This crash was detected when using GDB with the valgrind gdbserver.
To reproduce:
valgrind sleep 10000
In another window:
gdb
target remote | vgdb
p printf("make sleep print something\n")
=>
terminate called after throwing an instance of 'gdb_exception_error'
Aborted
The problem is that the valgrind gdbserver does not allow to change
registers when the inferior is blocked in a system call.
GDB then raises an exception. The exception causes the destructor
of
typedef std::unique_ptr<infcall_suspend_state, infcall_suspend_state_deleter>
infcall_suspend_state_up;
to be called. This destructor itself tries to restore the value of
the registers, and fails similarly. We must catch the exception in
the destructor to avoid crashing GDB.
If the destructor encounters a problem, no warning is produced if
there is an uncaught exception, as in this case, the user will already
be informed of a problem via this exception.
With this change, no crash anymore, and all the valgrind 3.15 tests
pass succesfully.
gdb/ChangeLog
2019-04-19 Philippe Waroquiers <philippe.waroquiers@skynet.be>
* inferior.h (struct infcall_suspend_state_deleter):
Catch exception in destructor to avoid crash.
event-loop.c does not need to include common/queue.h, so this removes
it.
gdb/ChangeLog
2019-04-19 Tom Tromey <tom@tromey.com>
* event-loop.c: Don't include "common/queue.h".
This changes remote_notif_state::notif_queue to be a std::list and
updates all the uses.
gdb/ChangeLog
2019-04-19 Tom Tromey <tom@tromey.com>
* remote.c (remote_target): Use delete.
* remote-notif.h: Include <list>, not "common/queue.h".
(notif_client_p): Remove typedef.
(remote_notif_state): Add constructor, destructor, initializer.
<notif_queue>: Now a std::list.
(remote_notif_state_xfree): Don't declare.
* remote-notif.c (remote_notif_process, handle_notification)
(remote_notif_state_allocate): Update.
(~remote_notif_state): Rename from remote_notif_state_xfree.
This changes objfile::static_links to be an htab_up, so that ~objfile
no longer has to explicitly destroy it.
Tested by the buildbot.
gdb/ChangeLog
2019-04-19 Tom Tromey <tom@tromey.com>
* symfile.c (reread_symbols): Update.
* objfiles.c (objfile_register_static_link)
(objfile_lookup_static_link): Update
(~objfile) Don't delete static_links.
* objfiles.h (struct objfile) <static_links>: Now an htab_up.
Allow disabling source-highlight dependency autodetection even
it exists in the system. More details on problem of automatic
dependencies:
https://wiki.gentoo.org/wiki/Project:Quality_Assurance/Automagic_dependencies
Noticed by Jeroen Roovers in https://bugs.gentoo.org/680238
* configure.ac: add --enable-source-highlight switch.
* configure: Regenerate.
* top.c (print_gdb_version): plumb --enable-source-highlight
status to "show configuration".
gdb/ChangeLog
2019-04-19 Sergei Trofimovich <siarheit@google.com>
* configure.ac: add --enable-source-highlight switch.
* configure: Regenerate.
* top.c (print_gdb_version): plumb --enable-source-highlight
status to "show configuration".
ada-lang.c is a bit too eager trying to decode unions in the Ada style
-- looking for discriminants and such. This causes crashes when
printing a non-Ada union in Ada mode, something that can easily happen
when printing a value from history or certain registers on AArch64.
This patch fixes the bug by changing ada-lang.c to only apply special
Ada treatment to types coming from an Ada CU. This in turn required a
couple of surprising changes.
First, some of the Ada code was already using HAVE_GNAT_AUX_INFO to
decide whether a type had already been fixed -- such types had
INIT_CPLUS_SPECIFIC called on them. This patch changes these spots to
use the "none" identifier instead.
This then required changing value_rtti_type to avoid changing the
language-specific object attached to an Ada type, which seems like a
good change regardless.
Tested on x86-64 Fedora 29.
gdb/ChangeLog
2019-04-19 Tom Tromey <tromey@adacore.com>
* ada-lang.c (ada_is_variant_part, ada_to_fixed_type_1):
Check ADA_TYPE_P.
(empty_record, ada_template_to_fixed_record_type_1)
(template_to_static_fixed_type)
(to_record_with_fixed_variant_part): Use INIT_NONE_SPECIFIC.
* cp-abi.c (value_rtti_type): Check HAVE_CPLUS_STRUCT.
* gdbtypes.h (INIT_NONE_SPECIFIC, ADA_TYPE_P): New
macros.
gdb/testsuite/ChangeLog
2019-04-19 Tom Tromey <tromey@adacore.com>
* gdb.ada/ptype_union.c: New file.
* gdb.ada/ptype_union.exp: New file.
PR symtab/24423 points out that control characters in a source file
cause a hang in the "list" command, a regression introduced by the
styling changes.
This patch, from the PR, fixes the bug. I've included a minimal
change to the "list" test that exercises this code.
I recall that this bug was discussed on gdb-patches, and I thought
there was a patch there as well, but I was unable to find it.
gdb/ChangeLog
2019-04-19 Ilya Yu. Malakhov <malakhov@mcst.ru>
PR symtab/24423:
* source.c (print_source_lines_base): Advance "iter" when a
control character is seen.
gdb/testsuite/ChangeLog
2019-04-19 Tom Tromey <tromey@adacore.com>
PR symtab/24423:
* gdb.base/list0.h (foo): Add a control-l character.
This crash was detected when using GDB with the valgrind gdbserver.
To reproduce:
valgrind sleep 10000
In another window:
gdb
target remote | vgdb
p printf("make sleep print something\n")
=>
terminate called after throwing an instance of 'gdb_exception_error'
Aborted
The problem is that the valgrind gdbserver does not allow to change
registers when the inferior is blocked in a system call.
GDB then raises an exception. The exception causes the destructor
of
typedef std::unique_ptr<infcall_suspend_state, infcall_suspend_state_deleter>
infcall_suspend_state_up;
to be called. This destructor itself tries to restore the value of
the registers, and fails similarly. We must catch the exception in
the destructor to avoid crashing GDB.
If the destructor encounters a problem, no warning is produced if
there is an uncaught exception, as in this case, the user will already
be informed of a problem via this exception.
With this change, no crash anymore, and all the valgrind 3.15 tests
pass succesfully.
gdb/ChangeLog
2019-04-19 Philippe Waroquiers <philippe.waroquiers@skynet.be>
* inferior.h (struct infcall_suspend_state_deleter):
Catch exception in destructor to avoid crash.
gdb/ChangeLog
2019-04-19 Philippe Waroquiers <philippe.waroquiers@skynet.be>
* cli/cli-cmds.c (_initialize_cli_cmds): Move "shell" "!" alias
close to the add_com "shell".
PR 24464
* config/tc-rx.h (md_relax_frag): Pass the max_iterations variable
to the relaxation function.
* config/tc-rx.c (rx_relax_frag): Add new parameter - the maximum
number of iterations. Make sure that our internal iteration limit
does not exceed this external iteration limit.
This tidies the remaining --no-print-map-discarded fails. h8300-elf
warns on a section without flags, and s12z doesn't support
--gc-sections.
bfd/
* elf32-s12z.c (elf_backend_can_gc_sections): Don't define
ld/
* testsuite/ld-gc/skip-map-discarded.s: Add section attributes.
* testsuite/lib/ld-lib.exp (check_gc_sections_available): Add
s12z to list of targets not supporting --gc-sections.
When running break-probes.exp with native-gdbserver, we run into:
...
FAIL: gdb.base/break-probes.exp: run til our library loads (the program exited)
FAIL: gdb.base/break-probes.exp: call (int) foo(23)
...
due to the fact that we're trying to match:
...
Inferior loaded /data/gdb_versions/devel/build/gdb/testsuite/outputs/gdb.base\
/break-probes/break-probes-solib.so
...
using pattern:
...
Inferior loaded $sysroot$binfile_lib
...
which expands into:
...
Inferior loaded //data/gdb_versions/devel/build/gdb/testsuite/outputs/gdb.base\
/break-probes/break-probes-solib.so
...
Fix by setting sysroot to "" in local-board.exp.
Tested on x86_64-linux with native-gdbserver.
gdb/testsuite/ChangeLog:
2019-04-18 Tom de Vries <tdevries@suse.de>
PR gdb/24433
* boards/local-board.exp: Set sysroot to "".
It seemed to me that process_stratum_target::stratum ought to be
"final".
Tested by rebuilding, let me know what you think.
gdb/ChangeLog
2019-04-18 Tom Tromey <tromey@adacore.com>
* process-stratum-target.h (class process_stratum_target)
<stratum>: Add "final".
When debugging any of the testcases added by this commit, which do a
vfork in a thread with "set follow-fork-mode child" + "set
detach-on-fork on", we run into this assertion:
...
src/gdb/nat/x86-linux-dregs.c:146: internal-error: \
void x86_linux_update_debug_registers(lwp_info*): \
Assertion `lwp_is_stopped (lwp)' failed.
...
The assert is caused by the following: the vfork-child exit or exec
event is handled by handle_vfork_child_exec_or_exit, which calls
target_detach to detach from the vfork parent. During target_detach
we call linux_nat_target::detach, which:
#1 - stops all the threads
#2 - waits for all the threads to be stopped
#3 - detaches all the threads
However, during the second step we run into this code in
stop_wait_callback:
...
/* If this is a vfork parent, bail out, it is not going to report
any SIGSTOP until the vfork is done with. */
if (inf->vfork_child != NULL)
return 0;
...
and we don't wait for the threads to be stopped, which results in this
assert in x86_linux_update_debug_registers triggering during the third
step:
...
gdb_assert (lwp_is_stopped (lwp));
...
The fix is to reset the vfork parent's vfork_child field before
calling target_detach in handle_vfork_child_exec_or_exit. There's
already similar code for the other paths handled by
handle_vfork_child_exec_or_exit, so this commit refactors the code a
bit so that all paths share the same code.
The new tests cover both a vfork child exiting, and a vfork child
execing, since both cases would trigger the assertion.
The new testcases also exercise following the vfork children with "set
detach-on-fork off", since it doesn't seem to be tested anywhere.
Tested on x86_64-linux, using native and native-gdbserver.
gdb/ChangeLog:
2019-04-18 Tom de Vries <tdevries@suse.de>
Pedro Alves <palves@redhat.com>
PR gdb/24454
* infrun.c (handle_vfork_child_exec_or_exit): Reset vfork parent's
vfork_child field before calling target_detach.
gdb/testsuite/ChangeLog:
2019-04-18 Tom de Vries <tdevries@suse.de>
Pedro Alves <palves@redhat.com>
PR gdb/24454
* gdb.threads/vfork-follow-child-exec.c: New file.
* gdb.threads/vfork-follow-child-exec.exp: New file.
* gdb.threads/vfork-follow-child-exit.c: New file.
* gdb.threads/vfork-follow-child-exit.exp: New file.
gas * config/tc-msp430.c (msp430_make_init_symbols): Define
__crt0_run_{preinit,init,fini}_array symbols if
.{preinit,init,fini}_array sections exist.
* testsuite/gas/msp430/fini-array.d: New test.
* testsuite/gas/msp430/init-array.d: New test.
* testsuite/gas/msp430/preinit-array.d: New test.
* testsuite/gas/msp430/fini-array.s: New test source.
* testsuite/gas/msp430/init-array.s: New test source.
* testsuite/gas/msp430/preinit-array.s: New test source.
* testsuite/gas/msp430/msp430.exp: Add new tests to driver.
ld * config/tc-msp430.c (msp430_make_init_symbols): Define __crt0_init_bss
symbol when .lower.bss or .either.bss sections exist.
Define __crt0_movedata when .lower.data or .either.data sections exist.
* testsuite/gas/msp430/either-data-bss-sym.d: New test.
* testsuite/gas/msp430/low-data-bss-sym.d: New test.
* testsuite/gas/msp430/either-data-bss-sym.s: New test source.
* testsuite/gas/msp430/low-data-bss-sym.s: New test source.
* testsuite/gas/msp430/msp430.exp: Run new tests.
Enable large code model when running -mdata-region={upper,either}
tests.
* config/tc-msp430.c (msp430_make_init_symbols): Define __crt0_init_bss
symbol when .lower.bss or .either.bss sections exist.
Define __crt0_movedata when .lower.data or .either.data sections exist.
* testsuite/gas/msp430/either-data-bss-sym.d: New test.
* testsuite/gas/msp430/low-data-bss-sym.d: New test.
* testsuite/gas/msp430/either-data-bss-sym.s: New test source.
* testsuite/gas/msp430/low-data-bss-sym.s: New test source.
* testsuite/gas/msp430/msp430.exp: Run new tests.
Enable large code model when running -mdata-region={upper,either}
tests.
gas * config/tc-msp430.c (options): New OPTION_UNKNOWN_INTR_NOPS,
OPTION_NO_UNKNOWN_INTR_NOPS and do_unknown_interrupt_nops.
(md_parse_option): Handle OPTION_UNKNOWN_INTR_NOPS and
OPTION_NO_UNKNOWN_INTR_NOPS by setting do_unknown_interrupt_nops
accordingly.
(md_show_usage): Likewise.
(md_shortopts): Add "mu" for OPTION_UNKNOWN_INTR_NOPS and
"mU" for OPTION_NO_UNKNOWN_INTR_NOPS.
(md_longopts): Likewise.
(warn_eint_nop): Update comment.
(warn_unsure_interrupt): Don't warn if prev_insn_is_nop or
prev_insn_is_dint or we are assembling for 430 ISA.
(msp430_operands): Only call warn_unsure_interrupt if
do_unknown_interrupt_nops == TRUE.
* testsuite/gas/msp430/nop-unknown-intr.s: New test source file.
* testsuite/gas/msp430/nop-unknown-intr-430.d: New test.
* testsuite/gas/msp430/nop-unknown-intr-430x.d: New test.
* testsuite/gas/msp430/nop-unknown-intr-430x-ignore.d: New test.
* testsuite/gas/msp430/nop-unknown-intr-430.l: Warning output for new
test.
* testsuite/gas/msp430/nop-unknown-intr-430x.l: Likewise.
* testsuite/gas/msp430/msp430.exp: Add new tests to driver.
After commit 35add35 ("gdb: Fix failure in gdb.base/complex-parts.exp
for x86-32"), dwarf2_init_complex_target_type can crash if "tt" is
nullptr. This patch avoids the problem by checking for this case.
No test case because I don't know a good way to write one; it was
found by an internal AdaCore test case that apparently uses a 16 bit
floating point type.
gdb/ChangeLog:
* dwarf2read.c (dwarf2_init_complex_target_type): Check "tt"
against nullptr before use.
gdb/ChangeLog
2019-04-17 Tom Tromey <tromey@adacore.com>
* dwarf2read.c (dwarf2_init_complex_target_type): Check "tt"
against nullptr before use.
All debug output needs to go via debug functions to ensure it writes to the
correct output stream.
gdb/ChangeLog:
* nat/linux-waitpid.c (linux_debug): Call debug_vprintf.
gdb/gdbserver/ChangeLog:
* ax.c (ax_vdebug): Call debug_printf.
* debug.c (debug_write): New function.
* debug.h (debug_write): New declaration.
* linux-low.c (sigchld_handler): Call debug_write.
A comment in debug.h (written in 2014) states: "We declare debug format
variables here, and debug_threads but no other debug content variables
(e.g., not remote_debug) because while this file is not currently used by
IPA it may be some day, and IPA may have its own set of debug content
variables".
This has resulted in remote_debug being declared in many .c/.h files
throughout gdbserver.
It would be much simplier to define it one place. The most logical place to
define it is in debug.h, surrounded by #define guards. If IPA is changed,
then at that point the variable can be moved elsewhere.
gdb/gdbserver/ChangeLog:
* debug.c (remote_debug): Add definition.
* debug.h (remote_debug): Add declaration.
* hostio.c (remote_debug): Remove declaration.
* remote-utils.c (struct ui_file): Likewise.
(remote_debug): Likewise.
* remote-utils.h (remote_debug): Likewise,
* server.c (remote_debug): Remove definition.
Some testsuite cases (gdb.cp/nsalias.exp for example) construct dwarf2
debug info for fake functions to test that this debug info is handled
correctly.
We currently get an error trying to read from an invalid address while
creating breakpoints for these fake functions.
Other targets allow creating breakpoints on invalid addresses, and
only error when GDB actually tries to insert the breakpoints.
In order to make RISC-V behave in the same way as other targets, this
commit makes the failure to read memory during breakpoint creation
non-fatal, we then expect to see a failure when GDB tries to insert
the breakpoint, just like other targets.
Tested with a riscv64-linux native testsuite run.
gdb/ChangeLog:
* riscv-tdep.c (riscv_breakpoint_kind_from_pc): Hanndle case where
code read might fail, assume 4-byte breakpoint in that case.
Some of these tests were excluded for ns32k-netbsd, exclude for all
ns32k instead.
binutils/
* testsuite/binutils-all/copy-2.d: Don't run for ns32k-*-*.
* testsuite/binutils-all/copy-3.d: Likewise.
gas/
* testsuite/gas/all/gas.exp: Remove ns32k xfails.
* testsuite/gas/all/weakref1u.d: Don't run for ns32k-*-*.
ld/
* testsuite/ld-scripts/pr20302.d: Don't run for ns32k-*-*.
* testsuite/ld-scripts/section-match-1.d: Likewise.
* testsuite/ld-undefined/require-defined.exp: Likewise.
These are only used by dlx and ns32k.
* write.h: Don't include bit_fix.h.
(struct fix): Rearrange some fields. Delete fx_im_disp and
fx_bit_fixP. Use bitfields for fx_size and fx_pcrel_adjust.
* write.c (fix_new_internal): Don't init fx_im_disp and fx_bit_fixP.
(fixup_segment): Don't exclude overflow checks on fx_bit_fixP.
(print_fixup): Don't print im_disp.
* config/tc-cris.c (md_apply_fix): Remove tests of fx_bit_fixP
and fx_im_disp.
* config/tc-dlx.c (md_apply_fix): Remove wrong debug code. Set
fx_no_overflow when fx_bit_fixP.
* config/tc-dlx.h: Include bit_fix.h.
(TC_FIX_TYPE, tc_fix_data, TC_INIT_FIX_DATA): Define.
* config/tc-ns32k.c (fix_new_ns32k, fix_new_ns32k_exp): Set
fx_no_overflow when bit_fixP.
* config/tc-ns32k.h (TC_FIX_TYPE): Add fx_bit_fixP and fx_im_disp.
(fix_im_disp, fix_bit_fixP): Adjust to suit.
(TC_INIT_FIX_DATA, TC_FIX_DATA_PRINT): Likewise.
The AMD64 System V ABI specifies that when a function has a return type
classified as MEMORY, the caller provides space for the value and passes
the address to this space as the first argument to the function (before
even the "this" pointer). The classification of MEMORY is applied to
struct that are sufficiently large, or ones with unaligned fields.
The expression evaluator uses call_function_by_hand to call functions,
and the hand-built frame has to push arguments in a way that matches the
ABI of the called function. call_function_by_hand supports ABI-based
struct returns, based on the value of gdbarch_return_value, however on
AMD64 the implementation of the classifier incorrectly assumed that all
non-POD types (implemented as "all types with a base class") should be
classified as MEMORY and use the struct return.
This ABI mismatch resulted in issues when calling a function that returns
a class of size <16 bytes which has a base class, including issues such
as the "this" pointer being incorrect (as it was passed as the second
argument rather than the first).
This is now fixed by checking for field alignment rather than POD-ness,
and a testsuite is added to test expression evaluation for AMD64.
gdb/ChangeLog:
* amd64-tdep.c (amd64_classify_aggregate): Use cp_pass_by_reference
rather than a hand-rolled POD check when checking for forced MEMORY
classification.
gdb/testsuite/ChangeLog:
* gdb.arch/amd64-eval.cc: New file.
* gdb.arch/amd64-eval.exp: New file.
When writing registers to the kernel, check if regcache VG has been changed. If
so then update the thread's vector length, then write back the registers.
When reading registers from the kernel, ensure regcache VG register is updated.
The regcache registers should already be of the correct length.
Remove all the checks that error if the vector length has changed.
gdb/ChangeLog:
* aarch64-linux-nat.c (store_sveregs_to_thread): Set vector length.
* nat/aarch64-sve-linux-ptrace.c (aarch64_sve_set_vq): New function.
(aarch64_sve_regs_copy_to_reg_buf): Remove VG checks.
(aarch64_sve_regs_copy_from_reg_buf): Likewise.
* nat/aarch64-sve-linux-ptrace.h (aarch64_sve_set_vq): New declaration.
Override the thread_architecture method, similar to SPU. If the vector
length has changed, then find the arch using info, making sure the vector
length is passed down to the init routine.
In the init routine, ensure the arch has the correct vector length.
Example output. Program is stopped in thread 2, just before it calls prctl
to change the vector length
(gdb) info threads
Id Target Id Frame
1 Thread 0xffffbf6f4000 (LWP 3188) "sve_change" 0x0000ffffbf6ae130 in pthread_join ()
* 2 Thread 0xffffbf55e200 (LWP 3189) "sve_change" thread1 (arg=0xfeedface) at sve_change_size.c:28
(gdb) print $vg
$1 = 8
(gdb) print $z0.s.u
$2 = {623191333, 623191333, 623191333, 623191333, 0 <repeats 12 times>}
(gdb) n
29 int ret = prctl(PR_SVE_SET_VL, vl/2);
(gdb) n
30 printf ("Changed: ret\n", ret);
(gdb) print $vg
$4 = 4
(gdb) print $z0.s.u
$5 = {623191333, 623191333, 623191333, 623191333, 0, 0, 0, 0}
(gdb) thr 1
[Switching to thread 1 (Thread 0xffffbf6f4000 (LWP 3181))]
(gdb) print $vg
$6 = 8
(gdb) print $z0.s.u
$7 = {623191333, 623191333, 623191333, 623191333, 0 <repeats 12 times>}
gdb/ChangeLog:
* aarch64-linux-nat.c
(aarch64_linux_nat_target::thread_architecture): Add override.
* aarch64-tdep.c (aarch64_gdbarch_init): Ensure different tdesc for
each VQ.
Move the lookup_by_info to the top of the function to avoid unnecessarily
creating a new feature when the gdbarch already exists.
Add some additional cleanups that have no functional effect.
gdb/ChangeLog:
* aarch64-tdep.c (aarch64_gdbarch_init): Move gdbarch lookup.
GNU as' Arm backend assumes each mnemonic has a single entry in the instruction table but VLDR (system register) and VSTR (system register) are different instructions than VLDR and VSTR. It is thus necessary to add some form of demultiplexing in the parser. It starts by creating a new operand type OP_VLDR which indicate that the operand is either the existing OP_RVSD operand or a system register. The function parse_operands () then tries these two cases in order, calling the new parse_sys_vldr_vstr for the second case.
Since the encoding function is specified in the instruction table entry, it also need to have some sort of demultiplexing. This is done in do_vldr_vstr which either calls the existing do_neon_ldr_str () or calls the new do_t_vldr_vstr_sysreg ().
A new internal relocation is needed as well since the offset has a shorter range than in other Thumb coprocessor instructions. Disassembly also requires special care since VSTR (system register) reuse the STC encoding with the coprocessor number 15. Armv8.1-M Mainline ARM manual states that coprocessor 8, 14 and 15 are reserved for floating-point and MVE instructions a feature bit check is added if the coprocessor number is one of this value and we are trying to match a coprocessor instruction (eg. STC) to forbid the match.
ChangeLog entries are as follows:
*** bfd/ChangeLog ***
2019-04-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
* reloc.c (BFD_RELOC_ARM_T32_VLDR_VSTR_OFF_IMM): New internal
relocation.
* bfd-in2.h: Regenerate.
* libbfd.h: Likewise.
*** gas/ChangeLog ***
2019-04-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
* config/tc-arm.c (parse_sys_vldr_vstr): New function.
(OP_VLDR): New enum operand_parse_code enumerator.
(parse_operands): Add logic for OP_VLDR.
(do_t_vldr_vstr_sysreg): New function.
(do_vldr_vstr): Likewise.
(insns): Guard VLDR and VSTR by arm_ext_v4t for Thumb mode.
(md_apply_fix): Add bound check for VLDR and VSTR co-processor offset.
Add masking logic for BFD_RELOC_ARM_T32_VLDR_VSTR_OFF_IMM relocation.
* testsuite/gas/arm/archv8m_1m-cmse-main-bad.s: Add examples of bad
uses of VLDR and VSTR.
* testsuite/gas/arm/archv8m_1m-cmse-main-bad.l: Add error messages for
above bad uses.
* testsuite/gas/arm/archv8m_1m-cmse-main.s: Add examples of VLDR and
VSTR valid uses.
* testsuite/gas/arm/archv8m_1m-cmse-main.d: Add disassembly for the
above examples.
*** opcodes/ChangeLog ***
2019-04-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
* arm-dis.c (coprocessor_opcodes): Document new %J and %K format
specifier. Add entries for VLDR and VSTR of system registers.
(print_insn_coprocessor): Forbid coprocessor numbers 8, 14 and 15 in
coprocessor instructions on Armv8.1-M Mainline targets. Add handling
of %J and %K format specifier.
Like for CLRM, this patch aims to share as much logic with the similar looking VLDM/VSTM. This is achieved by adding 2 new enumerator values in enum reg_list_els for the single-precision and double-precision variants of VSCCLRM and extending parse_vfp_reg_list () to deal with these types.
These behave like the existing REGLIST_VFP_S and REGLIST_VFP_D types with extra logic to expect VPR as the last element in the register list.
The function is algo augmented with a new partial_match parameter to indicate if any register other than VPR had already been parsed in the register list so as to not try parsing the second variant if that's the case and return the right error message.
The rest of the patch is the usual encoding function, new disassembler table entries and format specifier and parsing, encoding and disassembling tests.
It is worth mentioning that the new entry in the disassembler table was added in the coprocessor-related table despite VSCCLRM always being available even in FPU-less configurations. The main reason for this is that VSCCLRM also match VLDMIA entry and must thus be tried first but coprocessor entries are tried before T32 entries. It also makes sense because it is in the same encoding space as coprocessor and VFP instructions and is thus the natural place for someone to look for this instruction.
Note: Both variants of VSCCLRM support D16-D31 registers but Armv8.1-M Mainline overall does not. I have thus decided not to implement support for these registers in order to keep the code simpler. It can always be added later if needed.
ChangeLog entries are as follows:
*** gas/ChangeLog ***
2019-04-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
* config/tc-arm.c (arm_typed_reg_parse): Fix typo in comment.
(enum reg_list_els): New REGLIST_VFP_S_VPR and REGLIST_VFP_D_VPR
enumerators.
(parse_vfp_reg_list): Add new partial_match parameter. Set
*partial_match to TRUE if at least one element in the register list has
matched. Add support for REGLIST_VFP_S_VPR and REGLIST_VFP_D_VPR
register lists which expect VPR as last element in the list.
(s_arm_unwind_save_vfp_armv6): Adapt call to parse_vfp_reg_list to new
prototype.
(s_arm_unwind_save_vfp): Likewise.
(enum operand_parse_code): New OP_VRSDVLST enumerator.
(parse_operands): Adapt call to parse_vfp_reg_list to new prototype.
Handle new OP_VRSDVLST case.
(do_t_vscclrm): New function.
(insns): New entry for VSCCLRM instruction.
* testsuite/gas/arm/archv8m_1m-cmse-main-bad.s: Add invalid VSCCLRM
instructions.
* testsuite/gas/arm/archv8m_1m-cmse-main-bad.l: Add error expectations
for above instructions.
* testsuite/gas/arm/archv8m_1m-cmse-main.s: Add tests for VSCCLRM
instruction.
* testsuite/gas/arm/archv8m_1m-cmse-main.d: Add expected disassembly
for above instructions.
*** opcodes/ChangeLog ***
2019-04-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
* arm-dis.c (coprocessor_opcodes): Document new %C format control code.
Add new entries for VSCCLRM instruction.
(print_insn_coprocessor): Handle new %C format control code.
The coprocessor_opcodes table is used both to share commonalities in coprocessor-related instructions and to avoid duplication between Arm and Thumb mode. However, some instructions do have differences between Arm and Thumb. For instance, vldmia allows PC as base register in Arm mode but not in Thumb mode.
In that very case the distinction becomes necessary because the encoding with PC as base register is used in Thumb mode to denote a VSCCLRM. It is thus necessary to distinguish what is Arm or Thumb only from what is shared. This patch adds an extra field to the coprocessor_opcodes table entries to indicate what mode is a given instruction available in. The print_insn_coprocessor then uses that field to determine if an entry that matched the mark and value checked should be allowed to match or not given the current mode.
ChangeLog entry is as follows:
*** opcodes/ChangeLog ***
2019-04-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
* arm-dis.c (enum isa): New enum.
(struct sopcode32): New structure.
(coprocessor_opcodes): change type of entries to struct sopcode32 and
set isa field of all current entries to ANY.
(print_insn_coprocessor): Change type of insn to struct sopcode32.
Only match an entry if its isa field allows the current mode.
Given the similarity between LDM/STM and CLRM register lists, most of the changes in this patch aim at sharing code between those two sets of instruction. Sharing is achieved both in parsing and encoding of those instructions.
In terms of parsing, parse_reg_list () is extended to take a type that describe what type of instruction is being parsed. The reg_list_els used for parse_vfp_reg_list () is reused for the type and that function is added an assert for the new REGLIST_CLRM and REGLIST_RN enumerators.
parse_reg_list () is then taught to accept APSR and reject SP and PC when parsing for a CLRM instruction. At last, caller of parse_reg_list () is updated accordingly and logic is added for the new OP_CLRMLST operand.
Encoding-wise, encode_thumb2_ldmstm () is reused to encode the variable bits of CLRM and is thus renamed encode_thumb2_multi (). A new do_io parameter is added to distinguish between LDM/STM and CLRM which guard all the LDM/STM specific code of the function.
Finally objdump is told how to disassemble CLRM, again reusing the logic to print the LDM/STM register list (format specifier 'm'). Tests are also added in the form of negative tests to check parsing and encoding/disassembling tests.
ChangeLog entries are as follows:
*** gas/ChangeLog ***
2019-04-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
* config/tc-arm.c (enum reg_list_els): Define earlier and add
REGLIST_RN and REGLIST_CLRM enumerators.
(parse_reg_list): Add etype parameter to distinguish between regular
core register list and CLRM register list. Add logic to
recognize CLRM register list.
(parse_vfp_reg_list): Assert type is not for core register list.
(s_arm_unwind_save_core): Update call to parse_reg_list to new
prototype.
(enum operand_parse_code): Declare OP_CLRMLST enumerator.
(parse_operands): Update call to parse_reg_list to new prototype. Add
logic for OP_CLRMLST.
(encode_thumb2_ldmstm): Rename into ...
(encode_thumb2_multi): This. Add do_io parameter. Add logic to
encode CLRM and guard LDM/STM only code by do_io.
(do_t_ldmstm): Adapt to use encode_thumb2_multi.
(do_t_push_pop): Likewise.
(do_t_clrm): New function.
(insns): Define CLRM.
* testsuite/gas/arm/archv8m_1m-cmse-main-bad.d: New file.
* testsuite/gas/arm/archv8m_1m-cmse-main-bad.l: Likewise.
* testsuite/gas/arm/archv8m_1m-cmse-main-bad.s: Likewise.
* testsuite/gas/arm/archv8m_1m-cmse-main.d: Likewise.
* testsuite/gas/arm/archv8m_1m-cmse-main.s: Likewise.
*** opcodes/ChangeLog ***
2019-04-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
* arm-dis.c (thumb_opcodes): Document %n control code. Add entry for
CLRM.
(print_insn_thumb32): Add logic to print %n CLRM register list.
s patch is part of a series of patches to add support for Armv8.1-M Mainline instructions to binutils.
This patch adds support to the Scalar low overhead loop instructions:
LE
WLS
DLS
We also add a new assembler resolvable relocation bfd_reloc_code_real enum for the 12-bit branch offset used in these instructions.
ChangeLog entries are as follows:
*** bfd/ChnageLog ***
2019-04-12 Sudakshina Das <sudi.das@arm.com>
* reloc.c (BFD_RELOC_ARM_THUMB_LOOP12): New.
* bfd-in2.h: Regenerated.
* libbfd.h: Regenerated.
*** gas/ChangeLog ***
2019-04-12 Sudakshina Das <sudi.das@arm.com>
Andre Vieira <andre.simoesdiasvieira@arm.com>
* config/tc-arm.c (operand_parse_code): Add OP_LR and OP_oLR
for the LR operand and optional LR operand.
(parse_operands): Add switch cases for OP_LR and OP_oLR for
both type checking and value checking.
(encode_thumb32_addr_mode): New entries for DLS, WLS and LE.
(v8_1_loop_reloc): New helper function for handling labels
for the low overhead loop instructions.
(do_t_loloop): New function to encode DLS, WLS and LE.
(insns): New entries for WLS, DLS and LE.
(md_pcrel_from_section): New switch case
for BFD_RELOC_ARM_THUMB_LOOP12.
(md_appdy_fix): Likewise.
(tc_gen_reloc): Likewise.
* testsuite/gas/arm/armv8_1-m-tloop.s: New.
* testsuite/gas/arm/armv8_1-m-tloop.d: New.
* testsuite/gas/arm/armv8_1-m-tloop-bad.s: New.
* testsuite/gas/arm/armv8_1-m-tloop-bad.d: New.
* testsuite/gas/arm/armv8_1-m-tloop-bad.l: New.
*** opcodes/ChangeLog ***
2019-04-12 Sudakshina Das <sudi.das@arm.com>
* arm-dis.c (print_insn_thumb32): Updated to accept new %P
and %Q patterns.
