| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
This adds a new Python function, gdb.execute_mi, that can be used to
invoke an MI command but get the output as a Python object, rather
than a string. This is done by implementing a new ui_out subclass
that builds a Python object.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=11688
Reviewed-By: Eli Zaretskii <eliz@gnu.org>
|
|
When testing gdb.arch/disp-step-insn-reloc.exp with clang in an x86_64
machine, the compiled test case would segfault when returning from
the function can_relocate_call, with a suggestion of a broken stack.
The example assembly in the commment was the following:
f:
MOV $1, %[ok]
JMP end
set_point0:
CALL f ; tracepoint here.
end:
And the segmentation fault happening at the final "ret" instruction of
can_relocate_call. Looking at the disassembled version of the later
half of the important function, we see:
Clang version (f starting at 11a4):
00000000000011ae <set_point0>:
11ae: e8 f1 ff ff ff callq 11a4 <can_relocate_call+0x14>
11b3: 89 45 fc mov %eax,-0x4(%rbp)
11b6: 83 7d fc 01 cmpl $0x1,-0x4(%rbp)
11ba: 0f 85 0a 00 00 00 jne 11ca <set_point0+0x1c>
11c0: e8 5b 00 00 00 callq 1220 <pass>
11c5: e9 05 00 00 00 jmpq 11cf <set_point0+0x21>
11ca: e8 61 00 00 00 callq 1230 <fail>
11cf: 48 83 c4 10 add $0x10,%rsp
11d3: 5d pop %rbp
11d4: c3 retq
11d5: 66 66 2e 0f 1f 84 00 data16 nopw %cs:0x0(%rax,%rax,1)
11dc: 00 00 00 00
gcc version (f starting at 401125):
000000000040112c <set_point0>:
40112c: e8 f4 ff ff ff callq 401125 <can_relocate_call+0x11>
401131: 89 45 fc mov %eax,-0x4(%rbp)
401134: 83 7d fc 01 cmpl $0x1,-0x4(%rbp)
401138: 75 07 jne 401141 <set_point0+0x15>
40113a: e8 c7 ff ff ff callq 401106 <pass>
40113f: eb 05 jmp 401146 <set_point0+0x1a>
401141: e8 c7 ff ff ff callq 40110d <fail>
401146: 90 nop
401147: c9 leaveq
401148: c3 retq
The epilogue of set_point0 (11cf for clang, 401146 for gcc) is the main
difference: GCC's version uses the leaveq instruction, which resets rsp
based on rbp, while clang adds the same constant to rsp that it
subtracted in the prologue. Clang fails because the return address that
is added by the "call f" instruction isn't accounted for.
This commit fixes that by adding a return instruction to f, which leaves
the rsp as the compilers would expect.
Approved-By: Andrew Burgess <aburgess@redhat.com>
|
|
Add a new proc Term::get_line_with_attrs, similar to Term::get_line, that
annotates a tuiterm line with the active attributes.
For instance, the line representing the TUI status window with attribute mode
standout looks like this with Term::get_line:
...
exec No process In: ... L?? PC: ??
...
but like this with Term::get_line_with_attrs:
...
<reverse:1>exec No process In: ... L?? PC: ?? <reverse:0>
...
Also add Term::dump_screen_with_attrs, a Term::dump_screen variant that uses
Term::get_line_with_attrs instead of Term::get_line.
Tested by re-running the TUI test-cases (gdb.tui/*.exp and gdb.python/tui*.exp)
on x86_64-linux.
|
|
Factor out new proc Term::_reset_attrs.
Tested by re-running the TUI test-cases (gdb.tui/*.exp and gdb.python/tui*.exp)
on x86_64-linux.
|
|
547ce8f00b fixed an issue where dynamic types were not being resolved
correctly prior to printing a value. The same issue was discovered when
printing the value using mi-mode, which was not covered by the fix.
Porting the fix to the mi-mode code path resolved the issue.
However, it was discovered that a later patch series, ending
2fc3b8a4cb8, independently fixed the issue in both the cli- and mi-mode
code paths, making the original fix unneeded.
This commit removes this extra frame switch and adds test coverage for
the mi-mode scenario to protect against any future divergence in this
area.
GDB built with GCC 11.
No test suite regressions detected. Compilers: GCC 12.1.0, ACfL 22.1,
Intel 22.1; Platforms: x86_64, aarch64.
Approved-By: Tom Tromey <tom@tromey.com>
|
|
This commit adds a test for the following commit:
commit e86e87f77fd5d8afb3e714f1d9e09e0ff5b4e6ff
Date: Tue Nov 28 16:23:32 2006 +0000
* symtab.c (find_pc_sect_line): Do not return a line before
the start of a symtab.
We have been carrying a test for that commit in the Fedora GDB tree
since that commit was added to GDB. I don't know why the test wasn't
added along with the original commit, but as was written, the test is
pretty gross, it uses objcopy to pull the .text section from an object
file, which was then injected into another source file within a .asm
statement...
... these days we can just make use of the DWARF assembler to achieve
the same results, so I've rewritten the test and think it is worth
adding this to upstream GDB.
The original patch was about about how we find the best symtab and
line table entry, and what to do when GDB can't find a good match.
The new test creates a CU with two functions, only one of which is
covered by the line table. With the above patch reverted GDB returns
an invalid address.
With the above patch reverted I did run the testsuite to see what
other tests might already be exercising this functionality, and I
found two tests:
gdb.dwarf2/dw2-step-out-of-function-no-stmt.exp
gdb.dwarf2/dw2-vendor-extended-opcode.exp
These are pretty similar, they either create minimal, or no line table
for one of the functions in the source file, and as a consequence GDB
returns an unexpected address at some point during the test.
However, both of those tests are really focused on other issues, so I
think this new test does add some value. Plus the new test is not
large, so it's not a huge cost to also run this new test.
Reviewed-By: Tom Tromey <tom@tromey.com>
|
|
Noticed that in breakpoint.c, in one place, we do this:
gdb_printf (_("warning: Error removing "
"breakpoint %d\n"),
old_loc->owner->number);
Instead of using the `warning` function. There are a number of
differences between using gdb_printf like this and calling `warning`,
the main one is probably that real warnings are sent to gdb_stderr,
while the above gdb_printf call will go to gdb_stdout.
In this commit I:
1. Change to call `warning`, we can drop the "warning: " prefix from
the string in breakpoint.c,
2. Update the warning text, I now start with a lower case 'e', which
I believe is the GDB style for warnings,
3. And I have included the address of the bp_location in the warning
messsage,
4. Finally, I update all the tests (2) that include this error
message.
Reviewed-By: Tom Tromey <tom@tromey.com>
|
|
It was pointed out on the mailing list that the new tests added in
this commit:
commit 4de4e48514fc47aeb4ca95cd4091e2a333fbe9e1
Date: Tue Jan 24 15:35:45 2023 +0000
gdb/python: extend the Python Disassembler API to allow for styling
will fail when GDB is built with Python 3.6 or earlier. This is
because the error that is emitted when a function argument is missing
changed in Python 3.7, instead of an error like this:
Python Exception <class 'TypeError'>: function missing required argument 'style' (pos 1)
earlier versions of Python emit:
Python Exception <class 'TypeError'>: Required argument 'style' (pos 1) not found
and the new tests didn't allow for this.
This commit fixes this by allowing either pattern. I've tested this
building GDB against Python 3.7.9 and 3.6.15, with this commit all
tests in gdb.python/py-disasm.exp now pass.
|
|
parameters (PR 30414)
gdb.fortran/lbound-ubound.exp reads the expected lbound and ubound
values by reading some output from the inferior. This is racy when
running on boards where the inferior I/O is on a separate TTY than
GDB's, such as native-gdbserver.
I sometimes see this behavior:
(gdb) continue
Continuing.
Breakpoint 2, do_test (lb=..., ub=...) at /home/jenkins/workspace/binutils-gdb_master_linuxbuild/platform/jammy-amd64/target_board/nati
ve-gdbserver/src/binutils-gdb/gdb/testsuite/gdb.fortran/lbound-ubound.F90:45
45 print *, "" ! Test Breakpoint
(gdb) Remote debugging from host ::1, port 37496
Expected GDB Output:
LBOUND = (-8, -10)
UBOUND = (-1, -2)
APB: Run a test here
APB: Expected lbound '(-8, -10)'
APB: Expected ubound ''
What happened is that expect read the output from GDB before the output
from the inferior, triggering this gdb_test_multiple clause:
-re "$gdb_prompt $" {
set found_prompt true
if {$found_dealloc_breakpoint
|| ($expected_lbound != "" && $expected_ubound != "")} {
# We're done.
} else {
exp_continue
}
}
So it set found_prompt, but the gdb_test_multiple kept going because
found_dealloc_breakpoint is false (this is the flag indicating that the
test is finished) and we still don't have expected_lbound and
expected_ubound. Then, expect reads in the inferior I/O, triggering
this clause:
-re ".*LBOUND = (\[^\r\n\]+)\r\n" {
set expected_lbound $expect_out(1,string)
if {!$found_prompt} {
exp_continue
}
}
This sets expected_lbound, but since found_prompt is true, we don't do
exp_continue, and exit the gdb_test_multiple, without having an
expected_ubound.
Change the test to read the values from the lb and ub function
parameters instead. As far as I understand, this still exercises what
we want to test. These variables contain the return values of the
lbound and ubound functions as computed by the program. We'll use them
to check the return values of the lbound and ubound functions as
computed by GDB.
Change-Id: I3c4d3d17d9291870a758a42301d15a007821ebb5
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30414
|
|
I noticed a trailing whitespace and some indentation errors in lib/tuiterm.exp.
Fix these.
Tested by re-running the TUI test-cases (gdb.tui/*.exp and gdb.python/tui*.exp)
on x86_64-linux.
|
|
"&str" is an important type in Rust -- it's the type of string
literals. However, the compiler puts it in the DWARF in a funny way.
The slice itself is present and named "&str". However, the Rust
parser doesn't look for types with names like this, but instead tries
to construct them from components. In this case it tries to make a
pointer-to-"str" -- but "str" isn't always available, and in any case
that wouldn't yield the best result.
This patch adds a special case for &str.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=22251
Reviewed-By: Andrew Burgess <aburgess@redhat.com>
|
|
Currently, for a source file containing only 5 lines, we also show line
numbers 6 and 7 if they're in scope of the source window:
...
0 +-compact-source.c----------------+
1 |___3_{ |
2 |___4_ return 0; |
3 |___5_} |
4 |___6_ |
5 |___7_ |
6 +---------------------------------+
...
Fix this by not showing line numbers not in a source file, such that we have instead:
...
0 +-compact-source.c----------------+
1 |___3_{ |
2 |___4_ return 0; |
3 |___5_} |
4 | |
5 | |
6 +---------------------------------+
...
Tested on x86_64-linux.
Suggested-By: Simon Marchi <simon.marchi@efficios.com>
Approved-By: Tom Tromey <tom@tromey.com>
|
|
Approved-by: Andrew Burgess <aburgess@redhat.com>
|
|
Run black on gdb.python/py-disasm.py file and commit the changes.
|
|
Rewrite gdb_supported_languages as a caching proc that actually
queries GDB for the list of supported languages, rather than just
containing a hard-coded list of languages.
There's only one test that uses this proc right now,
gdb.python/py-function.exp, and that still passes after this change,
with no changes in the test names.
|
|
After this commit:
commit a68f7e9844208ad8cd498f89b5100084ece7d0f6
Date: Tue May 9 10:28:42 2023 +0100
gdb/testsuite: extend special '^' handling to gdb_test_multiple
buildbot notified me of a regression on s390 in the test:
gdb.base/break-main-file-remove-fail.exp
the failure looks like this:
print /d ((int (*) (void *, size_t)) munmap) (16781312, 4096)
warning: Error removing breakpoint 0
$2 = 0
(gdb) FAIL: gdb.base/break-main-file-remove-fail.exp: cmdline: get integer valueof "((int (*) (void *, size_t)) munmap) (16781312, 4096)"
On the mailing list it has been reported that this failure also
impacts arm, aarch64, and possibly ppc/ppc64 too.
The above commit changed get_integer_valueof so that no output is
expected between the command and the '$2 = 0' line. In this case the
'warning: Error removing breakpoint 0' output is causing the
get_integer_valueof call to fail.
The reason for this warning is that this test deliberately calls
munmap on a page of the inferior's code. The test is checking that
GDB can handle the situation where a s/w breakpoint can't be
removed (due to the page no longer being readable/writable).
The test that is supposed to trigger the warning is later in the test
script when we delete a breakpoint.
So why do some targets trigger the warning earlier during the inferior
call?
The impacted targets use AT_ENTRY_POINT as their strategy for handling
inferior calls, that is, the trampoline that calls the inferior
function is placed at the program's entry point, e.g. often the _start
label.
If this location happens to be on the same page as the page that the
test script unmaps then, when the inferior function call returns, GDB
will not be able to remove the temporary breakpoint that is inserted
to catch the inferior function call returning! As a result we end up
seeing the warning earlier than expected.
I did wonder if this means I should relax the pattern in
get_integer_valueof - just accept that there might be additional
output from GDB which we should ignore.
However, I don't think this the right way to go. With the change in
a68f7e984420 we are now stricter for GDB emitting additional,
unexpected, output, and I think that is a good thing.
So, I think, in this case, in order to handle the possible extra
output, we should implement something like get_integer_valueof
directly in the gdb.base/break-main-file-remove-fail.exp test script.
This local version will handle the possible warning output.
After this the test should pass again on the impacted targets.
|
|
This commit extends the Python Disassembler API to allow for styling
of the instructions.
Before this commit the Python Disassembler API allowed the user to do
two things:
- They could intercept instruction disassembly requests and return a
string of their choosing, this string then became the disassembled
instruction, or
- They could call builtin_disassemble, which would call back into
libopcode to perform the disassembly. As libopcode printed the
instruction GDB would collect these print requests and build a
string. This string was then returned from the builtin_disassemble
call, and the user could modify or extend this string as needed.
Neither of these approaches allowed for, or preserved, disassembler
styling, which is now available within libopcodes for many of the more
popular architectures GDB supports.
This commit aims to fill this gap. After this commit a user will be
able to do the following things:
- Implement a custom instruction disassembler entirely in Python
without calling back into libopcodes, the custom disassembler will
be able to return styling information such that GDB will display
the instruction fully styled. All of GDB's existing style
settings will affect how instructions coming from the Python
disassembler are displayed in the expected manner.
- Call builtin_disassemble and receive a result that represents how
libopcode would like the instruction styled. The user can then
adjust or extend the disassembled instruction before returning the
result to GDB. Again, the instruction will be styled as expected.
To achieve this I will add two new classes to GDB,
DisassemblerTextPart and DisassemblerAddressPart.
Within builtin_disassemble, instead of capturing the print calls from
libopcodes and building a single string, we will now create either a
text part or address part and store these parts in a vector.
The DisassemblerTextPart will capture a small piece of text along with
the associated style that should be used to display the text. This
corresponds to the disassembler calling
disassemble_info::fprintf_styled_func, or for disassemblers that don't
support styling disassemble_info::fprintf_func.
The DisassemblerAddressPart is used when libopcodes requests that an
address be printed, and takes care of printing the address and
associated symbol, this corresponds to the disassembler calling
disassemble_info::print_address_func.
These parts are then placed within the DisassemblerResult when
builtin_disassemble returns.
Alternatively, the user can directly create parts by calling two new
methods on the DisassembleInfo class: DisassembleInfo.text_part and
DisassembleInfo.address_part.
Having created these parts the user can then pass these parts when
initializing a new DisassemblerResult object.
Finally, when we return from Python to gdbpy_print_insn, one way or
another, the result being returned will have a list of parts. Back in
GDB's C++ code we walk the list of parts and call back into GDB's core
to display the disassembled instruction with the correct styling.
The new API lives in parallel with the old API. Any existing code
that creates a DisassemblerResult using a single string immediately
creates a single DisassemblerTextPart containing the entire
instruction and gives this part the default text style. This is also
what happens if the user calls builtin_disassemble for an architecture
that doesn't (yet) support libopcode styling.
This matches up with what happens when the Python API is not involved,
an architecture without disassembler styling support uses the old
libopcodes printing API (the API that doesn't pass style info), and
GDB just prints everything using the default text style.
The reason that parts are created by calling methods on
DisassembleInfo, rather than calling the class constructor directly,
is DisassemblerAddressPart. Ideally this part would only hold the
address which the part represents, but in order to support backwards
compatibility we need to be able to convert the
DisassemblerAddressPart into a string. To do that we need to call
GDB's internal print_address function, and to do that we need an
gdbarch.
What this means is that the DisassemblerAddressPart needs to take a
gdb.Architecture object at creation time. The only valid place a user
can pull this from is from the DisassembleInfo object, so having the
DisassembleInfo act as a factory ensures that the correct gdbarch is
passed over each time. I implemented both solutions (the one
presented here, and an alternative where parts could be constructed
directly), and this felt like the cleanest solution.
Reviewed-By: Eli Zaretskii <eliz@gnu.org>
Reviewed-By: Tom Tromey <tom@tromey.com>
|
|
This commit is a refactor ahead of the next change which will make
disassembler styling available through the Python API.
Unfortunately, in order to make the styling support available, I think
the easiest solution is to make a very small change to the existing
API.
The current API relies on returning a DisassemblerResult object to
represent each disassembled instruction. Currently GDB allows the
DisassemblerResult class to be sub-classed, which could mean that a
user tries to override the various attributes that exist on the
DisassemblerResult object.
This commit removes this ability, effectively making the
DisassemblerResult class final.
Though this is a change to the existing API, I'm hoping this isn't
going to cause too many issues:
- The Python disassembler API was only added in the previous release
of GDB, so I don't expect it to be widely used yet, and
- It's not clear to me why a user would need to sub-class the
DisassemblerResult type, I allowed it in the original patch
because at the time I couldn't see any reason to NOT allow it.
Having prevented sub-classing I can now rework the tail end of the
gdbpy_print_insn function; instead of pulling the results out of the
DisassemblerResult object by calling back into Python, I now cast the
Python object back to its C++ type (disasm_result_object), and access
the fields directly from there. In later commits I will be reworking
the disasm_result_object type in order to hold information about the
styled disassembler output.
The tests that dealt with sub-classing DisassemblerResult have been
removed, and a new test that confirms that DisassemblerResult can't be
sub-classed has been added.
Reviewed-By: Eli Zaretskii <eliz@gnu.org>
Reviewed-By: Tom Tromey <tom@tromey.com>
|
|
The cooked index scanner has special code to handle forward DIE
references. However, a bug report lead to the discovery that this
code does not work -- the "deferred_entry::spec_offset" field is
written to but never used, i.e., the lookup is done using the wrong
key.
This patch fixes the bug and adds a regression test.
The test in the bug itself used a thread_local variable, which
provoked a failure at runtime. This test instead uses "maint print
objfiles" and then inspects to ensure that the entry in question has a
parent. This lets us avoid a clang dependency in the test.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30271
|
|
Currently
print -elements=3 -- "AAAAAA"
prints complete string, which is not what the user asked for.
Fix two buggy tests exposed by the fix, and add a new test.
Reviewed-by: Keith Seitz <keiths@redhat.com>
|
|
A co-worker here at AdaCore discovered that the Pragma Import series
caused a rgression. When debugging gnat1, gdb started asking for
overload resolution like:
(gdb) call pp(n)
Multiple matches for pp
[0] cancel
[1] pp (types.union_id) at ../../gcc/gcc/ada/treepr.adb:511
[2] treepr.pp (types.union_id) at ../../gcc/gcc/ada/treepr.adb:511
This worked before the series, and is strange anyway, because the
matches refer to the same function.
This patch adds a test case for this situation and fixes the bug by
pruning identical functions in remove_extra_symbols.
|
|
Ada can import C APIs and also export Ada constructs to C via Pragma
Import and Pragma Export. This patch adds support for these to gdb,
by arranging to either defer some aspects of a symbol to the
underlying C symbol (for Import) or by introducing a second symbol
(for Export). A somewhat tricky approach is needed, both because gdb
doesn't generally handle symbol aliasing, and because Ada treats
symbol names in an unusual way (as compared to the rest of gdb).
|
|
I noticed that gdb's DAP code did not provide a way to see register
values. DAP defines a "register" scope, which this patch implements.
This patch also adds the missing (and optional) "presentationHint" to
scopes.
|
|
A co-worker at AdaCore noticed that calling a function without
debuginfo yields:
(gdb) print plus_one(23)
'pck.plus_one' has unknown return type; cast the call to its declared return type
However, this also happens if you follow the directions and add the
cast.
This patch fixes the problem and adds a regression test.
|
|
Add the DisassemblerResult.__str__ method. This gives the same result
as the DisassemblerResult.string attribute, but can be useful
sometimes depending on how the user is trying to print the object.
There's a test for the new functionality.
|
|
Add a __repr__ method for the DisassembleInfo and DisassemblerResult
types, and add some tests for these new methods.
|
|
The commit:
commit 08ec06d6440745ef9204d39197aa1e732df41056
Date: Wed Mar 29 10:41:07 2023 +0100
gdb/testsuite: special case '^' in gdb_test pattern
Added some special handling of '^' to gdb_test -- a leading '^' will
cause the command regexp to automatically be included in the expected
output pattern.
It was pointed out that the '-wrap' flag of gdb_test_multiple is
supposed to work in the same way as gdb_test, and that the recent
changes for '^' had not been replicated for gdb_test_multiple. This
patch addresses this issue.
So, after this commit, the following two constructs should have the
same meaning:
gdb_test "command" "^output" "test name"
gdb_test_multiple "command" "test name" {
-re -wrap "^output" {
pass $gdb_test_name
}
}
In both cases the '^' will case gdb.exp to inject a regexp that
matches 'command' after the '^' and before the 'output', this is in
addition to adding the $gdb_prompt pattern after 'output' in the
normal way.
The special '^' handling is only applied when '-wrap' is used, as this
is the only mode that aims to mimic gdb_test.
While working on this patch I realised that I could actually improve
the logic for the special '^' handling in the case where the expected
output pattern is empty. I replicated these updates for both gdb_test
and gdb_test_multiple in order to keep these two paths in sync.
There were a small number of tests that needed adjustment after this
change, mostly just removing command regexps that are now added
automatically, but the gdb.base/settings.exp case was a little weird
as it turns out trying to match a single blank line is probably harder
now than it used to be -- still, I suspect this is a pretty rare case,
so I think the benefits (improved anchoring) outweigh this small
downside (IMHO).
|
|
I spotted this behaviour:
(gdb) p $_gdb_maint_setting("xxx")
First argument of $_gdb_maint_setting must be a valid setting of the 'show' command.
Notice that GDB claims I need to use a setting from the 'show'
command, which isn't correct for $_gdb_maint_setting, in this case I
need to use a setting from 'maintenance show'.
This same issue is present for $_gdb_maint_setting_str.
This commit fixes this minor issue.
Approved-By: Simon Marchi <simon.marchi@efficios.com>
|
|
When running test-case gdb.dwarf2/opt-out-not-implptr.exp with target board
unix/-m32 we have:
...
(gdb) print noptr^M
$1 = {0, <optimized out>, <optimized out>, <optimized out>}^M
(gdb) FAIL: gdb.dwarf2/opt-out-not-implptr.exp: print noptr
...
The problem happens when evaluating this dwarf expression:
...
<45> DW_AT_location : 13 byte block: 10 86 ea d7 d0 96 8e cf 92 5c 9f 93 8 \
(DW_OP_constu: 6639779683436459270; DW_OP_stack_value; DW_OP_piece: 8)
...
The DW_OP_constu pushes a value with "generic type" on to the DWARF stack, and
the "generic type" has the size of an address on the target machine, which for
target board unix/-m32 is 4 bytes. Consequently, the constant is abbreviated.
Next, the DW_OP_piece declares that the resulting 4-byte value is 8 bytes
large, and we hit this clause in rw_pieced_value:
...
/* Use zeroes if piece reaches beyond stack value. */
if (p->offset + p->size > stack_value_size_bits)
break;
...
and consequently get a zero.
We could just add require is_target_64 to the test-case, but instead, add a
32-bit case and require is_target_64 just for the 64-bit case.
Tested on x86_64-linux.
|
|
I ran test-case gdb.dwarf2/opt-out-not-implptr.exp with make-check-all.sh, and
with target board dwarf64 ran into:
...
FAIL: gdb.dwarf2/opt-out-not-implptr.exp: print noptr
...
due to is_target_64 failing because of:
...
builtin_spawn -ignore SIGHUP gcc -fno-stack-protector \
-fdiagnostics-color=never -w -c -gdwarf64 -g -o is_64_target.o \
is_64_target.c^M
gcc: error: '-gdwarf64' is ambiguous; use '-gdwarf-64' for DWARF version or \
'-gdwarf -g64' for debug level^M
compiler exited with status 1
...
The FAIL is the same FAIL I run into with target board unix/-m32: is_target_64
fails for both cases.
The reason that is_target_64 is failing for target board dwarf64, is because
of using system compiler 7.5.0 which doesn't support -gdwarf64.
Fix this by making is_target_64 use nodebug instead of debug for compilation.
Tested on x86_64-linux.
|
|
I. Auto-detected width (xterm vs. ansi)
Say we have a terminal with a width of 40 chars:
...
$ echo $COLUMNS
40
...
With TERM=xterm, we report a width of 40 chars:
...
$ TERM=xterm gdb
(gdb) show width
Number of characters gdb thinks are in a line is 40.
...
And with TERM=ansi, a width of 39 chars:
...
$ TERM=ansi gdb
(gdb) show width
Number of characters gdb thinks are in a line is 39.
...
Gdb uses readline to auto-detect screen size, and readline decides in the
TERM=ansi case that the terminal does not have reliable auto-wrap, and
consequently decides to hide the last terminal column from the readline user
(in other words GDB), hence we get 39 instead of 40.
II. Types of wrapping
Looking a bit more in detail inside gdb, it seems there are two types of
wrapping:
- readline wrapping (in other words, prompt edit wrapping), and
- gdb output wrapping (can be observed by issuing "info sources").
This type of wrapping attempts to wrap some of the gdb output earlier
than the indicated width, to not break lines in inconvenient places.
III. Readline wrapping, auto-detected screen size
Let's investigate readline wrapping with the auto-detected screen widths.
First, let's try with xterm:
...
$ TERM=xterm gdb
(gdb) 7890123456789012345678901234567890
123
...
That looks as expected, wrapping occurs after 40 chars.
Now, let's try with ansi:
...
$ TERM=ansi gdb
(gdb) 78901234567890123456789012345678
90123
...
It looks like wrapping occurred after 38, while readline should be capable of
wrapping after 39 chars.
This is caused by readline hiding the last column, twice.
In more detail:
- readline detects the screen width: 40,
- readline hides the last column, setting the readline screen width to 39,
- readline reports 39 to gdb as screen width,
- gdb sets its width setting to 39,
- gdb sets readline screen width to 39,
- readline hides the last column, again, setting the readline screen width to
38.
This is reported as PR cli/30346.
IV. gdb output wrapping, auto-detected screen size
Say we set the terminal width to 56. With TERM=xterm, we have:
...
/home/abuild/rpmbuild/BUILD/glibc-2.31/csu/elf-init.c,
/data/vries/hello.c,
...
but with TERM=ansi:
...
/home/abuild/rpmbuild/BUILD/glibc-2.31/csu/elf-init.c, /
data/vries/hello.c,
...
So what happened here? With TERM=ansi, the width setting is auto-detected to
55, and gdb assumes the terminal inserts a line break there, which it doesn't
because the terminal width is 56.
This is reported as PR cli/30411.
V. Fix PRs
Fix both mentioned PRs by taking into account the hidden column when readline
reports the screen width in init_page_info, and updating chars_per_line
accordingly.
Note that now we report the same width for both TERM=xterm and TERM=ansi,
which is much clearer.
The point where readline respectively expects or ensures wrapping is still
indicated by "maint info screen", for xterm:
...
Number of characters readline reports are in a line is 40.
...
and ansi:
...
Number of characters readline reports are in a line is 39.
...
VI. Testing
PR cli/30346 is covered by existing regression tests gdb.base/wrap-line.exp
and gdb.tui/wrap-line.exp, so remove the KFAILs there.
I didn't manage to come up with a regression test for PR cli/30411. Perhaps
that would be easier if we had a maintenance command that echoes its arguments
while applying gdb output wrapping.
Tested on x86_64-linux.
PR cli/30346
PR cli/30411
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30346
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30411
|
|
A user reported a bug where printing a certain array of integer types
would result in the nonsensical:
(gdb) p l_126
$1 = {6639779683436459270, <synthetic pointer>, <synthetic pointer>, <synthetic pointer>}
I tracked this down to some issues in the DWARF expression code.
First, check_pieced_synthetic_pointer did not account for the
situation where a location expression does not describe all the bits
of a value -- in this case it returned true, meaning there is a
synthetic pointer, but in fact these bits are optimized out. (It
turns out this incorrect output had already been erroneously tested
for as well.)
Next, rw_pieced_value did not mark these bits as optimized-out,
either.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30296
|
|
gdb's debuginfod progress messages include the size of the file being
downloaded if the size information is available at the time the message
is printed. For example:
Downloading 10 MB separate debug info for /lib64/libxyz.so
This size information is omitted if it's not available at the time of
printing:
Downloading separate debug info for /lib64/libxyz.so
A pattern in crc_mismatch.exp fails to be matched if a progress message
includes a file size. Add a wildcard to the pattern so that it matches
the progress message whether or not it includes the file size.
|
|
Currently, when a local software watchpoint goes out of scope, GDB sets
the watchpoint's disposition to `delete at next stop' and then normal
stops (i.e., stop and wait for the next GDB command). When GDB normal
stops, it automatically deletes the breakpoints with their disposition
set to `delete at next stop'.
Suppose a Python script decides not to normal stop when a local
software watchpoint goes out of scope, the watchpoint will not be
automatically deleted even when its disposition is set to
`delete at next stop'.
Since GDB single-steps the program and tests the watched expression
after each instruction, not deleting the watchpoint causes the
watchpoint to be hit many more times than it should, as reported in
PR python/29603.
This was happening because the watchpoint is not deleted or disabled
when going out of scope.
This commit fixes this issue by disabling the watchpoint when going out
of scope. It also adds a test to ensure this feature isn't regressed in
the future.
Calling `breakpoint_auto_delete' on all kinds of stops (in
`fetch_inferior_event') seem to solve this issue, but is in fact
inappropriate, since `breakpoint_auto_delete' goes over all breakpoints
instead of just going through the bpstat chain (which only contains the
breakpoints that were hit right now).
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29603
Change-Id: Ia85e670b2bcba2799219abe4b6be3b582387e383
|
|
Andrew pointed out that the behaviour as tested in gdb.tui/compact-source.exp
is incorrect:
...
0 +-compact-source.c--------------------------------------------------------+
1 |___3_{ |
2 |___4_ return 0; |
3 |___5_} |
4 |___6_ |
5 |___7_ |
6 |___8_ |
7 |___9_ |
8 +-------------------------------------------------------------------------+
...
The last line number in the source file is 5, and there are 7 lines to display
source lines, so if we'd scroll all the way down, the first line number in the
source window would be 5, and the last one would be 11.
To represent 11 we'd need 2 digits, so we expect to see ___04_ here instead of
___4_, even though all line numbers currently in the src window (3-9) can be
represented with only 1 digit.
Fix this in tui_source_window::set_contents, by updating the computation of
max_line_nr:
...
- int max_line_nr = std::max (lines_in_file, last_line_nr_in_window);
+ int max_line_nr = lines_in_file + nlines - 1;
...
Tested on x86_64-linux.
Co-Authored-By: Andrew Burgess <aburgess@redhat.com>
Approved-By: Tom Tromey <tom@tromey.com>
|
|
While working on another patch I did this:
(gdb) set debug expression 1
(gdb) set language rust
(gdb) p "foo"
Operation: OP_AGGREGATE
Type: &str
Fatal signal: Segmentation fault
... etc ...
The problem is that the second field of the rust_aggregate_operation
is created as a nullptr, this can be seen in rust-parse.c. in the
function rust_parser::parse_string().
However, in expop.h, in the function dump_for_expression, we make the
assumption that the expressions will never be nullptr.
I did consider moving the nullptr handling into a new function
rust_aggregate_operation::dump, however, as the expression debug
dumping code is not exercised as much as it might be, I would rather
that this code be hardened and able to handle a nullptr without
crashing, so I propose that we add nullptr handling into the general
dump_for_expression function. The behaviour is now:
(gdb) set debug expression 1
(gdb) set language rust
(gdb) p "foo"
Operation: OP_AGGREGATE
Type: &str
nullptr
Vector:
String: data_ptr
Operation: UNOP_ADDR
Operation: OP_STRING
String: foo
String: length
Operation: OP_LONG
Type: usize
Constant: 3
evaluation of this expression requires the target program to be active
(gdb)
There's a new test to check for this case.
Reviewed-By: Tom Tromey <tom@tromey.com>
|
|
Consider a hello.c, with less than 10 lines:
...
$ wc -l hello.c
8 hello.c
...
and compiled with -g into an a.out.
With compact-source off:
...
$ gdb -q a.out \
-ex "set tui border-kind ascii" \
-ex "maint set tui-left-margin-verbose on" \
-ex "set tui compact-source off" \
-ex "tui enable"
...
we get:
...
+-./data/hello.c-----------------------+
|___000005_{ |
|___000006_ printf ("hello\n"); |
|___000007_ return 0; |
|___000008_} |
|___000009_ |
|___000010_ |
|___000011_ |
...
but with compact-source on:
...
+-./data/hello.c-----------------------+
|___5{ |
|___6 printf ("hello\n"); |
|___7 return 0; |
|___8} |
|___9 |
|___1 |
|___1 |
...
There are a couple of problems with compact-source.
First of all the documentation mentions:
...
The default display uses more space for line numbers and starts the
source text at the next tab stop; the compact display uses only as
much space as is needed for the line numbers in the current file, and
only a single space to separate the line numbers from the source.
...
The bit about the default display and the next tab stop looks incorrect. The
source doesn't start at a tab stop, instead it uses a single space to separate
the line numbers from the source.
Then the documentation mentions that there's single space in the compact
display, but evidently that's missing.
Then there's the fact that the line numbers "10" and "11" are both abbreviated
to "1" in the compact case.
The abbreviation is due to allocating space for <lines in source>, which is
8 for this example, and takes a single digit. The line numbers though
continue past the end of the file, so fix this by allocating space for
max (<lines in source>, <last line in window>), which in this example takes 2
digits.
The missing space is due to some confusion about what the "1" here in
tui_source_window::set_contents represent:
...
double l = log10 ((double) offsets->size ());
m_digits = 1 + (int) l;
...
It could be the trailing space that's mentioned in tui-source.h:
...
/* How many digits to use when formatting the line number. This
includes the trailing space. */
int m_digits;
...
Then again, it could be part of the calculation for the number of digits
needed for printing. With this minimal example:
...
int main () {
for (int i = 8; i <= 11; ++i) {
double l = log10 ((double) i);
printf ("%d %d\n", i, (int)l);
}
return 0;
}
...
we get:
...
$ ./a.out
8 0
9 0
10 1
11 1
...
which shows that the number of digits needed for printing i is
"1 + (int)log10 ((double) i)".
Fix this by introducing named variables needed_digits and trailing_space, each
adding 1.
With the fixes, we get for compact-source on:
...
+-./data/hello.c-----------------------+
|___05_{ |
|___06_ printf ("hello\n"); |
|___07_ return 0; |
|___08_} |
|___09_ |
|___10_ |
|___11_ |
|...
Also fix the documentation and help text to actually match effect of
compact-source.
Tested on x86_64-linux.
|
|
When stopped inside an inline function, trying to jump to a different line
of the same function currently results in a warning about jumping to another
function. Fix this by taking inline functions into account.
Before:
Breakpoint 1, function_inline (x=510) at jump-inline.cpp:22
22 a = a + x; /* inline-funct */
(gdb) j 21
Line 21 is not in `function_inline(int)'. Jump anyway? (y or n)
After:
Breakpoint 2, function_inline (x=510) at jump-inline.cpp:22
22 a = a + x; /* inline-funct */
(gdb) j 21
Continuing at 0x400679.
Breakpoint 1, function_inline (x=510) at jump-inline.cpp:21
21 a += 1020 + a; /* increment-funct */
This was regression-tested on X86-64 Linux.
Co-Authored-by: Cristian Sandu <cristian.sandu@intel.com>
Approved-By: Andrew Burgess <aburgess@redhat.com>
|
|
Python pretty-printers haven't applied to static members for quite
some time. I tracked this down to the call to cp_print_value_fields
in cp_print_static_field -- it doesn't let pretty-printers have a
chance to print the value. This patch fixes the problem.
The way that static members are handled is very weird to me. I tend
to think this should be done more globally, like in value_print.
However, I haven't made any big change.
Reviewed-by: Keith Seitz <keiths@redhat.com>
Tested-by: Keith Seitz <keiths@redhat.com>
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30057
|
|
The DAP scopes request examines the symbols in a block tree, but
neglects to omit types. This patch fixes the problem.
|
|
After this commit:
commit e2f620135d92f7cd670af4e524fffec7ac307666
Date: Thu Mar 30 13:26:25 2023 +0100
gdb/testsuite: change newline patterns used in gdb_test
It was pointed out in PR gdb/30403 that the same patterns can be found
in other lib/gdb.exp procs and that it would probably be a good idea
if these procs remained in sync with gdb_test. Actually, the bug
specifically calls out gdb_test_multiple when using with '-wrap', but
I found a couple of other locations in gdb_continue_to_breakpoint,
gdb_test_multiline, get_valueof, and get_local_valueof.
In all these locations one or both of the following issues are
addressed:
1. A leading pattern of '[\r\n]*' is pointless. If there is a
newline it will be matched, but if there is not then the testsuite
doesn't care. Also, as expect is happy to skip non-matched output
at the start of a pattern, if there is a newline expect is happy to
skip over it before matching the rest. As such, this leading
pattern is removed.
2. Using '\[\r\n\]*$gdb_prompt' means that we will swallow
unexpected blank lines at the end of a command's output, but also,
if the pattern from the test script ends with a '\r', '\n', or '.'
then these will partially match the trailing newline, with the
remainder of the newline matched by the pattern from gdb.exp. This
split matching doesn't add any value, it's just something that has
appeared as a consequence of how gdb.exp was originally written. In
this case the '\[\r\n\]*' is replaced with '\r\n'.
I've rerun the testsuite and fixed the regressions that I saw, these
were places where GDB emits a blank line at the end of the command
output, which we now need to explicitly match in the test script, this
was for:
gdb.dwarf2/dw2-out-of-range-end-of-seq.exp
gdb.guile/guile.exp
gdb.python/python.exp
Or a location where the test script was matching part of the newline
sequence, while gdb.exp was previously matching the remainder of the
newline sequence. Now we rely on gdb.exp to match the complete
newline sequence, this was for:
gdb.base/commands.exp
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30403
|
|
I noticed in gdb.base/page.exp:
...
set fours [string repeat 4 40]
...
but then shortly afterwards:
...
[list 1\r\n 2\r\n 3\r\n 444444444444444444444444444444]
...
Summarize the long string in the same way using string repeat:
...
[list 1\r\n 2\r\n 3\r\n [string repeat 4 30]]
...
Tested on x86_64-linux.
|
|
Tighten the expected output pattern in the test script:
gdb.debuginfod/build-id-no-debug-warning.exp
While working on some other patch I broke GDB such that this warning:
warning: "FILENAME": separate debug info file has no debug info
(which is generated in build-id.c) didn't actually include the
FILENAME any more -- yet this test script continued to pass. It turns
out that this script doesn't actually check for FILENAME.
This commit extends the test pattern to check for the full warning
string, including FILENAME, and also removes some uses of '.*' to make
the test stricter.
|
|
Add a test-case that tests prompt edit wrapping in CLI, both
for TERM=xterm and TERM=ansi, both with auto-detected and hard-coded width.
In the TERM=ansi case with auto-detected width we run into PR cli/30346, so
add a KFAIL for that failure mode.
Tested on x86_64-linux.
|
|
Add a test-case that tests prompt edit wrapping behaviour in the tuiterm, both
for CLI and TUI, both with auto-detected and hard-coded width.
In the CLI case with auto-detected width we run into PR cli/30411, so add a
KFAIL for that failure mode.
Tested on x86_64-linux.
|
|
This reverts commit 476410b3bca1389ee69e9c8fa18aaee16793a56d.
One of Simon's recent commits (2a740b3ba4c9f39c86dd75e0914ee00942cab471)
changed the way recording a remote target works and fixed the underlying
issue of the bug, so the KFails can be removed from the test.
Approved-By: Tom Tromey <tom@tromey.com>
|
|
SUMMARY
The '--simple-values' argument to '-stack-list-arguments' and similar
GDB/MI commands does not take reference types into account, so that
references to arbitrarily large structures are considered "simple" and
printed. This means that the '--simple-values' argument cannot be used
by IDEs when tracing the stack due to the time taken to print large
structures passed by reference.
DETAILS
Various GDB/MI commands ('-stack-list-arguments', '-stack-list-locals',
'-stack-list-variables' and so on) take a PRINT-VALUES argument which
may be '--no-values' (0), '--all-values' (1) or '--simple-values' (2).
In the '--simple-values' case, the command is supposed to print the
name, type, and value of variables with simple types, and print only the
name and type of variables with compound types.
The '--simple-values' argument ought to be suitable for IDEs that need
to update their user interface with the program's call stack every time
the program stops. However, it does not take C++ reference types into
account, and this makes the argument unsuitable for this purpose.
For example, consider the following C++ program:
struct s {
int v[10];
};
int
sum(const struct s &s)
{
int total = 0;
for (int i = 0; i < 10; ++i) total += s.v[i];
return total;
}
int
main(void)
{
struct s s = { { 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 } };
return sum(s);
}
If we start GDB in MI mode and continue to 'sum', the behaviour of
'-stack-list-arguments' is as follows:
(gdb)
-stack-list-arguments --simple-values
^done,stack-args=[frame={level="0",args=[{name="s",type="const s &",value="@0x7fffffffe310: {v = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10}}"}]},frame={level="1",args=[]}]
Note that the value of the argument 's' was printed, even though 's' is
a reference to a structure, which is not a simple value.
See https://github.com/microsoft/MIEngine/pull/673 for a case where this
behaviour caused Microsoft to avoid the use of '--simple-values' in
their MIEngine debug adapter, because it caused Visual Studio Code to
take too long to refresh the call stack in the user interface.
SOLUTIONS
There are two ways we could fix this problem, depending on whether we
consider the current behaviour to be a bug.
1. If the current behaviour is a bug, then we can update the behaviour
of '--simple-values' so that it takes reference types into account:
that is, a value is simple if it is neither an array, struct, or
union, nor a reference to an array, struct or union.
In this case we must add a feature to the '-list-features' command so
that IDEs can detect that it is safe to use the '--simple-values'
argument when refreshing the call stack.
2. If the current behaviour is not a bug, then we can add a new option
for the PRINT-VALUES argument, for example, '--scalar-values' (3),
that would be suitable for use by IDEs.
In this case we must add a feature to the '-list-features' command
so that IDEs can detect that the '--scalar-values' argument is
available for use when refreshing the call stack.
PATCH
This patch implements solution (1) as I think the current behaviour of
not printing structures, but printing references to structures, is
contrary to reasonable expectation.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29554
|
|
Initialize wpoffset_to_wpnumto avoid TCL error which happens in some aarch64 types.
ERROR: in testcase /root/build/gdb/testsuite/../../../binutils-gdb/gdb/testsuite/gdb.base/watchpoint-unaligned.exp
ERROR: can't read "wpoffset_to_wpnum(1)": no such element in array
ERROR: tcl error code TCL READ VARNAME
ERROR: tcl error info:
can't read "wpoffset_to_wpnum(1)": no such element in array
while executing
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30340
Reviewed-by: Luis Machado <luis.machado@arm.com>
Reviewed-By: Andrew Burgess <aburgess@redhat.com>
|
|
In gdb.dwarf2/dw2-abs-hi-pc.exp we do:
...
set sources [lmap i $sources { expr { "$srcdir/$subdir/$i" } }]
...
The use of expr is not idiomatic. Fix this by using set instead:
...
set sources [lmap i $sources { set tmp $srcdir/$subdir/$i }]
...
Reported-By: Tom Tromey <tom@tromey.com>
Reviewed-By: Andreas Schwab <schwab@suse.de>
|
|
Noticed in passing that the prepare_for_testing call in
gdb.linespec/cp-completion-aliases.exp does not pass the 'c++' flag,
despite this being a C++ test.
I guess, as the source file has the '.cc' extension, all the compilers
are doing the right thing anyway -- the source file uses templates, so
is definitely being compiled as C++.
I noticed this when I tried to set CXX_FOR_TARGET (but not
CC_FOR_TARGET) and spotted that this script was still using the C
compiler.
Fixed in this commit by adding the 'c++' flag for prepare_for_testing.
|
