Currently, the breakpoint documentation refers to some commands taking breakpoint
"ranges" as arguments. We discussed this with Pedro and concluded that it would
be more accurate to speak in terms of breakpoint "lists", whose elements can optionally
be ranges. I also fixed a couple of minor mistakes in the docs.
gdb/ChangeLog:
* breakpoint.c (_initialize_breakpoint): Update the help description
of the 'commands' command to indicate that it takes a list argument.
gdb/doc/ChangeLog:
* gdb.texinfo (Breakpoints): Reword documentation to speak in terms of
space-separated breakpoint lists. Also add a missing @table command
and @cindex for breakpoint lists.
gdb/testsuite/ChangeLog:
* gdb.base/help.exp: Update match pattern for testing 'help commands'.
When defining a new macro, "command" is not recognized as an alias for
"commands":
(gdb) define breakmain
Type commands for definition of "breakmain".
End with a line saying just "end".
>break main
>command
>echo "IN MAIN\n"
>end
(gdb)
There is a special case for while-stepping, where 'ws' and 'stepping' are
recognized explicitely. Instead of adding more special cases, this change
uses cli-decode.
gdb/ChangeLog:
* cli/cli-decode.c (find_command_name_length): Make it extern.
* cli/cli-decode.h (find_command_name_length): Declare.
* cli/cli-script.c (command_name_equals, line_first_arg):
New functions.
(process_next_line): Use cli-decode to parse command names.
(build_command_line): Make args a constant pointer.
gdb/testsuite/ChangeLog:
* gdb.base/define.exp: Add test for command abbreviations
in define.
This patch addresses BZ 21005, which is gdb failing to recognize an rdrand
instruction.
It enables support for both rdrand and rdseed and handles extended register
addressing (R8~R15) for 16-bit, 32-bit and 64-bit.
gdb/ChangeLog
2017-02-06 Luis Machado <lgustavo@codesourcery.com>
* NEWS: Mention support for record/replay of Intel 64 rdrand and
rdseed instructions.
i386-tdep.c (i386_process_record): Handle Intel 64 rdrand and rseed.
gdb/testsuite/ChangeLog:
2017-02-06 Luis Machado <lgustavo@codesourcery.com>
* gdb.reverse/insn-reverse.c: Include insn-reverse-x86.c.
* gdb.reverse/insn-reverse-x86.c: New file.
gdb/ChangeLog:
2017-02-06 Ivo Raisr <ivo.raisr@oracle.com>
PR tdep/20936
Provide and use sparc32 and sparc64 target description XML files.
* features/sparc/sparc32-cp0.xml, features/sparc/sparc32-cpu.xml,
features/sparc/sparc32-fpu.xml: New files for sparc 32-bit.
* features/sparc/sparc64-cp0.xml, features/sparc/sparc64-cpu.xml,
features/sparc/sparc64-fpu.xml: New files for sparc 64-bit.
* features/sparc/sparc32-solaris.xml: New file.
* features/sparc/sparc64-solaris.xml: New file.
* features/sparc/sparc32-solaris.c: Generated.
* features/sparc/sparc64-solaris.c: Generated.
* sparc-tdep.h: Account for differences in target descriptions.
* sparc-tdep.c (sparc32_register_name): Use target provided registers.
(sparc32_register_type): Use target provided registers.
(validate_tdesc_registers): New function.
(sparc32_gdbarch_init): Use tdesc_has_registers.
Set pseudoregister functions.
* sparc64-tdep.c (sparc64_register_name): Use target provided registers.
(sparc64_register_type): Use target provided registers.
(sparc64_init_abi): Set pseudoregister functions.
gdb/doc/ChangeLog:
2017-02-06 Ivo Raisr <ivo.raisr@oracle.com>
PR tdep/20936
* gdb.texinfo: (Standard Target Features): Document SPARC features.
(Sparc Features): New node.
gdb/testsuite/ChangeLog:
2017-02-06 Ivo Raisr <ivo.raisr@oracle.com>
PR tdep/20936
* gdb.xml/tdesc-regs.exp: Provide sparc core registers for the tests.
While looking into PR rust/21097, I found that ptype of a
single-element enum in Rust did not always format the result properly.
In particular, it would leave out the members of a tuple struct.
Further testing showed that it also did the wrong thing for ordinary
struct members as well.
This patch fixes these problems. I'm marking it as being associated
with the PR, since that is where the discovery was made; but this
doesn't actually fix that PR (which I think ultimately is due to a
Rust compiler bug).
Built and regtested on x86-64 Fedora 25, using the system Rust
compiler. I'm checking this in.
2017-02-03 Tom Tromey <tom@tromey.com>
PR rust/21097:
* rust-lang.c (rust_print_type) <TYPE_CODE_UNION>: Handle enums
with a single member.
2017-02-03 Tom Tromey <tom@tromey.com>
PR rust/21097:
* gdb.rust/simple.exp: Add new tests.
This commit fixes a "-gdb-set logging redirect on" crash by not
handling "logging redirect on" on the fly.
Previous discussion here:
https://sourceware.org/ml/gdb-patches/2017-01/msg00467.html
Code for handling "logging redirect on" on the fly was added here:
https://sourceware.org/ml/gdb-patches/2010-08/msg00202.html
Meanwhile, MI gained support for logging, but flipping redirect "on"
on the fly was not considered. The result is that this sequence of
commands crashes GDB:
-gdb-set logging on
-gdb-set logging redirect on
Program received signal SIGSEGV, Segmentation fault.
0x00000000008dd7bc in gdb_flush (file=0x2a097f0) at /home/pedro/gdb/mygit/cxx-convertion/src/gdb/ui-file.c:95
194 file->to_flush (file);
(top-gdb) bt
#0 0x00000000008dd7bc in gdb_flush(ui_file*) (file=0x2a097f0) at /home/pedro/gdb/mygit/cxx-convertion/src/gdb/ui-file.c:95
#1 0x00000000007b5f34 in gdb_wait_for_event(int) (block=0) at /home/pedro/gdb/mygit/cxx-convertion/src/gdb/event-loop.c:752
#2 0x00000000007b52b6 in gdb_do_one_event() () at /home/pedro/gdb/mygit/cxx-convertion/src/gdb/event-loop.c:322
#3 0x00000000007b5362 in start_event_loop() () at /home/pedro/gdb/mygit/cxx-convertion/src/gdb/event-loop.c:371
#4 0x000000000082704a in captured_command_loop(void*) (data=0x0) at /home/pedro/gdb/mygit/cxx-convertion/src/gdb/main.c:325
#5 0x00000000007b8d7c in catch_errors(int (*)(void*), void*, char*, return_mask) (func=0x827008 <captured_command_loop(void*)>, func_args=0x0, errstring=0x11dee51 "", mask=RETURN_MASK_ALL) at /home/pedro/gdb/mygit/cxx-convertion/src/gdb/exceptions.c:236
#6 0x000000000082839b in captured_main(void*) (data=0x7fffffffd820) at /home/pedro/gdb/mygit/cxx-convertion/src/gdb/main.c:1148
During symbol reading, cannot get low and high bounds for subprogram DIE at 24065.
#7 0x00000000008283c4 in gdb_main(captured_main_args*) (args=0x7fffffffd820) at /home/pedro/gdb/mygit/cxx-convertion/src/gdb/main.c:1158
#8 0x0000000000412d4d in main(int, char**) (argc=4, argv=0x7fffffffd928) at /home/pedro/gdb/mygit/cxx-convertion/src/gdb/gdb.c:32
The handling of redirect on the fly is not really a use case we need
to handle, IMO. Its inconsistent (other "set logging foo" commands
aren't handled on the fly), and complicates the code significantly.
Instead of complicating it further for MI, go back to the original
idea of warning, only:
https://sourceware.org/ml/gdb-patches/2010-08/msg00083.html
New test included.
gdb/ChangeLog:
2017-02-02 Pedro Alves <palves@redhat.com>
* cli/cli-logging.c (maybe_warn_already_logging): New factored out
from ...
(set_logging_overwrite): ... here.
(logging_no_redirect_file): Delete.
(set_logging_redirect): Don't handle redirection on the fly.
Instead warn that "logging off" / "logging on" is necessary.
(pop_output_files): Delete references to logging_no_redirect_file.
(show_logging_command): Always speak in terms of what will happen
once logging is reenabled.
gdb/testsuite/ChangeLog:
2017-02-02 Pedro Alves <palves@redhat.com>
* gdb.mi/mi-logging.exp: Add "redirect while already logging"
tests.
When a variable's location is expressed as DW_OP_implicit_value, but the
given value is longer than needed, which bytes should be used? GDB's
current logic was introduced with a patch from 2011 and uses the "least
significant" bytes:
https://sourceware.org/ml/gdb-patches/2011-08/msg00123.html
Now consider a sub-value from such a location at a given offset, accessed
through DW_OP_implicit_pointer. Which bytes should be used for that? The
patch above *always* uses the last bytes on big-endian targets, ignoring
the offset.
E.g., given the code snippet
const char foo[] = "Hello, world!";
const char *a = &foo[0];
const char *b = &foo[7];
assume that `foo' is described as DW_OP_implicit_value and `a' and `b'
each as DW_OP_implicit_pointer into that value. Then with current GDB
`*a' and `*b' yield the same result -- the string's zero terminator.
This patch basically reverts the portion of the patch above that deals
with DW_OP_implicit_value. This fixes the offset handling and also goes
back to dropping the last instead of the first bytes on big-endian targets
if the implicit value is longer than needed. The latter aspect of the
change probably doesn't matter for actual programs, but simplifies the
logic.
The patch also cleans up the original code a bit and adds appropriate test
cases.
gdb/testsuite/ChangeLog:
* gdb.dwarf2/dw2-op-stack-value.exp: Adjust expected result of
taking a 2-byte value out of a 4-byte DWARF implicit value on
big-endian targets.
* gdb.dwarf2/nonvar-access.exp: Add more comments to existing
logic. Add test cases for DW_OP_implicit.
gdb/ChangeLog:
* dwarf2loc.c (dwarf2_evaluate_loc_desc_full): For
DWARF_VALUE_LITERAL, no longer ignore the offset on big-endian
targets. And if the implicit value is longer than needed, extract
the first bytes instead of the "least significant" ones.
If GDB is running when gdb_skip_xml_tests is called with
--target_board=native-extended-gdbserer.exp, it fails with:
(gdb) FAIL: ....exp: set tdesc filename .../trivial.xml (got interactive prompt)
monitor exit
Diagnose this in gdb_skip_xml_tests to generate a more meaningful error message:
ERROR: tcl error sourcing ....exp.
ERROR: GDB must not be running in gdb_skip_xml_tests.
while executing
[...]
testsuite/
* lib/gdb.exp (gdb_skip_xml_tests): Error if GDB is running.
Parts of gdb.btrace/enable.exp are only valid for native debug. The check for
skip_gdbserver_tests is done while GDB is running, though, which causes it to
fail with --target_board=native-extended-gdbserver. Exit GDB before that check.
testsuite/
* gdb.btrace/enable.exp: Call gdb_exit before skip_gdbserver_tests.
With --target_board=native-extended-gdbserver non-stop tests are failing with
UNTESTED: gdb.btrace/non-stop.exp: failed to run to main
Fix that by adding '-ex "set non-stop on"' to GDBFLAGS before restarting.
testsuite/
* gdb.btrace/non-stop.exp: Add '-ex "set non-stop on"' to GDBFLAGS.
We may silently skip gdb.btrace tests if
- the target does not support record-btrace
- the target does not support TSX
- the target does not support gdbserver
- we fail to compile the test
- we fail to run to main
Add unsupported/untested messages for each of those.
testsuite/
* gdb.btrace/buffer-size.exp: Add unsupported/untested message if
the test is skipped.
* gdb.btrace/data.exp: Likewise.
* gdb.btrace/delta.exp: Likewise.
* gdb.btrace/dlopen.exp: Likewise.
* gdb.btrace/enable-running.exp: Likewise.
* gdb.btrace/enable.exp: Likewise.
* gdb.btrace/exception.exp: Likewise.
* gdb.btrace/function_call_history.exp: Likewise.
* gdb.btrace/gcore.exp: Likewise.
* gdb.btrace/instruction_history.exp: Likewise.
* gdb.btrace/multi-thread-step.exp: Likewise.
* gdb.btrace/nohist.exp: Likewise.
* gdb.btrace/non-stop.exp: Likewise.
* gdb.btrace/reconnect.exp: Likewise.
* gdb.btrace/record_goto-step.exp: Likewise.
* gdb.btrace/record_goto.exp: Likewise.
* gdb.btrace/rn-dl-bind.exp: Likewise.
* gdb.btrace/segv.exp: Likewise.
* gdb.btrace/step.exp: Likewise.
* gdb.btrace/stepi.exp: Likewise.
* gdb.btrace/tailcall-only.exp: Likewise.
* gdb.btrace/tailcall.exp: Likewise.
* gdb.btrace/tsx.exp: Likewise.
* gdb.btrace/unknown_functions.exp: Likewise.
* gdb.btrace/vdso.exp: Likewise.
When recording is started for a running thread, GDB was able to start tracing
but then failed to read registers to insert the initial entry for the current
PC. We don't really need that initial entry if we don't know where exactly we
started recording. Skip that step to allow recording to be started while
threads are running.
If we do run into errors, we need to undo the tracing enable to not leak this
thread. The operation did not complete so our caller won't clean up this
thread.
For the BTRACE_FORMAT_PT btrace format, we don't need that initial entry since
it will be recorded in the trace. We can omit the call to btrace_add_pc.
gdb/
* btrace.c (btrace_enable): Do not call btrace_add_pc for
BTRACE_FORMAT_PT or if can_access_registers_ptid returns false.
(btrace_fetch): Assert can_access_registers_ptid.
* record-btrace.c (require_btrace_thread, record_btrace_info): Call
validate_registers_access.
testsuite/
* gdb.btrace/enable-running.c: New.
* gdb.btrace/enable-running.exp: New.
This patch allows examination of the registers FS_BASE and GS_BASE
for Linux Systems running on 64bit. Tests for simple read and write
of the new registers is also added with this patch.
2017-01-27 Walfred Tedeschi <walfred.tedeschi@intel.com>
Richard Henderson <rth@redhat.com>
gdb/ChangeLog:
* amd64-linux-nat.c (PTRACE_ARCH_PRCTL): New define.
(amd64_linux_fetch_inferior_registers): Add case to fetch FS_BASE
GS_BASE for older kernels.
(amd64_linux_store_inferior_registers): Add case to store FS_BASE
GS_BASE for older kernels.
* amd64-linux-tdep.c (amd64_linux_gregset_reg_offset): Add FS_BASE
and GS_BASE to the offset table.
(amd64_linux_register_reggroup_p): Add FS_BASE and GS_BASE to the
system register group.
* amd64-nat.c (amd64_native_gregset_reg_offset): Implements case
for older kernels.
* amd64-tdep.c (amd64_init_abi): Add segment registers for the
amd64 ABI.
* amd64-tdep.h (amd64_regnum): Add AMD64_FSBASE_REGNUM and
AMD64_GSBASE_REGNUM.
(AMD64_NUM_REGS): Set to AMD64_GSBASE_REGNUM + 1.
* features/Makefile (amd64-linux.dat, amd64-avx-linux.dat)
(amd64-mpx-linux.dat, amd64-avx512-linux.dat, x32-linux.dat)
(x32-avx-linux.dat, x32-avx512-linux.dat): Add
i386/64bit-segments.xml in those rules.
* features/i386/64bit-segments.xml: New file.
* features/i386/amd64-avx-mpx-linux.xml: Add 64bit-segments.xml.
* features/i386/amd64-avx-linux.xml: Add 64bit-segments.xml.
* features/i386/amd64-avx512-linux.xml: Add 64bit-segments.xml.
* features/i386/amd64-mpx-linux.xml: Add 64bit-segments.xml.
* features/i386/x32-avx512-linux.xml: Add 64bit-segments.xml.
* features/i386/x32-avx-linux.xml: Add 64bit-segments.xml.
* features/i386/amd64-linux.xml: Add 64bit-segments.xml.
* features/i386/amd64-avx-linux.c: Regenerated.
* features/i386/amd64-avx-mpx-linux.c: Regenerated.
* features/i386/amd64-avx-mpx.c: Regenerated.
* features/i386/amd64-avx512-linux.c: Regenerated.
* features/i386/amd64-linux.c: Regenerated.
* features/i386/amd64-mpx-linux.c: Regenerated.
* features/i386/i386-avx-mpx-linux.c: Regenerated.
* features/i386/i386-avx-mpx.c: Regenerated.
* features/i386/x32-avx-linux.c: Regenerated.
* features/i386/x32-avx512-linux.c: Regenerated.
* regformats/i386/amd64-avx-linux.dat: Regenerated.
* regformats/i386/amd64-avx-mpx-linux.dat: Regenerated.
* regformats/i386/amd64-avx512-linux.dat: Regenerated.
* regformats/i386/amd64-linux.dat: Regenerated.
* regformats/i386/amd64-mpx-linux.dat: Regenerated.
* regformats/i386/x32-avx-linux.dat: Regenerated.
* regformats/i386/x32-avx512-linux.dat: Regenerated.
* regformats/i386/x32-linux.dat: Regenerated.
gdb/doc/ChangeLog:
* gdb.texinfo (i386 Features): Add system segment registers
as feature.
gdb/gdbserver/ChangeLog:
* linux-x86-low.c (x86_64_regmap): Add fs_base and gs_base
to the register table.
(x86_fill_gregset): Add support for old kernels for the
fs_base and gs_base system registers.
(x86_store_gregset): Likewise.
* configure.srv (srv_i386_64bit_xmlfiles): Add 64bit-segments.xml.
gdb/testsuite/ChangeLog:
* gdb.arch/amd64-gs_base.c: New file.
* gdb.arch/amd64-gs_base.exp: New file.
Change-Id: I2e0eeb93058a2320d4d3b045082643cfe4aff963
Signed-off-by: Walfred Tedeschi <walfred.tedeschi@intel.com>
With my debug build of Python (--with-pydebug), many tests fails because
of the same issue. Python scripts are loaded by the tests using this
pattern:
(gdb) python exec (open ('file.py').read ())
This causes Python to output this warning:
__main__:1: ResourceWarning: unclosed file <_io.TextIOWrapper name='file.py' mode='r' encoding='ANSI_X3.4-1968'>
and the test to fail because of that extra output. Instead of using the
open + read + exec trick which leaks the file and causes the warning,
why not just source the files?
(gdb) source file.py
This patch changes this, and standardizes the test names of the tests I
touched to "load python file" (some of them were empty, others were
overly complicated).
gdb/testsuite/ChangeLog:
* gdb.python/py-bad-printers.exp: Load python file using "source".
* gdb.python/py-events.exp: Likewise.
* gdb.python/py-evsignal.exp: Likewise.
* gdb.python/py-evthreads.exp: Likewise.
* gdb.python/py-frame-args.exp: Likewise.
* gdb.python/py-framefilter-invalidarg.exp: Likewise.
* gdb.python/py-framefilter-mi.exp: Likewise.
* gdb.python/py-framefilter.exp: Likewise.
* gdb.python/py-mi.exp: Likewise.
* gdb.python/py-pp-maint.exp: Likewise.
* gdb.python/py-pp-registration.exp: Likewise.
* gdb.python/py-prettyprint.exp: Likewise.
(run_lang_tests): Likewise.
* gdb.python/py-typeprint.exp: Likewise.
Exercising aarch64-elf with a custom debug stub i noticed a few failures in
both gdb.base/breakpoint-in-ro-region.exp and gdb.base/memattr.exp:
FAIL: gdb.base/breakpoint-in-ro-region.exp: create read-only mem region covering main
FAIL: gdb.base/breakpoint-in-ro-region.exp: writing to read-only memory fails
FAIL: gdb.base/breakpoint-in-ro-region.exp: inserting software breakpoint in read-only memory fails
FAIL: gdb.base/memattr.exp: create mem region 1
FAIL: gdb.base/memattr.exp: create mem region 2
FAIL: gdb.base/memattr.exp: create mem region 3
FAIL: gdb.base/memattr.exp: create mem region 4
FAIL: gdb.base/memattr.exp: create mem region 5
FAIL: gdb.base/memattr.exp: info mem (1)
FAIL: gdb.base/memattr.exp: mem1 cannot be read
FAIL: gdb.base/memattr.exp: mem2 cannot be written
FAIL: gdb.base/memattr.exp: mem2 can be read
FAIL: gdb.base/memattr.exp: disable mem 1
FAIL: gdb.base/memattr.exp: mem 1 was disabled
FAIL: gdb.base/memattr.exp: enable mem 1
FAIL: gdb.base/memattr.exp: mem 1 was enabled
FAIL: gdb.base/memattr.exp: disable mem 2 4
FAIL: gdb.base/memattr.exp: mem 2 and 4 were disabled
FAIL: gdb.base/memattr.exp: enable mem 2-4
FAIL: gdb.base/memattr.exp: mem 2-4 were enabled
FAIL: gdb.base/memattr.exp: mem 1 to 5 were disabled
FAIL: gdb.base/memattr.exp: mem 1 to 5 were enabled
FAIL: gdb.base/memattr.exp: delete mem 1
FAIL: gdb.base/memattr.exp: mem 1 was deleted
FAIL: gdb.base/memattr.exp: delete mem 2 4
FAIL: gdb.base/memattr.exp: mem 2 and 4 were deleted
FAIL: gdb.base/memattr.exp: mem 2-4 were deleted
These failures don't show up with gdbserver or native gdb on Linux because
they don't export any memory maps, therefore the vector of memory regions is
empty.
Outside of that scenario, we can't guarantee the absence of memory regions
reported by the target upon a connection. In our particular target, we
provide a memory map and the memory regions vector ceases to be empty.
With a non-empty memory regions vector, manipulating memory regions will cause
gdb to be more verbose and output text. For example:
memattr.c:require_user_regions
/* Otherwise, let the user know how to get back. */
if (from_tty)
warning (_("Switching to manual control of memory regions; use "
"\"mem auto\" to fetch regions from the target again."));
memattr.c:create_mem_region
if ((lo >= n->lo && (lo < n->hi || n->hi == 0))
|| (hi > n->lo && (hi <= n->hi || n->hi == 0))
|| (lo <= n->lo && ((hi >= n->hi && n->hi != 0) || hi == 0)))
{
printf_unfiltered (_("overlapping memory region\n"));
return;
}
In my particular case i got both of the above messages.
In order to fix this, i've moved the delete_memory proc from
gdb.base/memattr.exp to a new file lib/memory.exp and made lib/gdb.exp
load that file.
For both gdb.base/breakpoint-in-ro-region.exp and gdb.base/memattr.exp the
patch clears all existing memory regions after running to main. That way we
are guaranteed to have a clean state for memory regions so the tests can
exercise whatever they want and have an expected output pattern.
Regression checked on x86-64/Ubuntu 16.04.
gdb/testsuite/ChangeLog:
2017-01-26 Luis Machado <lgustavo@codesourcery.com>
* lib/memory.exp: New file.
* lib/gdb.exp: Load memory.exp.
* gdb.base/memattr.exp (delete_memory): Move proc to
lib/memory.exp and rename to delete_memory_regions.
Replace delete_memory with delete_memory_regions.
Cleanup memory regions before tests.
* gdb.base/breakpoint-in-ro-region.exp: Cleanup memory regions
before tests.
Changes in v2:
- Renamed arch-specific files to insn-reverse-<arch>.c.
- Adjusted according to reviews.
This patch prepares things for an upcoming testcase for record/replay support
on x86. As is, gdb.reverse/insn-reverse.c is divided into sections guarded by
a few #if blocks, and right now it only handles arm/aarch64.
If we move forward with requiring more tests for record/replay on different
architectures, i think this has the potential to become cluttered with a lot
of differing arch-specific code in the same file.
I've broken up the main file into other files with arch-specific bits
(insn-reverse-<arch>.c). The main file will hold the generic pieces that will
take care of calling the tests.
The arch-specific c files are then included at the top of the generic c file.
I've also added a generic initialize function since we need to run pre-test
checks on x86 to make sure the rdrand/rdseed instructions are supported,
otherwise we will run into a SIGILL.
The arch-specific files will implement their own initialize function with
whatever makes sense. Right now the aarch64 and arm files have an empty
initialization function.
Does this look reasonable?
gdb/testsuite/ChangeLog:
2017-01-26 Luis Machado <lgustavo@codesourcery.com>
* gdb.reverse/insn-reverse.c: Move arm and aarch64 code to their own
files.
(initialize): New function conditionally defined.
(testcases): Move within conditional block.
(main): Call initialize.
* gdb.reverse/insn-reverse-aarch64.c: New file, based on aarch64 bits
of gdb.reverse/insn-reverse.c.
* gdb.reverse/insn-reverse-arm.c: New file, based on arm bits of
gdb.reverse/insn-reverse.c.
Hi,
GDB calls some APIs from opcodes to do disassembly and provide some
call backs. This model makes troubles on C++ exception unwinding,
because GDB is a C++ program, and opcodes is still compiled as C.
As we can see, frame #10 and #12 are C++, while #frame 11 is C,
#10 0x0000000000544228 in memory_error (err=TARGET_XFER_E_IO, memaddr=<optimized out>) at ../../binutils-gdb/gdb/corefile.c:237
#11 0x00000000006b0a54 in print_insn_aarch64 (pc=0, info=0xffffffffeeb0) at ../../binutils-gdb/opcodes/aarch64-dis.c:3185
#12 0x0000000000553590 in gdb_pretty_print_insn (gdbarch=gdbarch@entry=0xbbceb0, uiout=uiout@entry=0xbc73d0, di=di@entry=0xffffffffeeb0,
insn=0xffffffffed40, insn@entry=0xffffffffed90, flags=flags@entry=0,
C++ exception unwinder can't go across frame #11 unless it has
unwind table. However, C program on many architectures doesn't
have it in default. As a result, GDB aborts, which is described
in PR 20939.
This is not the first time we see this kind of problem. We've
had a commit 89525768cd
"Propagate GDB/C++ exceptions across readline using sj/lj-based TRY/CATCH".
We can fix the disassembly bug in a similar way, this is the option one.
Since opcodes is built with gdb, we fix this problem in a different
way as we did for the same issue with readline. Instead of throwing
exception in dis_asm_memory_error, we record the failed memory
address, and throw exception when GDB returns from opcodes disassemblers.
gdb:
2017-01-26 Yao Qi <yao.qi@linaro.org>
Pedro Alves <palves@redhat.com>
PR gdb/20939
* disasm.c (gdb_disassembler::dis_asm_memory_error): Don't
call memory_error, save memaddr instead.
(gdb_disassembler::print_insn): If gdbarch_print_insn returns
negative, cal memory_error.
* disasm.h (gdb_disassembler) <m_err_memaddr>: New field.
gdb/testsuite:
2017-01-26 Yao Qi <yao.qi@linaro.org>
* gdb.base/all-architectures.exp.in (do_arch_tests): Test
disassemble on address 0.
This patch adds a DW_OP_implicit_value in dwarf assembler, and uses
dwarf assembler in implptr-64bit.exp. Using dwarf assembler in
implptr-64bit.exp exposes some limitations in dwarf assembler,
- some variables are not evaluated in the caller's context, so we
can not pass variable to assembler, like this
Dwarf::assemble $asm_file {
cu {
version $dwarf_version
addr_size $addr_size
is_64 $is_64
} {
}
and
{DW_AT_type :$struct_label "DW_FORM_ref$ref_addr_size"}
this limitation is fixed by adding "uplevel" and "subst".
- dwarf assembler doesn't emit DW_FORM_ref_addr for label referencing.
this limitation is fixed by adding a new character "%",
{ type %$int_label }
this means we want to emit DW_FORM_ref_addr for label referencing.
- we can't set the form of label referencing offset in dwarf assembler.
Nowadays, dwarf assembler guesses the form of labels, which is
DW_FORM_ref4. However, in implptr-64bit.exp, both DW_FORM_ref4
and DW_FORM_ref8 is used (see REF_ADDR in implptr-64bit.S). This
patch adds the flexibility of setting the form of label reference.
Both of them below are valid,
{DW_AT_type :$struct_label}
{DW_AT_type :$struct_label DW_FORM_ref8}
the former form is the default DW_FORM_ref4.
I compared the .debug_info of objects without and with this patch
applied. There is no changes except abbrev numbers.
gdb/testsuite:
2017-01-25 Andreas Arnez <arnez@linux.vnet.ibm.com>
Yao Qi <yao.qi@linaro.org>
* gdb.dwarf2/implptr-64bit.exp: Use dwarf assembler.
* gdb.dwarf2/implptr-64bit.S: Remove.
* lib/dwarf.exp (Dwarf): Handle character "%". Evaluate some
variables in caller's context. Add DW_OP_implicit_value.
DW_OP_GNU_implicit_pointer refers to a DIE with an offset of different
sizes in different dwarf versions. In v2, the size is the pointer size,
while in v3 and above, it is the ref_addr size. This patch fixes
dwarf assembler to emit the correct size of offset. We've already fixed
this size issue in gdb,
https://sourceware.org/ml/gdb-patches/2011-09/msg00451.html
gdb/testsuite:
2017-01-25 Yao Qi <yao.qi@linaro.org>
* lib/dwarf.exp (Dwarf::_location): Handle
DW_OP_GNU_implicit_pointer with proper size.
Some leftover uppercase test names in py-xmethods.exp. The patch also
replaces two "continue" calls with untested calls to make things a bit more
clear.
gdb/testsuite/ChangeLog:
2017-01-20 Luis Machado <lgustavo@codesourcery.com>
* gdb.python/py-xmethods.exp: Fix test names starting with lowercase
and add untested calls.
I noticed gdb.python/python.exp failing on aarch64-elf like so:
FAIL: gdb.python/python.exp: Test decode_line func1 line number
This particular test expects the line number for func1 to be 19, hardcoded.
In my aarch64-elf tests gdb thinks func1 is at line 20, making the test fail.
The following patch addresses this by reading the line number information from
GDB and comparing it against the python decoded symtab information.
gdb/testsuite/ChangeLog:
2017-01-20 Luis Machado <lgustavo@codesourcery.com>
* gdb.python/python.exp: Check line number against what GDB thinks
the line number is for func1.
While casting works as expected with expression debugging turned off,
this seems to be an indication that the D language parser function is
doing something wrong in the building of the expression.
Without changing the grammar, using UNOP_CAST_TYPE is the right thing to
do here, as the TypeExp handler has already wrapped the type around a
pair of OP_TYPE opcodes.
gdb/ChangeLog:
* d-exp.y (CastExpression): Emit UNOP_CAST_TYPE.
gdb/testsuite/ChangeLog:
* gdb.dlang/debug-expr.exp: New file.
This applies the second part of GDB's End of Year Procedure, which
updates the copyright year range in all of GDB's files.
gdb/ChangeLog:
Update copyright year range in all GDB files.
I recently see the test fails like this,
(gdb) PASS: gdb.gdb/selftest.exp: step over argv initialization
list^M
487 std::vector<struct cmdarg> cmdarg_vec;^M
(gdb) FAIL: gdb.gdb/selftest.exp: unknown source line (after step over argv initialization)
step^M
std::vector<cmdarg, std::allocator<cmdarg> >::vector (this=0x7fffffffdc10) at ../../binutils-gdb/gdb/main.c:487^M
487 std::vector<struct cmdarg> cmdarg_vec;^M
(gdb) FAIL: gdb.gdb/selftest.exp: step into xmalloc call
These fails are caused by using std::vector in commit
f60ee22ea1. selttest.exp should match
the source code of GDB. It is a maintenance pain, so this patch
removes do_steps_and_nexts.
gdb/testsuite:
2016-12-19 Yao Qi <yao.qi@linaro.org>
* gdb.gdb/selftest.exp (do_steps_and_nexts): Remove.
(test_with_self): Don't call do_steps_and_nexts, and remove
code about stepping into xmalloc.
I build GDB with all targets enabled, and "set architecture rx",
GDB crashes,
(gdb) set architecture rx
Program received signal SIGSEGV, Segmentation fault.
append_flags_type_flag (type=0x20cc360, bitpos=bitpos@entry=0, name=name@entry=0xd27529 "C") at ../../binutils-gdb/gdb/gdbtypes.c:4926
4926 name);
(gdb) bt 10
#0 append_flags_type_flag (type=0x20cc360, bitpos=bitpos@entry=0, name=name@entry=0xd27529 "C") at ../../binutils-gdb/gdb/gdbtypes.c:4926
#1 0x00000000004ce725 in rx_gdbarch_init (info=..., arches=<optimized out>) at ../../binutils-gdb/gdb/rx-tdep.c:1051
#2 0x00000000006b05a4 in gdbarch_find_by_info (info=...) at ../../binutils-gdb/gdb/gdbarch.c:5269
#3 0x000000000060eee4 in gdbarch_update_p (info=...) at ../../binutils-gdb/gdb/arch-utils.c:557
#4 0x000000000060f8a8 in set_architecture (ignore_args=<optimized out>, from_tty=1, c=<optimized out>) at ../../binutils-gdb/gdb/arch-utils.c:531
#5 0x0000000000593d0b in do_set_command (arg=<optimized out>, arg@entry=0x20bee81 "rx ", from_tty=from_tty@entry=1, c=c@entry=0x20b1540)
at ../../binutils-gdb/gdb/cli/cli-setshow.c:455
#6 0x00000000007665c3 in execute_command (p=<optimized out>, p@entry=0x20bee70 "set architecture rx ", from_tty=1) at ../../binutils-gdb/gdb/top.c:666
#7 0x00000000006935f4 in command_handler (command=0x20bee70 "set architecture rx ") at ../../binutils-gdb/gdb/event-top.c:577
#8 0x00000000006938d8 in command_line_handler (rl=<optimized out>) at ../../binutils-gdb/gdb/event-top.c:767
#9 0x0000000000692c2c in gdb_rl_callback_handler (rl=0x20be7f0 "") at ../../binutils-gdb/gdb/event-top.c:200
The cause is that we want to access some builtin types in gdbarch init, but
it is not initialized yet. I fix it by creating the type when it is to be
used. We've already done this in sparc, sparc64 and m68k.
gdb:
2016-12-09 Yao Qi <yao.qi@linaro.org>
PR tdep/20954
* rx-tdep.c (rx_psw_type): New function.
(rx_fpsw_type): New function.
(rx_register_type): Call rx_psw_type and rx_fpsw_type.
(rx_gdbarch_init): Move code to rx_psw_type and
rx_fpsw_type.
gdb/testsuite:
2016-12-09 Yao Qi <yao.qi@linaro.org>
* gdb.base/all-architectures.exp.in: Remove kfail for "rx".
I build GDB for all targets enabled. When I "set architecture rl78",
GDB crashes,
(gdb) set architecture rl78
Program received signal SIGSEGV, Segmentation fault.
append_flags_type_flag (type=0x20cc0e0, bitpos=bitpos@entry=0, name=name@entry=0x11dba3f "CY") at ../../binutils-gdb/gdb/gdbtypes.c:4926
4926 name);
(gdb) bt 10
#0 append_flags_type_flag (type=0x20cc0e0, bitpos=bitpos@entry=0, name=name@entry=0x11dba3f "CY") at ../../binutils-gdb/gdb/gdbtypes.c:4926
#1 0x00000000004aaca8 in rl78_gdbarch_init (info=..., arches=<optimized out>) at ../../binutils-gdb/gdb/rl78-tdep.c:1410
#2 0x00000000006b05a4 in gdbarch_find_by_info (info=...) at ../../binutils-gdb/gdb/gdbarch.c:5269
#3 0x000000000060eee4 in gdbarch_update_p (info=...) at ../../binutils-gdb/gdb/arch-utils.c:557
#4 0x000000000060f8a8 in set_architecture (ignore_args=<optimized out>, from_tty=1, c=<optimized out>) at ../../binutils-gdb/gdb/arch-utils.c:531
#5 0x0000000000593d0b in do_set_command (arg=<optimized out>, arg@entry=0x20be851 "rl78", from_tty=from_tty@entry=1, c=c@entry=0x20b1540)
at ../../binutils-gdb/gdb/cli/cli-setshow.c:455
#6 0x00000000007665c3 in execute_command (p=<optimized out>, p@entry=0x20be840 "set architecture rl78", from_tty=1) at ../../binutils-gdb/gdb/top.c:666
#7 0x00000000006935f4 in command_handler (command=0x20be840 "set architecture rl78") at ../../binutils-gdb/gdb/event-top.c:577
#8 0x00000000006938d8 in command_line_handler (rl=<optimized out>) at ../../binutils-gdb/gdb/event-top.c:767
#9 0x0000000000692c2c in gdb_rl_callback_handler (rl=0x20be890 "") at ../../binutils-gdb/gdb/event-top.c:200
The cause is that we want to access some builtin types in gdbarch init, but
it is not initialized yet. I fix it by creating the type when it is to be
used. We've already done this in sparc, sparc64 and m68k.
gdb:
2016-12-09 Yao Qi <yao.qi@linaro.org>
PR tdep/20953
* rl78-tdep.c (rl78_psw_type): New function.
(rl78_register_type): Call rl78_psw_type.
(rl78_gdbarch_init): Move code to rl78_psw_type.
gdb/testsuite:
2016-12-09 Yao Qi <yao.qi@linaro.org>
* gdb.base/all-architectures.exp.in: Remove kfail for rl78.
This adds a test that exposes several problems fixed by earlier
patches:
#1 - Buffer overrun when host/target formats match, but sizes don't.
https://sourceware.org/ml/gdb-patches/2016-03/msg00125.html#2 - Missing handling for FR-V FR300.
https://sourceware.org/ml/gdb-patches/2016-03/msg00117.html#3 - BFD architectures with spaces in their names (v850).
https://sourceware.org/ml/binutils/2016-03/msg00108.html#4 - The OS ABI names with spaces issue.
https://sourceware.org/ml/gdb-patches/2016-03/msg00116.html#5 - Bogus HP/PA long double format.
https://sourceware.org/ml/gdb-patches/2016-03/msg00122.html#6 - Cris big endian internal error.
https://sourceware.org/ml/gdb-patches/2016-03/msg00126.html#7 - Several PowerPC bfd archs/machines not handled by gdb.
https://sourceware.org/bugzilla/show_bug.cgi?id=19797
And hopefully helps catch others in the future.
This started out as a test that simply did,
gdb -ex "print 1.0L"
to exercise #1 above.
Then to cover both 32-bit target / 64-bit host and the converse, I
thought of having the testcase print the floats twice, once with the
architecture set to "i386" and then to "i386:x86-64". This way it
wouldn't matter whether gdb was built as 32-bit or a 64-bit program.
Then I thought that other archs might have similar host/target
floatformat conversion issues as well. Instead of hardcoding some
architectures in the test file, I thought we could just iterate over
all bfd architectures and OS ABIs supported by the gdb build being
tested. This is what then exposed all the other problems listed
above...
With an --enable-targets=all, this exercises over 14 thousand
combinations. If left in a single test file, it all consistenly runs
in under a minute on my machine (An Intel i7-4810MQ @ 2.8 MHZ running
Fedora 23). Split in 8 chunks, as in this commit, it runs in around
25 seconds, with make -j8.
To avoid flooding the gdb.sum file, it avoids calling "pass" on each
tested combination/iteration. I'm explicitly not implementing that by
passing an empty message to gdb_test / gdb_test_multiple, because I
still want a FAIL to be logged in gdb.sum. So instead this puts the
internal passes in the gdb.log file, only, prefixed "IPASS:", for
internal pass. TBC, if some iteration fails, it'll still show up as
FAIL in gdb.sum. If this is an approach that takes on, I can see us
extending the common bits to support it for all testcases.
gdb/testsuite/ChangeLog:
2016-12-09 Pedro Alves <palves@redhat.com>
* gdb.base/all-architectures-0.exp: New file.
* gdb.base/all-architectures-1.exp: New file.
* gdb.base/all-architectures-2.exp: New file.
* gdb.base/all-architectures-3.exp: New file.
* gdb.base/all-architectures-4.exp: New file.
* gdb.base/all-architectures-5.exp: New file.
* gdb.base/all-architectures-6.exp: New file.
* gdb.base/all-architectures-7.exp: New file.
* gdb.base/all-architectures.exp.in: New file.
gdb.perf/skip-prologue.exp is intended to measure the performance of
skipping prologue with prologue analysis by setting breakpoints.
However, if program is compiled with debug info, GDB is smart to
skip prologue by line table from debug info, so prologue analysis
is not exercised at all.
This patch adds a parameter COMPILE to specify compiling with
debug information, otherwise, it is compiled without debug
information.
gdb/testsuite:
2016-12-09 Yao Qi <yao.qi@linaro.org>
* gdb.perf/skip-prologue.exp: Add parameter COMPILE.
This gets rid of more useless pattern matching cases in gdb.base/maint.exp.
gdb/testsuite/ChangeLog:
2016-12-02 Luis Machado <lgustavo@codesourcery.com>
* gdb.base/maint.exp: Use gdb_test instead of gdb_test_multiple when
possible.
Remove useless pattern-matching code.
New in v2:
- A few adjustments / simplifications were possible now that we
require C++11:
. Use std::unique_ptr to make the user_args_stack std::vector own
its elements:
static std::vector<std::unique_ptr<user_args>> user_args_stack;
. use vector::emplace_back to construct elements directly in the
corresponding vectors.
. use std::to_string instead of adding a gdb::to_string
replacement.
- Now includes a test.
Docs/NEWS are unchanged from v1 and have already been approved.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
I recently wrote a user-defined command that could benefit from
supporting an unlimited number of arguments:
http://palves.net/list-active-signal-handlers-with-gdb/
E.g., 'info signal-dispositions 1 2 3 4 5 6 7 8 9 10 11'
However, we currently only support up to 10 arguments passed to
user-defined commands ($arg0..$arg9).
I can't find a good reason for that, other than "old code with hard
coded limits". This patch removes that limit and modernizes the code
along the way:
- Makes the user_args struct a real C++ class that uses std::vector
for storage.
- Removes the "next" pointer from within user_args and uses a
std::vector to maintain a stack instead.
- Adds a new RAII-based scoped_user_args_level class to help
push/pop user args in the stack instead of using a cleanup.
gdb/ChangeLog:
2016-12-02 Pedro Alves <palves@redhat.com>
* NEWS: Mention that user commands now accept an unlimited number
of arguments.
* cli/cli-script.c: Include <vector>.
(struct string_view): New type.
(MAXUSERARGS): Delete.
(struct user_args): Now a C++ class.
(user_args_stack): New.
(struct scoped_user_args_level): New type.
(execute_user_command): Use scoped_user_args_level.
(arg_cleanup): Delete.
(setup_user_args): Deleted, and refactored as ...
(user_args::user_args): ... this new constructor. Limit of number
of arguments removed.
(insert_user_defined_cmd_args): Defer to user_args_stack.
(user_args::insert_args): New, bits based on old
insert_user_defined_cmd_args with limit of number of arguments
eliminated.
gdb/doc/ChangeLog:
2016-12-02 Pedro Alves <palves@redhat.com>
* gdb.texinfo (User-defined Commands): Limit on number of
arguments passed to user-defined commands removed; update.
gdb/testsuite/ChangeLog:
2016-12-02 Pedro Alves <palves@redhat.com>
* gdb.base/commands.exp (user_defined_command_manyargs_test): New
procedure.
(top level): Call it.
We're missing a test that makes sure that arguments to user-defined
commands are handled correctly when a user-defined command calls
another user-defined command / recurses.
The following patch changes that code, so add such a test first so we
can be confident won't be breaking this use case.
gdb/testsuite/ChangeLog:
2016-12-02 Pedro Alves <palves@redhat.com>
* gdb.base/commands.exp (user_defined_command_args_stack_test):
New procedure.
(top level): Call it.
It'd be handy to be able to iterate over command arguments in
user-defined commands, in order to support optional arguments
($arg0..$argN).
I thought I could make it work with "eval", but alas, it doesn't work
currently. E.g., with:
define test
set $i = 0
while $i < $argc
eval "print $arg%d", $i
set $i = $i + 1
end
end
we get:
(gdb) test 1
$1 = void
(gdb) test 1 2 3
$2 = void
$3 = void
$4 = void
(gdb)
The problem is that "eval" doesn't do user-defined command arguments
substitution after expanding its own argument. This patch fixes that,
which makes the example above work:
(gdb) test 1
$1 = 1
(gdb) test 1 2 3
$2 = 1
$3 = 2
$4 = 3
(gdb)
New test included, similar the above, but also exercises expanding
$argc.
I think this is likely to simplify many scripts out there, so I'm
adding an example to the manual and mentioning it in NEWS as well.
gdb/ChangeLog:
2016-12-02 Pedro Alves <palves@redhat.com>
PR cli/20559
* NEWS: Mention "eval" expands user-defined command arguments.
* cli/cli-script.c (execute_control_command): Adjust to rename.
(insert_args): Rename to ...
(insert_user_defined_cmd_args): ... this, and make extern.
* cli/cli-script.h (insert_user_defined_cmd_args): New
declaration.
* printcmd.c: Include "cli/cli-script.h".
(eval_command): Call insert_user_defined_cmd_args.
gdb/doc/ChangeLog:
2016-12-02 Pedro Alves <palves@redhat.com>
PR cli/20559
* gdb.texinfo (Define): Add example of using "eval" to process a
variable number of arguments.
(Output) <eval>: Add anchor.
gdb/testsuite/ChangeLog:
2016-12-02 Pedro Alves <palves@redhat.com>
PR cli/20559
* gdb.base/commands.exp (user_defined_command_args_eval): New
procedure.
(top level): Call it.
This reverts the timeout handling (removed by
018572b888) for gdb.cp/ovldbreak.exp until we
decide what to do about this particular function.
gdb/testsuite/ChangeLog:
2016-12-02 Luis Machado <lgustavo@codesourcery.com>
* gdb.base/ovldbreak.exp (take_gdb_out_of_choice_menu): Restore
timeout handling.
This patch adds support for DW_AT_main_subprogram.
This is PR symtab/16264.
DW_AT_main_subprogram is used to mark a program's entry point. GCC
can emit this, and I hope to change the Rust compiler to emit it as
well.
GDB already supports an older, pre-DWARF 4 convention adopted by
FORTRAN compilers, namely to emit DW_AT_calling_convention for the
"main" function. However, I think this support in GDB had a small
bug, in that it seems to rely on the DW_AT_name being read before
DW_AT_calling_convention. This patch fixes this as well.
Built and regtested on x86-64 Fedora 24 and the buildbot. New test
case included.
2016-12-02 Tom Tromey <tom@tromey.com>
PR symtab/16264:
* dwarf2read.c (struct partial_die_info) <main_subprogram>: New
member.
(add_partial_symbol): Call set_objfile_main_name.
(read_partial_die): Handle DW_AT_main_subprogram.
<DW_AT_calling_convention>: don't call set_objfile_main_name, but
set main_subprogram flag.
2016-12-02 Tom Tromey <tom@tromey.com>
* gdb.dwarf2/main-subprogram.c: New file.
* gdb.dwarf2/main-subprogram.exp: New file.
This fixes a few cases where the testcase is explicitly handling timeouts
inside gdb_test_multiple when it is not necessary.
It also converts two gdb_test_multiple calls to gdb_test_no_output calls
(also removing the timeout handling).
gdb/testsuite/ChangeLog:
2016-12-01 Luis Machado <lgustavo@codesourcery.com>
* gdb.base/maint.exp: Remove timeout handling for gdb_test_multiple.
* gdb.cp/gdb2495.exp: Likewise and convert gdb_test_multiple into
gdb_test_no_output for a couple of cases.
* gdb.cp/ovldbreak.exp: Remove timeout handling for gdb_test_multiple.
This fixes offender testcases that have test names starting with uppercase
when using gdb_test_multiple in a multi-line construct.
gdb/testsuite/ChangeLog
2016-12-01 Luis Machado <lgustavo@codesourcery.com>
* gdb.cp/gdb2495.exp: Replace gdb_test_multiple
with gdb_test_no_output.
Use command as test name.
This fixes offender testcases that have test names starting with uppercase
when using gdb_test_no_output in a multi-line construct.
gdb/testsuite/ChangeLog
2016-12-01 Luis Machado <lgustavo@codesourcery.com>
Fix test names starting with uppercase throughout the files.
* gdb.ada/assign_1.exp
* gdb.ada/boolean_expr.exp
* gdb.base/arrayidx.exp
* gdb.base/del.exp
* gdb.base/gcore-buffer-overflow.exp
* gdb.base/testenv.exp
* gdb.compile/compile.exp
* gdb.python/py-framefilter-invalidarg.exp
* gdb.python/py-framefilter.exp
This fixes offender testcases that have test names starting with uppercase
when using gdb_test_multiple in a single-line construct.
gdb/testsuite/ChangeLog
2016-12-01 Luis Machado <lgustavo@codesourcery.com>
Fix test names starting with uppercase throughout the files.
* gdb.arch/i386-bp_permanent.exp
* gdb.arch/i386-gnu-cfi.exp
* gdb.base/disasm-end-cu.exp
* gdb.base/macscp.exp
* gdb.base/pending.exp
* gdb.base/watch_thread_num.exp
* gdb.cp/exception.exp
* gdb.cp/gdb2495.exp
* gdb.cp/local.exp
* gdb.python/py-evsignal.exp
* gdb.python/python.exp
* gdb.trace/tracecmd.exp
Since we don't use suffix rules nor implicit rules in gdb, we can
disable them. The advantage is a slightly faster make [1].
Here are some numbers about the speedup. I ran this on my trusty old
Intel Q6600, so the time numbers are probably higher than what you'd get
on any recent hardware. I ran "make" in the gdb/ directory of an
already built repository (configured with --enable-targets=all). I
recorded the time of execution (average of 5). I then ran "make -d" and
recorded the number of printed lines, which gives a rough idea of the
number of operations done.
I compared the following configurations, to see the impact of both the
empty .SUFFIXES target and the empty pattern rules, as well as running
"make -r", which can be considered the "ideal" case.
A - baseline
B - baseline + .SUFFIXES
C - baseline + pattern rules
D - baseline + .SUFFIXES + pattern rules
E - baseline + make -r
config | time (s) | "make -d"
-----------------------------
A | 5.74 | 2396643
B | 1.19 | 298469
C | 2.81 | 1266573
D | 1.13 | 245489
E | 1.01 | 163914
We can see that the empty .SUFFIXES target has a bigger impact than the
empty pattern rules, but still it doesn't hurt to disable the implicit
pattern rules as well.
There are still some mentions of implicit rules I can't get rid of in
the "make -d" output. For example, it's trying to build .c files from
.w files:
Looking for an implicit rule for '/home/simark/src/binutils-gdb/gdb/infrun.c'.
Trying pattern rule with stem 'infrun'.
Trying implicit prerequisite '/home/simark/src/binutils-gdb/gdb/infrun.w'.
and trying to build Makefile.in from a bunch of extensions:
Looking for an implicit rule for 'Makefile.in'.
Trying pattern rule with stem 'Makefile.in'.
Trying implicit prerequisite 'Makefile.in.o'.
Trying pattern rule with stem 'Makefile.in'.
Trying implicit prerequisite 'Makefile.in.c'.
Trying pattern rule with stem 'Makefile.in'.
Trying implicit prerequisite 'Makefile.in.cc'.
... many more ...
If somebody knows how to disable them, we can do it, but at this point
the returns are minimal, so it is not that important.
I verified that both in-tree and out-of-tree builds work.
[1] Switching from explicit rules to pattern rules for files in
subdirectories actually made it slower, so this is kind of a way to
redeem myself. But it the end it's faster than it was previously,
so it was all worth it. :)
gdb/ChangeLog:
* disable-implicit-rules.mk: New file.
* Makefile.in: Include disable-implicit-rules.mk.
* data-directory/Makefile.in: Likewise.
* gnulib/Makefile.in: Likewise.
gdb/doc/ChangeLog:
* Makefile.in: Likewise.
gdb/gdbserver/ChangeLog:
* Makefile.in: Include disable-implicit-rules.mk.
gdb/testsuite/ChangeLog:
* Makefile.in: Include disable-implicit-rules.mk.
Martin Galvan
Jerome Guitton
Luis Machado
Ivo Raisr
Tom Tromey
Pedro Alves
Andreas Arnez
Markus Metzger
Walfred Tedeschi
Simon Marchi
Yao Qi
Iain Buclaw
Joel Brobecker
Doug Evans