The PowerPC ioctl value for ioctl_TCSETW is 0x802c7415. The variable
ioctl_TCSETW is defined in gdb/linux-record.h as an int. The TCSETW value
has the MSB set to one so it is a negative integer. The comparison of the
unsigned long value tmpulongest to a negative integer value for
ioctl_TCSETSW fails.
This patch changes the declarations for the ioctl_* values in struct
linux_record_tdep to unsigned long to fix the comparisons between
tmpulongest and the tdep->ioctl_* values.
An additional test gdb.reverse/test_ioctl_TCSETSW.exp is added to verify
the gdb record_linux_system_call() if statement for the ioctl TCSETSW
succeeds.
This patch has been tested on Power 10 and Intel with no test failures.
Carl Love [Fri, 10 Jun 2022 16:17:27 +0000 (16:17 +0000)]
PowerPC, correct the gdb ioctl values for TCGETS, TCSETS, TCSETSW and TCSETSF.
Some of the ioctl numbers are based on the size of kernel termios structure.
Currently the PowerPC GDB definitions are "hard coded" into the ioctl
number.
The current PowerPC values for TCGETS, TCSETS, TCSETSW and TCSETSF are
defined in gdb/ppc-linux-tdep.c as:
Where the termios structure size is in hex digits [5:4] as 0x3c.
The definition for the PowerPC termios structure is given in:
arch/powerpc/include/uapi/asm/termbits.h
The size of the termios data structure in this file is 0x2c not 0x3c.
This patch changes the hex digits for the size of the PowerPC termios size
in the ioctl values for TCGETS, TCSETS, TCSETSW and TCSETSF to 0x2c.
This patch also changes the hard coding to generate the number based on a
it easier to update the ioctl numbers.
Andrew Burgess [Thu, 9 Jun 2022 15:56:35 +0000 (16:56 +0100)]
gdb/testsuite: remove definition of true/false from gdb_compiler_info
Since pretty much forever the get_compiler_info function has included
these lines:
# Most compilers will evaluate comparisons and other boolean
# operations to 0 or 1.
uplevel \#0 { set true 1 }
uplevel \#0 { set false 0 }
These define global variables true (to 1) and false (to 0).
It seems odd to me that these globals are defined in
get_compiler_info, I guess maybe the original thinking was that if a
compiler had different true/false values then we would detect it there
and define true/false differently.
I don't think we should be bundling this logic into get_compiler_info,
it seems weird to me that in order to use $true/$false a user needs to
first call get_compiler_info.
It would be better I think if each test script that wants these
variables just defined them itself, if in the future we did need
different true/false values based on compiler version then we'd just
do:
if { [test_compiler_info "some_pattern"] } {
# Defined true/false one way...
} else {
# Defined true/false another way...
}
But given the current true/false definitions have been in place since
at least 1999, I suspect this will not be needed any time soon.
Given that the definitions of true/false are so simple, right now my
suggestion is just to define them in each test script that wants
them (there's not that many). If we ever did need more complex logic
then we can always add a function in gdb.exp that sets up these
globals, but that seems overkill for now.
There should be no change in what is tested after this commit.
Tom de Vries [Fri, 10 Jun 2022 06:59:05 +0000 (08:59 +0200)]
[gdb/testsuite] Fix XPASS with gcc-12 in gdb.base/vla-struct-fields.exp
With gcc-12, I get for test-case gdb.base/vla-struct-fields.exp:
...
(gdb) print inner_vla_struct_object_size == sizeof(inner_vla_struct_object)^M
$7 = 1^M
(gdb) XPASS: gdb.base/vla-struct-fields.exp: size of inner_vla_struct_object
...
Fix this by limiting the xfailing to gcc-11 and earlier. Also, limit the
xfailing to the equality test.
Tom de Vries [Fri, 10 Jun 2022 06:46:17 +0000 (08:46 +0200)]
[gdb/testsuite] Fix timeout in gdb.ada/ghost.exp
On openSUSE Tumbleweed with gcc-12, I run into a timeout:
...
(gdb) print value^M
Multiple matches for value^M
[0] cancel^M
[1] ada.strings.maps.value (<ref> ada.strings.maps.character_mapping; \
character) return character at a-strmap.adb:599^M
[2] pck.value at src/gdb/testsuite/gdb.ada/ghost/pck.ads:17^M
[3] system.object_reader.value (<ref> system.object_reader.object_symbol) \
return system.object_reader.uint64 at s-objrea.adb:2279^M
[4] system.traceback.symbolic.value (system.address) return string at \
s-trasym.adb:200^M
> FAIL: gdb.ada/ghost.exp: print value (timeout)
print ghost_value^M
Argument must be choice number^M
(gdb) FAIL: gdb.ada/ghost.exp: print ghost_value
...
Fix this by prefixing value (as well as the other printed values) with the
package name:
...
(gdb) print pck.value^M
...
Tom Tromey [Thu, 9 Jun 2022 15:40:51 +0000 (09:40 -0600)]
Minor fix to Python breakpoint event documentation
I noticed that the Python event documentation referred to the event's
"breakpoint" field as a function, whereas it is actually an attribute.
This patch fixes the error.
Andrew Burgess [Thu, 9 Jun 2022 10:18:51 +0000 (11:18 +0100)]
gdb/aarch64: fix 32-bit arm compatibility
GDB's ability to run 32-bit ARM processes on an AArch64 native target
is currently broken. The test gdb.multi/multi-arch.exp currently
fails with a timeout.
The cause of these problems is the following three functions:
What has happened, over time, is that these functions have been
modified, forgetting that any particular thread (running on the native
target) might be an ARM thread, or might be an AArch64 thread.
The problems always start with a line similar to this:
The problem with this line is that if 'inf->gdbarch' is an ARM
architecture, then gdbarch_tdep will return a pointer to an
arm_gdbarch_tdep object, not an aarch64_gdbarch_tdep object. The
result of the above cast will, as a consequence, be undefined.
In aarch64_linux_nat_target::thread_architecture, after the undefined
cast we then proceed to make use of TDEP, like this:
if (vq == tdep->vq)
return inf->gdbarch;
Obviously at this point the result is undefined, but, if this check
returns false we then proceed with this code:
As a consequence we will return an AArch64 gdbarch object for our ARM
thread! Things go downhill from there on.
There are similar problems, with similar undefined behaviour, in the
fetch_registers and store_registers functions.
The solution is to make use of a check like this:
if (gdbarch_bfd_arch_info (inf->gdbarch)->bits_per_word == 32)
If the word size is 32 then we know we have an ARM architecture. We
just need to make sure that we perform this check before trying to
read the tdep field.
In aarch64_linux_nat_target::thread_architecture a little reordering,
and the addition of the above check allows us to easily avoid the
undefined behaviour.
For fetch_registers and store_registers I made the decision to split
each of the functions into two new helper functions, and so
aarch64_linux_nat_target::fetch_registers now calls to either
aarch64_fetch_registers or aarch32_fetch_registers, and there's a
similar change for store_registers.
One thing I had to decide was whether to place the new aarch32_*
functions into the aarch32-linux-nat.c file. In the end I decided to
NOT place the functions there, but instead leave them in
aarch64-linux-nat.c, my reasoning was this:
The existing functions in that file are shared from arm-linux-nat.c
and aarch64-linux-nat.c, this generic code to support 32-bit ARM
debugging from either native target.
In contrast, the two new aarch32 functions I have added _only_ make
sense when debugging on an AArch64 native target. These function
shouldn't be called from arm-linux-nat.c at all, and so, if we places
the functions into aarch32-linux-nat.c, the functions would be built
into a 32-bit ARM GDB, but never used.
With that said, there's no technical reason why they couldn't go in
aarch32-linux-nat.c, so if that is preferred I'm happy to move them.
After this commit the gdb.multi/multi-arch.exp passes.
Yvan Roux [Thu, 9 Jun 2022 14:44:50 +0000 (16:44 +0200)]
gdb/arm: Document and fix exception stack offsets
Add a description of exception entry context stacking and fix next
frame offset (at 0xA8 relative to R0 location) as well as FPU
registers ones (starting at 0x68 relative to R0).
Pedro Alves [Thu, 9 Jun 2022 14:32:52 +0000 (15:32 +0100)]
Fix ARM_CC_FOR_TARGET handling
The previous patch that introduced the arm_cc_for_target procedure
moved the ARM_CC_FOR_TARGET global check to that procedure, but forgot
to tell tcl that ARM_CC_FOR_TARGET is a global. As a result,
specifying ARM_CC_FOR_TARGET on the command line actually does
nothing. This fixes it.
Yvan Roux [Thu, 9 Jun 2022 14:31:40 +0000 (16:31 +0200)]
gdb/arm: Terminate unwinding when LR is 0xffffffff
ARMv7-M Architecture Reference "A2.3.1 Arm core registers" states
that LR is set to 0xffffffff on reset.
ARMv8-M Architecture Reference "B3.3 Registers" states that LR is set
to 0xffffffff on warm reset if Main Extension is implemented,
otherwise the value is unknown.
The problem is associated with how compiler_info variable is cached
between calls to get_compiler_info.
Even before the above commit, get_compiler_info supported two
language, C and C++. Calling get_compiler_info would set the global
compiler_info based on the language passed as an argument to
get_compiler_info, and, in theory, compiler_info would not be updated
for the rest of the dejagnu run.
This obviously is slightly broken behaviour. If the first call to
get_compiler_info was for the C++ language then compiler_info would be
set based on the C++ compiler in use, while if the first call to
get_compiler_info was for the C language then compiler_info would be
set based on the C compiler.
This probably wasn't very noticable, assuming a GCC based test
environment then in most cases the C and C++ compiler would be the
same version.
However, if the user starting playing with CC_FOR_TARGET or
CXX_FOR_TARGET, then they might not get the behaviour they expect.
Except, to make matters worse, most of the time, the user probably
would get the behaviour they expected .... except when they didn't!
I'll explain:
In gdb.exp we try to avoid global variables leaking between test
scripts, this is done with the help of the two procs
gdb_setup_known_globals and gdb_cleanup_globals. All known globals
are recorded before a test script starts, and then, when the test
script ends, any new globals are deleted.
Normally, compiler_info is only set as a result of a test script
calling get_compiler_info or test_compiler_info. This means that the
compiler_info global will not exist when the test script starts, but
will exist when the test script end, and so, the compiler_info
variable is deleted at the end of each test.
This means that, in reality, the compiler_info is recalculated once
for each test script, hence, if a test script just checks on the C
compiler, or just checks on the C++ compiler, then compiler_info will
be correct and the user will get the behaviour they expect.
However, if a single test script tries to check both the C and C++
compiler versions then this will not work (even before the above
commit).
The situation is made worse be the behaviour or the load_lib proc.
This proc (provided by dejagnu) will only load each library once.
This means that if a library defines a global, then this global would
normally be deleted at the end of the first test script that includes
the library.
As future attempts to load the library will not actually reload it,
then the global will not be redefined and would be missing for later
test scripts that also tried to load that library.
To work around this issue we override load_lib in gdb.exp, this new
version adds all globals from the newly loaded library to the list of
globals that should be preserved (not deleted).
And this is where things get interesting for us. The library
trace-support.exp includes calls, at the file scope, to things like
is_amd64_regs_target, which cause get_compiler_info to be called.
This means that after loading the library the compiler_info global is
defined.
Our override of load_lib then decides that this new global has to be
preserved, and adds it to the gdb_persistent_globals array.
From that point on compiler_info will never be recomputed!
This commit addresses all the caching problems by doing the following:
Change the compiler_info global into compiler_info_cache global. This
new global is an array, the keys of this array will be each of the
supported languages, and the values will be the compiler version for
that language.
Now, when we call get_compiler_info, if the compiler information for
the specific language has not been computed, then we do that, and add
it to the cache.
Next, compiler_info_cache is defined by calling
gdb_persistent_global. This automatically adds the global to the list
of persistent globals. Now the cache will not be deleted at the end
of each test script.
This means that, for a single test run, we will compute the compiler
version just once for each language, this result will then be cached
between test scripts.
Finally, the legacy 'gcc_compiled' flag is now only set when we call
get_compiler_info with the language 'c'. Without making this change
the value of 'gcc_compiled' would change each time a new language is
passed to get_compiler_info. If the last language was e.g. Fortran,
then gcc_compiled might be left false.
Andrew Burgess [Wed, 8 Jun 2022 13:04:36 +0000 (14:04 +0100)]
gdb/testsuite: handle errors better in test_compiler_info
Now that get_compiler_info might actually fail (if the language is not
handled), then we should try to handle this failure better in
test_compiler_info.
After this commit, if get_compiler_info fails then we will return a
suitable result depending on how the user called test_compiler_info.
If the user does something like:
set version [test_compiler_info "" "unknown-language"]
Then test_compiler_info will return an empty string. My assumption is
that the user will be trying to match 'version' against something, and
the empty string hopefully will not match.
If the user does something like:
if { [test_compiler_info "some_pattern" "unknown-language"] } {
....
}
Then test_compiler_info will return false which seems the obvious
choice.
There should be no change in the test results after this commit.
Andrew Burgess [Wed, 8 Jun 2022 12:35:29 +0000 (13:35 +0100)]
gdb/testsuite: make 'c' default language for get/test compiler info
This commit is a minor cleanup for the two functions (in gdb.exp)
get_compiler_info and test_compiler_info.
Instead of using the empty string as the default language, and just
"knowing" that this means the C language. Make this explicit. The
language argument now defaults to "c" if not specified, and the if
chain in get_compiler_info that checks the language not explicitly
handles "c" and gives an error for unknown languages.
This is a good thing, now that the API appears to take a language, if
somebody does:
test_compiler_info "xxxx" "rust"
to check the version of the rust compiler then we will now give an
error rather than just using the C compiler and leaving the user
having to figure out why they are not getting the results they
expect.
After a little grepping, I think the only place we were explicitly
passing the empty string to either get_compiler_info or
test_compiler_info was in gdb_compile_shlib_1, this is now changed to
pass "c" as the default language.
There should be no changes to the test results after this commit.
Andrew Burgess [Wed, 8 Jun 2022 13:25:00 +0000 (14:25 +0100)]
gdb/testsuite: remove get_compiler_info calls from gdb.exp and dwarf.exp
We don't need to call get_compiler_info before calling
test_compiler_info; test_compiler_info includes a call to
get_compiler_info.
This commit cleans up lib/gdb.exp and lib/dwarf.exp a little by
removing some unneeded calls to get_compiler_info. We could do the
same cleanup throughout the testsuite, but I'm leaving that for
another day.
There should be no change in the test results after this commit.
gdb/testsuite: use test_compiler_info in gcc_major_version
The procedure gcc_major_version was earlier using the global variable
compiler_info to retrieve gcc's major version. This is discouraged and
(as can be read in a comment in compiler.c) compiler_info should be
local to get_compiler_info and test_compiler_info.
The preferred way of getting the compiler string is via calling
test_compiler_info without arguments. Gcc_major_version was changed to
do that.
Andrew Burgess [Thu, 9 Jun 2022 12:40:07 +0000 (13:40 +0100)]
gdb/testsuite: resolve duplicate test names in gdb.threads/tls.exp
While running the gdb.threads/tls.exp test with a GDB configured
without Python, I noticed some duplicate test names.
This is caused by a call to skip_python_tests that is within a proc
that is called multiple times by the test script. Each call to
skip_python_tests results in a call to 'unsupported', and this causes
the duplicate test names.
After this commit we now call skip_python_tests just once and place
the result into a variable. Now, instead of calling skip_python_tests
multiple times, we just check the variable.
There should be no change in what is tested after this commit.
Andrew Burgess [Tue, 7 Jun 2022 16:23:42 +0000 (17:23 +0100)]
gdb: make throw_perror_with_name static
The throw_perror_with_name function is not used outside of utils.c
right now. And as perror_with_name is just a wrapper around
throw_perror_with_name, then any future calls would be to
perror_with_name.
Lets make throw_perror_with_name static.
There should be no user visible changes after this commit.
Andrew Burgess [Tue, 7 Jun 2022 16:18:20 +0000 (17:18 +0100)]
gdb: remove trailing '.' from perror_with_name calls
I ran into this error while working on AArch64 GDB:
Unable to fetch VFP registers.: Invalid argument.
Notice the '.:' in the middle of this error message.
This is because of this call in aarch64-linux-nat.c:
perror_with_name (_("Unable to fetch VFP registers."));
The perror_with_name function take a string, and adds ': <message>' to
the end the string, so I don't think the string that we pass to
perror_with_name should end in '.'.
This commit removes all of the trailing '.' characters from
perror_with_name calls, which give more readable error messages.
I don't believe that any of these errors are tested in the
testsuite (after a little grepping).
Tom Tromey [Wed, 8 Jun 2022 00:55:51 +0000 (18:55 -0600)]
Move CU queue to dwarf2_per_objfile
The CU queue is a member of dwarf2_per_bfd, but it is only used when
expanding CUs. Also, the dwarf2_per_objfile destructor checks the
queue -- however, if the per-BFD object is destroyed first, this will
not work. This was pointed out Lancelot as fallout from the patch to
rewrite the registry system.
This patch avoids this problem by moving the queue to the per-objfile
object.
Andrew Burgess [Fri, 22 Apr 2022 10:23:02 +0000 (11:23 +0100)]
libopcodes: extend the styling within the i386 disassembler
The i386 disassembler is pretty complex. Most disassembly is done
indirectly; operands are built into buffers within a struct instr_info
instance, before finally being printed later in the disassembly
process.
Sometimes the operand buffers are built in a different order to the
order in which they will eventually be printed.
Each operand can contain multiple components, e.g. multiple registers,
immediates, other textual elements (commas, brackets, etc).
When looking for how to apply styling I guess the ideal solution would
be to move away from the operands being a single string that is built
up, and instead have each operand be a list of "parts", where each
part is some text and a style. Then, when we eventually print the
operand we would loop over the parts and print each part with the
correct style.
But it feels like a huge amount of work to move from where we are
now to that potentially ideal solution. Plus, the above solution
would be pretty complex.
So, instead I propose a .... different solution here, one that works
with the existing infrastructure.
As each operand is built up, piece be piece, we pass through style
information. This style information is then encoded into the operand
buffer (see below for details). After this the code can continue to
operate as it does right now in order to manage the set of operand
buffers.
Then, as each operand is printed we can split the operand buffer into
chunks at the style marker boundaries, with each chunk being printed
with the correct style.
For encoding the style information I use a single character, currently
\002, followed by the style encoded as a single hex digit, followed
again by the \002 character.
This of course relies on there not being more than 16 styles, but that
is currently true, and hopefully will remain true for the foreseeable
future.
The other major concern that has arisen around this work is whether
the escape character could ever be encountered in output naturally
generated by the disassembler. If this did happen then the escape
characters would be stripped from the output, and the wrong styling
would be applied.
However, I don't believe that this is currently a problem.
Disassembler content comes from a number of sources. First there's
content that copied directly from the i386-dis.c file, this is things
like register names, and other syntax elements (brackets, commas,
etc). We can easily check that the i386-dis.c file doesn't contain
our special character.
The next source of content are immediate operands. The text for these
operands is generated by calls into libc. By selecting a
non-printable character we can be confident that this is not something
that libc will generate as part of an immediate representation.
The other output that appears to be from the disassembler is operands
that contain addresses and (possibly) symbol names. It is quite
possible that a symbol name might contain any special character we
could imagine, so is this a problem?
I don't think it is, we don't actually print address and symbol
operands through the disassembler, instead, the disassembler calls
back to the user (objdump, gdb, etc) to print the address and symbol
on its behalf. This content is printed directly to the output stream,
it does not pass through the i386 disassembler output buffers. As a
result, we never check this particular output for styling escape
characters.
In some (not very scientific) benchmarking on my machine,
disassembling a reasonably large (142M) shared library, I'm not seeing
any significant slow down in disassembler speed with this change.
Most instructions are now being fully syntax highlighted when I
disassemble using the --disassembler-color=extended-color option. I'm
sure that there are probably still a few corner cases that need fixing
up, but we can come back to them later I think.
When disassembler syntax highlighting is not being used, then there
should be no user visible changes after this commit.
Per Power ISA Version 3.1B 3.3.12, isel uses BC field rather than CRB
field present in binutils sources. Also, per 1.6.2, BC has the same
semantics as BA and BB fields, so this should keep the same flags and
mask, only with the different offset.
opcodes/
* ppc-opc.c
(BC): Define new field, with the same definition as CRB field,
but with the PPC_OPERAND_CR_BIT flag present.
gas/
* testsuite/gas/ppc/476.d: Update.
* testsuite/gas/ppc/a2.d: Update.
* testsuite/gas/ppc/e500.d: Update.
* testsuite/gas/ppc/power7.d: Update.
<snip>
--- a/gas/testsuite/gas/ppc/476.d
+++ b/gas/testsuite/gas/ppc/476.d
@@ -209,7 +209,7 @@ Disassembly of section \.text:
.*: (7c 20 07 8c|8c 07 20 7c) ici 1
.*: (7c 03 27 cc|cc 27 03 7c) icread r3,r4
.*: (50 83 65 36|36 65 83 50) rlwimi r3,r4,12,20,27
-.*: (7c 43 27 1e|1e 27 43 7c) isel r2,r3,r4,28
+.*: (7c 43 27 1e|1e 27 43 7c) isel r2,r3,r4,4\*cr7\+lt
The above change breaks the gdb regression test gdb.arch/powerpc-power7.exp
on Power 7, Power 8, Power 9 and Power 10.
This patch updates the regression test gdb.arch/powerpc-power7.exp with
the new expected output for the isel instruction.
The patch has been tested on Power 7 and Power 10 to verify the patch fixes
the test.
Pedro Alves [Tue, 7 Jun 2022 19:11:32 +0000 (20:11 +0100)]
aarch64: Add fallback if ARM_CC_FOR_TARGET not set
On Aarch64, you can set ARM_CC_FOR_TARGET to point to the 32-bit
compiler to use when testing gdb.multi/multi-arch.exp and
gdb.multi/multi-arch-exec.exp. If you don't set it, then those
testcases don't run.
I guess that approximately nobody remembers to set ARM_CC_FOR_TARGET.
This commit adds a fallback. If ARM_CC_FOR_TARGET is not set, and
testing for Linux, try arm-linux-gnueabi-gcc,
arm-none-linux-gnueabi-gcc, arm-linux-gnueabihf-gcc as 32-bit
compilers, making sure that the produced executable runs on the target
machine before claiming that the compiler produces useful executables.
Alan Modra [Wed, 8 Jun 2022 00:19:09 +0000 (09:49 +0930)]
Don't encode reloc.size
I expect the encoded reloc.size field originally came from aout
r_length ecoding, but somehow went wrong for 64-bit relocs (which
should have been encoded as 3). Toss all that out, just use a byte
size instead. The changes outside of reloc.c in this patch should
make the code independent of how reloc.size is encoded.
Alan Modra [Wed, 8 Jun 2022 08:03:10 +0000 (17:33 +0930)]
Revert reloc howto nits
The "HOWTO size encoding" patch put 1 as the HOWTO size arg for
numerous howtos that are unused, describe dynamic relocs, are markers,
or otherwise are special purpose reloc howtos that don't care about
the size. The idea was to ensure no howto changed by inspecting
object files. Revert those changes, making them zero size.
Alan Modra [Tue, 7 Jun 2022 13:13:20 +0000 (22:43 +0930)]
HOWTO size encoding
This changes the HOWTO macro to encode the howto.size field from a
value given in bytes. This of course requires editing all target
uses of HOWTO, a major pain, but makes it a little nicer to specify
new target HOWTOs. Object files before/after this patch are
unchanged in .data and .rodata.
Alan Modra [Mon, 6 Jun 2022 04:57:17 +0000 (14:27 +0930)]
asan: double free sb_kill
oss-fuzz hits a flaky crash with a double-free. I think this is due
to gas static state not being reinitialised between testcases, a bug
with oss-fuzz not gas. Anyway, this patch should avoid the problem.
* input-scrub.c (input_scrub_push): Move init of sb_index..
(input_scrub_reinit): ..to here.
Tom Tromey [Tue, 5 Apr 2022 19:22:16 +0000 (13:22 -0600)]
Use subclasses of windows_process_info
This changes windows_process_info to use virtual methods for its
callbacks, and then changes the two clients of this code to subclass
this class to implement the methods.
I considered using CRTP here, but that would require making the new
structures visible to the compilation of of nat/windows-nat.c. This
seemed like a bit of a pain, so I didn't do it.
This change then lets us change all the per-inferior globals to be
members of the new subclass. Note that there can still only be a
single inferior -- currently there's a single global of the new type.
This is just another step toward possibly implementing multi-inferior
for Windows.
It's possible this could be cleaned up further... ideally I'd like to
move more of the data into the base class. However, because gdb
supports Cygwin and gdbserver does not, and because I don't have a way
to build or test Cygwin, larger refactorings are difficult.
Tom Tromey [Tue, 5 Apr 2022 19:23:59 +0000 (13:23 -0600)]
Turn some windows-nat.c static functions into methods
This patch turns some windows-nat.c static functions into methods on
windows_nat_target. This avoids having to reference the
windows_nat_target singleton in some more spots -- a minor code
cleanup.
Tom Tromey [Wed, 8 Sep 2021 14:20:29 +0000 (08:20 -0600)]
Allow ASLR to be disabled on Windows
On Windows, it is possible to disable ASLR when creating a process.
This patch adds code to do this, and hooks it up to gdb's existing
disable-randomization feature. Because the Windows documentation
cautions that this isn't available on all versions of Windows, the
CreateProcess wrapper function is updated to make the attempt, and
then fall back to the current approach if it fails.
Tom Tromey [Mon, 6 Jun 2022 15:56:30 +0000 (09:56 -0600)]
Constify solib_name_from_address
I noticed that solib_name_from_address returned a non-const string,
but it's more appropriate to return const. This patch implements
this. Tested by rebuilding.
Tom de Vries [Tue, 7 Jun 2022 07:59:54 +0000 (09:59 +0200)]
[gdb] Allow frv::fr300 in selftests
In skip_arch in gdb/selftest-arch.c we skip architecture fr300 because of
PR20946, but the PR has been fixed by commit 0ae60c3ef45 ("Prevent an abort in
the FRV disassembler if the target bfd name is unknown.") in Januari 2017.
Tom Tromey [Mon, 6 Jun 2022 19:07:37 +0000 (13:07 -0600)]
Consolidate "Python API" sections in NEWS
I noticed that the gdb NEWS file had two "Python API" sections in
"Changes since GDB 12". This patch consolidates the two. I chose to
preserve the second one, first because it is longer, and second
because I felt that user command changes should come before API
changes.
Tom Tromey [Thu, 26 May 2022 16:40:32 +0000 (10:40 -0600)]
Simplify varobj "change" logic
varobj used to store 'print_value' as a C string, where NULL was a
valid value, and so it had logic to handle this situation. However,
at some point this was changed to be a std::string, and so the code
can be simplified in this spot.
Tom Tromey [Thu, 26 May 2022 16:17:02 +0000 (10:17 -0600)]
Remove "-break-insert -r" tests
PR mi/14270 points out that mi-break.exp has some tests for an
unimplemented "-r" switch for "-break-insert". This switch was never
implemented, and is not documented -- though it is mentioned in a
comment in the documentation. This patch removes the test and the doc
comment.
Tom de Vries [Mon, 6 Jun 2022 17:27:46 +0000 (19:27 +0200)]
[gdb] Enable some more print_one_insn selftests
In print_one_insn_test we have this cluster of skipped tests:
...
case bfd_arch_ia64:
case bfd_arch_mep:
case bfd_arch_mips:
case bfd_arch_tic6x:
case bfd_arch_xtensa:
return;
...
Enable some of these, and document in more detail why they're enabled or
skipped.
Likewise, document bfd_arch_or1k because it's an odd case.
The insn is written to gdb_stdout, and there is code in the selftest to add a
newline after the insn, which writes to stream().
The stream() ui_file points into a string buffer, which the disassembler uses
before writing to gdb_stdout, so writing into it after the disassembler has
finished has no effect.
Fix this by using gdb_stdlog and debug_printf (which is what the unit test
infrastructure itself uses) instead, such that we have:
...
Running selftest print_one_insn::A6.
.shor 0x783e
Running selftest print_one_insn::A7.
trap_s 0x1
Running selftest print_one_insn::ARC600.
.shor 0x783e
Running selftest print_one_insn::ARC601.
Running selftest print_one_insn::ARC700.
trap_s 0x1
Running selftest print_one_insn::ARCv2.
trap_s 0x1
Running selftest print_one_insn::Loongarch32.
...
Note: I've also removed the printing of arch_name, which would give
us otherwise the redundant:
...
Running selftest print_one_insn::A6.
arc .shor 0x783e
Running selftest print_one_insn::A7.
arc trap_s 0x1
...
gdb/python: improve formatting of help text for user defined commands
the test that was added (gdb.python/py-doc-reformat.exp) was missing a
call to skip_python_tests. As a result, this test would fail for any
GDB built within Python support.
Tom Tromey [Sun, 5 Jun 2022 16:09:42 +0000 (10:09 -0600)]
Remove obsolete Python 2 comment
I found a comment that referred to Python 2, but that is now obsolete
-- the code it refers to is gone. I'm checking in this patch to
remove the comment.
There's a similar comment elsewhere, but I plan to remove that one in
another patch I'm going to submit shortly.
Tom de Vries [Sat, 4 Jun 2022 11:17:33 +0000 (13:17 +0200)]
[gdb/rust] Fix literal truncation
Make sure we error out on overflow instead of truncating in all cases.
I've used as overflow string: "Integer literal is too large", based
on what I found at
<rust-lang/rust>/src/test/ui/parser/int-literal-too-large-span.rs
but perhaps someone has a better idea.
Tested on x86_64-linux, with a build with --enable-targets=all.
Tom de Vries [Sat, 4 Jun 2022 11:17:33 +0000 (13:17 +0200)]
[gdb/pascal] Fix literal truncation
Make sure we error out on overflow instead of truncating in all cases.
The current implementation of parse_number contains a comment about PR16377,
but that's related to C-like languages. In absence of information of whether
the same fix is needed for pascal, take the conservative approach and keep
behaviour for decimals unchanged.
Tested on x86_64-linux, with a build with --enable-targets=all.
Tom de Vries [Sat, 4 Jun 2022 11:17:33 +0000 (13:17 +0200)]
[gdb/go] Fix literal truncation
Make sure we error out on overflow instead of truncating in all cases.
The current implementation of parse_number contains a comment about PR16377,
but that's related to C-like languages. In absence of information of whether
the same fix is needed for go, take the conservative approach and keep
behaviour for decimals unchanged.
Tested on x86_64-linux, with a build with --enable-targets=all.
Tom de Vries [Sat, 4 Jun 2022 11:17:33 +0000 (13:17 +0200)]
[gdb/fortran] Fix literal truncation
As mentioned in commit 5b758627a18 ("Make gdb.base/parse_number.exp test all
architectures"):
...
There might be a bug that 32-bit fortran truncates 64-bit values to
32-bit, given "p/x 0xffffffffffffffff" returns "0xffffffff".
...
More concretely, we have:
...
$ for arch in i386:x86-64 i386; do \
gdb -q -batch -ex "set arch $arch" -ex "set lang fortran" \
-ex "p /x 0xffffffffffffffff"; \
done
The target architecture is set to "i386:x86-64".
$1 = 0xffffffffffffffff
The target architecture is set to "i386".
$1 = 0xffffffff
...
Fix this by adding a range check in parse_number in gdb/f-exp.y.
Furthermore, make sure we error out on overflow instead of truncating in all
other cases.
Currently we have (decimal for 0x80000000):
...
(gdb) ptype 2147483648
type = unsigned int
...
According to C language rules, unsigned types cannot be used for decimal
constants, so the type should be long instead (reported in PR16377).
Fix this by making sure the type of 2147483648 is long.
The next interesting case is (decimal for 0x8000000000000000):
...
(gdb) ptype 9223372036854775808
type = unsigned long
...
According to the same rules, unsigned long is incorrect.
Current gcc uses __int128 as type, which is allowed, but we don't have that
available in gdb, so the strict response here would be erroring out with
overflow.
Older gcc without __int128 support, as well as clang use an unsigned type, but with
a warning. Interestingly, clang uses "unsigned long long" while gcc uses
"unsigned long", which seems the better choice.
Given that the compilers allow this as a convience, do the same in gdb
and keep type "unsigned long", and make this explicit in parser and test-case.
Furthermore, make sure we error out on overflow instead of truncating in all
cases.
Tom de Vries [Sat, 4 Jun 2022 11:17:32 +0000 (13:17 +0200)]
[gdb/testsuite] Test more values in gdb.base/parse_numbers.exp
Currently we only test value 0xffffffffffffffff in test-case
gdb.base/parse_numbers.exp.
Test more interesting values, both in decimal and hex format, as well as
negative decimals for language modula-2.
This results in an increase in total tests from 15572 to 847448 (55 times
more tests).
Balance out the increase in runtime by reducing the number of architectures
tested: only test one architecture per sizeof longlong/long/int/short
combination, while keeping the possibility intact to run with all
architectures (through setting a variable in the test-case)
Results in slight reduction of total tests: 15572 -> 13853.
Document interesting cases in the expected results:
- wrapping from unsigned to signed
- truncation
- PR16377: using unsigned types to represent decimal constants in C
Running the test-case with a gdb build with -fsanitize=undefined, we trigger
two UB errors in the modula-2 parser, filed as PR29163.
Tom de Vries [Sat, 4 Jun 2022 09:16:22 +0000 (11:16 +0200)]
[gdb/testsuite] Fix gdb.threads/manythreads.exp with check-read1
When running test-case gdb.threads/manythreads.exp with check-read1, I ran
into this hard-to-reproduce FAIL:
...
[New Thread 0x7ffff7318700 (LWP 31125)]^M
[Thread 0x7ffff7321700 (LWP 31124) exited]^M
[New T^C^M
^M
Thread 769 "manythreads" received signal SIGINT, Interrupt.^M
[Switching to Thread 0x7ffff6d66700 (LWP 31287)]^M
0x00007ffff7586a81 in clone () from /lib64/libc.so.6^M
(gdb) FAIL: gdb.threads/manythreads.exp: stop threads 1
...
The matching in the failing gdb_test_multiple is done in an intricate way,
trying to pass on some order and fail on another order.
Fix this by rewriting the regexps to match one line at most, and detecting
invalid order by setting and checking state variables.
Tom de Vries [Sat, 4 Jun 2022 09:11:37 +0000 (11:11 +0200)]
[gdb] Fix warning in print_one_insn::ez80-adl
When running selftest print_one_insn::ez80-adl we run into this warning:
...
Running selftest print_one_insn::ez80-adl.
warning: Unable to determine inferior's software breakpoint type: couldn't
find `_break_handler' function in inferior. Will be used default software \
breakpoint instruction RST 0x08.
...
Fix this by explicitly handling bfd_arch_z80 in print_one_insn_test.
Tom de Vries [Fri, 3 Jun 2022 13:34:50 +0000 (15:34 +0200)]
[gdb] Fix warning in foreach_arch selftests
When running the selftests, I run into:
...
$ gdb -q -batch -ex "maint selftest"
...
Running selftest execute_cfa_program::aarch64:ilp32.
warning: A handler for the OS ABI "GNU/Linux" is not built into this
configuration of GDB. Attempting to continue with the default aarch64:ilp32
settings.
...
and likewise for execute_cfa_program::i8086 and
execute_cfa_program::ia64-elf32.
The warning can easily be reproduced outside the selftests by doing:
...
$ gdb -q -batch -ex "set arch aarch64:ilp32"
...
and can be prevented by first doing "set osabi none".
Fix the warning by setting osabi to none while doing selftests that iterate
over all architectures.
This causes a regression in the print_one_insn selftests for the ARC
architecture.
The problem is pre-existing, and can be demonstrated (already without this
patch) using:
...
$ gdb -q -batch -ex "set osabi none" -ex "maint selftest print_one_insn::A6"
Running selftest print_one_insn::A6.
Self test failed: Cannot access memory at address 0x0
Ran 1 unit tests, 1 failed
$
...
For ARC, we use the generic case in print_one_insn_test, containing this code:
...
int kind = gdbarch_breakpoint_kind_from_pc (gdbarch, &pc);
...
insn = gdbarch_sw_breakpoint_from_kind (gdbarch, kind, &bplen);
...
The problem is that with osabi linux we trigger:
...
static int
arc_linux_breakpoint_kind_from_pc (struct gdbarch *gdbarch, CORE_ADDR *pcptr)
{
return trap_size;
}
...
but with osabi none:
...
arc_breakpoint_kind_from_pc (struct gdbarch *gdbarch, CORE_ADDR *pcptr)
{
size_t length_with_limm = gdb_insn_length (gdbarch, *pcptr);
...
which needs access to memory, and will consequently fail.
Fix this in print_one_insn_test, in the default case, by iterating over
supported osabi's to makes sure we trigger arc_linux_breakpoint_kind_from_pc
which will give us a usable instruction to disassemble.
Tom de Vries [Fri, 3 Jun 2022 13:34:50 +0000 (15:34 +0200)]
Revert "[gdb] Fix warning in foreach_arch selftests"
This reverts commit fc18b1c5afd ("[gdb] Fix warning in foreach_arch
selftests").
The commit introduced regressions for an --enable-targets=all build:
...
Running selftest print_one_insn::A6.^M
Self test failed: Cannot access memory at address 0x0^M
...
and while investigating those I realized that the commit fc18b1c5afd
complicates things by trying to set the current osabi.
So, revert the patch in preparation for a simpler solution.
Roland McGrath [Fri, 3 Jun 2022 05:59:23 +0000 (22:59 -0700)]
gdb: LoongArch: Remove nonportable #include
Don't use gregset.h in *-tdep.c since it's not usable on
hosts that don't have <sys/procfs.h>. It's not needed by
this file, and should only be needed by *-nat.c files.
Alan Modra [Fri, 3 Jun 2022 03:42:01 +0000 (13:12 +0930)]
Re: asan: mips_gprel_reloc segfault
Similarly for the elf mips support.
* elf32-mips.c (mips_elf_final_gp): Don't segfault on symbols
in any of the bfd_is_const_section sections.
* elf64-mips.c (mips_elf64_final_gp): Likewise.
* elfn32-mips.c (mips_elf_final_gp): Likewise.
Alan Modra [Fri, 3 Jun 2022 03:02:00 +0000 (12:32 +0930)]
asan: mips_gprel_reloc segfault
Not just the undefined section has a NULL owner, the absolute section
has too. Which means we can't find output_bfd for __gp. Also, may as
well test directly for output_bfd == NULL.
* coff-mips.c (mips_gprel_reloc): Don't segfault on any of
bfd_is_const_section sections.
Tom de Vries [Thu, 2 Jun 2022 18:40:06 +0000 (20:40 +0200)]
[gdb/testsuite] Detect change instead of init in gdb.mi/mi-var-block.exp
On openSUSE Tumbleweed with target board unix/-m32, I run into:
...
PASS: gdb.mi/mi-var-block.exp: step at do_block_test 2
Expecting: ^(-var-update \*[^M
]+)?(\^done,changelist=\[{name="foo",in_scope="true",type_changed="false",has_more="0"},
{name="cb",in_scope="true",type_changed="false",has_more="0"}\][^M
]+[(]gdb[)] ^M
[ ]*)
-var-update *^M
^done,changelist=[{name="foo",in_scope="true",type_changed="false",has_more="0"}]^M
(gdb) ^M
FAIL: gdb.mi/mi-var-block.exp: update all vars: cb foo changed (unexpected output)
...
The problem is that the test-case attempts to detect a change in the cb
variable caused by this initialization:
...
void
do_block_tests ()
{
int cb = 12;
...
but that only works if the stack location happens to be unequal to 12 before
the initialization.
Fix this by first initializing to 0, and then changing the value to 12:
...
- int cb = 12;
+ int cb = 0;
+ cb = 12;
...
and detecting that change.
Tom Tromey [Wed, 18 May 2022 16:13:32 +0000 (10:13 -0600)]
ODR warning for "main"
"main" is redeclared with a different type in maint.c. I think this
might have come from my first gdb patch, many many years ago. While I
wonder if this profiling code is actually useful at all any more, in
the meantime it's simple to fix the declaration.
Tom Tromey [Wed, 18 May 2022 16:10:57 +0000 (10:10 -0600)]
ODR warnings for "struct insn_decode_record_t"
"struct insn_decode_record_t" is defined in multiple .c files, causing
ODR warnings. This patch renames the types, and removes the use of
"typedef" here -- this is a C-ism that's no longer needed.
... may be the only declaration of "struct target_ops" in scope
(depending on the particular .c file) -- and this declares it in a
namespace, resulting in confusion.
This patch fixes the problem by adding a forward declaration.
Tiezhu Yang [Thu, 2 Jun 2022 06:51:17 +0000 (14:51 +0800)]
gdb: LoongArch: Implement the software_single_step gdbarch method
When execute the following command on LoongArch:
make check-gdb TESTS="gdb.base/branch-to-self.exp"
there exist the following failed testcases:
FAIL: gdb.base/branch-to-self.exp: single-step: si (timeout)
FAIL: gdb.base/branch-to-self.exp: break-cond: side=host: continue to breakpoint: continue to break (timeout)
FAIL: gdb.base/branch-to-self.exp: break-cond: side=host: p counter (timeout)
Implement the software_single_step gdbarch method to decode the current
branch instruction and determine the address of the next instruction on
LoongArch to fix the above failed testcases.
Signed-off-by: Tiezhu Yang <yangtiezhu@loongson.cn>
build_objfile_section_table () creates four synthetic sections per
objfile, which are collected by update_section_map () and passed to
std::sort (). When there are a lot of objfiles, for example, when
debugging JITs, the presence of these sections slows down the sorting
significantly.
The output of update_section_map () is used by find_pc_section (),
which can never return any of these sections: their size is 0, so they
cannot be accepted by bsearch_cmp ().
Filter them (and all the other empty sections) out in
insert_section_p (), which is used only by update_section_map ().
projects / thirdparty / binutils-gdb.git / log
