I was "lucky" enough that an unrelated patch changed how many symtabs
GDB expands in a plain run to main, and that triggered a latent issue
in this test:
PASS: gdb.base/maint.exp: maint print objfiles: header
PASS: gdb.base/maint.exp: maint print objfiles: psymtabs
FAIL: gdb.base/maint.exp: maint print objfiles: symtabs
The problem is in my case, expect is managing to alway put in the
buffer chunks like this:
Psymtabs:
../../../src/gdb/testsuite/gdb.base/break1.c at 0x1ed2280, ../../../src/gdb/testsuite/gdb.base/break.c at 0x1ed21d0,
Symtabs:
../../../src/gdb/testsuite/gdb.base/break.c at 0x1f044f0, /usr/include/stdio.h at 0x1ed25a0, /usr/include/libio.h at 0x1ed2510, /usr/include/bits/types.h at 0x1ed2480, /usr/lib/gcc/x86_64-redhat-linux/4.7.2/include/stddef.h at 0x1ed23f0,
Object file /usr/lib/debug/lib64/ld-2.15.so.debug: Objfile at 0x1f4bff0, bfd at 0x1f2d940, 0 minsyms
Psymtabs:
bsearch.c at 0x1f65340, ../sysdeps/x86_64/multiarch/init-arch.c at
0x1f65290, ...
Note: Psymtabs:/Symtabs:/Psymtabs:.
So, the loop matches the first Psymtabs in the buffer. Then we're
left with
../../../src/gdb/testsuite/gdb.base/break1.c at 0x1ed2280, ../../../src/gdb/testsuite/gdb.base/break.c at 0x1ed21d0,
Symtabs:
../../../src/gdb/testsuite/gdb.base/break.c at 0x1f044f0, /usr/include/stdio.h at 0x1ed25a0, /usr/include/libio.h at 0x1ed2510, /usr/include/bits/types.h at 0x1ed2480, /usr/lib/gcc/x86_64-redhat-linux/4.7.2/include/stddef.h at 0x1ed23f0,
Object file /usr/lib/debug/lib64/ld-2.15.so.debug: Objfile at 0x1f4bff0, bfd at 0x1f2d940, 0 minsyms
Psymtabs:
bsearch.c at 0x1f65340, ../sysdeps/x86_64/multiarch/init-arch.c at
0x1f65290, ...
In the next iteration, because the psymtabs regex comes first, we
match with the Psymtabs: line, then of course, end up with just
bsearch.c at 0x1f65340, ../sysdeps/x86_64/multiarch/init-arch.c at
0x1f65290, ...
in the buffer. The "Symtabs:" line is lost. expect then reads more
gdb output, and manages to again retrieve the same pattern. Rinse,
repeat, and the test never matches any "Symtab:" line.
We don't know the order the matches lines will appear, so the fix is
to consume one line at a time, and run it through all three milestone
regexes.
gdb/testsuite/
2013-11-20 Pedro Alves <palves@redhat.com>
* gdb.base/maint.exp (maint print objfiles): Consume one line at a
time, and run it through all three milestone regexes.
Ensure that certain commands (e.g. whatis/ptype) and sizeof intrinsic
have no side effects (variables cannot be altered).
2013-11-20 Sanimir Agovic <sanimir.agovic@intel.com>
testsuite/
* gdb.base/eval-avoid-side-effects.exp: New test.
Boundary length is simpler implemented by means of a pretty
printer. This simplifies users life when examining a bound register.
Changelog:
2013-11-20 Walfred Tedeschi <walfred.tedeschi@intel.com>
* python/lib/gdb/command/bound_register.py: New file.
* gdb/data-directory/Makefile.in: copy bond_register.py to the right path to
be initialized at gdb startup.
testsuite/
* gdb.python/py-pp-maint.exp: Consider new pretty-print added for registers.
Change-Id: Id4f39845e5ece56c370a1fd4343648909f08b731
Signed-off-by: Walfred Tedeschi <walfred.tedeschi@intel.com>
Conflicts:
gdb/ChangeLog
Bitfields are represented by intervals [start, begin]. It means that for an
interval comprised by only one bit start and end will be equal.
The present condition does not always hold. On the other hand in target-description.c
(tdesc_gdb_type) bitfield is created when "f->type" is null. The routine
maint_print_maint_print_c_tdesc_cmd is modified to follow the same strategy.
2013-11-20 Walfred Tedeschi <walfred.tedeschi@intel.com>
* target-descriptions.c (maint_print_maint_print_c_tdesc_cmd):
Modified logic of creating a bitfield to be in sync with
tdesc_gdb_type.
testsuite/
* gdb.xml/maint_print_struct.xml (bitfield): Added bitfield having
start and end equal 0.
Change-Id: I8c62db049995f0c0c30606d9696b86afe237cbb9
Hi,
In proc mi_child_regexp, \(,thread-id=\"\[0-9\]+\") is appended to
children_exp, while the first '\' is not necessary. This patch
is to remove it. With this patch applied, Emacs can find the right
left paren.
gdb/testsuite:
2013-11-19 Yao Qi <yao@codesourcery.com>
* lib/mi-support.exp (mi_child_regexp): Remove unnecessary '\'.
There are some format issues in lib/mi-support.exp, such as using
spaces instead of tab and trailing spaces. This patch is to fix them.
gdb/testsuite:
2013-11-19 Yao Qi <yao@codesourcery.com>
* lib/mi-support.exp: Fix format.
Variable 'whatever' is not used at all. This patch is to remove it.
gdb/testsuite:
2013-11-19 Yao Qi <yao@codesourcery.com>
* lib/mi-support.exp (mi_child_regexp): Remove 'whatever'.
(mi_list_varobj_children_range): Likewise.
Consider the following variable:
type Small is range -128 .. 127;
SR : Small := 48;
Trying to get its value as an integer within Python code yields:
(gdb) python sr = gdb.parse_and_eval('sr')
(gdb) python print int(sr)
Traceback (most recent call last):
File "<string>", line 1, in <module>
gdb.error: Cannot convert value to int.
Error while executing Python code.
This is happening because our variable is a range type, and
py-value's is_intlike does not handle TYPE_CODE_RANGE. This
patch fixes this.
gdb/ChangeLog:
* python/py-value.c (is_intlike): Add TYPE_CODE_RANGE handling.
gdb/testsuite/ChangeLog:
* gdb.ada/py_range: New testcase.
In the case where the GNAT runtime was built with full debugging info,
several of the exceptions defined there might have a name contain
the word "global". To make this less likely, this patch renames
the exception name, replacing "Global" by "Global_GDB". It still
keeps the exeption name relatively short, while it is unlikely that
the GNAT runtime has an exception whose name explicitly mentions GDB,
and even less likely that it contains "Global_GDB".
gdb/testsuite/ChangeLog:
* info_exc/const.ads (Aint_Global_GDB_E): Renames Aint_Global_E.
* info_exc/foo.adb: Adjust to new exception name.
* info_exc.exp: Adjust after exception renaming in const.ads.
Update "info exception global" test to test "info exceptions
global_gdb" instead.
* mi_exc_info/const.ads (Aint_Global_GDB_E): Renames Aint_Global_E.
* mi_exc_info/foo.adb (Adjust to new exception name.
* mi_exc_info.exp: Adjust after exception renaming in const.ads.
Update "-info-ada-exceptions global" test to test
"-info-ada-exceptions global_gdb" instead.
of inferior output for remote and native sessions.
* gdb.mi/mi-console.exp: Remove obsolete comment.
Check for semihosted inferior output pattern.
(semihosted_string): New function.
If the runtime has full debug info, then the non-standard exceptions
declared in the GNAT runtime will appear in the list of exceptions
printed by GDB ("info exceptions" or "-info-ada-exceptions").
This is valid output, so this patch allows for it.
gdb/testsuite/ChangeLog:
* gdb.ada/info_exc.exp: Allow other global exceptions to be
listed in the output of "info exceptions".
* gdb.ada/mi_exc_info.exp: Allow other global exceptions to be
listed in the output of "-info-ada-exceptions".
This patch fixes some spurious failures when the inferior is linked
against the shared version of libgnat by default, as appears to be
the case on many GNU/Linux distributions. When that happens, we have
to start the program in order to ensure that the GNAT runtime is
mapped to memory, in order for us to find the standard exceptions
(defined within the runtime). Otherwise, they will not be shown,
as expected, by the debugger.
gdb/testsuite/ChangeLog:
* gdb.ada/info_exc.exp: Start inferior before starting
the "info exceptions" tests.
* gdb.ada/mi_exc_info.exp: Start inferior before starting
the "-info-ada-exceptions" tests.
This patch fixes PR c++/16117.
gdb has an extension so that users can use expressions like FILE::NAME
to choose a variable of the given name from the given file. The bug
is that this extension takes precedence over ordinary C++ expressions
of the same form. You might think this is merely hypothetical, but
now that C++ headers commonly do not use an extension, it is more
common.
This patch fixes the bug by making two related changes. First, it
changes gdb to prefer the ordinary C++ meaning of a symbol over the
extended meaning. Second, it arranges for single-quoting of the
symbol to indicate a preference for the extension.
Built and regtested on x86-64 Fedora 18.
New test case included.
2013-11-15 Tom Tromey <tromey@redhat.com>
PR c++/16117:
* c-exp.y (lex_one_token): Add "is_quoted_name" argument.
(classify_name): Likewise. Prefer a field of "this" over a
filename.
(classify_inner_name, yylex): Update.
2013-11-15 Tom Tromey <tromey@redhat.com>
* gdb.texinfo (Variables): Note gdb rules for ambiguous cases.
Add example.
2013-11-15 Tom Tromey <tromey@redhat.com>
* gdb.cp/includefile: New file.
* gdb.cp/filename.exp: New file.
* gdb.cp/filename.cc: New file.
Consider the following code, compiled at -O2 on ppc-linux:
procedure Increment (Val : in out Float; Msg : String);
The implementation does not really matter in this case). In our example,
this function is being called from a function with Param_1 set to 99.0.
Trying to break inside that function, and running until reaching that
breakpoint yields:
(gdb) b increment
Breakpoint 1 at 0x100014b4: file callee.adb, line 6.
(gdb) run
Starting program: /[...]/foo
Breakpoint 1, callee.increment (val=99.0, val@entry=0.0, msg=...)
at callee.adb:6
6 if Val > 200.0 then
The @entry value for parameter "val" is incorrect, it should be 99.0.
The associated call-site parameter DIE looks like this:
.uleb128 0xc # (DIE (0x115) DW_TAG_GNU_call_site_parameter)
.byte 0x2 # DW_AT_location
.byte 0x90 # DW_OP_regx
.uleb128 0x21
.byte 0x3 # DW_AT_GNU_call_site_value
.byte 0xf5 # DW_OP_GNU_regval_type
.uleb128 0x3f
.uleb128 0x25
The DW_AT_GNU_call_site_value uses a DW_OP_GNU_regval_type
operation, referencing register 0x3f=63, which is $f31,
an 8-byte floating register. In that register, the value is
stored using the usual 8-byte float format:
(gdb) info float
f31 99.0 (raw 0x4058c00000000000)
The current code evaluating DW_OP_GNU_regval_type operations
currently is (dwarf2expr.c:execute_stack_op):
result = (ctx->funcs->read_reg) (ctx->baton, reg);
result_val = value_from_ulongest (address_type, result);
result_val = value_from_contents (type,
value_contents_all (result_val));
What the ctx->funcs->read_reg function does is read the contents
of the register as if it contained an address. The rest of the code
continues that assumption, thinking it's OK to then use that to
create an address/ulongest struct value, which we then re-type
to the type specified by DW_OP_GNU_regval_type.
We're getting 0.0 above because the read_reg implementations
end up treating the contents of the FP register as an integral,
reading only 4 out of the 8 bytes. Being a big-endian target,
we read the high-order ones, which gives us zero.
This patch fixes the problem by introducing a new callback to
read the contents of a register as a given type, and then adjust
the handling of DW_OP_GNU_regval_type to use that new callback.
gdb/ChangeLog:
* dwarf2expr.h (struct dwarf_expr_context_funcs) <read_reg>:
Extend the documentation a bit.
<get_reg_value>: New field.
* dwarf2loc.c (dwarf_expr_get_reg_value)
(needs_frame_get_reg_value): New functions.
(dwarf_expr_ctx_funcs, needs_frame_ctx_funcs): Add "get_reg_value"
callback.
* dwarf2-frame.c (get_reg_value): New function.
(dwarf2_frame_ctx_funcs): Add "get_reg_value" callback.
* dwarf2expr.c (execute_stack_op) <DW_OP_GNU_regval_type>:
Use new callback to compute result_val.
gdb/testsuite/ChangeLog:
* gdb.ada/O2_float_param: New testcase.
Pedro pointed out that it is handy for "make check" to print a summary
of the results. This happens in the check-single case and also if you
invoke runtest by hand.
This patch implements the same thing for check-parallel.
2013-11-14 Tom Tromey <tromey@redhat.com>
* Makefile.in (check-parallel): Print summary from gdb.sum.
dw2-case-insensitive.exp: p fuNC_lang fails on arm. The problem occurs
when thumb mode code is generated. On ARM last bit of function pointer
value indicates whether the target function is an ARM (if 0) or Thumb
(if 1) routine. The PC address should refer to actual address in
either case. This patch adds new compile unit and function labels to
code which act as address ranges of compile unit and functions in
debug information. Therefore address ranges will have correct
addresses and not the ones with an incremented least significant bit.
This patch has been tested on x86_64 and arm machines.
gdb/testsuite/ChangeLog:
2013-11-14 Omair Javaid <Omair.Javaid@linaro.org>
* gdb.dwarf2/dw2-case-insensitive-debug.S: Updated compile unit
and function label names.
* gdb.dwarf2/dw2-case-insensitive.c: Created function and
compile unit labels.
Frontend sometimes need to evaluate expressions that are
language-specific. For instance, Eclipse uses the following
expression to determine the size of an address on the target:
-data-evaluate-expression "sizeof (void*)"
Unfortunately, if the main of the program being debugged is not C,
this may not work. For instance, if the main is in Ada, you get...
-data-evaluate-expression "sizeof (void*)"
^error,msg="No definition of \"sizeof\" in current context."
... and apparently decides to stop the debugging session as a result.
The recommendation sent was to specifically set the language to C
before trying to evaluate the expression. Something such as:
1. save current language
2. set language c
3. -data-evaluate-expression "sizeof (void*)"
4. Restore language
This has the same disadvantages as the ones outlined in the "Context
Management" section of the GDB/MI documentation regarding setting
the current thread or the current frame, thus recommending the use of
general command-line switches such as --frame, or --thread instead.
This patch follows the same steps for the language, adding a similar
new command option: --language LANG. Example of use:
-data-evaluate-expression --language c "sizeof (void*)"
^done,value="4"
gdb/ChangeLog:
* mi/mi-parse.h (struct mi_parse) <language>: New field.
* mi/mi-main.c (mi_cmd_execute): Temporarily set language to
PARSE->LANGUAGE during command execution, if set.
* mi/mi-parse.c: Add "language.h" #include.
(mi_parse): Add parsing of "--language" command option.
* NEWS: Add entry mentioning the new "--language" command option.
gdb/testsuite/ChangeLog:
* gdb.mi/mi-language.exp: New file.
gdb/doc/ChangeLog:
* gdb.texinfo (Show): Add xref anchor for "show language" command.
(Context management): Place current subsection text into its own
subsubsection. Add new subsubsection describing the "--language"
command option.
I noticed today that multi-arch-exec.exp was failing in parallel mode.
The bug is that multi-arch-exec.c assumes the non-parallel directory
layout.
This patch fixes the problem using the same "BASEDIR" approach used in
other tests.
Tested both ways on x86-64 Fedora 18.
I'm checking this in.
2013-11-13 Tom Tromey <tromey@redhat.com>
* gdb.multi/multi-arch-exec.exp: Define BASEDIR when compiling.
* gdb.multi/multi-arch-exec.c (main): Use BASEDIR.
This patch implements a new GDB/MI command implementing the equivalent
of the "info exceptions" CLI command. The command syntax is:
-info-ada-exceptions [REGEXP]
Here is an example of usage (slightly formatted by hand to make it
easier to read):
-info-ada-exceptions ions\.a_
^done,ada-exceptions=
{nr_rows="2",nr_cols="2",
hdr=[{width="1",alignment="-1",col_name="name",colhdr="Name"},
{width="1",alignment="-1",col_name="address",colhdr="Address"}],
body=[{name="global_exceptions.a_global_exception",
address="0x0000000000613a80"},
{name="global_exceptions.a_private_exception",
address="0x0000000000613ac0"}]}
Also, in order to allow graphical frontends to easily determine
whether this command is available or not, the output of the
"-list-features" command has been augmented to contain
"info-ada-exceptions".
gdb/Changelog:
* mi/mi-cmds.h (mi_cmd_info_ada_exceptions): Add declaration.
* mi/mi-cmds.c (mi_cmds): Add entry for -info-ada-exceptions
command.
* mi/mi-cmd-info.c: #include "ada-lang.c" and "arch-utils.c".
(mi_cmd_info_ada_exceptions): New function.
* mi/mi-main.c (mi_cmd_list_features): Add "info-ada-exceptions".
gdb/testsuite/ChangeLog:
* gdb.ada/mi_exc_info: New testcase.
This patch adds a new command "info exceptions" whose purpose is to
provide the list of exceptions currently defined in the inferior.
The usage is:
(gdb) info exceptions [REGEXP]
Without argument, the command lists all exceptions. Otherwise,
only those whose name match REGEXP are listed.
For instance:
(gdb) info exceptions
All defined Ada exceptions:
constraint_error: 0x613dc0
program_error: 0x613d40
storage_error: 0x613d00
tasking_error: 0x613cc0
global_exceptions.a_global_exception: 0x613a80
global_exceptions.a_private_exception: 0x613ac0
The name of the command, as well as its output is part of a legacy
I inherited long ago. It's output being parsed by frontends such as
GPS, I cannot easily change it. Same for the command name.
The implementation is mostly self-contained, and is written in a way
that should make it easy to implement the GDB/MI equivalent. The
careful reviewer will notice that the code added in ada-lang.h could
normally be made private inside ada-lang.c. But these will be used
by the GDB/MI implementation. Rather than making those private now,
only to move them later, I've made them public right away.
gdb/ChangeLog:
* ada-lang.h: #include "vec.h".
(struct ada_exc_info): New.
(ada_exc_info): New typedef.
(DEF_VEC_O(ada_exc_info)): New vector.
(ada_exceptions_list): Add declaration.
* ada-lang.c (ada_is_exception_sym)
(ada_is_non_standard_exception_sym, compare_ada_exception_info)
(sort_remove_dups_ada_exceptions_list)
(ada_exc_search_name_matches, ada_add_standard_exceptions)
(ada_add_exceptions_from_frame, ada_add_global_exceptions)
(ada_exceptions_list_1, ada_exceptions_list)
(info_exceptions_command): New function.
(_initialize_ada_language): Add "info exception" command.
gdb/testsuite/ChangeLog:
* gdb.ada/info_exc: New testcase.
When using the GDB/MI commands to insert a catchpoint on a specific
Ada exception, any re-evaluation of that catchpoint (for instance
a re-evaluation performed after a shared library got mapped by the
inferior) fails. For instance, with any Ada program:
(gdb)
-catch-exception -e program_error
^done,bkptno="1",bkpt={[...]}
(gdb)
-exec-run
=thread-group-started,id="i1",pid="28315"
=thread-created,id="1",group-id="i1"
^running
*running,thread-id="all"
(gdb)
=library-loaded,[...]
&"warning: failed to reevaluate internal exception condition for catchpoint 1: No definition of \"exec\" in current context.\n"
&"warning: failed to reevaluate internal exception condition for catchpoint 1: No definition of \"exec\" in current context.\n"
[...]
The same is true if using an Ada exception catchpoint.
The problem comes from the fact that that we deallocate the strings
given as arguments to create_ada_exception_catchpoint, while the latter
just makes shallow copies of those strings, thus creating dandling
pointers.
This patch fixes the issue by passing freshly allocated strings to
create_ada_exception_catchpoint, while at the same time updating
create_ada_exception_catchpoint's documentation to make it clear
that deallocating the strings is no longer the responsibility of
the caller.
gdb/ChangeLog:
* ada-lang.c (create_ada_exception_catchpoint): Enhance
the documentation of fields "except_string" and "condition".
* mi/mi-cmd-catch.c (mi_cmd_catch_assert): Reallocate
CONDITION on the heap before passing it to
create_ada_exception_catchpoint.
(mi_cmd_catch_exception): Likewise for EXCEPTION_NAME and
CONDITION.
gdb/testsuite/ChangeLog:
* gdb.ada/mi_ex_cond: New testcase.
Tested on x86_64-linux. The "-break-list" test FAILs without
this patch.
* python/py-breakpoint.c (bppy_get_temporary): New function.
(bppy_init): New keyword: temporary. Parse it and set breakpoint
to temporary if True.
2013-11-07 Phil Muldoon <pmuldoon@redhat.com>
* gdb.python/py-breakpoint.exp: Add temporary breakpoint tests.
2013-11-07 Phil Muldoon <pmuldoon@redhat.com>
* gdb.texinfo (Breakpoints In Python): Document temporary
option in breakpoint constructor, and add documentation to the
temporary attribute.
This patch is to add a test case to on the performance of GDB handling
load and unload of shared library.
In V4:
- Handle malloc and dlopen failure,
- Document test parameters.
In V3, there are some changes,
- Adapt to perf test framework changes.
- Measure load and unload separately.
In V2, there are some changes,
- A new proc gdb_produce_source to produce source files. I tried to
move all source file generation code out of solib.exp, but
compilation step still needs to know the generated file names. I
have to hard-code the file names in compilation step, which is not
good to me, so I give up on this moving.
- SOLIB_NUMBER -> SOLIB_COUNT
- New variable SOLIB_DLCLOSE_REVERSED_ORDER to control the order of
iterating a list of shared libs to dlclose them.
- New variable GDB_PERFORMANCE to enable these perf test cases.
- Remove dlsym call in solib.c.
- Update solib.py for the updated framework.
gdb/testsuite/
* lib/gdb.exp (gdb_produce_source): New procedure.
* gdb.perf/solib.c: New.
* gdb.perf/solib.exp: New.
* gdb.perf/solib.py: New.
This patch adds a basic framework to do performance testing for GDB.
perftest.py is about the test case, testresult.py is about test
results, and how are they saved. reporter.py is about how results
are reported (in what format). measure.py is about measuring the
execution of tests by a collection of measurements.
In V5:
- Simplify perftest.exp.
In V4:
- Rename MeasurementCPUTime to MeasurementCpuTime,
- Add 'pass' in empty method,
- Simplify string comparison in perftest.exp.
- Rename GDB_PERFORMANCE to GDB_PERFTEST_MODE and rename
GDB_PERFORMANCE_TIMEOUT to GDB_PERFTEST_TIMEOUT.
In V3, there are some changes,
- Add wall time measurement, cpu time measurement and vmsize
measurement.
- Rename SingleStatisticTestCase to TestCaseWithBasicMeasurements,
which measures cpu time, wall time, and memory (vmsize).
- GDB_PERFORMANCE=run|compile|both to control the mode of perf
testing.
- New GDB_PERFORMANCE_TIMEOUT to specify the timeout.
- Split proc prepare to proc compile and startup.
- Disable GC while doing measurements.
In V2, there are several changes to address Doug and Sanimir's
comments.
- Add copyright header and docstring in perftest/__init__.py
- Remove config.py.
- Fix docstring format.
- Rename classes "SingleVariable" to "SingleStatistic".
- Don't extend gdb.Function in class TestCase. Add a new method run
to run the test case so that we can pass parameters to test.
- Allow to customize whether to warm up and to append test log.
- Move time measurement into test harness. Add a new class
Measurement for a specific measurement and a new class Measure to
measure them for a given test case.
- A new class ResultFactory to create instances of TestResult.
- New file lib/perftest.exp, which is to do some preparations and
cleanups to simplify each *.exp file.
- Skip compilation step if GDB_PERFORMANCE_SKIP_COMPILE is set.
gdb/testsuite/
2013-11-06 Yao Qi <yao@codesourcery.com>
* lib/perftest.exp: New.
* gdb.perf/lib/perftest/__init__.py: New.
* gdb.perf/lib/perftest/measure.py: New.
* gdb.perf/lib/perftest/perftest.py: New.
* gdb.perf/lib/perftest/reporter.py: New.
* gdb.perf/lib/perftest/testresult.py: New.
We add a new dir gdb.perf in testsuite for all performance tests.
However, current 'make check' logic will either run dejagnu in
directory testsuite or iterate all gdb.* directories which has *.exp
files. Both of them will run tests in gdb.perf. We want to achieve:
1) typical 'make check' should not run performance tests. In each perf
test case, GDB_PERFTEST_MODE is checked. If it doesn't exist, return.
2) run perf tests easily. We add a new makefile target 'check-perf'.
gdb:
2013-11-06 Yao Qi <yao@codesourcery.com>
* Makefile.in (check-perf): New target.
gdb/testsuite:
2013-11-06 Yao Qi <yao@codesourcery.com>
* Makefile.in (check-perf): New target.
* configure.ac (AC_OUTPUT): Output Makefile in gdb.perf.
* configure: Re-generated.
* gdb.perf/Makefile.in: New.
This switches "make check" to fully parallel mode.
One primary issue facing full parallelization is the overhead of
"runtest". On my machine, if I "touch gdb.base/empty.exp", making a
new file, and then "time runtest.exp", it takes 0.08 seconds.
Multiply this by the 1008 (in my configuration) tests and you get ~80
seconds. This is the overhead that would theoretically be present if
all tests were run in parallel.
However, the problem isn't nearly as bad as this, for two reasons.
First, you must divide by the number of jobs, assuming perfect
parallelization -- reasonably true for small -j numbers, based on the
results I see.
Second, the current test suite parallelization approach bundles the
tests, largely by directory, but also splitting up gdb.base into two
halves.
I was curious to see how the current bundling played out in practice,
so I ran "make -j1 check RUNTEST='/bin/time runtest'". This invokes
the parallel mode (thus the bundling) and then shows the time taken by
each invocation of runtest.
Then, I ran "/bin/time make -j3 check". (See below about -j2.)
The time for the entire -j3 test run was the same as the time for
"gdb.base1". What this means is that gdb.base1 is currently the
time-limiting run, preventing further parallelization gains.
So, I reason, whatever overhead we see from full parallelization will
only be seen by "-j1" and "-j2".
I then tried a -j2 test run. This does take longer than a -j3 build,
meaning that the gdb.base1 job finishes and then proceeds to other
runtest invocations.
Finally I tried a -j2 test run with the appended patch.
This was 9% slower than the -j2 run without the patch.
I think that is a reasonable slowdown for what is probably a rare
case. I believe this patch will yield faster test results for all -j
values greater than 2. For -j3 on my machine, the test suite is a few
seconds faster; I didn't try any larger -j values.
For -j1, I went ahead and changed the Makefile so that, if no -j
option is given, then the "check-single" mode is used. You can still
use "make -j1 check" to get single-job parallel-mode, though of course
there's no good reason to do so.
This change is likely to speed up the plain "make check" scenario a
little as we will now bypass dg-extract-results.sh.
One drawback of this change is that "make -jN check" is now much more
verbose. I generally only look at the .sum and .log files, but
perhaps this will bother some.
Another interesting question is scalability of the result. The
slowest test, which limits the scalability, took 80.78 seconds. The
mean of the remaining tests is 1.08 seconds. (Note that this is just
a rough estimate, since there are still outliers.)
This means we can run 80.78 / 1.08 =~ 74 tests in the time available.
And, in this data set (slightly older than the above, but materially
the same) there were 948 tests. So, I think the current test suite
should scale ok up to about -j12.
We could improve this number if need be by breaking up the biggest
tests.
2013-11-04 Tom Tromey <tromey@redhat.com>
* Makefile.in (TEST_DIRS): Remove.
(TEST_TARGETS, check-parallel): Rewrite.
(check-gdb.%, BASE1_FILES, BASE2_FILES, check-gdb.base%)
(subdir_do, subdirs): Remove.
(do-check-parallel, check/%): New targets.
(clean): Remove outputs, temp, and cache directories.
(saw_dash_j): New variable.
(CHECK_TARGET): Use it.
(check): Depend on all, site.exp. Rewrite.
(check-single): Remove dependencies.
(slow_tests, all_tests, reordered_tests): New variables.
A couple of Fission tests rely on the current directory layout. This
assumption is not valid in parallel mode.
This patch fixes the problem by removing the relative directory from
the .S files and instead having the tests set debug-file-directory
before opening the main file.
2013-11-04 Tom Tromey <tromey@redhat.com>
* gdb.dwarf2/fission-base.S: Remove "gdb.dwarf/".
* gdb.dwarf2/fission-base.exp: Set debug-file-directory
before loading binfile.
* gdb.dwarf2/fission-loclists.S: Remove "gdb.dwarf/".
* gdb.dwarf2/fission-loclists.exp: Set debug-file-directory
before loading binfile.
A few tests run an inferior that execs some other program. The name
of this exec'd program is compiled in. These tests assume the current
test suite directory layout, but fail in parallel mode.
This patch fixes these tests by letting the .exp files pass in the
directory names at compile time.
2013-11-04 Tom Tromey <tromey@redhat.com>
* gdb.base/foll-exec.c (main): Use BASEDIR.
* gdb.base/foll-exec.exp: Define BASEDIR during compilation.
* gdb.base/foll-vfork.c (main): Use BASEDIR.
* gdb.base/foll-vfork.exp: Define BASEDIR during compilation.
* gdb.multi/bkpt-multi-exec.c (main): Use BASEDIR.
* gdb.multi/bkpt-multi-exec.exp: Define BASEDIR during compilation.
argv0-symlink.exp doesn't work properly if standard_output_file puts
files into a per-test subdirectory. That's because it assumes that
files appear in $subdir, which is no longer true.
This patch fixes the problem by computing the correct directory at
runtime.
Tested both with and without GDB_PARALLEL on x86-64 Fedora 18.
2013-11-04 Tom Tromey <tromey@redhat.com>
* gdb.base/argv0-symlink.exp: Compute executable's directory
dynamically.
This introduces a new relative_filename proc to gdb.exp and changes
some tests to use it. This helps make these tests parallel-safe.
2013-11-04 Tom Tromey <tromey@redhat.com>
* gdb.base/fullname.exp: Use standard_output_file,
relative_filename.
* gdb.base/hashline1.exp: Use standard_testfile,
standard_output_file, relative_filename, clean_restart.
* gdb.base/hashline2.exp: Use standard_testfile,
standard_output_file.
* gdb.base/hashline3.exp: Use standard_testfile,
standard_output_file, relative_filename.
* lib/gdb.exp (relative_filename): New proc.
This updates the fileio test to be parallel-safe.
2013-11-04 Tom Tromey <tromey@redhat.com>
* gdb.base/fileio.c (test_open, test_write, test_read)
(test_lseek, test_close, test_stat, test_fstat)
(test_isatty, test_system, test_rename, test_unlink):
Use OUTDIR define.
* gdb.base/fileio.exp: Define OUTDIR during compilation.
Use standard_output_file.
This fixes the "checkpoint" test to use the standard output directory.
This makes the test be parallel-safe.
2013-11-04 Tom Tromey <tromey@redhat.com>
* gdb.base/checkpoint.c (main): Use PI_TXT and COPY1_TXT
defines.
* gdb.base/checkpoint.exp: Define PI_TXT and COPY1_TXT during
compilation. Use prepare_for_testing, standard_output_file.
This makes more changes in gdb.base to make it parallel-safe. I think
the changes in this particular patch are relatively straightforward,
so I've grouped them all together.
2013-11-04 Tom Tromey <tromey@redhat.com>
* gdb.base/advance.exp: Use standard_testfile and
prepare_for_testing.
* gdb.base/bigcore.exp: Use standard_output_file. "cd" to
appropriate directory when local.
* gdb.base/dump.exp: Use standard_output_file. Update all
"dump" and "restore" filenames.
* gdb.base/interact.exp: Use standard_output_file.
* gdb.base/jit-so.exp: Don't download file when local.
* gdb.base/jit.exp (compile_jit_test): Don't download file
when local.
* gdb.base/list.exp: Use gdb_remote_download.
* gdb.base/maint.exp: Use standard_output_file.
* gdb.base/prelink.exp: Use standard_output_file.
* gdb.base/save-bp.exp: Use standard_output_file.
* gdb.base/sepdebug.exp: Use standard_testfile,
standard_output_file.
(test_different_dir): Don't declare objdir.
* gdb.base/solib-search.exp: Use standard_output_file.
* gdb.base/step-line.exp: Use gdb_remote_download.
* gdb.base/trace-commands.exp: Use standard_output_file.
