -B picks up new ld
gcc configured using --with-ld always uses that particular version of
ld if it exists. This patch adds a check that the ld version reported
directly from ld-new matches that reported via gcc. I chose to
compare the ld --version output rather than a more strict test (for
example by gcc --print-prog-name=ld), because the version test allows
a user to run the ld testsuite after installing a new ld (in the place
expected by gcc).
Sample output:
Running target unix
Using /usr/share/dejagnu/baseboards/unix.exp as board description file for target.
Using /usr/share/dejagnu/config/unix.exp as generic interface file for target.
Using /home/alan/src/binutils-gdb/ld/testsuite/config/default.exp as tool-and-target-specific interface file.
ERROR: ************************************************************************
ERROR: Your compiler driver ignores -B when choosing ld.
ERROR: You will not be testing the new ld in many of the following tests.
ERROR: It seems you will be testing /usr/bin/x86_64-w64-mingw32-ld instead.
ERROR: ************************************************************************
WARNING: Assuming target board is the local machine (which is probably wrong).
You may need to set your DEJAGNU environment variable.
Running /home/alan/src/binutils-gdb/ld/testsuite/ld-aarch64/aarch64-elf.exp ...
* testsuite/lib/ld-lib.exp (run_host_cmd): Check that gcc -B
works.
Alan Modra [Thu, 19 Apr 2018 01:49:35 +0000 (11:19 +0930)]
PR22978, TLS local-dynamic incorrectly linked on hppa-linux
We were emitting dynamic relocs on the second word of a TLS GD GOT
entry pair (the dtprel offset), without the addend necessary when no
symbol is present on the dynamic reloc. Unfortunately the simple
solution of providing the proper addend doesn't work due to an hppa
glibc ld.so bug that ignores such addends. So instead optimize the
relocs. The dtprel offset is known at link time for locally defined
symbols (the only case where we'll end up with no symbol on a dynamic
reloc) so we can omit the dynamic reloc in that case.
Furthermore, we can omit a dynamic reloc on the first word of a TLS GD
GOT entry pair (the module id) if the symbol is local and we are
producing an executable. Similarly, a tprel reloc on a TLS IE GOT
entry is not needed for local symbols in an executable. So the
condition for TLS GOT relocs can become bfd_link_dll(info) rather than
bfd_link_pic(info) as needed for normal GOT relocs.
This all presumes hppa ld.so doesn't need to differentiate TLS GD GOT
pairs from TLS LD GOT pairs, which is currently true.
PR 22978
* elf32-hppa.c (got_relocs_needed): Add extra param to special
case both dtprel and tprel relocs.
(allocate_dynrelocs): Adjust conditions for got relocs.
(elf32_hppa_relocate_section): Likewise for local sym got relocs.
Emit dynamic relocs on TLS GOT entries for shared libraries,
not when pic. Omit dynamic reloc on dtprel entry when local,
and on tprel entry when local and executable.
Richard Bunt [Fri, 20 Apr 2018 03:02:35 +0000 (23:02 -0400)]
Add test case for a known hang in infrun
The hang occurs when GDB tries to call inferior functions on two
different threads with scheduler-locking turned on. The first call works
fine, with the call to infrun_async(1) causing the signal_handler to be
marked and the event to be handled, but then the event loop resets the
"ready" member to zero, while leaving infrun_is_async set to 1. As a
result, GDB hangs if the user switches to another thread and calls a
second function because calling infrun_async(1) a second time has no
effect, meaning the inferior call events are never handled.
The added test case provokes the above issue.
gdb/testsuite/ChangeLog:
* gdb.threads/multiple-successive-infcall.c: New test.
* gdb.threads/multiple-successive-infcall.exp: New file.
Fix some typos.
Remove obsolete comment about dispatch to thread_apply_command,
rather tell that thread_command either switches to a thread,
or prints the current thread.
2018-04-19 Philippe Waroquiers <philippe.waroquiers@skynet.be>
Simon Marchi [Thu, 19 Apr 2018 17:19:41 +0000 (13:19 -0400)]
Fix dependency tracking in gdbserver subdirectories
The dependency tracking (the thing that knows which source file included
which other source file during last build to know what to rebuild when
an included file changes) is broken for gdbserver subdirectories (arch
and common).
The dependency tracking files are created in the form
arch/.deps/i386.Po
but we try to include
.deps/arch/i386.Po
An easy smoke test is too "touch" the gdb/features/i386/32bit-core.c
file in the source directory and try to rebuild gdbserver. This file is
included by gdb/arch/i386.c, so it should cause
gdb/gdbserver/arch/i386.o in the build directory to be rebuilt. It
currently isn't rebuilt, but is with this patch applied.
This patch copies the technique used in GDB to transform the dep file
paths to the proper form.
Also, while testing using the depcomp method of dependency tracking (by
just hacking the condition), I noticed that depcomp was not found. The
path to depcomp seems to be missing a "..".
gdb/gdbserver/ChangeLog:
* Makefile.in (depcomp): Add "..".
(all_deps_files): New and use it.
Fix second bug where --icf=safe triggers segfault when linking ARM.
When checking a R_ARM_TARGET[12] relocation, we need a valid target
pointer, but the garbage collection code was passing a NULL instead.
The previous fix for this bug fixed the call to
scan.global_reloc_may_be_function_pointer, but missed the similar
call to scan.local_reloc_may_be_function_pointer.
gold/
PR gold/23046
* gc.h (gc_process_relocs): Pass target to
scan.local_reloc_may_be_function_pointer.
Alan Modra [Thu, 19 Apr 2018 04:43:41 +0000 (14:13 +0930)]
PR22537, Segmentation fault with static PIE
The only stub type that makes sense for undefined symbols, or those
defined in shared libraries, is a plt call stub. This patch arranges
to have "destination" set to -1 on such symbols, making for an easy
test in hppa_type_of_stub.
PR 22537
* elf32-hppa.c (elf32_hppa_size_stubs): Init "destination" to -1.
(hppa_type_of_stub): Don't return a long branch stub for
symbols other than those defined statically.
Alan Hayward [Wed, 18 Apr 2018 19:06:14 +0000 (20:06 +0100)]
Add feature reference in .dat files
For all targets which use the newer style target descriptions, add a
"feature" marker in the dat files.
Update regdat.sh to parse feature, but do not use it (yet).
H.J. Lu [Wed, 18 Apr 2018 01:15:13 +0000 (18:15 -0700)]
x86: Don't define elf32_bed/elf64_bed variables
Define elf32_bed and elf64_bed before including "elf32-target.h" and
"elf64-target.h" to avoid local elf32_bed and elf64_bed variables.
* elf32-i386.c (elf32_bed): Define before including
"elf32-target.h".
* elf64-x86-64.c (elf64_bed): Define before including
"elf64-target.h".
(elf32_bed): Define before including "elf32-target.h".
Michael Eager [Tue, 17 Apr 2018 21:47:13 +0000 (14:47 -0700)]
[MicroBlaze] PIC data text relative
Andrew Sadek <andrew.sadek.se@gmail.com>
A new implemented feature in GCC Microblaze that allows Position
Independent Code to run using Data Text Relative addressing instead
of using Global Offset Table.
Its aim was to make 'PIC' more efficient and flexible as elf size
excess performance overhead were noticed when using GOT due to the
indirect addressing.
include/ChangeLog:
* bfdlink.h (Add flag): Add new flag @ 'bfd_link_info' struct.
* elf/microblaze.h (Add 3 new relocations):
R_MICROBLAZE_TEXTPCREL_64, R_MICROBLAZE_TEXTREL_64
and R_MICROBLAZE_TEXTREL_32_LO for relax function.
bfd/ChangeLog:
* bfd/reloc.c (2 new BFD relocations):
BFD_RELOC_MICROBLAZE_64_TEXTPCREL &
BFD_RELOC_MICROBLAZE_64_TEXTPCREL
* bfd/bfd-in2.h: Regenerate
* bfd/libbfd.h: Regenerate
* bfd/elf32-microblaze.c (Handle new relocs): define 'HOWTO' of 3
new relocs and handle them in both relocate and relax functions.
(microblaze_elf_reloc_type_lookup): add mapping between for new
bfd relocs.
(microblaze_elf_relocate_section): Handle new relocs in case of
elf relocation.
(microblaze_elf_relax_section): Handle new relocs for elf relaxation.
gas/ChangeLog:
* gas/config/tc-microblaze.c (Handle new relocs directives in
assembler): Handle new relocs from compiler output.
(imm_types): add new imm types for data text relative addressing
TEXT_OFFSET, TEXT_PC_OFFSET
(md_convert_frag): conversion for BFD_RELOC_MICROBLAZE_64_TEXTPCREL,
BFD_RELOC_MICROBLAZE_64_TEXTPCREL
(md_apply_fix): apply fix for BFD_RELOC_MICROBLAZE_64_TEXTPCREL,
BFD_RELOC_MICROBLAZE_64_TEXTPCREL
(md_estimate_size_before_relax): estimate size for
BFD_RELOC_MICROBLAZE_64_TEXTPCREL,
BFD_RELOC_MICROBLAZE_64_TEXTPCREL
(tc_gen_reloc): generate relocations for
BFD_RELOC_MICROBLAZE_64_TEXTPCREL,
BFD_RELOC_MICROBLAZE_64_TEXTPCREL
ld/ChangeLog:
* ld/lexsup.c (Add 2 ld options):
(ld_options): add disable-multiple-abs-defs @ 'ld_options' array
(parse_args): parse new option and pass flag to 'link_info' struct.
* ld/ldlex.h (Add enum): add new enum @ 'option_values' enum.
* ld/ld.texinfo (Add new option): Add description for
'disable-multiple-abs-defs'
* ld/main.c: Initialize flags with false @ 'main'. Handle
disable-multiple-abs-defs @ 'mutiple_definition'.
Tom Tromey [Thu, 12 Apr 2018 14:05:16 +0000 (08:05 -0600)]
Conditionally drop the discriminant field in quirk_rust_enum
While debugging the crash that Jan reported, I noticed that in some
situations we could end up with a situation where one branch of a Rust
enum type ended up with a field count of -1.
The fix is simple: only conditionally drop the discriminant field when
rewriting the enum variants.
I couldn't find a way to test this; I only noticed it while debugging
the DWARF reader.
2018-04-17 Tom Tromey <tom@tromey.com>
* dwarf2read.c (quirk_rust_enum): Conditionally drop the
discriminant field.
Andreas Arnez [Tue, 17 Apr 2018 17:31:58 +0000 (19:31 +0200)]
Don't print symbol declaration's line number in rbreak output
This commit:
b744723f57 -- Show line numbers in output for "info var/func/type"
adds the symbol declaration's line number to the output of certain GDB
commands. It also (inadvertently) changes the `rbreak' command's output,
like this:
(gdb) rbreak foo
Breakpoint 1 at 0x40049b: file rbreak.c, line 6.
4: static int foo1(void);
Breakpoint 2 at 0x4004b1: file rbreak.c, line 12.
10: static int foo2(void);
(gdb)
where the function declaration is now prefixed by its source line number,
followed by a colon. But without showing the declaration's file name, the
line number is useless and can possibly cause severe confusion.
No declaration line number was shown before. Instead, the function
declaration started at the first column:
(gdb) rbreak foo
Breakpoint 1 at 0x40049b: file rbreak.c, line 6.
static int foo1(void);
Breakpoint 2 at 0x4004b1: file rbreak.c, line 12.
static int foo2(void);
(gdb)
This old behavior is restored, fixing some FAILs in fullpath-expand.exp,
realname-expand.exp, and pr10179.exp.
In order to distinguish when to print location information, the meaning of
print_symbol_info()'s parameter `last' is changed. Now NULL means to skip
any filename or line number information. Previously NULL meant to always
print the filename.
gdb/ChangeLog:
* symtab.c (print_symbol_info): Skip printing filename and line
number when `last' is NULL.
(symtab_symbol_info): Use empty string instead of NULL for first
invocation of print_symbol_info.
(rbreak_command): Pass NULL to `last' parameter of
print_symbol_info.
Nick Clifton [Tue, 17 Apr 2018 16:47:51 +0000 (17:47 +0100)]
Fix illegal memory accesses trigeered when linking corrupt input files.
PR 23055
* aoutx.h (find_nearest_line): Check that the symbol name exists
and is long enough, before attempting to see if it is for a .o
file.
* hash.c (bfd_hash_hash): Add an assertion that the string is not
NULL.
* linker.c (bfd_link_hash_lookup): Fail if the table or string are
NULL.
(_bfd_generic_link_add_archive_symbols): Fail if an archive entry
has no name.
* simple-object-xcoff.c (simple_object_xcoff_find_sections): Use
ulong_type to avoid warning about 32-bit shift.
2018-01-11 Richard Biener <rguenther@suse.de>
Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
PR lto/81968
* simple-object-common.h (struct simple_object_functions):
Change copy_lto_debug_sections callback signature.
* simple-object-elf.c (SHN_HIRESERVE, SHT_SYMTAB_SHNDX,
SHF_INFO_LINK): Add defines.
(simple_object_elf_copy_lto_debug_sections): Instead of
leaving not to be copied sections empty unnamed SHT_NULL
remove them from the target section headers and adjust section
reference everywhere. Handle SHN_XINDEX in the symbol table
processing properly.
* simple-object.c (handle_lto_debug_sections): Change
interface to return a modified string and handle renaming
of relocation sections.
2018-01-10 Daniel van Gerpen <daniel@vangerpen.de>
* argv.c (expandargv): Correct check for dynamically
allocated argv.