2024-01-12 23:30:44 +08:00
|
|
|
# Copyright 2012-2024 Free Software Foundation, Inc.
|
2016-06-18 02:24:08 +08:00
|
|
|
|
|
|
|
|
# This program is free software; you can redistribute it and/or modify
|
|
|
|
|
# it under the terms of the GNU General Public License as published by
|
|
|
|
|
# the Free Software Foundation; either version 3 of the License, or
|
|
|
|
|
# (at your option) any later version.
|
|
|
|
|
#
|
|
|
|
|
# This program is distributed in the hope that it will be useful,
|
|
|
|
|
# but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
|
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
|
# GNU General Public License for more details.
|
|
|
|
|
#
|
|
|
|
|
# You should have received a copy of the GNU General Public License
|
|
|
|
|
# along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
|
|
|
|
2022-06-21 02:32:52 +08:00
|
|
|
# Optionally test a Python API here as well.
|
|
|
|
|
load_lib gdb-python.exp
|
|
|
|
|
|
2020-02-11 20:23:25 +08:00
|
|
|
standard_testfile jit-reader-host.c
|
2016-06-18 02:24:08 +08:00
|
|
|
|
2023-01-23 01:05:00 +08:00
|
|
|
require {is_any_target "i?86-*-*" "x86_64-*-*"} is_lp64_target
|
2016-06-18 02:24:08 +08:00
|
|
|
|
2023-01-09 02:47:51 +08:00
|
|
|
require allow_shlib_tests isnative
|
2016-06-18 02:24:08 +08:00
|
|
|
|
2020-06-03 23:18:52 +08:00
|
|
|
# Increase this to see more detail.
|
|
|
|
|
set test_verbose 0
|
|
|
|
|
|
2016-06-18 02:24:08 +08:00
|
|
|
set jit_host_src $srcfile
|
|
|
|
|
set jit_host_bin $binfile
|
|
|
|
|
|
|
|
|
|
# We inject the complete path to jit-reader.h into the source file
|
|
|
|
|
# lest we end up (incorrectly) building against a system-installed
|
|
|
|
|
# version.
|
|
|
|
|
set jit_reader_header [standard_output_file "../../../../../gdb/jit-reader.h"]
|
|
|
|
|
set jit_reader_flag "-DJIT_READER_H=\"$jit_reader_header\""
|
|
|
|
|
|
|
|
|
|
if { [gdb_compile "${srcdir}/${subdir}/${jit_host_src}" "${jit_host_bin}" \
|
|
|
|
|
executable [list debug additional_flags=$jit_reader_flag]] != "" } {
|
2016-12-24 00:52:18 +08:00
|
|
|
untested "failed to compile"
|
2016-06-18 02:24:08 +08:00
|
|
|
return -1
|
|
|
|
|
}
|
|
|
|
|
|
2020-02-11 20:23:25 +08:00
|
|
|
set jit_reader jit-reader
|
2016-06-18 02:24:08 +08:00
|
|
|
set jit_reader_src ${jit_reader}.c
|
|
|
|
|
set jit_reader_bin [standard_output_file ${jit_reader}.so]
|
|
|
|
|
|
|
|
|
|
if { [gdb_compile_shlib "${srcdir}/${subdir}/${jit_reader_src}" "${jit_reader_bin}" \
|
|
|
|
|
[list debug additional_flags=$jit_reader_flag]] != "" } {
|
2016-12-24 00:52:18 +08:00
|
|
|
untested "failed to compile"
|
2016-06-18 02:24:08 +08:00
|
|
|
return -1
|
|
|
|
|
}
|
|
|
|
|
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
# Test "info registers" in the current frame, expecting RSP's value to
|
|
|
|
|
# be SP.
|
|
|
|
|
|
|
|
|
|
proc info_registers_current_frame {sp} {
|
|
|
|
|
global hex decimal
|
|
|
|
|
|
|
|
|
|
set any "\[^\r\n\]*"
|
|
|
|
|
|
2019-10-16 22:53:37 +08:00
|
|
|
set neg_decimal "-?$decimal"
|
2020-02-11 20:46:27 +08:00
|
|
|
|
|
|
|
|
set expected \
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
[multi_line \
|
2019-10-16 22:53:37 +08:00
|
|
|
"rax $hex +$neg_decimal" \
|
|
|
|
|
"rbx $hex +$neg_decimal" \
|
|
|
|
|
"rcx $hex +$neg_decimal" \
|
|
|
|
|
"rdx $hex +$neg_decimal" \
|
|
|
|
|
"rsi $hex +$neg_decimal" \
|
|
|
|
|
"rdi $hex +$neg_decimal" \
|
2018-06-23 01:40:47 +08:00
|
|
|
"rbp $hex +$hex" \
|
|
|
|
|
"rsp $sp +$sp" \
|
2019-10-16 22:53:37 +08:00
|
|
|
"r8 $hex +$neg_decimal" \
|
|
|
|
|
"r9 $hex +$neg_decimal" \
|
|
|
|
|
"r10 $hex +$neg_decimal" \
|
|
|
|
|
"r11 $hex +$neg_decimal" \
|
|
|
|
|
"r12 $hex +$neg_decimal" \
|
|
|
|
|
"r13 $hex +$neg_decimal" \
|
|
|
|
|
"r14 $hex +$neg_decimal" \
|
|
|
|
|
"r15 $hex +$neg_decimal" \
|
2018-06-23 01:40:47 +08:00
|
|
|
"rip $hex +$hex$any" \
|
|
|
|
|
"eflags $hex +\\\[$any\\\]" \
|
2019-10-16 22:53:37 +08:00
|
|
|
"cs $hex +$neg_decimal" \
|
|
|
|
|
"ss $hex +$neg_decimal" \
|
|
|
|
|
"ds $hex +$neg_decimal" \
|
|
|
|
|
"es $hex +$neg_decimal" \
|
|
|
|
|
"fs $hex +$neg_decimal" \
|
|
|
|
|
"gs $hex +$neg_decimal" \
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
]
|
2020-02-11 20:46:27 +08:00
|
|
|
|
|
|
|
|
# There may be more registers.
|
|
|
|
|
append expected ".*"
|
|
|
|
|
|
|
|
|
|
gdb_test "info registers" $expected
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
}
|
|
|
|
|
|
2016-06-18 02:24:08 +08:00
|
|
|
proc jit_reader_test {} {
|
|
|
|
|
global jit_host_bin
|
|
|
|
|
global jit_reader_bin
|
2020-06-03 23:18:52 +08:00
|
|
|
global test_verbose
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
global hex decimal
|
|
|
|
|
|
|
|
|
|
set any "\[^\r\n\]*"
|
2016-06-18 02:24:08 +08:00
|
|
|
|
|
|
|
|
clean_restart $jit_host_bin
|
|
|
|
|
gdb_load_shlib $jit_reader_bin
|
|
|
|
|
|
2020-06-03 23:18:52 +08:00
|
|
|
if {$test_verbose > 0} {
|
2016-06-18 02:24:08 +08:00
|
|
|
gdb_test_no_output "set debug jit 1"
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
gdb_test_no_output "jit-reader-load ${jit_reader_bin}" "jit-reader-load"
|
|
|
|
|
gdb_run_cmd
|
|
|
|
|
gdb_test "" "Program received signal SIGTRAP, .*" "expect SIGTRAP"
|
|
|
|
|
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
# Test the JIT reader unwinder.
|
|
|
|
|
with_test_prefix "with jit-reader" {
|
|
|
|
|
|
|
|
|
|
with_test_prefix "before mangling" {
|
|
|
|
|
gdb_test "bt" \
|
|
|
|
|
[multi_line \
|
Fix double-free when creating more than one block in JIT debug info reader
A double-free happens when using a JIT debug info reader that creates
more than one block. In the loop that frees blocks in finalize_symtab,
at the very end, the gdb_block_iter_tmp variable is set initially, but
not changed as the loop advances. If we have two blocks, the first
iteration frees the first block, the second iteration frees the second
block, but the third iteration tries to free the second block again, as
gdb_block_iter_tmp keeps pointing on the second block.
Fix it by assigning the gdb_block_iter_tmp variable in the loop.
I have improved the jit-reader.exp test to cover this case, by adding a
second "JIT-ed" function and creating a block for it. I have renamed
the existing function to something I find a bit more descriptive. There
are no significant changes to jit-reader.exp itself, only updates
following the renaming. The important changes are in jithost.c
(generate a new function) and in jitreader.c (create a gdb_block for
that function).
This was found because of an ASan report:
$ ./gdb testsuite/outputs/gdb.base/jit-reader/jit-reader -ex "jit-reader-load /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jitreader.so" -ex r
Reading symbols from testsuite/outputs/gdb.base/jit-reader/jit-reader...
Starting program: /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jit-reader
=================================================================
==1751048==ERROR: AddressSanitizer: heap-use-after-free on address 0x604000042eb8 at pc 0x5650ef8eec88 bp 0x7ffe52767290 sp 0x7ffe52767280
READ of size 8 at 0x604000042eb8 thread T0
#0 0x5650ef8eec87 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:768
#1 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#2 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#3 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#4 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#5 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#6 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
0x604000042eb8 is located 40 bytes inside of 48-byte region [0x604000042e90,0x604000042ec0)
freed by thread T0 here:
#0 0x7fbbe57376b0 in __interceptor_free /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:122
#1 0x5650ef8f350b in xfree<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/common-utils.h:62
#2 0x5650ef8eeca9 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:769
#3 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#4 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
previously allocated by thread T0 here:
#0 0x7fbbe5737cd8 in __interceptor_calloc /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:153
#1 0x5650eef662f3 in xcalloc /home/simark/src/binutils-gdb/gdb/alloc.c:100
#2 0x5650ef8f34ea in xcnew<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/poison.h:122
#3 0x5650ef8ed467 in jit_block_open_impl /home/simark/src/binutils-gdb/gdb/jit.c:557
#4 0x7fbbda98620a in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:60
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
gdb/ChangeLog:
* jit.c (finalize_symtab): Set gdb_block_iter_tmp in loop.
gdb/testsuite/ChangeLog:
* gdb.base/jit-reader.exp (jit_reader_test): Rename
jit_function_00 to jit_function_stack_mangle.
* gdb.base/jithost.c (jit_function_t): Rename to...
(jit_function_stack_mangle_t): ... this.
(jit_function_add_t): New typedef.
(jit_function_00_code): Rename to...
(jit_function_stack_mangle_code): ... this, make static.
(jit_function_add_code): New.
(main): Generate "add" function and call it. Adjust to changes
in jithost_abi.
* gdb.base/jithost.h (struct jithost_abi_bounds): New.
(struct jithost_abi) <begin, end>: Remove fields.
<object, function_stack_mangle, function_add>: New fields.
* gdb.base/jitreader.c (struct reader_state) <code_begin,
code_end>: Remove fields.
<func_stack_mangle>: New field.
(read_debug_info): Adjust to renaming, create block for "add"
function.
(read_sp, unwind_frame, get_frame_id): Adjust to other changes.
2019-12-17 05:30:49 +08:00
|
|
|
"#0 ${any} in jit_function_stack_mangle ${any}" \
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
"#1 ${any} in main ${any}" \
|
|
|
|
|
] \
|
|
|
|
|
"bt works"
|
|
|
|
|
|
|
|
|
|
set sp_before_mangling \
|
|
|
|
|
[get_hexadecimal_valueof "\$sp" 0 "get sp"]
|
|
|
|
|
|
Fix double-free when creating more than one block in JIT debug info reader
A double-free happens when using a JIT debug info reader that creates
more than one block. In the loop that frees blocks in finalize_symtab,
at the very end, the gdb_block_iter_tmp variable is set initially, but
not changed as the loop advances. If we have two blocks, the first
iteration frees the first block, the second iteration frees the second
block, but the third iteration tries to free the second block again, as
gdb_block_iter_tmp keeps pointing on the second block.
Fix it by assigning the gdb_block_iter_tmp variable in the loop.
I have improved the jit-reader.exp test to cover this case, by adding a
second "JIT-ed" function and creating a block for it. I have renamed
the existing function to something I find a bit more descriptive. There
are no significant changes to jit-reader.exp itself, only updates
following the renaming. The important changes are in jithost.c
(generate a new function) and in jitreader.c (create a gdb_block for
that function).
This was found because of an ASan report:
$ ./gdb testsuite/outputs/gdb.base/jit-reader/jit-reader -ex "jit-reader-load /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jitreader.so" -ex r
Reading symbols from testsuite/outputs/gdb.base/jit-reader/jit-reader...
Starting program: /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jit-reader
=================================================================
==1751048==ERROR: AddressSanitizer: heap-use-after-free on address 0x604000042eb8 at pc 0x5650ef8eec88 bp 0x7ffe52767290 sp 0x7ffe52767280
READ of size 8 at 0x604000042eb8 thread T0
#0 0x5650ef8eec87 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:768
#1 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#2 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#3 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#4 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#5 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#6 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
0x604000042eb8 is located 40 bytes inside of 48-byte region [0x604000042e90,0x604000042ec0)
freed by thread T0 here:
#0 0x7fbbe57376b0 in __interceptor_free /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:122
#1 0x5650ef8f350b in xfree<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/common-utils.h:62
#2 0x5650ef8eeca9 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:769
#3 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#4 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
previously allocated by thread T0 here:
#0 0x7fbbe5737cd8 in __interceptor_calloc /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:153
#1 0x5650eef662f3 in xcalloc /home/simark/src/binutils-gdb/gdb/alloc.c:100
#2 0x5650ef8f34ea in xcnew<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/poison.h:122
#3 0x5650ef8ed467 in jit_block_open_impl /home/simark/src/binutils-gdb/gdb/jit.c:557
#4 0x7fbbda98620a in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:60
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
gdb/ChangeLog:
* jit.c (finalize_symtab): Set gdb_block_iter_tmp in loop.
gdb/testsuite/ChangeLog:
* gdb.base/jit-reader.exp (jit_reader_test): Rename
jit_function_00 to jit_function_stack_mangle.
* gdb.base/jithost.c (jit_function_t): Rename to...
(jit_function_stack_mangle_t): ... this.
(jit_function_add_t): New typedef.
(jit_function_00_code): Rename to...
(jit_function_stack_mangle_code): ... this, make static.
(jit_function_add_code): New.
(main): Generate "add" function and call it. Adjust to changes
in jithost_abi.
* gdb.base/jithost.h (struct jithost_abi_bounds): New.
(struct jithost_abi) <begin, end>: Remove fields.
<object, function_stack_mangle, function_add>: New fields.
* gdb.base/jitreader.c (struct reader_state) <code_begin,
code_end>: Remove fields.
<func_stack_mangle>: New field.
(read_debug_info): Adjust to renaming, create block for "add"
function.
(read_sp, unwind_frame, get_frame_id): Adjust to other changes.
2019-12-17 05:30:49 +08:00
|
|
|
gdb_test "up" "#1 $any in main $any\r\n$any function_stack_mangle $any" \
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
"move up to caller"
|
|
|
|
|
|
|
|
|
|
set caller_sp \
|
|
|
|
|
[get_hexadecimal_valueof "\$sp" 0 "get caller sp"]
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
# Step over the instruction that mangles the stack pointer.
|
|
|
|
|
# While that confuses GDB's built-in unwinder, the JIT
|
|
|
|
|
# reader's unwinder understands the mangling and should thus
|
|
|
|
|
# be able to unwind at that location.
|
|
|
|
|
with_test_prefix "after mangling" {
|
Fix double-free when creating more than one block in JIT debug info reader
A double-free happens when using a JIT debug info reader that creates
more than one block. In the loop that frees blocks in finalize_symtab,
at the very end, the gdb_block_iter_tmp variable is set initially, but
not changed as the loop advances. If we have two blocks, the first
iteration frees the first block, the second iteration frees the second
block, but the third iteration tries to free the second block again, as
gdb_block_iter_tmp keeps pointing on the second block.
Fix it by assigning the gdb_block_iter_tmp variable in the loop.
I have improved the jit-reader.exp test to cover this case, by adding a
second "JIT-ed" function and creating a block for it. I have renamed
the existing function to something I find a bit more descriptive. There
are no significant changes to jit-reader.exp itself, only updates
following the renaming. The important changes are in jithost.c
(generate a new function) and in jitreader.c (create a gdb_block for
that function).
This was found because of an ASan report:
$ ./gdb testsuite/outputs/gdb.base/jit-reader/jit-reader -ex "jit-reader-load /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jitreader.so" -ex r
Reading symbols from testsuite/outputs/gdb.base/jit-reader/jit-reader...
Starting program: /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jit-reader
=================================================================
==1751048==ERROR: AddressSanitizer: heap-use-after-free on address 0x604000042eb8 at pc 0x5650ef8eec88 bp 0x7ffe52767290 sp 0x7ffe52767280
READ of size 8 at 0x604000042eb8 thread T0
#0 0x5650ef8eec87 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:768
#1 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#2 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#3 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#4 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#5 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#6 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
0x604000042eb8 is located 40 bytes inside of 48-byte region [0x604000042e90,0x604000042ec0)
freed by thread T0 here:
#0 0x7fbbe57376b0 in __interceptor_free /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:122
#1 0x5650ef8f350b in xfree<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/common-utils.h:62
#2 0x5650ef8eeca9 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:769
#3 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#4 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
previously allocated by thread T0 here:
#0 0x7fbbe5737cd8 in __interceptor_calloc /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:153
#1 0x5650eef662f3 in xcalloc /home/simark/src/binutils-gdb/gdb/alloc.c:100
#2 0x5650ef8f34ea in xcnew<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/poison.h:122
#3 0x5650ef8ed467 in jit_block_open_impl /home/simark/src/binutils-gdb/gdb/jit.c:557
#4 0x7fbbda98620a in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:60
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
gdb/ChangeLog:
* jit.c (finalize_symtab): Set gdb_block_iter_tmp in loop.
gdb/testsuite/ChangeLog:
* gdb.base/jit-reader.exp (jit_reader_test): Rename
jit_function_00 to jit_function_stack_mangle.
* gdb.base/jithost.c (jit_function_t): Rename to...
(jit_function_stack_mangle_t): ... this.
(jit_function_add_t): New typedef.
(jit_function_00_code): Rename to...
(jit_function_stack_mangle_code): ... this, make static.
(jit_function_add_code): New.
(main): Generate "add" function and call it. Adjust to changes
in jithost_abi.
* gdb.base/jithost.h (struct jithost_abi_bounds): New.
(struct jithost_abi) <begin, end>: Remove fields.
<object, function_stack_mangle, function_add>: New fields.
* gdb.base/jitreader.c (struct reader_state) <code_begin,
code_end>: Remove fields.
<func_stack_mangle>: New field.
(read_debug_info): Adjust to renaming, create block for "add"
function.
(read_sp, unwind_frame, get_frame_id): Adjust to other changes.
2019-12-17 05:30:49 +08:00
|
|
|
gdb_test "si" "in jit_function_stack_mangle .*" "step over stack mangling"
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
|
|
|
|
|
set sp_after_mangling \
|
|
|
|
|
[get_hexadecimal_valueof "\$sp" 0 "get sp"]
|
|
|
|
|
|
|
|
|
|
gdb_assert {$sp_before_mangling != $sp_after_mangling} \
|
|
|
|
|
"sp is mangled"
|
|
|
|
|
|
|
|
|
|
# Check that the jit unwinder manages to backtrace through
|
|
|
|
|
# the mangled stack pointer.
|
|
|
|
|
gdb_test "bt" \
|
|
|
|
|
[multi_line \
|
Fix double-free when creating more than one block in JIT debug info reader
A double-free happens when using a JIT debug info reader that creates
more than one block. In the loop that frees blocks in finalize_symtab,
at the very end, the gdb_block_iter_tmp variable is set initially, but
not changed as the loop advances. If we have two blocks, the first
iteration frees the first block, the second iteration frees the second
block, but the third iteration tries to free the second block again, as
gdb_block_iter_tmp keeps pointing on the second block.
Fix it by assigning the gdb_block_iter_tmp variable in the loop.
I have improved the jit-reader.exp test to cover this case, by adding a
second "JIT-ed" function and creating a block for it. I have renamed
the existing function to something I find a bit more descriptive. There
are no significant changes to jit-reader.exp itself, only updates
following the renaming. The important changes are in jithost.c
(generate a new function) and in jitreader.c (create a gdb_block for
that function).
This was found because of an ASan report:
$ ./gdb testsuite/outputs/gdb.base/jit-reader/jit-reader -ex "jit-reader-load /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jitreader.so" -ex r
Reading symbols from testsuite/outputs/gdb.base/jit-reader/jit-reader...
Starting program: /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jit-reader
=================================================================
==1751048==ERROR: AddressSanitizer: heap-use-after-free on address 0x604000042eb8 at pc 0x5650ef8eec88 bp 0x7ffe52767290 sp 0x7ffe52767280
READ of size 8 at 0x604000042eb8 thread T0
#0 0x5650ef8eec87 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:768
#1 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#2 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#3 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#4 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#5 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#6 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
0x604000042eb8 is located 40 bytes inside of 48-byte region [0x604000042e90,0x604000042ec0)
freed by thread T0 here:
#0 0x7fbbe57376b0 in __interceptor_free /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:122
#1 0x5650ef8f350b in xfree<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/common-utils.h:62
#2 0x5650ef8eeca9 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:769
#3 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#4 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
previously allocated by thread T0 here:
#0 0x7fbbe5737cd8 in __interceptor_calloc /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:153
#1 0x5650eef662f3 in xcalloc /home/simark/src/binutils-gdb/gdb/alloc.c:100
#2 0x5650ef8f34ea in xcnew<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/poison.h:122
#3 0x5650ef8ed467 in jit_block_open_impl /home/simark/src/binutils-gdb/gdb/jit.c:557
#4 0x7fbbda98620a in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:60
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
gdb/ChangeLog:
* jit.c (finalize_symtab): Set gdb_block_iter_tmp in loop.
gdb/testsuite/ChangeLog:
* gdb.base/jit-reader.exp (jit_reader_test): Rename
jit_function_00 to jit_function_stack_mangle.
* gdb.base/jithost.c (jit_function_t): Rename to...
(jit_function_stack_mangle_t): ... this.
(jit_function_add_t): New typedef.
(jit_function_00_code): Rename to...
(jit_function_stack_mangle_code): ... this, make static.
(jit_function_add_code): New.
(main): Generate "add" function and call it. Adjust to changes
in jithost_abi.
* gdb.base/jithost.h (struct jithost_abi_bounds): New.
(struct jithost_abi) <begin, end>: Remove fields.
<object, function_stack_mangle, function_add>: New fields.
* gdb.base/jitreader.c (struct reader_state) <code_begin,
code_end>: Remove fields.
<func_stack_mangle>: New field.
(read_debug_info): Adjust to renaming, create block for "add"
function.
(read_sp, unwind_frame, get_frame_id): Adjust to other changes.
2019-12-17 05:30:49 +08:00
|
|
|
"#0 ${any} in jit_function_stack_mangle ${any}" \
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
"#1 ${any} in main ${any}" \
|
|
|
|
|
] \
|
|
|
|
|
"bt works"
|
|
|
|
|
|
|
|
|
|
with_test_prefix "current frame" {
|
|
|
|
|
info_registers_current_frame $sp_after_mangling
|
|
|
|
|
|
|
|
|
|
gdb_test "info frame" \
|
Fix double-free when creating more than one block in JIT debug info reader
A double-free happens when using a JIT debug info reader that creates
more than one block. In the loop that frees blocks in finalize_symtab,
at the very end, the gdb_block_iter_tmp variable is set initially, but
not changed as the loop advances. If we have two blocks, the first
iteration frees the first block, the second iteration frees the second
block, but the third iteration tries to free the second block again, as
gdb_block_iter_tmp keeps pointing on the second block.
Fix it by assigning the gdb_block_iter_tmp variable in the loop.
I have improved the jit-reader.exp test to cover this case, by adding a
second "JIT-ed" function and creating a block for it. I have renamed
the existing function to something I find a bit more descriptive. There
are no significant changes to jit-reader.exp itself, only updates
following the renaming. The important changes are in jithost.c
(generate a new function) and in jitreader.c (create a gdb_block for
that function).
This was found because of an ASan report:
$ ./gdb testsuite/outputs/gdb.base/jit-reader/jit-reader -ex "jit-reader-load /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jitreader.so" -ex r
Reading symbols from testsuite/outputs/gdb.base/jit-reader/jit-reader...
Starting program: /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jit-reader
=================================================================
==1751048==ERROR: AddressSanitizer: heap-use-after-free on address 0x604000042eb8 at pc 0x5650ef8eec88 bp 0x7ffe52767290 sp 0x7ffe52767280
READ of size 8 at 0x604000042eb8 thread T0
#0 0x5650ef8eec87 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:768
#1 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#2 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#3 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#4 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#5 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#6 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
0x604000042eb8 is located 40 bytes inside of 48-byte region [0x604000042e90,0x604000042ec0)
freed by thread T0 here:
#0 0x7fbbe57376b0 in __interceptor_free /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:122
#1 0x5650ef8f350b in xfree<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/common-utils.h:62
#2 0x5650ef8eeca9 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:769
#3 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#4 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
previously allocated by thread T0 here:
#0 0x7fbbe5737cd8 in __interceptor_calloc /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:153
#1 0x5650eef662f3 in xcalloc /home/simark/src/binutils-gdb/gdb/alloc.c:100
#2 0x5650ef8f34ea in xcnew<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/poison.h:122
#3 0x5650ef8ed467 in jit_block_open_impl /home/simark/src/binutils-gdb/gdb/jit.c:557
#4 0x7fbbda98620a in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:60
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
gdb/ChangeLog:
* jit.c (finalize_symtab): Set gdb_block_iter_tmp in loop.
gdb/testsuite/ChangeLog:
* gdb.base/jit-reader.exp (jit_reader_test): Rename
jit_function_00 to jit_function_stack_mangle.
* gdb.base/jithost.c (jit_function_t): Rename to...
(jit_function_stack_mangle_t): ... this.
(jit_function_add_t): New typedef.
(jit_function_00_code): Rename to...
(jit_function_stack_mangle_code): ... this, make static.
(jit_function_add_code): New.
(main): Generate "add" function and call it. Adjust to changes
in jithost_abi.
* gdb.base/jithost.h (struct jithost_abi_bounds): New.
(struct jithost_abi) <begin, end>: Remove fields.
<object, function_stack_mangle, function_add>: New fields.
* gdb.base/jitreader.c (struct reader_state) <code_begin,
code_end>: Remove fields.
<func_stack_mangle>: New field.
(read_debug_info): Adjust to renaming, create block for "add"
function.
(read_sp, unwind_frame, get_frame_id): Adjust to other changes.
2019-12-17 05:30:49 +08:00
|
|
|
"Stack level 0, frame at $sp_before_mangling.*in jit_function_stack_mangle.*"
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
with_test_prefix "caller frame" {
|
Fix double-free when creating more than one block in JIT debug info reader
A double-free happens when using a JIT debug info reader that creates
more than one block. In the loop that frees blocks in finalize_symtab,
at the very end, the gdb_block_iter_tmp variable is set initially, but
not changed as the loop advances. If we have two blocks, the first
iteration frees the first block, the second iteration frees the second
block, but the third iteration tries to free the second block again, as
gdb_block_iter_tmp keeps pointing on the second block.
Fix it by assigning the gdb_block_iter_tmp variable in the loop.
I have improved the jit-reader.exp test to cover this case, by adding a
second "JIT-ed" function and creating a block for it. I have renamed
the existing function to something I find a bit more descriptive. There
are no significant changes to jit-reader.exp itself, only updates
following the renaming. The important changes are in jithost.c
(generate a new function) and in jitreader.c (create a gdb_block for
that function).
This was found because of an ASan report:
$ ./gdb testsuite/outputs/gdb.base/jit-reader/jit-reader -ex "jit-reader-load /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jitreader.so" -ex r
Reading symbols from testsuite/outputs/gdb.base/jit-reader/jit-reader...
Starting program: /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jit-reader
=================================================================
==1751048==ERROR: AddressSanitizer: heap-use-after-free on address 0x604000042eb8 at pc 0x5650ef8eec88 bp 0x7ffe52767290 sp 0x7ffe52767280
READ of size 8 at 0x604000042eb8 thread T0
#0 0x5650ef8eec87 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:768
#1 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#2 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#3 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#4 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#5 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#6 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
0x604000042eb8 is located 40 bytes inside of 48-byte region [0x604000042e90,0x604000042ec0)
freed by thread T0 here:
#0 0x7fbbe57376b0 in __interceptor_free /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:122
#1 0x5650ef8f350b in xfree<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/common-utils.h:62
#2 0x5650ef8eeca9 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:769
#3 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#4 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
previously allocated by thread T0 here:
#0 0x7fbbe5737cd8 in __interceptor_calloc /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:153
#1 0x5650eef662f3 in xcalloc /home/simark/src/binutils-gdb/gdb/alloc.c:100
#2 0x5650ef8f34ea in xcnew<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/poison.h:122
#3 0x5650ef8ed467 in jit_block_open_impl /home/simark/src/binutils-gdb/gdb/jit.c:557
#4 0x7fbbda98620a in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:60
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
gdb/ChangeLog:
* jit.c (finalize_symtab): Set gdb_block_iter_tmp in loop.
gdb/testsuite/ChangeLog:
* gdb.base/jit-reader.exp (jit_reader_test): Rename
jit_function_00 to jit_function_stack_mangle.
* gdb.base/jithost.c (jit_function_t): Rename to...
(jit_function_stack_mangle_t): ... this.
(jit_function_add_t): New typedef.
(jit_function_00_code): Rename to...
(jit_function_stack_mangle_code): ... this, make static.
(jit_function_add_code): New.
(main): Generate "add" function and call it. Adjust to changes
in jithost_abi.
* gdb.base/jithost.h (struct jithost_abi_bounds): New.
(struct jithost_abi) <begin, end>: Remove fields.
<object, function_stack_mangle, function_add>: New fields.
* gdb.base/jitreader.c (struct reader_state) <code_begin,
code_end>: Remove fields.
<func_stack_mangle>: New field.
(read_debug_info): Adjust to renaming, create block for "add"
function.
(read_sp, unwind_frame, get_frame_id): Adjust to other changes.
2019-12-17 05:30:49 +08:00
|
|
|
gdb_test "up" "#1 $any in main $any\r\n$any function_stack_mangle $any" \
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
"up to caller"
|
|
|
|
|
|
|
|
|
|
# Since the JIT unwinder only provides RIP/RSP/RBP,
|
|
|
|
|
# all other registers should show as "<not saved>".
|
2020-02-11 20:46:27 +08:00
|
|
|
|
|
|
|
|
set expected \
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
[multi_line \
|
|
|
|
|
"rax <not saved>" \
|
|
|
|
|
"rbx <not saved>" \
|
|
|
|
|
"rcx <not saved>" \
|
|
|
|
|
"rdx <not saved>" \
|
|
|
|
|
"rsi <not saved>" \
|
|
|
|
|
"rdi <not saved>" \
|
2018-06-23 01:40:47 +08:00
|
|
|
"rbp $hex +$hex" \
|
|
|
|
|
"rsp $caller_sp +$caller_sp" \
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
"r8 <not saved>" \
|
|
|
|
|
"r9 <not saved>" \
|
|
|
|
|
"r10 <not saved>" \
|
|
|
|
|
"r11 <not saved>" \
|
|
|
|
|
"r12 <not saved>" \
|
|
|
|
|
"r13 <not saved>" \
|
|
|
|
|
"r14 <not saved>" \
|
|
|
|
|
"r15 <not saved>" \
|
2018-06-23 01:40:47 +08:00
|
|
|
"rip $hex +$hex $any" \
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
"eflags <not saved>" \
|
|
|
|
|
"cs <not saved>" \
|
|
|
|
|
"ss <not saved>" \
|
|
|
|
|
"ds <not saved>" \
|
|
|
|
|
"es <not saved>" \
|
|
|
|
|
"fs <not saved>" \
|
|
|
|
|
"gs <not saved>" \
|
|
|
|
|
]
|
|
|
|
|
|
2020-02-11 20:46:27 +08:00
|
|
|
# There may be more registers.
|
|
|
|
|
append expected ".*"
|
|
|
|
|
|
|
|
|
|
gdb_test "info registers" $expected
|
|
|
|
|
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
# Make sure that "info frame" doesn't crash.
|
|
|
|
|
gdb_test "info frame" "Stack level 1, .*in main.*"
|
|
|
|
|
|
|
|
|
|
# ... and that neither does printing a pseudo
|
|
|
|
|
# register.
|
|
|
|
|
gdb_test "print /x \$ebp" " = $hex" "print pseudo register"
|
|
|
|
|
|
|
|
|
|
# There's no way for the JIT reader API to support
|
|
|
|
|
# modifyiable values.
|
|
|
|
|
gdb_test "print \$rbp = -1" \
|
|
|
|
|
"Attempt to assign to an unmodifiable value\." \
|
|
|
|
|
"cannot assign to register"
|
|
|
|
|
}
|
2022-02-01 22:49:30 +08:00
|
|
|
|
2023-01-09 02:46:05 +08:00
|
|
|
if { [allow_python_tests] } {
|
2022-02-01 22:49:30 +08:00
|
|
|
gdb_test "python print(gdb.objfiles())" \
|
2022-02-02 23:54:03 +08:00
|
|
|
"$any<gdb.Objfile filename=<< JIT compiled code at $hex >>>$any" \
|
2022-02-01 22:49:30 +08:00
|
|
|
"python gdb.Objfile.__repr__ ()"
|
|
|
|
|
|
|
|
|
|
gdb_test "python print(list(map(lambda objf : objf.filename, gdb.objfiles())))" \
|
2022-02-02 23:54:03 +08:00
|
|
|
"$any'<< JIT compiled code at $hex >>'$any" \
|
2022-02-01 22:49:30 +08:00
|
|
|
"python gdb.Objfile.filename"
|
2023-01-18 19:34:37 +08:00
|
|
|
|
|
|
|
|
gdb_test "python print( \[o for o in gdb.objfiles() if o.filename.startswith('<< JIT compiled code')\]\[0\].build_id )" \
|
|
|
|
|
"None" \
|
|
|
|
|
"python gdb.Objfile.build_id"
|
2022-02-01 22:49:30 +08:00
|
|
|
}
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
# Now unload the jit reader, and ensure that backtracing really
|
|
|
|
|
# doesn't work without it.
|
|
|
|
|
with_test_prefix "without jit-reader" {
|
|
|
|
|
gdb_test_no_output "jit-reader-unload ${jit_reader_bin}" \
|
|
|
|
|
"jit-reader-unload"
|
|
|
|
|
|
|
|
|
|
# Check that we're no longer using the JIT unwinder, and that
|
|
|
|
|
# the built-in unwinder cannot backtrace through the mangled
|
|
|
|
|
# stack pointer.
|
|
|
|
|
gdb_test "bt" \
|
2021-06-08 23:39:05 +08:00
|
|
|
"Backtrace stopped: Cannot access memory at address $sp_after_mangling" \
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
"bt shows error"
|
|
|
|
|
|
|
|
|
|
gdb_test "info frame" "Cannot access memory at address.*" \
|
|
|
|
|
"info frame shows error"
|
|
|
|
|
info_registers_current_frame $sp_after_mangling
|
|
|
|
|
gdb_test "up" "Initial frame selected; you cannot go up\\." \
|
|
|
|
|
"cannot go up"
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
with_test_prefix "with jit-reader again" {
|
|
|
|
|
gdb_test_no_output "jit-reader-load ${jit_reader_bin}" "jit-reader-load"
|
|
|
|
|
|
|
|
|
|
# Check that the jit unwinder manages to backtrace through
|
|
|
|
|
# the mangled stack pointer.
|
|
|
|
|
gdb_test "bt" \
|
|
|
|
|
[multi_line \
|
Fix double-free when creating more than one block in JIT debug info reader
A double-free happens when using a JIT debug info reader that creates
more than one block. In the loop that frees blocks in finalize_symtab,
at the very end, the gdb_block_iter_tmp variable is set initially, but
not changed as the loop advances. If we have two blocks, the first
iteration frees the first block, the second iteration frees the second
block, but the third iteration tries to free the second block again, as
gdb_block_iter_tmp keeps pointing on the second block.
Fix it by assigning the gdb_block_iter_tmp variable in the loop.
I have improved the jit-reader.exp test to cover this case, by adding a
second "JIT-ed" function and creating a block for it. I have renamed
the existing function to something I find a bit more descriptive. There
are no significant changes to jit-reader.exp itself, only updates
following the renaming. The important changes are in jithost.c
(generate a new function) and in jitreader.c (create a gdb_block for
that function).
This was found because of an ASan report:
$ ./gdb testsuite/outputs/gdb.base/jit-reader/jit-reader -ex "jit-reader-load /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jitreader.so" -ex r
Reading symbols from testsuite/outputs/gdb.base/jit-reader/jit-reader...
Starting program: /home/simark/build/binutils-gdb/gdb/testsuite/outputs/gdb.base/jit-reader/jit-reader
=================================================================
==1751048==ERROR: AddressSanitizer: heap-use-after-free on address 0x604000042eb8 at pc 0x5650ef8eec88 bp 0x7ffe52767290 sp 0x7ffe52767280
READ of size 8 at 0x604000042eb8 thread T0
#0 0x5650ef8eec87 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:768
#1 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#2 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#3 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#4 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#5 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#6 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
0x604000042eb8 is located 40 bytes inside of 48-byte region [0x604000042e90,0x604000042ec0)
freed by thread T0 here:
#0 0x7fbbe57376b0 in __interceptor_free /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:122
#1 0x5650ef8f350b in xfree<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/common-utils.h:62
#2 0x5650ef8eeca9 in finalize_symtab /home/simark/src/binutils-gdb/gdb/jit.c:769
#3 0x5650ef8eef88 in jit_object_close_impl /home/simark/src/binutils-gdb/gdb/jit.c:797
#4 0x7fbbda986278 in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:71
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
previously allocated by thread T0 here:
#0 0x7fbbe5737cd8 in __interceptor_calloc /build/gcc/src/gcc/libsanitizer/asan/asan_malloc_linux.cc:153
#1 0x5650eef662f3 in xcalloc /home/simark/src/binutils-gdb/gdb/alloc.c:100
#2 0x5650ef8f34ea in xcnew<gdb_block> /home/simark/src/binutils-gdb/gdb/gdbsupport/poison.h:122
#3 0x5650ef8ed467 in jit_block_open_impl /home/simark/src/binutils-gdb/gdb/jit.c:557
#4 0x7fbbda98620a in read_debug_info /home/simark/src/binutils-gdb/gdb/testsuite/gdb.base/jitreader.c:60
#5 0x5650ef8ef56b in jit_reader_try_read_symtab /home/simark/src/binutils-gdb/gdb/jit.c:850
#6 0x5650ef8effe3 in jit_register_code /home/simark/src/binutils-gdb/gdb/jit.c:948
#7 0x5650ef8f2c92 in jit_event_handler(gdbarch*) /home/simark/src/binutils-gdb/gdb/jit.c:1396
#8 0x5650ef0d137e in handle_jit_event /home/simark/src/binutils-gdb/gdb/breakpoint.c:5470
[snip]
gdb/ChangeLog:
* jit.c (finalize_symtab): Set gdb_block_iter_tmp in loop.
gdb/testsuite/ChangeLog:
* gdb.base/jit-reader.exp (jit_reader_test): Rename
jit_function_00 to jit_function_stack_mangle.
* gdb.base/jithost.c (jit_function_t): Rename to...
(jit_function_stack_mangle_t): ... this.
(jit_function_add_t): New typedef.
(jit_function_00_code): Rename to...
(jit_function_stack_mangle_code): ... this, make static.
(jit_function_add_code): New.
(main): Generate "add" function and call it. Adjust to changes
in jithost_abi.
* gdb.base/jithost.h (struct jithost_abi_bounds): New.
(struct jithost_abi) <begin, end>: Remove fields.
<object, function_stack_mangle, function_add>: New fields.
* gdb.base/jitreader.c (struct reader_state) <code_begin,
code_end>: Remove fields.
<func_stack_mangle>: New field.
(read_debug_info): Adjust to renaming, create block for "add"
function.
(read_sp, unwind_frame, get_frame_id): Adjust to other changes.
2019-12-17 05:30:49 +08:00
|
|
|
"#0 ${any} in jit_function_stack_mangle ${any}" \
|
Extend JIT-reader test and fix GDB problems that exposes
The jit-reader.exp test isn't really exercising the jit-reader's
unwinder API at all. This commit address that, and then fixes GDB
problems exposed.
- The custom JIT reader provided for the jit-reader.exp testcase
always rejects the jitted function's frame...
This is because the custom JIT reader in the testcase never ever
sets state->code_begin/end, so the bounds check in
gdb.base/jitreader.c:unwind_frame:
if (this_ip >= state->code_end || this_ip < state->code_begin)
return GDB_FAIL;
tends to fail, unless you're "lucky" (because it references
uninitialized data).
The result is that GDB is always actually using a built-in unwinder
for the jitted function.
- The provided unwinder doesn't do anything that GDB's built-in
unwinder can't do.
IOW, we can't really tell whether the JIT reader's unwinder is
working or not.
I fixed that by making the jitted function mangle its own stack
pointer with a xor, and then teaching the jit unwinder to demangle
it back (another xor). So now "backtrace" with GDB's built-in
unwinder fails while with the jit unwinder, it succeeds.
- GDB crashes after unloading the JIT reader, and flushing frames...
I made the testcase use the "flushregs" command after unloading the
JIT reader, to force the JIT frames to be flushed. However, that
crashes GDB...
When reinit_frame_cache tears down a frame's cache, it calls its
unwinder's dealloc_cache method, which for JIT frames ends up in
jit.c:jit_dealloc_cache. This function calls each of the frame's
gdb_reg_value's "free" pointer:
for (i = 0; i < gdbarch_num_regs (frame_arch); i++)
if (priv_data->registers[i] && priv_data->registers[i]->free)
priv_data->registers[i]->free (priv_data->registers[i]);
and the problem is these gdb_reg_value instances have been returned
by the JIT reader that has been already unloaded, and their "free"
function pointers likely point to functions in the DSO that has
already been unloaded...
A fix for that could be to call reinit_frame_cache in
jit_reader_unload_command _before_ unloading the jit reader DSO so
that the jit reader is given a chance to clean up the gdb_reg_values
before it is unloaded. However, the fix for the point below makes
this unnecessary, because it stops jit.c from keeping around
gdb_reg_values in the first place.
- However, it still makes sense to clear the frame cache when loading
or unloading a JIT unwinder.
This makes testing a JIT unwinder a bit simpler.
- Not only the frame cache actually -- gdb is not unloading the
jit-registered objfiles when the JIT reader is unloaded, and not
loading the already-registered descriptors when a JIT reader is
loaded.
The new test exercises unloading the jit reader, loading it back
again, and then making sure the JIT reader's unwinder works again.
Without the unload/re-load of already-read descriptors, the newly
loaded JIT would have no idea where the new function is, because
it's stored at symbol read time.
- I added a couple "info frame" calls to the test, and that
crashes GDB...
The problem is that jit_frame_prev_register assumes it'll only be
called for raw registers, so when it gets a pseudo register number,
the "priv->registers[reg]" access is really an out-of-bounds access.
To fix that, I made jit_frame_prev_register use
gdbarch_pseudo_register_read_value for reading the pseudo-registers.
However, that works with a regcache and we don't have one. To fix
that, I made the JIT unwinder store a regcache in its cache instead
of an array of gdb_reg_value pointers.
gdb/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
* jit.c (jit_reader_load_command): Call reinit_frame_cache and
jit_inferior_created_hook.
(jit_reader_unload_command): Call reinit_frame_cache and
jit_inferior_exit_hook.
* jit.c (struct jit_unwind_private) <registers>: Delete field.
<regcache>: New field.
(jit_unwind_reg_set_impl): Set the register's value in the
regcache. Free the passed-in gdb_reg_value.
(jit_dealloc_cache): Adjust to free the regcache.
(jit_frame_sniffer): Allocate a regcache instead of an array of
gdb_reg_value pointers.
(jit_frame_this_id): Adjust.
(jit_frame_prev_register): Read raw registers off of the regcache
instead of from the gdb_reg_value pointer array. Use
gdbarch_pseudo_register_read_value to read pseudo registers.
* regcache.c (regcache_raw_set_cached_value): New function,
factored out from ...
(regcache_raw_write): ... here.
* regcache.h (regcache_raw_set_cached_value): Declare.
gdb/testsuite/ChangeLog:
2016-07-01 Pedro Alves <palves@redhat.com>
* gdb.base/jit-reader.exp (info_registers_current_frame): New
procedure.
(jit_reader_test): Test the jit reader's unwinder.
* gdb.base/jithost.c (jit_function_00_code): New global.
(main): Use memcpy to fill in the mmapped code, instead of poking
bytes manually here.
* gdb.base/jitreader.c (enum register_mapping) <AMD64_RBP>: New
value.
(read_debug_info): Save the function's range.
(read_sp): New function.
(unwind_frame): Use it. Also unwind RBP.
(get_frame_id): Use read_sp.
(gdb_init_reader): Use calloc instead of malloc.
* lib/gdb.exp (get_hexadecimal_valueof): Add optional 'test'
parameter. Use gdb_test_multiple.
2016-07-01 18:56:39 +08:00
|
|
|
"#1 ${any} in main ${any}" \
|
|
|
|
|
]
|
|
|
|
|
}
|
2022-06-21 02:32:52 +08:00
|
|
|
|
2023-01-09 02:46:05 +08:00
|
|
|
if {[allow_python_tests]} {
|
2022-06-21 02:32:52 +08:00
|
|
|
gdb_test "python print(any(\[not x.is_file for x in gdb.objfiles()\]))" \
|
|
|
|
|
"True" \
|
|
|
|
|
"at least one non-file objfile"
|
|
|
|
|
gdb_test "python print(any(\[x.is_file for x in gdb.objfiles()\]))" \
|
|
|
|
|
"True" \
|
|
|
|
|
"at least one file-based objfile"
|
|
|
|
|
}
|
2022-12-08 19:30:25 +08:00
|
|
|
|
|
|
|
|
with_test_prefix "test dwarf unwinder" {
|
|
|
|
|
# Check that the DWARF unwinder does not crash in presence of
|
|
|
|
|
# JIT objfiles.
|
|
|
|
|
gdb_test "up"
|
|
|
|
|
gdb_breakpoint "*function_add" temporary
|
|
|
|
|
gdb_test "cont" ".*Temporary breakpoint ${any} in jit_function_add .*"
|
|
|
|
|
gdb_test "bt" \
|
|
|
|
|
[multi_line \
|
|
|
|
|
"#0 ${any} in jit_function_add ${any}" \
|
|
|
|
|
"#1 ${any} in main ${any}" \
|
|
|
|
|
]
|
|
|
|
|
}
|
2016-06-18 02:24:08 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
jit_reader_test
|
