Mike Frysinger [Sun, 6 Nov 2022 09:56:39 +0000 (16:56 +0700)]
sim: modules.c: move generation to top-level
In order to compile arch objects from the top-level, we need to
generate the modules.c file, so move that logic up to the top
level first. The deps are a bit imperfect currently due to the
common/ files not being shared. That'll improve as we share the
sources more.
Mike Frysinger [Sun, 6 Nov 2022 14:40:56 +0000 (21:40 +0700)]
sim: build: hoist lists of hw devices up
We need these in the top-level to generate libsim.a, but also in the
subdirs to generate hw-config.h. Move it to the local.mk, and pass
it down when running recursive make. This avoids duplication, and
makes it available to both. We can simplify this once we move the
various steps up to the top-level too.
Mike Frysinger [Sun, 6 Nov 2022 14:24:24 +0000 (21:24 +0700)]
sim: build: hoist lists of common objects up
In order to create libsim.a in the common dir, we need the list of
objects for each target. To avoid duplicating the list with the
recursive make in each port, pass it down as a variable. This is
a temporary hack until the top-level creates libsim.a for ports.
Mike Frysinger [Thu, 10 Nov 2022 18:35:42 +0000 (01:35 +0700)]
sim: igen: fix hang when decoding boolean rule constants
The parser for boolean rules fails to skip over the , separator in
the options which makes it hang forever. No dc files in the tree
use boolean rules atm which is why no one noticed.
Mike Frysinger [Thu, 10 Nov 2022 08:53:25 +0000 (15:53 +0700)]
sim: ppc: rename ppc-instructions to powerpc.igen
To make it clear this is an input to the igen tool, rename it with an
igen extension. This matches the other files in the ppc dir (altivec
& e500 igen files), and the other igen ports (mips, mn10300, v850).
H.J. Lu [Fri, 4 Nov 2022 20:47:59 +0000 (13:47 -0700)]
i386: Check invalid (%dx) usage
(%dx) isn't a valid memory address in any modes. It is used as a special
memory operand for input/output port address in AT&T syntax and should
only be used with input/output instructions. Update i386_att_operand to
set i.input_output_operand to true for (%dx) and issue an error if (%dx)
is used with non-input/output instructions.
PR gas/29751
* config/tc-i386.c (_i386_insn): Add input_output_operand.
(md_assemble): Issue an error if input/output memory operand is
used with non-input/output instructions.
(i386_att_operand): Set i.input_output_operand to true for
(%dx).
* testsuite/gas/i386/inval.l: Updated.
* testsuite/gas/i386/x86-64-inval.l: Likewise.
* testsuite/gas/i386/inval.s: Add tests for invalid (%dx) usage.
* testsuite/gas/i386/x86-64-inval.s: Likewise.
Simon Marchi [Thu, 4 Aug 2022 15:49:12 +0000 (11:49 -0400)]
gdb: make "start" breakpoint inferior-specific
I saw this failure on a CI:
(gdb) add-inferior
[New inferior 2]
Added inferior 2
(gdb) PASS: gdb.threads/vfork-multi-inferior.exp: method=non-stop: add-inferior
inferior 2
[Switching to inferior 2 [<null>] (<noexec>)]
(gdb) PASS: gdb.threads/vfork-multi-inferior.exp: method=non-stop: inferior 2
kill
The program is not being run.
(gdb) file /home/jenkins/workspace/binutils-gdb_master_linuxbuild/platform/jammy-amd64/target_board/unix/tmp/tmp.GYATAXR8Ku/gdb/testsuite/outputs/gdb.threads/vfork-multi-inferior/vfork-multi-inferior-sleep
Reading symbols from /home/jenkins/workspace/binutils-gdb_master_linuxbuild/platform/jammy-amd64/target_board/unix/tmp/tmp.GYATAXR8Ku/gdb/testsuite/outputs/gdb.threads/vfork-multi-inferior/vfork-multi-inferior-sleep...
(gdb) run &
Starting program: /home/jenkins/workspace/binutils-gdb_master_linuxbuild/platform/jammy-amd64/target_board/unix/tmp/tmp.GYATAXR8Ku/gdb/testsuite/outputs/gdb.threads/vfork-multi-inferior/vfork-multi-inferior-sleep
(gdb) PASS: gdb.threads/vfork-multi-inferior.exp: method=non-stop: run inferior 2
inferior 1
[Switching to inferior 1 [<null>] (<noexec>)]
(gdb) PASS: gdb.threads/vfork-multi-inferior.exp: method=non-stop: inferior 1
kill
The program is not being run.
(gdb) file /home/jenkins/workspace/binutils-gdb_master_linuxbuild/platform/jammy-amd64/target_board/unix/tmp/tmp.GYATAXR8Ku/gdb/testsuite/outputs/gdb.threads/vfork-multi-inferior/vfork-multi-inferior
Reading symbols from /home/jenkins/workspace/binutils-gdb_master_linuxbuild/platform/jammy-amd64/target_board/unix/tmp/tmp.GYATAXR8Ku/gdb/testsuite/outputs/gdb.threads/vfork-multi-inferior/vfork-multi-inferior...
(gdb) break should_break_here
Breakpoint 1 at 0x11b1: file /home/jenkins/workspace/binutils-gdb_master_linuxbuild/platform/jammy-amd64/target_board/unix/src/binutils-gdb/gdb/testsuite/gdb.threads/vfork-multi-inferior.c, line 25.
(gdb) PASS: gdb.threads/vfork-multi-inferior.exp: method=non-stop: break should_break_here
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
start
Temporary breakpoint 2 at 0x11c0: -qualified main. (2 locations)
Starting program: /home/jenkins/workspace/binutils-gdb_master_linuxbuild/platform/jammy-amd64/target_board/unix/tmp/tmp.GYATAXR8Ku/gdb/testsuite/outputs/gdb.threads/vfork-multi-inferior/vfork-multi-inferior
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
Thread 2.1 "vfork-multi-inf" hit Temporary breakpoint 2, main () at /home/jenkins/workspace/binutils-gdb_master_linuxbuild/platform/jammy-amd64/target_board/unix/src/binutils-gdb/gdb/testsuite/gdb.threads/vfork-multi-inferior-sleep.c:23
23 sleep (30);
(gdb) FAIL: gdb.threads/vfork-multi-inferior.exp: method=non-stop: start inferior 1
What happens is:
1. We start inferior 2 with "run&", it runs very slowly, takes time to
get to main
2. We switch to inferior 1, and run "start"
3. The temporary breakpoint inserted by "start" applies to all inferiors
4. Inferior 2 hits that breakpoint and GDB reports that hit
To avoid this, breakpoints inserted by "start" should be
inferior-specific. However, we don't have a nice way to make
inferior-specific breakpoints yet. It's possible to make
pspace-specific breakpoints (for example how the internal_breakpoint
constructor does) by creating a symtab_and_line manually. However,
inferiors can share program spaces (usually on particular embedded
targets), so we could have a situation where two inferiors run the same
code in the same program space. In that case, it would just not be
possible to insert a breakpoint in one inferior but not the other.
A simple solution that should work all the time is to add a condition to
the breakpoint inserted by "start", to check the inferior reporting the
hit is the expected one. This is what this patch implements.
Add a test that does:
- start in background inferior 1 that sleeps before reaching its main
function (using a sleep in a global C++ object's constructor)
- start inferior 2 with the "start" command, which also sleeps before
reaching its main function
- validate that we hit the breakpoint in inferior 2
Without the fix, we hit the breakpoint in inferior 1 pretty much all the
time. There could be some unfortunate scheduling causing the test not
to catch the bug, for instance if the scheduler decides not to schedule
inferior 1 for a long time, but it would be really rare. If the bug is
re-introduced, the test will catch it much more often than not, so it
will be noticed.
Reviewed-By: Bruno Larsen <blarsen@redhat.com> Approved-By: Pedro Alves <pedro@palves.net>
Change-Id: Ib0148498a476bfa634ed62353c95f163623c686a
Commit 041de3d73aa changed the output format of all error messages when
GDB couldn't determine a compatible overload for a given function, but
it was only supposed to change if the failure happened due to incomplete
types. This commit removes the stray . that was added
Aaron Merey [Mon, 24 Oct 2022 18:05:06 +0000 (14:05 -0400)]
gdb/debuginfod: Improve progress updates
If the download size is known, a progress bar is displayed along with
the percentage of completion and the total download size.
Downloading separate debug info for /lib/libxyz.so
[############ ] 25% (10.01 M)
If the download size is not known, a progress indicator is displayed
with a ticker ("###") that moves across the screen at a rate of 1 tick
every 0.5 seconds.
Downloading separate debug info for /lib/libxyz.so
[ ### ]
If the output stream is not a tty, batch mode is enabled, the screen is
too narrow or width has been set to 'unlimited', then only a static
description of the download is printed. No bar or ticker is displayed.
Downloading separate debug info for /lib/libxyz.so...
In any case, if the size of the download is known at the time the
description is printed then it will be included in the description.
Downloading 10.01 MB separate debug info for /lib/libxyz.so...
Simon Marchi [Mon, 24 Oct 2022 20:18:43 +0000 (16:18 -0400)]
gdb: add special handling for frame level 0 in frame_info_ptr
I noticed this problem while preparing the initial submission for the
ROCm GDB port. One particularity of this patch set is that it does not
support unwinding frames, that requires support of some DWARF extensions
that will come later. It was still possible to run to a breakpoint and
print frame #0, though.
When rebasing on top of the frame_info_ptr work, GDB started tripping on
a prepare_reinflate call, making it not possible anymore to event print
the frame when stopping on a breakpoint. One thing to know about frame
0 is that its id is lazily computed when something requests it through
get_frame_id. See:
So, up to that prepare_reinflate call, frame 0's id was not computed,
and prepare_reinflate, calling get_frame_id, forces it to be computed.
Computing the frame id generally requires unwinding the previous frame,
which with my ROCm GDB patch fails. An exception is thrown and the
printing of the frame is simply abandonned.
Regardless of this ROCm GDB problem (which is admittedly temporary, it
will be possible to unwind with subsequent patches), we want to avoid
prepare_reinflate to force the computing of the frame id, for the same
reasons we lazily compute it in the first place.
In addition, frame 0's id is subject to change across a frame cache
reset. This is why save_selected_frame and restore_selected_frame have
special handling for frame 0:
For this last reason, we also need to handle frame 0 specially in
prepare_reinflate / reinflate. Because the frame id of frame 0 can
change across a frame cache reset, we must not rely on the frame id from
that frame to reinflate it. We should instead just re-fetch the current
frame at that point.
This patch adds a frame_info_ptr::m_cached_level field, set in
frame_info_ptr::prepare_reinflate, so we can tell if a frame is frame 0.
There are cases where a frame_info_ptr object wraps a sentinel frame,
for which frame_relative_level returns -1, so I have chosen the value -2
to represent "invalid frame level", for when the frame_info_ptr object
is empty.
In frame_info_ptr::prepare_reinflate, only cache the frame id if the
frame level is not 0. It's fine to cache the frame id for the sentinel
frame, it will be properly handled by frame_find_by_id later.
In frame_info_ptr::reinflate, if the frame level is 0, call
get_current_frame to get the target's current frame. Otherwise, use
frame_find_by_id just as before.
This patch should not have user-visible changes with upstream GDB. But
it will avoid forcing the computation of frame 0's when calling
prepare_reinflate. And, well, it fixes the upcoming ROCm GDB patch
series.
Change-Id: I176ed7ee9317ddbb190acee8366e087e08e4d266 Reviewed-By: Bruno Larsen <blarsen@redhat.com>
Simon Marchi [Fri, 21 Oct 2022 15:49:44 +0000 (11:49 -0400)]
gdb: add missing prepare_reinflate call in print_frame_info
print_frame_info calls frame_info_ptr::reinflate, but not
frame_info_ptr::prepare_reinflate, add the call to prepare_reinflate.
It works right now, because all callers of print_frame_info that could
possibly lead to the pretty printers being called, and the frame_info
objects being invalidated, do call prepare_reinflate themselves. And
since the cached frame id is copied when passing a frame_info_ptr by
value, print_frame_info does have a cached frame id on entry. So
technically, this change isn't needed. But I don't think it's good for
a function to rely on its callers to have called prepare_reinflate, if
it intends to call reinflate.
Change-Id: Ie332b2d5479aef46f83fdc1120c7c83f4e84d1b0 Reviewed-By: Bruno Larsen <blarsen@redhat.com>
Simon Marchi [Mon, 24 Oct 2022 20:16:09 +0000 (16:16 -0400)]
gdb: use frame_id_p instead of comparing to null_frame_id in frame_info_ptr::reinflate
The assertion
gdb_assert (m_cached_id != null_frame_id);
is always true, as comparing equal to null_frame_id is always false
(it's the first case in frame_id::operator==, not sure why it's not this
way, but that's what it is).
Replace the comparison with a call to frame_id_p.
Approved-By: Tom Tromey <tom@tromey.com>
Change-Id: I93986e6a85ac56353690792552e5b3b4cedec7fb
Simon Marchi [Mon, 24 Oct 2022 19:57:26 +0000 (15:57 -0400)]
gdb: remove manual frame_info reinflation code in backtrace_command_1
With the following patch applied (gdb: use frame_id_p instead of
comparing to null_frame_id in frame_info_ptr::reinflate), I would get:
$ ./gdb -q -nx --data-directory=data-directory testsuite/outputs/gdb.base/bt-selected-frame/bt-selected-frame -ex "b breakpt" -ex r -ex "bt full"
Reading symbols from testsuite/outputs/gdb.base/bt-selected-frame/bt-selected-frame...
Breakpoint 1 at 0x1131: file /home/smarchi/src/binutils-gdb/gdb/testsuite/gdb.base/bt-selected-frame.c, line 22.
Starting program: /home/smarchi/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/bt-selected-frame/bt-selected-frame
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
Breakpoint 1, breakpt () at /home/smarchi/src/binutils-gdb/gdb/testsuite/gdb.base/bt-selected-frame.c:22
22 }
#0 breakpt () at /home/smarchi/src/binutils-gdb/gdb/testsuite/gdb.base/bt-selected-frame.c:22
No locals.
/home/smarchi/src/binutils-gdb/gdb/frame-info.c:42: internal-error: reinflate: Assertion `frame_id_p (m_cached_id)' failed.
This is because the code in backtrace_command_1 to manually reinflate
`fi` steps overs frame_info_ptr's toes.
When calling
fi.prepare_reinflate ();
`fi` gets properly filled with the cached frame id. But when this
happens:
fi = frame_find_by_id (frame_id);
`fi` gets replaced by a brand new frame_info_ptr that doesn't have a
cached frame id. Then this is called without a cached frame id:
fi.reinflate ();
That doesn't cause any problem currently, since
- the gdb_assert in the reinflate method doesn't actually do anything
(the following patch fixes that)
- `fi.m_ptr` will always be non-nullptr, since we just got it from
frame_find_by_id, so reinflate will not do anything, it won't try to
use m_cached_id
Fix that by removing the code to manually re-fetch the frame. That
should be taken care of by frame_info_ptr::reinflate.
Note that the old code checked if we successfully re-inflated the frame
or not, and if not it did emit a warning. The equivalent in
frame_info_ptr::reinflate asserts that the frame has been successfully
re-inflated. It's not clear if / when this can happen, but if it can
happen, we'll need to find a solution to this problem globally
(everywhere a frame_info_ptr can be re-inflated), not just here. So I
propose to leave it like this, until it does become a problem.
Reviewed-By: Bruno Larsen <blarsen@redhat.com>
Change-Id: I07b783d94e2853e0a2d058fe7deaf04eddf24835
Simon Marchi [Fri, 21 Oct 2022 20:06:59 +0000 (16:06 -0400)]
gdb: move frame_info_ptr method implementations to frame-info.c
I don't see any particular reason why the implementations of the
frame_info_ptr object are in the header file. It only seems to add some
complexity. Since we can't include frame.h in frame-info.h, we have to
add declarations of functions defined in frame.c, in frame-info.h. By
moving the implementations to a new frame-info.c, we can avoid that.
Change-Id: I435c828f81b8a3392c43ef018af31effddf6be9c Reviewed-By: Bruno Larsen <blarsen@redhat.com> Reviewed-By: Tom Tromey <tom@tromey.com>
Simon Marchi [Mon, 7 Nov 2022 14:55:23 +0000 (09:55 -0500)]
gdb: add prepare_reinflate/reinflate around print_frame_args in info_frame_command_core
I noticed this crash:
$ ./gdb --data-directory=data-directory -nx -q \
testsuite/outputs/gdb.python/pretty-print-call-by-hand/pretty-print-call-by-hand \
-x testsuite/outputs/gdb.python/pretty-print-call-by-hand/pretty-print-call-by-hand.py \
-ex "b g" -ex r
(gdb) info frame
Stack level 0, frame at 0x7fffffffdd80:
rip = 0x555555555160 in g
(/home/simark/src/binutils-gdb/gdb/testsuite/gdb.python/pretty-print-call-by-hand.c:41); saved rip = 0x5555555551a3
called by frame at 0x7fffffffdda0
source language c.
Arglist at 0x7fffffffdd70, args: mt=mytype is 0x555555556004 "hello world",
depth=10
Fatal signal: Segmentation fault
This is another case of frame_info being invalidated under a function's
feet. The stack trace when the frame_info get invalidated looks like:
... many frames to pretty print the arg, that eventually invalidate the frame_infos ...
#35 0x00005568d0a8ab24 in print_frame_arg (fp_opts=..., arg=0x7ffc3216bcb0) at /home/simark/src/binutils-gdb/gdb/stack.c:489
#36 0x00005568d0a8cc75 in print_frame_args (fp_opts=..., func=0x621000233210, frame=..., num=-1, stream=0x60b000000300)
at /home/simark/src/binutils-gdb/gdb/stack.c:898
#37 0x00005568d0a9536d in info_frame_command_core (fi=..., selected_frame_p=true) at /home/simark/src/binutils-gdb/gdb/stack.c:1682
print_frame_args knows that print_frame_arg can invalidate frame_info
objects, and therefore calls prepare_reinflate/reinflate. However,
info_frame_command_core has a separate frame_info_ptr instance (it is
passed by value / copy). So info_frame_command_core needs to know that
print_frame_args can invalidate frame_info objects, and therefore needs
to prepare_reinflate/reinflate as well. Add those calls, and enhance
the gdb.python/pretty-print-call-by-hand.exp test to test that command.
Reviewed-By: Bruno Larsen <blarsen@redhat.com>
Change-Id: I9edaae06d62e97ffdb30938d364437737238a960
Simon Marchi [Fri, 21 Oct 2022 15:57:15 +0000 (11:57 -0400)]
gdb: clear other.m_cached_id in frame_info_ptr's move ctor
We do it in the move assignment operator, so I think it makes sense to
do it here too for consistency. I don't think it's absolutely necessary
to clear the other object's fields (in other words, copy constructor and
move constructor could be the same), as there is no exclusive resource
being transfered. The important thing is to leave the moved-from object
in an unknown, but valid state. But still, I think that clearing the
fields of the moved-from object is not a bad idea, it helps ensure we
don't rely on the moved-from object after.
Change-Id: Iee900ff9d25dad51d62765d694f2e01524351340 Reviewed-By: Bruno Larsen <blarsen@redhat.com>
Bruno Larsen [Wed, 5 Oct 2022 12:22:56 +0000 (14:22 +0200)]
gdb/c++: Improve error messages in overload resolution
When resolving overloaded functions, GDB relies on knowing relationships
between types, i.e. if a type inherits from another. However, some
compilers may not add complete information for given types as a way to
reduce unnecessary debug information. In these cases, GDB would just say
that it couldn't resolve the method or function, with no extra
information.
The problem is that sometimes the user may not know that the type
information is incomplete, and may just assume that there is a bug in
GDB. To improve the user experience, we attempt to detect if the
overload match failed because of an incomplete type, and warn the user
of this.
This commit also adds a testcase confirming that the message is only
triggered in the correct scenario. This test was not developed as an
expansion of gdb.cp/overload.cc because it needed the dwarf assembler,
and porting all of overload.cc seemed unnecessary.
Bruno Larsen [Wed, 12 Oct 2022 12:51:10 +0000 (14:51 +0200)]
gdb/testsuite: allowed for function_range to deal with mangled functions
When calling get_func_info inside a test case, it would cause failures
if the function was printed using a C++ style mangled name. The current
patch fixes this by allowing for mangled names along with the current
rules.
Alan Modra [Thu, 10 Nov 2022 01:18:01 +0000 (11:48 +1030)]
Sanity check reloc count in get_reloc_upper_bound
The idea here is the stop tools from allocating up to 32G per section
for the arelent pointer array, only to find a little later that the
section reloc count was fuzzed. This usually doesn't hurt much (on
systems that allow malloc overcommit) except when compiled with asan.
We already do this for ELF targets, and while fixing the logic
recently I decided other targets ought to do the same.
Lancelot SIX [Wed, 9 Nov 2022 23:02:44 +0000 (23:02 +0000)]
gdb/testsuite: Fix rtld-step-nodebugsym.exp
The test case introduced in bafcc335266 (Fix stepping in rtld without
debug symbol) fails on some systems as reported by PR/29768. This can
be seen if the system does not have debug info for the libc:
(gdb) step^M
Single stepping until exit from function main,^M
which has no line number information.^M
hello world[Inferior 1 (process 48203) exited normally]^M
(gdb) PASS: gdb.base/rtld-step-nodebugsym.exp: step
continue^M
The program is not being run.^M
(gdb) FAIL: gdb.base/rtld-step-nodebugsym.exp: continue until exit (the program is no longer running)
Without glibc debug info, GDB steps until the program finishes, and
then "gdb_continue_to_end" fails.
As this test was designed to check that GDB does not crash in the "step"
command, the continue does not carry real meaning to the test.
Replace it by "print 0" so we still check that after the step command
GDB is still alive, which is what we care about.
Tested on Ubuntu-22.04 x86_64, with and without libc6-dbg.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29768 Approved-By: Simon Marchi <simon.marchi@efficios.com>
Mike Frysinger [Wed, 9 Nov 2022 19:44:38 +0000 (02:44 +0700)]
sim: ppc: drop support for dgen -L option
Nothing passes this to dgen, and even if it did, nothing would happen
because the generated spreg.[ch] files don't include any references
back to the original data table. So drop it to simplify.
Since we know we'll return 0 by default, we don't have to output case
statements for readonly or length fields whose values are also zero.
This is the most common case by far and thus generates a much smaller
switch table in the end.
Mike Frysinger [Wed, 9 Nov 2022 19:15:34 +0000 (02:15 +0700)]
sim: ppc: collapse is_valid switch table more
Instead of writing:
case 1:
return 1;
case 2:
return 1;
...etc...
Output a single return so we get:
case 1:
case 2:
case ...
return 1;
This saves ~100 lines of code. Hopefully the compiler was already
smart enough to optimize to the same code, but if not, this probably
helps there too :).
Mike Frysinger [Wed, 9 Nov 2022 19:12:42 +0000 (02:12 +0700)]
sim: ppc: pull default switch return out
This saves a single line for the same result. By itself, it's not
interesting, but we can further optimize the generated output and
completely omit the switch table in some cases. Which we'll do in
follow up commits.
Lancelot SIX [Fri, 4 Nov 2022 15:14:38 +0000 (15:14 +0000)]
Fix stepping in rtld without debug symbol
Commit be6276e0aed "Allow debugging of runtime loader / dynamic linker"
introduced a small regression when stepping into the runtime loader /
dynamic linker from function we do not have debug information for. This
is reported in PR/29747.
This can be shown by the following example (given by Simon Marchi in
buzilla bug report):
$ cat test.c
#include <stdio.h>
int main()
{
printf("Hi\n");
return 0;
}
$ gcc test.c -O0 -o test
$ ./gdb -q -nx --data-directory=data-directory test -ex start -ex s
Reading symbols from test...
(No debugging symbols found in test)
Temporary breakpoint 1 at 0x1151
Starting program: .../binutils-gdb/gdb/test
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
Temporary breakpoint 1, 0x0000555555555151 in main ()
Single stepping until exit from function main,
which has no line number information.
/home/smarchi/src/binutils-gdb/gdb/infrun.c:6960:64: runtime error: member call on null pointer of type 'struct symbol'
The crash happens here:
#0 __sanitizer::Die () at ../../../../src/libsanitizer/sanitizer_common/sanitizer_termination.cpp:50
#1 0x00007ffff5dd7128 in __ubsan::__ubsan_handle_type_mismatch_v1_abort (Data=<optimized out>, Pointer=<optimized out>) at ../../../../src/libsanitizer/ubsan/ubsan_handlers.cpp:148
#2 0x000055556183e1a7 in process_event_stop_test (ecs=0x7fffffffccd0) at .../binutils-gdb/gdb/infrun.c:6960
#3 0x0000555561838ea4 in handle_signal_stop (ecs=0x7fffffffccd0) at .../binutils-gdb/gdb/infrun.c:6615
#4 0x000055556182f77b in handle_inferior_event (ecs=0x7fffffffccd0) at .../binutils-gdb/gdb/infrun.c:5866
we dereference, ecs->event_thread->control.step_start_function which is
nullptr.
This patch changes this condition so it evaluates to true if
ecs->event_thread->control.step_start_function is nullptr since this
matches the behaviour before be6276e0aed.
Tested on ubuntu-22.04 x86_64.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29747 Reviewed-By: Bruno Larsen <blarsen@redhat.com> Approved-By: Kevin Buettner <kevinb@redhat.com>
Mike Frysinger [Wed, 9 Nov 2022 17:19:45 +0000 (00:19 +0700)]
sim: restore lstat & mkdir func checks
When merging ppc configure checks into the top-level, these 2 funcs
were accidentally dropped (probably due to incorrect resolution of
conflicts). Restore them since the ppc code utilizes them both.
Mike Frysinger [Wed, 9 Nov 2022 17:11:17 +0000 (00:11 +0700)]
sim: ppc: drop obsolete USE_WIN32API check
This controls only one thing: how to call mkdir(). The gnulib code
already has a mkdir module that provides this exact logic for us, so
punt the code entirely.
gdbserver: do not report btrace support if target does not announce it
Gdbserver unconditionally reports support for btrace packets. Do not
report the support, if the underlying target does not say it supports
it. Otherwise GDB would query the server with btrace-related packets
unnecessarily.
Tom Tromey [Mon, 7 Nov 2022 18:12:35 +0000 (11:12 -0700)]
Allow 'ptype/o' for assembly
PR exp/28359 points out that 'ptype/o' does not work when the current
language is "asm".
I tracked this down to a hard-coded list of languages in typeprint.c.
This patch replaces this list with a method on 'language_defn'
instead. If all languages are ever updated to have this feature, the
method could be removed; but in the meantime this lets each language
control what happens.
I looked at having each print_type method simply modify the flags
itself, but this doesn't work very well with the feature that disables
method-printing by default (but allows it via a flag).
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=28359 Approved-By: Andrew Burgess <aburgess@redhat.com> Approved-By: Keith Seitz <keiths@redhat.com>
Mike Frysinger [Wed, 9 Nov 2022 15:33:27 +0000 (22:33 +0700)]
sim: ppc: add missing parens with e500 macro
This macro expansion was missing a set of outer-most parenthesis which
some compilers would complain about depending on how the macro is used.
This is just standard good macro hygiene too.
Mike Frysinger [Wed, 9 Nov 2022 15:31:47 +0000 (22:31 +0700)]
sim: ppc: drop useless linking of helper tools
We've never run these helper programs directly. The igen program
includes the relevant source files directly and runs the code that
way. So stop wasting developer CPU time linking programs that are
never run. We leave the rules in place for people who need to test
and debug the specific bits of code every now & then.
Jan Beulich [Wed, 9 Nov 2022 10:09:34 +0000 (11:09 +0100)]
x86/Intel: don't accept malformed EXTRQ / INSERTQ
Operand swapping was mistakenly suppressed when the first two operands
were immediate ones, not taking into account overall operand count. This
way EXTRQ / INSERTQ would have been accepted also with kind-of-AT&T
operand order.
For the testcase being extended, in order to not move around "GAS
LISTING" expectations, suppress pagination.
Alan Modra [Wed, 9 Nov 2022 06:07:42 +0000 (16:37 +1030)]
Re: Fuzzed files in archives
Like commit ffbe89531c2e this avoids more silliness writing output
that is going to be deleted. bfd_close and bfd_close_all_done differ
in that only the former calls _bfd_write_contents.
* objcopy.c (copy_archive): Don't call bfd_close for elements
that are going to be deleted, call bfd_close_all_done instead.
Do the same for the archive itself.
RISC-V: xtheadfmemidx: Use fp register in mnemonics
Although the encoding for scalar and fp registers is identical,
we should follow common pratice and use fp register names
when referencing fp registers.
The xtheadmemidx extension consists of indirect load/store instructions
which all load to or store from fp registers.
Let's use fp register names in this case and adjust the test cases
accordingly.
gas/
* testsuite/gas/riscv/x-thead-fmemidx-fail.l: Updated since rd need to
be float register.
* testsuite/gas/riscv/x-thead-fmemidx-fail.s: Likewise.
* testsuite/gas/riscv/x-thead-fmemidx.d: Likewise.
* testsuite/gas/riscv/x-thead-fmemidx.s: Likewise.
opcodes/
* riscv-opc.c (riscv_opcodes): Updated since rd need to be float register.
Signed-off-by: Christoph Müllner <christoph.muellner@vrull.eu>
Simon Marchi [Fri, 4 Nov 2022 14:07:09 +0000 (10:07 -0400)]
gdb/linux-nat: get core count using /sys/devices/system/cpu/possible
I get this test failure on my CI;
FAIL: gdb.base/info-os.exp: get process list
The particularity of this setup is that builds are done in containers
who are allocated 4 CPUs on a machine that has 40. The code in
nat/linux-osdata.c fails to properly fetch the core number for each
task.
linux_xfer_osdata_processes uses `sysconf (_SC_NPROCESSORS_ONLN)`, which
returns 4, so it allocates an array of 4 integers. However, the core
numbers read from /proc/pid/task/tid/stat, by function
linux_common_core_of_thread, returns a value anywhere between 0 and 39.
The core numbers above 3 are therefore ignored, many processes end up
with no core value, and the regexp in the test doesn't match (it
requires an integer as the core field).
The way this the CPUs are exposed to the container is that the container
sees 40 CPUs "present" and "possible", but only 4 arbitrary CPUs
actually online:
The solution proposed in this patch is to find out the number of
possible CPUs using /sys/devices/system/cpu/possible. In practice, this
will probably always contain `0-N`, where N is the number of CPUs, minus
one. But the documentation [1] doesn't such guarantee, so I'll assume
that it can contain a more complex range list such as `2,4-31,32-63`,
like the other files in that directory can have. The solution is to
iterate over these numbers to find the highest possible CPU id, and
use that that value plus one as the size of the array to allocate.
Simon Marchi [Fri, 4 Nov 2022 13:39:12 +0000 (09:39 -0400)]
gdbsupport, gdb: add read_text_file_to_string, use it in linux_common_core_of_thread
I would like to add more code to nat/linux-osdata.c that reads an entire
file from /proc or /sys and processes it as a string afterwards. I
would like to avoid duplicating the somewhat error-prone code that reads
an entire file to a buffer. I think we should have a utility function
that does that.
Add read_file_to_string to gdbsupport/filestuff.{c,h}, and make
linux_common_core_of_thread use it. I want to make the new function
return an std::string, and because strtok doesn't play well with
std::string (it requires a `char *`, std::string::c_str returns a `const
char *`), change linux_common_core_of_thread to use std::string methods
instead.
Approved-By: Tom Tromey <tom@tromey.com>
Change-Id: I1793fda72a82969c28b944a84acb953f74c9230a
Tom de Vries [Tue, 8 Nov 2022 17:47:24 +0000 (18:47 +0100)]
[gdb/cli] Make quit really quit after remote connection closed
Consider a hello world a.out, started using gdbserver:
...
$ gdbserver --once 127.0.0.1:2345 ./a.out
Process ./a.out created; pid = 15743
Listening on port 2345
...
that we can connect to using gdb:
...
$ gdb -ex "target remote 127.0.0.1:2345"
Remote debugging using 127.0.0.1:2345
Reading /home/vries/a.out from remote target...
...
0x00007ffff7dd4550 in _start () from target:/lib64/ld-linux-x86-64.so.2
(gdb)
...
After that, we can for instance quit with confirmation:
...
(gdb) quit
A debugging session is active.
Inferior 1 [process 16691] will be killed.
Quit anyway? (y or n) y
$
...
Or, kill with confirmation and quit:
...
(gdb) kill
Kill the program being debugged? (y or n) y
[Inferior 1 (process 16829) killed]
(gdb) quit
$
...
Or, monitor exit, kill with confirmation, and quit:
...
(gdb) monitor exit
(gdb) kill
Kill the program being debugged? (y or n) y
Remote connection closed
(gdb) quit
$
...
But when doing monitor exit followed by quit with confirmation, we get the gdb
prompt back, requiring us to do quit once more:
...
(gdb) monitor exit
(gdb) quit
A debugging session is active.
Inferior 1 [process 16944] will be killed.
Quit anyway? (y or n) y
Remote connection closed
(gdb) quit
$
...
So, the first quit didn't quit. This happens as follows:
- quit_command calls query_if_trace_running
- a TARGET_CLOSE_ERROR is thrown
- it's caught in remote_target::get_trace_status, but then
rethrown because it's TARGET_CLOSE_ERROR
- catch_command_errors catches the error, at which point the quit command
has been aborted.
The TARGET_CLOSE_ERROR is defined as:
...
/* Target throwing an error has been closed. Current command should be
aborted as the inferior state is no longer valid. */
TARGET_CLOSE_ERROR,
...
so in a way this is expected behaviour. But aborting quit because the inferior
state (which we've already confirmed we're not interested in) is no longer
valid, and having to type quit again seems pointless.
Furthermore, the purpose of not catching errors thrown by
query_if_trace_running as per commit 2f9d54cfcef ("make -gdb-exit call
disconnect_tracing too, and don't lose history if the target errors on
"quit""), was to make sure that error (_("Not confirmed.") had effect.
Fix this in quit_command by catching only the TARGET_CLOSE_ERROR exception
during query_if_trace_running and reporting it:
...
(gdb) monitor exit
(gdb) quit
A debugging session is active.
Inferior 1 [process 19219] will be killed.
Quit anyway? (y or n) y
Remote connection closed
$
...
Tested on x86_64-linux.
PR server/15746
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=15746 Approved-By: Tom Tromey <tom@tromey.com>
Tom de Vries [Tue, 8 Nov 2022 17:40:25 +0000 (18:40 +0100)]
[gdb/testsuite] Remove test-case from test name
Remove test-cases from test-names, such that we don't have the redundant:
...
PASS: gdb.base/corefile.exp: backtrace in corefile.exp
...
but simply:
...
PASS: gdb.base/corefile.exp: backtrace
...
Fixed all instances found using:
...
$ grep ":.*:.*\.exp" gdb.sum
...
Tom de Vries [Tue, 8 Nov 2022 15:28:11 +0000 (16:28 +0100)]
[gdb/testsuite] Fix find_core_file for core named core
With test-case gdb.base/bigcore.exp I run into:
...
(gdb) PASS: gdb.base/bigcore.exp: get inferior pid
signal SIGABRT^M
Continuing with signal SIGABRT.^M
^M
Program terminated with signal SIGABRT, Aborted.^M
The program no longer exists.^M
(gdb) PASS: gdb.base/bigcore.exp: signal SIGABRT
UNTESTED: gdb.base/bigcore.exp: can't generate a core file
...
due to find_core_file returning "".
There is a core file name core:
...
$ ls ./outputs/gdb.base/bigcore
bigcore bigcore.corefile core gdb.cmd.1 gdb.in.1 gdbserver.cmd.1
...
but it's not found.
The problem is this statement:
...
lappend files [list ${::testfile}.core core]
...
which adds a single list item "${::testfile}.core core".
Fix this in the most readable way:
...
lappend files ${::testfile}.core
lappend files core
...
The intention of this code seems to be to indicate that this insn
should not be used and produces undefined behavior, so instead of
setting registers to bogus values, call Unpredictable. This fixes
build warnings due to 32-bit/64-bit type conversions, and outputs
a log message for users at runtime instead of silent corruption.
Mike Frysinger [Tue, 8 Nov 2022 07:09:06 +0000 (14:09 +0700)]
sim: drop unused CORE_ADDR_TYPE
This hasn't been used by gdb in decades, and doesn't make sense with
a standalone sim program/library where the ABI is fixed. So punt it
to simplify the code.
Mike Frysinger [Mon, 7 Nov 2022 16:07:43 +0000 (23:07 +0700)]
sim: v850: drop subdir configure logic
We've been using this only to set the default word size to 32. We
can easily move this into the makefile via a -D compiler flag and
clean up the build logic quite a bit.
Mike Frysinger [Mon, 7 Nov 2022 16:05:46 +0000 (23:05 +0700)]
sim: mn10300: drop subdir configure logic
We've been using this only to set the default word size to 32. We
can easily move this into the makefile via a -D compiler flag and
clean up the build logic quite a bit.
Mike Frysinger [Mon, 7 Nov 2022 16:04:25 +0000 (23:04 +0700)]
sim: or1k: drop subdir configure logic
We've been using this only to set the default word size to 32. We
can easily move this into the makefile via a -D compiler flag and
clean up the build logic quite a bit.
Mike Frysinger [Mon, 7 Nov 2022 15:59:30 +0000 (22:59 +0700)]
sim: bpf: drop subdir configure logic
We've been using this only to set the default word size to 64. We
can easily move this into the makefile via a -D compiler flag and
clean up the build logic quite a bit.
Mike Frysinger [Mon, 7 Nov 2022 15:56:58 +0000 (22:56 +0700)]
sim: riscv: drop subdir configure logic
We've been using this only to set the default word size to 32-vs-64
based on the $target. We can easily merge this with the top-level
configure script to clean things up a bit.
This patch changes the GDB build system in order to use libtool to
link the several built executables. This makes it possible to refer
to libtool libraries (.la files) in CLIBS.
As an application of the above,
BFD now refers to ../libbfd/libbfd.la
OPCODES now refers to ../opcodes/libopcodes.la
LIBBACKTRACE_LIB now refers to ../libbacktrace/libbacktrace.la
LIBCTF now refers to ../libctf/libctf.la
NOTE1: The addition of libtool adds a few new configure-time options
to GDB. Among these, --enable-shared and --disable-shared, which were
previously ignored. Now GDB shall honor these options when linking,
picking up the right version of the referred libtool libraries
automagically.
NOTE2: I have not tested the insight build.
NOTE3: For regenerating configure I used an environment with Autoconf
2.69 and Automake 1.15.1. This should match the previously
used version as announced in the configure script.
NOTE4: Now the installed shared objects libbfd.so, libopcodes.so and
libctf.so are used by gdb if binutils is installed with
--enable-shared.
Testing performed:
- --enable-shared and --disable-shared (the default in binutils) work
as expected: the linked executables link with the archive or shared
libraries transparently.
- Makefile.in modified for EXEEXT = .exe. It installs the binaries
just fine. The installed gdb.exe runs fine.
- Native build regtested in x86_64. No regressions found.
- Cross build for aarch64-linux-gnu built to exercise
program_transform_name and friends. The installed
aarch64-linux-gnu-gdb runs fine.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29372 Approved-By: Simon Marchi <simon.marchi@efficios.com>
Simon Marchi [Mon, 7 Nov 2022 14:38:12 +0000 (09:38 -0500)]
gdb/testsuite: use a more unique name in gdb.mi/mi-breakpoint-multiple-locations.exp
I see failures in this test, due to the function name "add" being too
generic, and unexpected breakpoint locations being found in my
libstdc++, such as (wrapped for readability):
Pedro Alves [Fri, 14 Oct 2022 19:17:36 +0000 (20:17 +0100)]
Don't explicitly set clone child ptrace options
linux_handle_extended_wait calls target_post_attach if we're handling
a PTRACE_EVENT_CLONE, and libthread_db.so isn't active.
target_post_attach just calls linux_init_ptrace_procfs to set the
lwp's ptrace options. However, this is completely unnecessary,
because, as man ptrace [1] says, options are inherited:
"Flags are inherited by new tracees created and "auto-attached" via
active PTRACE_O_TRACEFORK, PTRACE_O_TRACEVFORK, or PTRACE_O_TRACECLONE
options."
Mike Frysinger [Mon, 7 Nov 2022 14:30:10 +0000 (21:30 +0700)]
sim: build: add a proper var for enabled arches
The install code was using $SUBDIRS to track all enabled arches. This
works, but isn't great if we want to add a subdir that isn't an arch
port, or as we merge the subdirs into the top-level. Create a new var
explicitly to track the list of enabled arches instead.
Christophe Lyon [Fri, 4 Nov 2022 10:06:47 +0000 (11:06 +0100)]
configure: require libzstd >= 1.4.0
gas uses ZSTD_compressStream2 which is only available with libzstd >=
1.4.0, leading to build errors when an older version is installed.
This patch updates the check libzstd presence to check its version is
>= 1.4.0. However, since gas seems to be the only component requiring
such a recent version this may imply that we disable ZSTD support for
all components although some would still benefit from an older
version.
I ran 'autoreconf -f' in all directories containing a configure.ac
file, using vanilla autoconf-2.69 and automake-1.15.1. I noticed
several errors from autoheader in readline, as well as warnings in
intl, but they are unrelated to this patch.
Clément Chigot [Fri, 4 Nov 2022 15:52:05 +0000 (16:52 +0100)]
ld/testsuite: skip tests related to -shared when disabled
Call the helper function "check_shared_lib_support" to ensure -shared
is enabled before launching ld-shared, ld-elfweak and ld-elfvers.
This allows to catch custom targets explicitly disabling it.
Tsukasa OI [Sat, 10 Sep 2022 08:32:17 +0000 (08:32 +0000)]
RISC-V: Remove RV32EF conflict
Despite that the RISC-V ISA Manual version 2.2 prohibited "RV32EF", later
versions beginning with the version 20190608-Base-Ratified removed this
restriction. Because the 'E' extension is still a draft, the author chose
to *just* remove the conflict (not checking the ISA version).
Note that, because RV32E is only used with a soft-float calling convention,
there's no valid official ABI for RV32EF. It means, even if we can assemble
a program with -march=rv32ef -mabi=ilp32e, floating-point registers are kept
in an unmanaged state (outside ABI management).
The purpose of this commit is to suppress unnecessary errors while parsing
an ISA string and/or disassembling, not to allow hard-float with RVE.
bfd/ChangeLog:
* elfxx-riscv.c (riscv_parse_check_conflicts): Accept RV32EF
because only older specifications disallowed it.
gas/ChangeLog:
* testsuite/gas/riscv/march-fail-rv32ef.d: Remove as not directly
prohibited.
* testsuite/gas/riscv/march-fail-rv32ef.l: Likewise.
Mike Frysinger [Sun, 6 Nov 2022 14:08:04 +0000 (21:08 +0700)]
sim: build: stop passing down SIM_PRIMARY_TARGET
This was needed when the install step was run in subdirs, but now
that we process that entirely in the top-level, we don't need to
pass this down, so drop it.
Tom Tromey [Wed, 21 Sep 2022 16:46:51 +0000 (10:46 -0600)]
Deprecate MI version 1
MI version 1 is long since obsolete. Rather than remove it
immediately (though I did send a patch for that), instead let's
deprecate it in GDB 13 and then remove it for GDB 14.
This version of the patch incorporates Simon's warning change, and
Luis' recommendation to mention the gdb versions here.
Mike Frysinger [Thu, 3 Nov 2022 11:19:13 +0000 (18:19 +0700)]
sim: run: move linking into top-level
Automake will run each subdir individually before moving on to the next
one. This means that the linking phase, a single threaded process, will
not run in parallel with anything else. When we have to link ~32 ports,
that's 32 link steps that don't take advantage of parallel systems. On
my really old 4-core system, this cuts a multi-target build from ~60 sec
to ~30 sec. We eventually want to move all compile+link steps to this
common dir anyways, so might as well move linking now for a nice speedup.
We use noinst_PROGRAMS instead of bin_PROGRAMS because we're taking care
of the install ourselves rather than letting automake process it.
Mike Frysinger [Sat, 5 Nov 2022 07:35:00 +0000 (14:35 +0700)]
sim: build: move install steps to the top-level
We still have to maintain custom install rules due to how we rename
arch-specific files with an arch prefix in their name, but we can at
least unify the logic in the common dir.
Mike Frysinger [Sat, 5 Nov 2022 03:02:58 +0000 (10:02 +0700)]
sim: ppc: drop unused /dev/zero logic
Nothing in the tree checks this option, or has checked for decades.
The pre-cvs-import ChangeLog suggests this was added & removed back
then, but can't be sure as that history doesn't exist in the VCS.
Mike Frysinger [Sat, 5 Nov 2022 02:52:29 +0000 (09:52 +0700)]
sim: ppc: delete unused host bitsize settings
Nothing checks this define anywhere, so drop all the logic. We don't
want this to be a configure option in the first place as all such usage
should be automatic & following proper types.
Mike Frysinger [Sat, 5 Nov 2022 02:29:17 +0000 (09:29 +0700)]
sim: ppc: inline the sim-packages option
This has only ever had a single option that's enabled by default.
The objects it adds are pretty small and don't add overhead at
runtime if it isn't used, so just enable it all the time to make
the build code simpler.