For some reason, macro_expand_next does not return a
gdb::unique_xmalloc_ptr<char>, like its counterparts macro_expand and
macro_expand_once. This patch fixes that.
macro_buffer::release now returns a gdb::unique_xmalloc_ptr<char> too,
which required updating the other callers. The `.release (). release
()` in macro_stringify looks a bit funny, but it's because one release
is for the macro_buffer, and the other is for the unique ptr.
I removed the ATTRIBUTE_UNUSED_RESULT on macro_buffer::release, I don't
really understand why it's there. I don't see how this method could be
called without using the result, that would be an obvious memory leak.
The commit that introduced it (4e4a8b932b "Add ATTRIBUTE_UNUSED_RESULT
to macro_buffer") doesn't give any details.
gdb/ChangeLog:
* c-exp.y (scan_macro_expansion): Don't free `expansion`.
(lex_one_token): Update.
* macroexp.c (struct macro_buffer) <release>: Return
gdb::unique_xmalloc_ptr<char>.
(macro_stringify): Update.
(macro_expand): Update.
(macro_expand_next): Return gdb::unique_xmalloc_ptr<char>.
* macroexp.h (macro_expand_next): Likewise.
Change-Id: I67a74d0d479d2c20cdc82161ead7c54cea034f56
I started to look into changing the callbacks in macroexp.h to use
gdb::function_view. However, I noticed that the passed lookup function
was always `standard_macro_lookup`, which looks up a macro in a
`macro_scope` object. Since that doesn't look like a very useful
abstraction, it would be simpler to just pass the scope around and have
the various functions call standard_macro_lookup themselves. This is
what this patch does.
gdb/ChangeLog:
* macroexp.h (macro_lookup_ftype): Remove.
(macro_expand, macro_expand_once, macro_expand_next): Remove
lookup function parameters, add scope parameter.
* macroexp.c (scan, substitute_args, expand, maybe_expand,
macro_expand, macro_expand_once, macro_expand_next): Likewise.
* macroscope.h (standard_macro_lookup): Change parameter type
to macro_scope.
* macroscope.c (standard_macro_lookup): Likewise.
* c-exp.y (lex_one_token): Update.
* macrocmd.c (macro_expand_command): Likewise.
(macro_expand_once_command): Likewise.
Change-Id: Id2431b1489359e1b0274dc2b81e5ea5d225d730c
gdb.base/structs2.exp fails to run with Clang, because of:
gdb compile failed, /home/pedro/gdb/mygit/src/gdb/testsuite/gdb.base/structs2.c:16:14: warning:
implicit conversion from 'int' to 'signed char' changes value from 130 to
-126 [-Wconstant-conversion]
param_reg (130, 120, 33000, 32000);
~~~~~~~~~ ^~~
/home/pedro/gdb/mygit/src/gdb/testsuite/gdb.base/structs2.c:16:24: warning:
implicit conversion from 'int' to 'short' changes value from 33000 to
-32536 [-Wconstant-conversion]
param_reg (130, 120, 33000, 32000);
~~~~~~~~~ ^~~~~
2 warnings generated.
=== gdb Summary ===
# of untested testcases 1
Fix it by passing actual negative numbers.
gdb/testsuite/ChangeLog:
* gdb.base/structs2.c (main): Adjust second parem_reg call to
explicitly write negative numbers.
* gdb.base/structs2.exp: Adjust expected output.
gdb.base/charset.exp fails to run with Clang, because of:
gdb compile failed, /home/pedro/gdb/mygit/src/gdb/testsuite/gdb.base/charset.c:144:20: warning:
implicit conversion from 'int' to 'char' changes value from 162 to -94
[-Wconstant-conversion]
11, 162, 17);
^~~
/home/pedro/gdb/mygit/src/gdb/testsuite/gdb.base/charset.c:151:16: warning:
implicit conversion from 'int' to 'char' changes value from 167 to -89
[-Wconstant-conversion]
167,
^~~
/home/pedro/gdb/mygit/src/gdb/testsuite/gdb.base/charset.c:168:16: warning:
implicit conversion from 'int' to 'char' changes value from 167 to -89
[-Wconstant-conversion]
167,
^~~
3 warnings generated.
=== gdb Summary ===
# of untested testcases 1
Fix it by changing init_string to take unsigned char parameters.
gdb/testsuite/ChangeLog:
* gdb.base/charset.c (init_string): Change all char parameters to
unsigned char parameters.
The gdb.base/call-sc.exp, gdb.base/structs.exp and
gdb.base/structs2.exp testcases still try compiling the sources with
-DNO_PROTOTYPES, but the corresponding sources don't have any #ifdef
NO_PROTOTYPES any longer. Those were removed throughout years ago.
OTOH, gdb.base/ovlymgr.h does check for NO_PROTOTYPES, but no .exp
file compiles it with -DNO_PROTOTYPES.
gdb.base/reread.exp and gdb.base/varargs.exp set a 'prototypes'
global, which is a stale bit left behind when the "try-compiling
without and then with -DNO_PROTOTYPES" logic was around.
gdb/testsuite/ChangeLog:
* gdb.base/call-sc.exp (start_scalars_test): Use
prepare_for_testing and don't try compiling with -DNO_PROTOTYPES.
* gdb.base/overlays.c: Remove references to PARAMS.
* gdb.base/ovlymgr.h (PARAMS): Delete, and remove all references.
* gdb.base/reread.exp: Don't set 'prototypes' global.
* gdb.base/structs.exp (start_structs_test): Use
prepare_for_testing and don't try compiling with -DNO_PROTOTYPES.
* gdb.base/structs2.exp: Don't set 'prototypes' global. Use
prepare_for_testing and don't try compiling with -DNO_PROTOTYPES.
Don't issue "set width 0". Remove gdb_stop_suppressing_tests
call.
* gdb.base/varargs.exp: Don't set 'prototypes' global.
D10V support was removed years ago, but the gdb.base/d10vovly.c file
stayed behind. Looking a bit closer, I can't find anywhere that
references gdb.base/m32rovly.c either.
Both gdb.base/m32rovly.c and gdb.base/d10vovly.c seem to be older
copies of gdb.base/ovlymgr.c, that are exactly the same, except for
some cosmetic differences, and for missing _ovly_debug_event. Note
that gdb.base/ovlymgr.c has the #ifdef __M32R__ bits too. Note also
that gdb.base/overlays.exp is currently only supported on m32r, and
that uses ovlymgr.c not gdb.base/m32rovly.c.
gdb/testsuite/ChangeLog:
* gdb.base/d10vovly.c: Delete.
* gdb.base/m32rovly.c: Delete.
* gdb.base/ovlymgr.c: Remove all code guarded by __D10V__.
I noticed that fetch_inferior_event receives the client_data parameter
from its caller, inferior_event_handler, but doesn't actually need it.
This patch removes it. In turn, inferior_event_handler doesn't use its
parameter, so remove it too.
The `data` argument used when registering
remote_async_inferior_event_handler is changed to NULL, to avoid
confusion. It could make people think that the value passed is used
somewhere, when in fact it's not.
gdb/ChangeLog:
* inf-loop.c (inferior_event_handler): Remove client_data param.
* inf-loop.h (inferior_event_handler): Likewise.
* infcmd.c (step_1): Adjust.
* infrun.c (proceed): Adjust.
(fetch_inferior_event): Remove client_data param.
(infrun_async_inferior_event_handler): Adjust.
* infrun.h (fetch_inferior_event): Remove `void *` param.
* linux-nat.c (handle_target_event): Adjust.
* record-btrace.c (record_btrace_handle_async_inferior_event):
Adjust.
* record-full.c (record_full_async_inferior_event_handler):
Adjust.
* remote.c (remote_async_inferior_event_handler): Adjust.
Change-Id: I3c2aa1eb0ea3e0985df096660d2dcd794674f2ea
It seemed cleaner to me for tui_win_info::name to be pure virtual.
This meant adding a name method to the locator window; but this too
seems like an improvement.
gdb/ChangeLog
2020-07-01 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (struct tui_win_info) <name>: Now pure virtual.
* tui/tui-stack.h (struct tui_locator_window) <name>: New method.
This merges the tui_gen_win_info base class with tui_win_info;
renaming the resulting class to tui_win_info.
gdb/ChangeLog
2020-07-01 Tom Tromey <tom@tromey.com>
* tui/tui-wingeneral.c (tui_win_info::refresh_window): Move from
tui_gen_win_info.
(tui_win_info::make_window): Merge with
tui_gen_win_info::make_window.
(tui_win_info::make_visible): Move from tui_gen_win_info.
* tui/tui-win.c (tui_win_info::max_width): Move from
tui_gen_win_info.
* tui/tui-layout.h (class tui_layout_window) <m_window>: Change
type.
<window_factory>: Likewise.
* tui/tui-layout.c (tui_win_info::resize): Move from
tui_gen_win_info.
(make_standard_window): Change return type.
(get_locator_window, tui_get_window_by_name): Likewise.
(tui_layout_window::apply): Remove a cast.
* tui/tui-data.h (MIN_WIN_HEIGHT): Move earlier.
(struct tui_win_info): Merge with tui_gen_win_info.
(struct tui_gen_win_info): Remove.
tui_locator_window is the last remaining concrete child class of
tui_gen_win_info. It seems a bit cleaner to me to flatten the
hierarchy a bit; this patch prepares for that by changing
tui_locator_window to derive from tui_win_info.
gdb/ChangeLog
2020-07-01 Tom Tromey <tom@tromey.com>
* tui/tui-stack.h (struct tui_locator_window): Derive from
tui_win_info.
<do_scroll_horizontal, do_scroll_vertical>: New methods.
<can_box>: New method.
The tui_locator_window constructor initializes the first character of
two of its members. However, this is actually an error, since these
were changed to be std::string. This removes the erroneous code.
gdb/ChangeLog
2020-07-01 Tom Tromey <tom@tromey.com>
* tui/tui-stack.h (struct tui_locator_window): Remove body.
There's no deep reason that tui_data_item_window should derive from
tui_gen_win_info -- it currently uses a curses window to render, but
that isn't truly needed, and it adds some hacks to other parts of the
TUI.
This patch changes tui_data_item_window so that it doesn't have a base
class, and updates the register window. This simplifies the code and
enables a subsequent cleanup.
gdb/ChangeLog
2020-07-01 Tom Tromey <tom@tromey.com>
* tui/tui-regs.c (tui_data_window::display_registers_from)
(tui_data_window::display_registers_from)
(tui_data_window::first_data_item_displayed)
(tui_data_window::delete_data_content_windows): Update.
(tui_data_window::refresh_window, tui_data_window::no_refresh):
Remove.
(tui_data_window::check_register_values): Update.
(tui_data_item_window::rerender): Add parameters. Update.
(tui_data_item_window::refresh_window): Remove.
* tui/tui-data.h (struct tui_gen_win_info) <no_refresh>: No longer
virtual.
* tui/tui-regs.h (struct tui_data_item_window): Don't derive from
tui_gen_win_info.
<refresh_window, max_height, min_height>: Remove.
<rerender>: Add parameters.
<x, y, visible>: New members.
(struct tui_data_window) <refresh_window, no_refresh>: Remove.
<m_item_width>: New member.
tui_data_item_window::item_no is misnamed -- it only can be used for a
register, but it references a "display" number as well. (Based on
other comments I've seen in the past -- most since deleted -- I think
there were plans at one point to display variables in this window as
well. However, this was never implemented.)
This patch renames this member to be more correct.
gdb/ChangeLog
2020-07-01 Tom Tromey <tom@tromey.com>
* tui/tui-regs.c (tui_data_window::show_register_group)
(tui_data_window::check_register_values): Update.
* tui/tui-regs.h (struct tui_data_item_window) <regno>: Rename
from item_no.
tui_data_window::show_register_group had a useless "if" -- the
condition could never be false. This patch removes it.
gdb/ChangeLog
2020-07-01 Tom Tromey <tom@tromey.com>
* tui/tui-regs.c (tui_data_window::show_register_group): Remove
useless "if".
The "name" member of tui_data_window was set, but never used. This
removes it.
gdb/ChangeLog
2020-07-01 Tom Tromey <tom@tromey.com>
* tui/tui-regs.c (tui_data_window::show_register_group): Update.
* tui/tui-regs.h (struct tui_data_item_window) <name>: Remove.
This moves some code out of tui-data.h, to more closely related
places. Some unused forward declarations are also removed.
gdb/ChangeLog
2020-07-01 Tom Tromey <tom@tromey.com>
* tui/tui-stack.c (SINGLE_KEY): Move from tui-data.h
* tui/tui-winsource.h (enum tui_line_or_address_kind)
(struct tui_line_or_address): Move from tui-data.h.
* tui/tui-win.c (DEFAULT_TAB_LEN): Move from tui-data.h.
* tui/tui-data.h (DEFAULT_TAB_LEN): Move to tui-win.c.
(tui_cmd_window, tui_source_window_base, tui_source_window)
(tui_disasm_window): Don't declare.
(enum tui_line_or_address_kind, struct tui_line_or_address): Move
to tui-winsource.h.
(SINGLE_KEY): Move to tui-stack.c.
tui_expand_tabs only has a single caller. This patch removes this
function, in favor of a tab-expanding variant of string_file. This
simplifies the code somewhat.
gdb/ChangeLog
2020-07-01 Tom Tromey <tom@tromey.com>
* tui/tui-regs.h (struct tui_data_item_window) <content>: Now a
std::string.
* tui/tui-regs.c (class tab_expansion_file): New.
(tab_expansion_file::write): New method.
(tui_register_format): Change return type. Use
tab_expansion_file.
(tui_get_register, tui_data_window::display_registers_from)
(tui_data_item_window::rerender): Update.
* tui/tui-io.h (tui_expand_tabs): Don't declare.
* tui/tui-io.c (tui_expand_tabs): Remove.
tui_reggroup_completer has an "XXXX" comment suggesting the use of
complete_on_enum. This patch implements this suggestion.
gdb/ChangeLog
2020-07-01 Tom Tromey <tom@tromey.com>
* tui/tui-regs.c (tui_reggroup_completer): Use complete_on_enum.
LLD from 11 onwards (https://reviews.llvm.org/D81784) uses -1 to
represent a relocation in .debug_line referencing a discarded symbol.
Recognize -1 to fix gdb.base/break-on-linker-gcd-function.exp when the
linker is a newer LLD.
gdb/ChangeLog:
* dwarf2/read.c (lnp_state_machine::check_line_address): Test -1.
Currently, GDB rejects the (die) reference form while it accepts exprloc
form. It is allowed in DWARF standard. "Table 7.5: Attribute encodings"
in DWARF5 standard. Flang compiler assigns (die) reference to
DW_AT_associated and DW_AT_allocated for some cases.
gdb/ChangeLog
* dwarf2/read.c (set_die_type): Removed conditions to restrict
forms for DW_AT_associated and DW_AT_allocated attributes,
which is already checked in function attr_to_dynamic_prop.
Tom de Vries detected that some python tests were broken as
they were still using gdb_py_test_multiple that was replaced
by gdb_test_multiline. Repair these tests by using the new function.
gdb/testsuite/ChangeLog
2020-06-30 Philippe Waroquiers <philippe.waroquiers@skynet.be>
* gdb.python/py-breakpoint.exp: use gdb_test_multiline instead
of gdb_py_test_multiple.
* gdb.python/py-cmd.exp: Likewise.
* gdb.python/py-events.exp: Likewise.
* gdb.python/py-function.exp: Likewise.
* gdb.python/py-inferior.exp: Likewise.
* gdb.python/py-infthread.exp: Likewise.
* gdb.python/py-linetable.exp: Likewise.
* gdb.python/py-parameter.exp: Likewise.
* gdb.python/py-value.exp: Likewise.
Tom de Vries pointed out that some Rust tests were failing after the
variant part rewrite. He sent an executable, which helped track down
this bug.
quirk_rust_enum was passing 1 to alloc_rust_variant in one case.
However, a comment earlier says:
/* We don't need a range entry for the discriminant, but we do
need one for every other field, as there is no default
variant. */
In this case, we must pass -1 for this parameter. That is what this
patch implements.
gdb/ChangeLog
2020-06-30 Tom Tromey <tromey@adacore.com>
* dwarf2/read.c (quirk_rust_enum): Correctly call
alloc_rust_variant for default-less enum.
gdb's copy of basic_string_view includes a to_string method. However,
according to cppreference, this is not a method on the real
std::basic_string_view:
https://en.cppreference.com/w/cpp/string/basic_string_view
This difference matters because gdb_string_view.h will use the
standard implementation when built with a C++17 or later. This caused
PR build/26183.
This patch fixes the problem by changing the method to be a standalone
helper function, and then rewriting the uses. Tested by rebuilding
with a version of GCC that defaults to C++17.
(Note that the build still is not clean; and also I noticed that the
libstdc++ string_view forbids the use of nullptr ... I wonder if gdb
violates that.)
gdb/ChangeLog
2020-06-30 Tom Tromey <tromey@adacore.com>
PR build/26183:
* ada-lang.c (ada_lookup_name_info::ada_lookup_name_info): Use
gdb::to_string.
gdbsupport/ChangeLog
2020-06-30 Tom Tromey <tromey@adacore.com>
PR build/26183:
* gdb_string_view.h (basic_string_view::to_string): Remove.
(gdb::to_string): New function.
In gdb_default_target_compile, we use dejagnu's default_target_compile, unless
we need support for languages that are not yet support in the used dejagnu
version, in which case we use a local default_target_compile:
gdb_default_target_compile_1.
However, there's another reason to use the local default_target_compile: when
early_flags is used, because there's no dejagnu release available yet
supporting this.
Fix this by detecting and handling early_flags in gdb_default_target_compile.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2020-06-30 Tom de Vries <tdevries@suse.de>
PR testsuite/26175
* lib/future.exp (gdb_default_target_compile): Detect and handle
early_flags.
I spotted something that looks wrong in the doc of
gdbarch_displaced_step_copy_insn.
It says that if the function returns NULL, it means that it has emulated
the behavior of the instruction and written the result to REGS.
However, it says below that the function may return NULL to indicate
that the instruction can't be single-stepped out-of-line, in which case
the core steps the instruction in-line. The two are contradictory.
The right one is the latter, if the function returns NULL, the core
falls back to in-line stepping. I checked all the implementations of
this function and they all agree with this.
gdb/ChangeLog:
* gdbarch.sh (displaced_step_copy_insn): Update doc.
* gdbarch.h: Re-generate.
Change-Id: I98163cdd38970cde4c77680e249b10f5d2d5bf9b
This patch makes a few adjustments to the simavr.exp to handle tests
that error out more gracefully. I put all the changes in the same patch
because right now it's in a very bad shape, so it's very painful to do
small incremental adjustements. I found that with these changes, it
becomes reasonably stable.
For example, the gdb.base/step-over-syscall.exp test is a bit buggy
(stuff for another patch...), in that it calls gdb_load (through
clean_restart) with a file that doesn't exist. The gdb_load
implementation in simavr.exp gets called with a file that doesn't exist,
and simavr (expectedly) fails to start.
When this happens, we currently leave the $simavr_spawn_id variable set.
However, because the simavr process has terminated, its spawn id is
closed. When the next test tries to close the previous connection to
simavr, it fails to, and this error is thrown:
ERROR: close: spawn id exp86 not open
while executing
"close -i $simavr_spawn_id"
(procedure "gdb_load" line 18)
invoked from within
In other words, any test ran after step-over-syscall.exp will have
simavr.exp's gdb_load fail on it.
This patch tries to make sure that simavr.exp's gdb_load only leaves
simavr_spawn_id set if everything went fine. If we couldn't start
simavr, don't see the expected output, or fail to connect to it, we
close the spawn id (if needed) and unset simavr_spawn_id.
As an additional precaution, I added a catch around the "close the
previous connection" code. Ideally, this shouldn't be necessary, but I
bet there are other similar scenarios where we might try to close an
already close spawn id. So I think displaying a warning is better than
messing up all following tests.
Also, the board never wait'ed for the simavr process, resulting in tons
of zombie simavr processes hanging around. This patch adds some calls
to "wait" whenever we close the connection (or realize it is already
closed), which seems to fix the problem.
Finally I switched a `gdb_expect` to bare `expect`, where we wait for
the "listening" message from simavr. I found it necessary because
gdb_expect (through remote_expect) adds a `-i <gdb spawn id> -timeout
10`. So if for some reason the GDB process has crashed in the mean
time, this expect will unexpectedly error out with a `spawn id <gdb
spawn id> not open`. Therefore, change `gdb_expect` to `expect` so that
we only deal with simavr's spawn id here.
Here are the results on TESTS="gdb.base/*.exp" before:
# of expected passes 4816
# of unexpected failures 4433
# of known failures 2
# of unresolved testcases 300
# of untested testcases 143
# of unsupported tests 7
# of paths in test names 2
# of duplicate test names 10
and after:
# of expected passes 21957
# of unexpected failures 1564
# of expected failures 8
# of unknown successes 1
# of known failures 31
# of unresolved testcases 532
# of untested testcases 153
# of unsupported tests 28
# of paths in test names 2
# of duplicate test names 32
I tested using simavr's current master (7c254e2081b5).
gdb/testsuite/ChangeLog:
* boards/simavr.exp (gdb_load): Catch errors when closing
previous connection. Close connection, wait for process and
unset simavr_spawn_id on failure.
Change-Id: I695fc765e1c22e1d1dc0b08e0f5141244986b868
Since commit 26783bce15 "[gdb/testsuite] Don't abort testrun for invalid
command in test-case" we don't abort the testrun when encountering an invalid
command. However, since we don't report errors in the summary, there's a
chance that the error goes unnoticed.
Make the invalid command error more visible by marking the test-case
unresolved, such that we have f.i.:
...
PASS: gdb.python/py-breakpoint.exp: test_bkpt_internal: Test watchpoint write
UNRESOLVED: gdb.python/py-breakpoint.exp: test_bkpt_eval_funcs: \
testcase aborted due to invalid command name: gdb_py_test_multiple
ERROR: tcl error sourcing py-breakpoint.exp.
ERROR: invalid command name "gdb_py_test_multiple"
while executing
...
=== gdb Summary ===
nr of expected passes 56
nr of unresolved testcases 1
...
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2020-06-29 Tom de Vries <tdevries@suse.de>
* lib/gdb.exp (unknown): Make test-case unresolved.
Before commit a8654e7d78 'Fixes PR 25475: ensure exec-file-mismatch "ask"
always asks in case of mismatch', there was a difference in behaviour in
test-case gdb.server/solib-list.exp.
If the executable did not contain debug info (as is usually the case), gdb
would detect a mismatch but not ask for confirmation:
...
(gdb) target remote localhost:2346^M
Remote debugging using localhost:2346^M
warning: Mismatch between current exec-file solib-list^M
and automatically determined exec-file /lib64/ld-2.26.so^M
exec-file-mismatch handling is currently "ask"^M
Reading symbols from /lib64/ld-2.26.so...^M
Reading symbols from /usr/lib/debug/lib64/ld-2.26.so.debug...^M
0x00007ffff7dd7ea0 in _start () at rtld.c:745^M
745 }^M
(gdb) PASS: gdb.server/solib-list.exp: non-stop 0: target remote
...
If the executable did contain debug info (as happens to be the case for
openSUSE), gdb would detect a mismatch and ask for confirmation:
...
(gdb) PASS: gdb.server/solib-list.exp: non-stop 0: file binfile
target remote localhost:2346^M
Remote debugging using localhost:2346^M
warning: Mismatch between current exec-file solib-list^M
and automatically determined exec-file /lib64/ld-2.26.so^M
exec-file-mismatch handling is currently "ask"^M
Load new symbol table from "/lib64/ld-2.26.so"? (y or n) y^M
Reading symbols from /lib64/ld-2.26.so...^M
Reading symbols from /usr/lib/debug/lib64/ld-2.26.so.debug...^M
0x00007ffff7dd7ea0 in _start () at rtld.c:745^M
745 }^M
(gdb) PASS: gdb.server/solib-list.exp: non-stop 0: target remote
...
After commit a8654e7d78, the confirmation is now also asked in case there's
no debug info.
Tighten the test-case by verifying that the confirmation question is asked, as
suggested in the log message of commit a8654e7d78:
...
we can remove the bypass introduced by Tom in 6b9374f1, in order to always
answer to the 'load' question.
...
gdb/testsuite/ChangeLog:
2020-06-29 Tom de Vries <tdevries@suse.de>
PR gdb/25475
* gdb.server/solib-list.exp: Verify that the symbol reload
confirmation question is asked.
The cmd_type function only has a single caller, which is in the CLI
implementation code. This patch removes the function, and moves the
cmd_types enum definition from command.h to cli-decode.h, fixing an 18
year old FIXME.
gdb/ChangeLog
2020-06-28 Tom Tromey <tom@tromey.com>
* command.h (cmd_types): Remove.
(cmd_type): Don't declare.
* cli/cli-decode.h (enum cmd_types): Uncomment. No longer a
typedef.
* cli/cli-cmds.c (setting_cmd): Use cmd->type directly.
* cli/cli-decode.c (cmd_type): Remove.
This converts the get_inferior_io_terminal and
set_inferior_io_terminal free functions to inferior methods.
Since the related commands are called "tty", "{set,show}
inferior-tty", and MI's "-inferior-tty-{set,show}", to make the
connection between the commands and the code more obvious, the methods
are named set_tty/tty instead of set_io_terminal/io_terminal.
gdb/ChangeLog:
* fork-child.c (prefork_hook): Adjust.
* infcmd.c (set_inferior_io_terminal, get_inferior_io_terminal):
Delete.
(set_inferior_tty_command, show_inferior_tty_command): Adjust.
* inferior.c (inferior::set_tty, inferior::tty): New methods.
* inferior.h (set_inferior_io_terminal, get_inferior_io_terminal):
Remove declarations.
(struct inferior) <set_tty, tty>: New methods.
(struct inferior) <terminal>: Rename to ...
(struct inferior) <m_terminal>: ... this and make private.
* main.c (captured_main_1): Adjust.
* mi/mi-cmd-env.c (mi_cmd_inferior_tty_set): Adjust.
(mi_cmd_inferior_tty_show): Adjust.
* nto-procfs.c (nto_procfs_target::create_inferior): Adjust.
* windows-nat.c (windows_nat_target::create_inferior): Adjust.
Version 2, handles the comments of Simon and Pedro.
Note that gdb_test_multiline and gdb_py_test_multiple are using
the "input line" as the test name, and so when there is a duplicated
input line (such as a line containing "end"), we have duplicated test
names => as gdb_test_multiline and gdb_py_test_multiple are identical,
as indicated in FIXME, move this to gdb.exp, and make the test name unique
by adding the inputnr to the pass message for each input.
2020-06-26 Philippe Waroquiers <philippe.waroquiers@skynet.be>
* lib/gdb.exp (gdb_test_multiline): New, moved from gdb-guile.exp,
have a input seq nr in each pass message.
* lib/gdb-guile.exp (gdb_test_multiline): Move to gdb.exp.
* lib/gdb-python.exp (gdb_py_test_multiple): Remove.
* gdb.python/python.exp: Make test names unique,
use gdb_test_multiline instead of gdb_py_test_multiple,
use $gdb_test_name.
* gdb.guile/guile.exp: Make test names unique, use $gdb_test_name
This fixes test runs and compilation when --disable-libctf,
--disable-static, or --enable-shared are passed.
Changes since v2: Use GCC_ENABLE and fix indentation. Fix prototype
using 'void'. Use 'unsupported' and gdb_caching_proc.
Changes since v3: Adapt to upstream changes providing skip_ctf_tests.
Changes since v4: Adapt to upstream changes in the seven months (!)
since I last looked at this.
gdb/ChangeLog
* configure.ac: Add --enable-libctf: handle --disable-static
properly.
* acinclude.m4: sinclude ../config/enable.m4.
* Makefile.in (aclocal_m4_deps): Adjust accordingly.
(LIBCTF): Substitute in.
(CTF_DEPS): New, likewise.
(CLIBS): libctf needs symbols from libbfd: move earlier.
(CDEPS): Use CTF_DEPS, not LIBCTF, now LIBCTF can include rpath
flags.
* ctfread.c: Surround in ENABLE_LIBCTF.
(elfctf_build_psymtabs) [!ENABLE_LIBCTF]: New stub.
* configure: Regenerate.
* config.in: Likewise.
gdb/testsuite/ChangeLog
* configure.ac: Add --enable-libctf.
* aclocal.m4: sinclude ../config/enable.m4.
* Makefile.in (site.exp): Add enable_libctf to site.exp.
* lib/gdb.exp (skip_ctf_tests): Use it.
* gdb.base/ctf-constvars.exp: Error message tweak.
* gdb.base/ctf-ptype.exp: Likewise.
* configure: Regenerate.
Clang fails to compile the file gdb/testsuite/gdb.cp/try_catch.cc
with the following error:
warning: result of comparison against a string literal is
unspecified (use strncmp instead) [-Wstring-compare]
This commit fixes the error, replacing the pointer comparison with
a call to strcmp. This commit also adds a final check: the test
program is run to the final return statement, and the value of
"test" is checked to ensure it is still "true" at that point.
gdb/testsuite/ChangeLog:
* gdb.cp/try_catch.cc: Include string.h.
(main): Replace comparison against string literal with
strcmp, avoiding build failure with -Wstring-compare.
Add "marker test-complete".
* gdb.cp/try_catch.exp: Run the test to the above marker,
then verify that the value of "test" is still true.
gdb/doc/ChangeLog:
2020-06-26 Eli Zaretskii <eliz@gnu.org>
* gdb.texinfo (Shell Commands): More accurate description of use
of $SHELL. Reported by Sandra Loosemore <sandra@codesourcery.com>.
This changes the inferior::terminal field to be a unique pointer, so its
deallocation is automatically managed.
gdb/ChangeLog:
* inferior.h (struct inferior) <terminal>: Change type to
gdb::unique_xmalloc_ptr<char>.
* inferior.c (inferior::~inferior): Don't free inf->terminal.
* infcmd.c (set_inferior_io_terminal): Don't free terminal
field, adjust to unique pointer.
(get_inferior_io_terminal): Adjust to unique pointer.
Change-Id: Iedb6459b4f9eeae812b0cb9d514b5707d5107cdb
Currently the 'info all-registers' command only loops over those
registers that are known to GDB. Any registers that are unknown, that
is, are mentioned in the target description, but are not something GDB
otherwise knows, will not be displayed.
This feels wrong, so this commit fixes this mistake. The output of
'info all-registers' now matches 'info registers all'.
gdb/ChangeLog:
* riscv-tdep.c (riscv_print_registers_info): Loop over all
registers, not just the known core set of registers.
gdb/testsuite/ChangeLog:
* gdb.arch/riscv-tdesc-regs.exp: New test cases.
Making use of the previous commit, record information about unknown
registers in the target description, and use this to resolve two
issues.
1. Some targets (QEMU) are reporting three register fflags, frm, and
fcsr, twice, once in the FPU feature, and once in the CSR feature.
GDB does create two registers with identical names, but this
is (sort of) fine, we only ever use the first one, and as both
registers access the same target state things basically work OK.
The only real problem is that the register names show up twice in
'info registers all' output.
In this commit we spot the duplicates of these registers and then
return NULL when asked for the name of these registers. This
causes GDB to hide these registers from the user, fixing this
problem.
2. Some targets (QEMU) advertise CSRs that GDB then can't read. The
problem is these targets also say these CSRs are part of the
save/restore register groups.
This means that before an inferior call GDB tries to save all of
these CSRs, and a failure to read one causes the inferior call to
be abandoned.
We already work around this issue to some degree, known CSRs are
removed from the save/restore groups, despite what the target might
say. However, any unknown CSRs are (currently) not removed in this
way.
After this commit we keep a log of the register numbers for all
unknown CSRs, then when asked about the register groups, we
override the group information for unknown CSRs, removing them from
the save and restore groups.
gdb/ChangeLog:
* riscv-tdep.c (riscv_register_name): Return NULL for duplicate
fflags, frm, and fcsr registers.
(riscv_register_reggroup_p): Remove unknown CSRs from save and
restore groups.
(riscv_tdesc_unknown_reg): New function.
(riscv_gdbarch_init): Pass riscv_tdesc_unknown_reg to
tdesc_use_registers.
* riscv-tdep.h (struct gdbarch_tdep): Add
unknown_csrs_first_regnum, unknown_csrs_count,
duplicate_fflags_regnum, duplicate_frm_regnum, and
duplicate_fcsr_regnum fields.
gdb/testsuite/ChangeLog:
* gdb.arch/riscv-tdesc-regs.exp: Extend test case.
This commit adds a new step to the processing of a target description
done in tdesc_use_registers, this new step is about how unknown
registers are processed.
Currently an architecture looks through the target description and
calls tdesc_numbered_register for each register is was expecting (or
hoping) to find. This builds up a map from GDB's register numbers to
the tdesc_reg object. Later the architecture calls
tdesc_use_registers.
In tdesc_use_registers we build a hash with keys being all the
tdesc_reg object pointers, from this hash we remove all of the
tdesc_reg objects that were assigned register numbers using
tdesc_numbered_register.
Finally we walk through all of the tdesc_reg objects, and if it was
not already assigned a number we assign that register the next
available number.
The problem with this is that the architecture has no visibility of
which unknown registers exist, and which tdesc_feature the register
came from, in some cases this might be important.
For example, on RISC-V GDB overrides the use of
tdesc_register_reggroup_p, with riscv_register_reggroup_p to modify
some of the register group choices. In this function GDB wants to
treat all registers from a particular feature in a certain way. This
is fine for registers that GDB knows might be in that feature, but for
unknown registers the RISC-V parts of GDB have no easy way to figure
out which unknown registers exist, and what numbers they were
assigned.
We could figure this information out by probing the register
structures after calling tdesc_use_registers, but this would be
horrible, much better to have tdesc_use_registers tell the
architecture about unknown registers.
This is what this commit does. A new phase of tdesc_use_registers,
just before the unknown registers are assigned a number, we loop over
each tdesc_reg object, if it has not been assigned a number then we
figure out what number would be assigned and then call back into the
architecture passing the tdesc_feature, register name, and the
proposed register number.
The architecture is free to return the proposed register number, or it
can return a different number (which has a result identical to having
called tdesc_numbered_register). Alternatively the architecture can
return -1 to indicate the register should be numbered later.
After calling the callback for every tdesc_reg object any registers
still don't have a number assigned (because the architecture returned
-1), then a new register number is assigned, which might be different
from the proposed number that was suggested earlier.
This commit adds the general target-description parts of this
mechanism. No targets are currently using this code. The RISC-V
target will make use of this in the next commit.
There should be no user visible changes after this commit.
gdb/ChangeLog:
* target-descriptions.c (tdesc_use_registers): Add new parameter a
callback, use the callback (when not null) to help number unknown
registers.
* target-descriptions.h (tdesc_unknown_register_ftype): New typedef.
(tdesc_use_registers): Add extra parameter to declaration.
For the RISC-V target it is desirable if the three floating pointer
status CSRs fflags, frm, and fcsr can be placed into either the FPU
feature or the CSR feature. This allows different targets to build
the features in a way that better reflects their target.
The change to support this within GDB is fairly simple, so this is
done in this commit, and some tests added to check this new
functionality.
gdb/ChangeLog:
* riscv-tdep.c (value_of_riscv_user_reg): Moved to here from later
in the file.
(class riscv_pending_register_alias): Likewise.
(riscv_register_feature::register_info): Change 'required_p' field
to 'required', and change its type. Add 'check' member function.
(riscv_register_feature::register_info::check): Define new member
function.
(riscv_xreg_feature): Change initialisation of 'required' field.
(riscv_freg_feature): Likewise.
(riscv_virtual_feature): Likewise.
(riscv_csr_feature): Likewise.
(riscv_check_tdesc_feature): Take extra parameter, the csr
tdesc_feature, rewrite the function to use the new
riscv_register_feature::register_info::check function.
(riscv_gdbarch_init): Pass the csr tdesc_feature where needed.
gdb/testsuite/ChangeLog:
* gdb.arch/riscv-tdesc-loading-01.xml: New file.
* gdb.arch/riscv-tdesc-loading-02.xml: New file.
* gdb.arch/riscv-tdesc-loading-03.xml: New file.
* gdb.arch/riscv-tdesc-loading-04.xml: New file.
* gdb.arch/riscv-tdesc-loading.exp: New file.
There is currently a bug in the RISC-V CSR/FPU feature files. The
CSRs containing the FPU status registers are mentioned in both the FPU
feature file and the CSR feature file.
My original thinking when adding the FPU feature file was that it made
more sense to group the FPU status registers with the other FPU
state. This opened up the possibility of debugging very
simple (possibly simulator only) targets that had little more than CPU
and FPU available for GDB to access.
When I then added code to automatically generate the CSR XML file I
forgot to filter out the FPU status CSRs, so these registers were
mentioned twice.
Now for GDB's default RISC-V target descriptions this doesn't actually
matter. I did consider adding the CSRs to the default target
description, but in the end I didn't bother. The reasoning again was
simplicity; the default target description is only to be used when the
target doesn't supply its own description, and NOT supplying the CSRs
actually serves to encourage targets to supply an accurate
description. Combine this with the fact that the CSRs change from
revision to revision, sometimes in non-backward compatible ways, then
having a "default" set of CSRs just feels like a path to confusion and
complaints.
However, having a broken CSR XML file in the GDB source tree has had
one negative effect, QEMU has copied this file into its source tree,
and is using this as its description that it passes to GDB. That is
QEMU announces the FPU status registers twice, once in the FPU
feature, and once in the CSR feature.
This commit starts along the path back to sanity by deleting the
default CSR XML files from within GDB. These files were not used in
any way by current GDB, so there is absolutely no loss of
functionality with this change.
gdb/ChangeLog:
* features/Makefile: Remove all references to the deleted files
below.
* features/riscv/32bit-csr.c: Deleted.
* features/riscv/32bit-csr.xml: Deleted.
* features/riscv/64bit-csr.c: Deleted.
* features/riscv/64bit-csr.xml: Deleted.
* features/riscv/rebuild-csr-xml.sh: Deleted.
First, consider the RISC-V register $x1. This register has an alias
$ra. When GDB processes an incoming target description we allow the
target to use either register name to describe the target.
However, within GDB's UI we want to use the $ra alias in preference to
the $x1 architecture name.
To achieve this GDB overrides the tdesc_register_name callback with
riscv_register_name. In riscv_register_name we ensure that we always
return the preferred name, so in this case "ra".
To ensure the user can still access the register as $x1 if they want
to, when in riscv_check_tdesc_feature we spot that the target has
supplied the register, we add aliases for every name except the
preferred one, so in this case we add the alias "x1".
This scheme seems to work quite well, the targets have the flexibility
to be architecture focused if they wish (using x0 - x31) while GDB is
still using the ABI names ra, sp, gp, etc.
When this code was originally added there was an attempt made to
include the CSRs in the same scheme. At the time the CSRs only had
two names, one pulled from riscv-opc.h, and one generated in GDB that
had the pattern csr%d.
The idea was that if the remote targets description described the CSRs
as csr%d then GDB would rename these back to the real CSR name. This
code was only included because if followed the same pattern as the
x-regs and f-regs, not because I was actually aware of any target that
did this.
However, recent changes to add additional CSR aliases has made me
rethink the position here.
Lets consider the CSR $dscratch0. This register has an alias
'csr1970' (1970 is 0x7b2, which is the offset of the CSR register into
the CSR address space). However, this register was originally called
just 'dscratch', and so, after recent commits, this register also has
the alias 'dscratch'.
As the riscv-opc.h file calls this register 'dscratch0' GDB's
preferred name for this register is 'dscratch0'.
So, if the remote target description includes the register
'dscratch0', then GDB will add the aliases 'dscratch', and 'csr1970'.
In the UI GDB will describe the register as 'dscratch0', and all it
good.
The problem I see in this case is where the target describes the
register as 'dscratch'. In this case GDB will still spot the register
and add the aliases 'dscratch', and 'csr1970', GDB will then give the
register the preferred name 'dscratch0'.
I don't like this. For the CSRs I think that we should stick with the
naming scheme offered by the remote target description. As the RISC-V
specification evolves and CSR register names evolve, insisting on
referring to registers by the most up to date name makes it harder for
a target to provide a consistent target description for an older
version of the RISC-V architecture spec.
In this precise case the target offers 'dscratch', which is from an
older version of the RISC-V specification, the newer version of the
spec has two registers 'dscratch0' and 'dscratch1'. If we insist on
using 'dscratch0' it is then a little "weird" (or seems so to me) when
'dscratch1' is missing.
This patch makes a distinction between the x and f registers and the
other register sets. For x and f we still make use of the renaming
scheme, forcing GDB to prefer the ABI name. But after this patch the
CSR register group, and also the virtual register group, will always
prefer to use the name given in the target description, adding other
names as aliases, but not making any other name the preferred name.
gdb/ChangeLog:
* riscv-tdep.c (struct riscv_register_feature::register_info): Fix
whitespace error for declaration of names member variable.
(struct riscv_register_feature): Add new prefer_first_name member
variable, and fix whitespace error in declaration of registers.
(riscv_xreg_feature): Initialize prefer_first_name field.
(riscv_freg_feature): Likewise.
(riscv_virtual_feature): Likewise.
(riscv_csr_feature): Likewise.
(riscv_register_name): Expand on comments. Remove register name
modifications for CSR and virtual registers.
gdb/testsuite/ChangeLog:
* gdb.arch/riscv-tdesc-regs.exp: Extend test case.
This commit does two things:
1. Makes use of the DECLARE_CSR_ALIAS definitions in riscv-opc.h to
add additional aliases for CSRs.
2. Only creates aliases for registers that are actually present on
the target (as announced in the target XML description).
This means that the 'csr%d' aliases that exist will only be created
for those CSRs the target actually has, which is a nice improvement,
as accessing one of the CSRs that didn't exist would cause GDB to
crash with this error:
valprint.c:1560: internal-error: bool maybe_negate_by_bytes(const gdb_byte*, unsigned int, bfd_endian, gdb::byte_vector*): Assertion `len > 0' failed.
When we look at the DECLARE_CSR_ALIAS lines in riscv-opc.h, these can
be split into three groups:
DECLARE_CSR_ALIAS(misa, 0xf10, CSR_CLASS_I, PRIV_SPEC_CLASS_1P9, PRIV_SPEC_CLASS_1P9P1)
The 'misa' register used to exist of offset 0xf10, but was moved to
its current offset (0x301) in with privilege spec 1.9.1. We don't
want GDB to create an alias called 'misa' as we will already have a
'misa' register created by the DECLARE_CSR(misa ....) call earlier in
riscv-opc.h
DECLARE_CSR_ALIAS(ubadaddr, CSR_UTVAL, CSR_CLASS_I, PRIV_SPEC_CLASS_1P9, PRIV_SPEC_CLASS_1P10)
DECLARE_CSR_ALIAS(sbadaddr, CSR_STVAL, CSR_CLASS_I, PRIV_SPEC_CLASS_1P9, PRIV_SPEC_CLASS_1P10)
DECLARE_CSR_ALIAS(sptbr, CSR_SATP, CSR_CLASS_I, PRIV_SPEC_CLASS_1P9, PRIV_SPEC_CLASS_1P10)
DECLARE_CSR_ALIAS(mbadaddr, CSR_MTVAL, CSR_CLASS_I, PRIV_SPEC_CLASS_1P9, PRIV_SPEC_CLASS_1P10)
DECLARE_CSR_ALIAS(mucounteren, CSR_MCOUNTINHIBIT, CSR_CLASS_I, PRIV_SPEC_CLASS_1P9, PRIV_SPEC_CLASS_1P10)
These aliases are all CSRs that were removed in privilege spec 1.10,
and whose addresses were reused by new CSRs. The names meaning of the
old names is totally different to the new CSRs that have taken their
place. I don't believe we should add these as aliases into GDB. If
the new CSR exists in the target then that should be enough.
DECLARE_CSR_ALIAS(dscratch, CSR_DSCRATCH0, CSR_CLASS_I, PRIV_SPEC_CLASS_1P9, PRIV_SPEC_CLASS_1P11)
In privilege spec 1.11 the 'dscratch' register was renamed to
'dscratch0', however the meaning of the register didn't change.
Adding the 'dscratch' alias makes sense I think.
Looking then at the final PRIV_SPEC_CLASS_* field for each alias then
we can see that currently we only want to take the alias from
PRIV_SPEC_CLASS_1P11. For now then this is what I'm using to filter
the aliases within GDB.
In the future there's no telling how DECLARE_CSR_ALIAS will be used.
I've heard it said that future RISC-V privilege specs will not reuse
CSR offsets again. But it could happen. We just don't know.
If / when it does we may need to revisit how aliases are created for
GDB, but for now this seems to be OK.
gdb/ChangeLog:
* riscv-tdep.c (riscv_create_csr_aliases): Handle csr aliases from
riscv-opc.h.
(class riscv_pending_register_alias): New class.
(riscv_check_tdesc_feature): Take vector of pending aliases and
populate it as appropriate.
(riscv_setup_register_aliases): Delete.
(riscv_gdbarch_init): Create vector of pending aliases and pass it
to riscv_check_tdesc_feature in all cases. Use the vector to
create the register aliases.
gdb/testsuite/ChangeLog:
* gdb.arch/riscv-tdesc-regs-32.xml: New file.
* gdb.arch/riscv-tdesc-regs-64.xml: New file.
* gdb.arch/riscv-tdesc-regs.c: New file.
* gdb.arch/riscv-tdesc-regs.exp: New file.
gdbarch_static_transform_name is completely Solaris-specific or rather
specific to the Studio compilers. Studio cc has deprecated Stabs support
in the 12.4 release back in 2015, GCC has defaulted to DWARF-2 on Solaris
7+ since 2004 and Stabs themselves are pretty much obsolete, so the whole
code can go.
Tested on sparcv9-sun-solaris2.11 and x86_64-pc-linux-gnu with
--enable-targets=all.
* sol2-tdep.c (sol2_static_transform_name): Remove.
(sol2_init_abi): Don't register it.
* gdbarch.sh (static_transform_name): Remove.
* gdbarch.c, gdbarch.h: Regenerate.
* dbxread.c (read_dbx_symtab) <'S'>: Remove call to
gdbarch_static_transform_name.
* mdebugread.c (parse_partial_symbols) <'S'>: Likewise.
* stabsread.c (define_symbol) <'X'>: Remove.
(define_symbol) <'S'>: Remove gdbarch_static_transform_name
handling.
<'V'>: Likewise.
* xcoffread.c (scan_xcoff_symtab): Remove gdbarch.
<'S'>: Remove call to gdbarch_static_transform_name.
The gdb.mi/mi-exec-run.exp test never completed/timed out on Solaris for
quite some time:
FAIL: gdb.mi/mi-exec-run.exp: inferior-tty=main: mi=main: force-fail=1: run failure detected (timeout)
This is for gdb trying to exec mi-exec-run.nox, a copy of mi-exec-run
with execute permissions removed.
The process tree at this point looks like this:
21254 /vol/gcc/bin/expect -- /vol/gcc/share/dejagnu/runtest.exp GDB_PARALLEL=yes --outdir=outputs/gdb.mi/mi-exec-run-vfork gdb.mi/mi-exec-run.exp
21300 <defunct>
21281 <defunct>
21294 $obj/gdb/testsuite/../../gdb/gdb -nw -nx -data-directory $obj/gdb/testsuite/../data-directory -i=mi
21297 $obj/gdb/testsuite/../../gdb/gdb -nw -nx -data-directory $obj/gdb/testsuite/../data-directory -i=mi
The parent gdb hangs here:
21294: $obj/gdb/testsuite/../../gdb/gdb -nw
------------ lwp# 1 / thread# 1 ---------------
0000000000000000 SYS#0 ()
0000000000daeccd procfs_target::create_inferior(char const*, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, char**, int) () + 97 (procfs.c:2853)
0000000000ca63a7 run_command_1(char const*, int, run_how) () + 349 (basic_string.h:187)
0000000000ca6516 start_command(char const*, int) () + 26 (infcmd.c:584)
0000000000b3ca8e do_const_cfunc(cmd_list_element*, char const*, int) () + f (cli-decode.c:96)
0000000000b3ed77 cmd_func(cmd_list_element*, char const*, int) () + 32 (cli-decode.c:2113)
0000000000f2d219 execute_command(char const*, int) () + 455 (top.c:657)
0000000000d4ad77 mi_execute_cli_command(char const*, int, char const*) () + 242 (basic_string.h:187)
0000000000d4ae80 mi_cmd_exec_run(char const*, char**, int) () + ba (mi-main.c:473)
with these process flags
21294: $obj/gdb/testsuite/../../gdb/gdb -nw
data model = _LP64 flags = VFORKP|ORPHAN|MSACCT|MSFORK
sigpend = 0x00004103,0x00000000,0x00000000
/1: flags = 0
sigmask = 0xffbffeff,0xffffffff,0x000000ff
cursig = SIGKILL
/2: flags = DETACH|STOPPED|ASLEEP lwp_park(0x0,0x0,0x0)
why = PR_SUSPENDED
sigmask = 0x000a2002,0x00000000,0x00000000
[...]
while the child sits at
21297: $obj/gdb/testsuite/../../gdb/gdb -nw
00007fffbf078a0b execve (7fffbffff756, 7fffbfffec58, 7fffbfffec90, 0)
00007fffbef84cf6 execvpex () + f6
00007fffbef84f45 execvp () + 15
0000000000d60a44 fork_inferior(char const*, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, char**, void (*)(), gdb::function_view<void (int)>, void (*)(), char const*, void (*)(char const*, char* const*, char* const*)) () + 47f (fork-inferior.c:423)
0000000000daeccd procfs_target::create_inferior(char const*, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, char**, int) () + 97 (procfs.c:2853)
0000000000ca63a7 run_command_1(char const*, int, run_how) () + 349 (basic_string.h:187)
0000000000ca6516 start_command(char const*, int) () + 26 (infcmd.c:584)
0000000000b3ca8e do_const_cfunc(cmd_list_element*, char const*, int) () + f (cli-decode.c:96)
0000000000b3ed77 cmd_func(cmd_list_element*, char const*, int) () + 32 (cli-decode.c:2113)
0000000000f2d219 execute_command(char const*, int) () + 455 (top.c:657)
0000000000d4ad77 mi_execute_cli_command(char const*, int, char const*) () + 242 (basic_string.h:187)
0000000000d4ae80 mi_cmd_exec_run(char const*, char**, int) () + ba (mi-main.c:473)
with
21297: $obj/gdb/testsuite/../../gdb/gdb -nw
data model = _LP64 flags = MSACCT|MSFORK
exitset = 0x00000000 0x04000000 0x00000000 0x00000000
0x00000000 0x00000000 0x00000000 0x00000000
/1: flags = STOPPED|ISTOP execve(0x7fffbffff756,0x7fffbfffec58,0x7fffbfffec90,0x0)
why = PR_SYSEXIT what = execve
We have a deadlock here: the execve in the child cannot return until the
parent has handled the PR_SYSEXIT while the parent cannot run with a
vfork'ed child as documented in proc(4):
The child of a vfork(2) borrows the parent's address space. When a
vfork(2) is executed by a traced process, all watched areas established
for the parent are suspended until the child terminates or performs an
exec(2). Any watched areas established independently in the child are
cancelled when the parent resumes after the child's termination or
exec(2). PCWATCH fails with EBUSY if applied to the parent of a
vfork(2) before the child has terminated or performed an exec(2). The
PR_VFORKP flag is set in the pstatus structure for such a parent
process.
In that situation, the parent cannot be killed even with SIGKILL (as
runtest will attempt once the timeout occurs; the pending signal can be
seen in the pflags output above), so the whole test hangs until one
manually kills the child process.
Fortunately, there's an easy way out: when using fork instead of vfork,
the problem doesn't occur, and this is what the current patch does: it
calls fork_inferior with a dummy pre_trace_fun arg.
Tested on amd64-pc-solaris2.11 and sparcv9-sun-solaris2.11.
* procfs.c (procfs_pre_trace): New function.
(procfs_target::create_inferior): Pass it to fork_inferior.
Linux/sparc* currently links Solaris-specific files (sparc-sol2-tdep.o,
sparc64-sol2-tdep.o, sol2-tdep.o) for no apparent reason. It has no
business doing so, and none of the functions/variables defined there are
used explicitly. If support for the Solaris OSABI were desired, this
should be done using --enable-targets instead.
Since neither sparc{32,64}_sol2_init_abi currently declared in common
headers (sparc*-tdep.h) are used outside their source files, they are made
static and the declarations removed.
Tested on sparcv9-sun-solaris2.11 and sparc64-unknown-linux-gnu.
* configure.tgt <sparc-*-linux*> (gdb_target_obs): Remove
sparc-sol2-tdep.o, sol2-tdep.o, sparc64-sol2-tdep.o.
<sparc64-*-linux*> (gdb_target_obs): Remove sparc64-sol2-tdep.o,
sol2-tdep.o, sparc-sol2-tdep.o.
* sparc-sol2-tdep.c (sparc32_sol2_init_abi): Make static.
* sparc-tdep.h (sparc32_sol2_init_abi): Remove.
* sparc64-sol2-tdep.c (sparc64_sol2_init_abi): Make static.
* sparc64-tdep.h (sparc64_sol2_init_abi): Remove.
There's some overlap and duplication between 32 and 64-bit Solaris/SPARC
and x86 tdep files, in particular
sol2_core_pid_to_str
*_sol2_sigtramp_p
sol2_skip_solib_resolver
*_sol2_static_transform_name (forgotten on amd64)
set_gdbarch_sofun_address_maybe_missing (likewise)
This patch avoids this by centralizing common code in sol2-tdep.c.
While sparc_sol2_pc_in_sigtramp and sparc_sol2_static_transform_name
were declared in the shared sparc-tdep.h, they were only used in Solaris
files.
Tested on amd64-pc-solaris2.11, i386-pc-solaris2.11,
sparcv9-sun-solaris2.11, and sparc-sun-solaris2.11, and
sparc64-unknown-linux-gnu.
* amd64-sol2-tdep.c (amd64_sol2_sigtramp_p): Remove.
(amd64_sol2_init_abi): Use sol2_sigtramp_p.
Call sol2_init_abi.
Remove calls to set_gdbarch_skip_solib_resolver,
set_gdbarch_core_pid_to_str.
* i386-sol2-tdep.c (i386_sol2_sigtramp_p): Remove.
(i386_sol2_static_transform_name): Remove.
(i386_sol2_init_abi): Call sol2_init_abi.
Remove calls to set_gdbarch_sofun_address_maybe_missing,
set_gdbarch_static_transform_name,
set_gdbarch_skip_solib_resolver, set_gdbarch_core_pid_to_str.
Use sol2_sigtramp_p.
* sol2-tdep.c (sol2_pc_in_sigtramp): New function.
(sol2_sigtramp_p): New function.
(sol2_static_transform_name): New function.
(sol2_skip_solib_resolver, sol2_core_pid_to_str): Make static.
(sol2_init_abi): New function.
* sol2-tdep.h (sol2_sigtramp_p, sol2_init_abi): Declare.
(sol2_skip_solib_resolver, sol2_core_pid_to_str): Remove.
* sparc-sol2-tdep.c (sparc_sol2_pc_in_sigtramp): Remove.
(sparc32_sol2_sigtramp_frame_sniffer): Just call sol2_sigtramp_p.
(sparc_sol2_static_transform_name): Remove.
(sparc32_sol2_init_abi): Call sol2_init_abi.
Remove calls to set_gdbarch_sofun_address_maybe_missing,
set_gdbarch_static_transform_name,
set_gdbarch_skip_solib_resolver,
set_gdbarch_core_pid_to_str.
* sparc-tdep.h (sparc_sol2_pc_in_sigtramp)
(sparc_sol2_static_transform_name): Remove
* sparc64-sol2-tdep.c (sparc64_sol2_sigtramp_frame_sniffer): Just
call sol2_sigtramp_p.
(sparc64_sol2_init_abi): Call sol2_init_abi.
Remove calls to set_gdbarch_sofun_address_maybe_missing,
set_gdbarch_static_transform_name,
set_gdbarch_skip_solib_resolver, set_gdbarch_core_pid_to_str.
Some testcases want to compile .c files with a C++ compiler. So they
pass the "c++" option to gdb_compile. That works fine with GCC, but
with Clang, it results in:
gdb compile failed, clang-5.0: warning: treating 'c' input as 'c++' when in C++ mode, this behavior is deprecated [-Wdeprecated]
and the testcase is skipped with UNTESTED.
A previous patch fixed a case like that in
gdb.compile/compile-cplus.exp, by adding -Wno-deprecated to the build
options. However, there are other testcases that use the same
pattern, and all fail for the same reason. For example:
gdb.base/info-types-c++.exp
gdb.base/max-depth-c++.exp
gdb.base/msym-lang.exp
gdb.base/whatis-ptype-typedefs.exp
gdb.btrace/rn-dl-bind.exp
Fix this in a central place, within gdb_compile, by passing "-x c++"
to the compiler driver when we're compiling/linking C++.
This revealed that gdb.compile/compile-cplus.exp and
gdb.arch/amd64-entry-value-paramref.exp tests are compiling an
assembly file with the "c++" option, which would now fail to compile,
with the C++ compiler not grokking the assembly, of course. We just
need to not pass "c++" and all the other related C++ options when
compiling an assembly file.
gdb/testsuite/ChangeLog:
2020-06-24 Pedro Alves <palves@redhat.com>
* gdb.arch/amd64-entry-value-paramref.exp: Use
prepare_for_testing_full and don't pass "c++" for the .S file
build spec.
* gdb.compile/compile-cplus.exp: Don't compile $srcfile3 with
$options, since it's an assembly file. Remove -Wno-deprecated.
* lib/gdb.exp (gdb_compile): Pass "-x c++" explicitly when
compiling C++ programs.
Simon Marchi
Simon Marchi
Pedro Alves
Tom Tromey
Fangrui Song
Alok Kumar Sharma
Philippe Waroquiers
Tom Tromey
Tom de Vries
Nick Alcock
Gary Benson
Eli Zaretskii
Andrew Burgess
Rainer Orth