binutils-gdb/gdb/testsuite/gdb.base/break-unload-file.exp

158 lines
4.6 KiB
Plaintext
Raw Normal View History

# Copyright 2014-2017 Free Software Foundation, Inc.
Stale breakpoint instructions, spurious SIGTRAPS. Without the code portion of the patch, we get these failures: FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue FAIL: gdb.base/break-unload-file.exp: always-inserted on: hbreak: continue FAIL: gdb.base/sym-file.exp: stale bkpts: continue to breakpoint: end here They all looks like random SIGTRAPs: continue Continuing. Program received signal SIGTRAP, Trace/breakpoint trap. 0x0000000000400541 in foo () at ../../../src/gdb/testsuite/gdb.base/break-unload-file.c:21 21 } (gdb) FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue (This is a regression caused by the remove-symbol-file command series.) break-unload-file.exp is about having breakpoints inserted, and then doing "file". I caught this while writing a test that does "file PROGRAM", while PROGRAM was already loaded, which internally does "file" first, because I wanted to force a breakpoint_re_set, but the test is more explicit in case GDB ever optimizes out that re-set. The problem is that unloading the file with "file" ends up in disable_breakpoints_in_freed_objfile, which marks all breakpoint locations of the objfile as both shlib_disabled, _and_ clears the inserted flag, without actually removing the breakpoints from the inferior. Now, usually, in all-stop, breakpoints will already be removed from the inferior before the user can issue the "file" command, but, with non-stop, or breakpoints always-inserted on mode, breakpoints stay inserted even while the user has the prompt. In the latter case, then, if we let the program continue, and it executes the address where we had previously set the breakpoint, it'll actually execute the breakpoint instruction that we left behind... Now, one issue is that the intent of disable_breakpoints_in_freed_objfile is really to handle the unloading of OBJF_USERLOADED objfiles. These are objfiles that were added with add-symbol-file and that are removed with remove-symbol-file. "add-symbol-file"'s docs in the manual clearly say these commands are used to let GDB know about dynamically loaded code: You would use this command when @var{filename} has been dynamically loaded (by some other means) into the program that is running. Similarly, the online help says: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. So it makes sense to, like when shared libraries are unloaded through the generic solib machinery, mark the breakpoint locations as shlib_disabled. But, the "file" command is not about dynamically loaded code, it's about the main program. So the patch makes disable_breakpoints_in_freed_objfile skip all objfiles but OBJF_USERLOADED ones, thus skipping the main objfile. Then, the reason that disable_breakpoints_in_freed_objfile was clearing the inserted flag isn't clear, but likely to avoid breakpoint removal errors, assuming remove-symbol-file was called after the dynamic object was already unmapped from the inferior. In that case, it'd okay to simply clear the inserted flag, but not so if the user for example does remove-symbol-file to remove the library because he made a mistake in the library's address, and wants to re-do add-symbol-file with the correct address. To address all that, I propose an alternative implementation, that handles both cases. The patch includes changes to sym-file.exp to cover them. This implementation leaves the inserted flag alone, and handles breakpoint insertion/removal failure gracefully when the locations are in OBJF_USERLOADED objfiles, just like we handle insertion/removal failure gracefully for locations in shared libraries. To try to make sure we aren't patching back stale shadow memory contents into the inferior, in case the program mapped a different library at the same address where we had the breakpoint, without the user having had a chance of remove-symbol-file'ing before, this adds a new memory_validate_breakpoint function that checks if the breakpoint instruction is still in memory. ppc_linux_memory_remove_breakpoint does this unconditionally for all memory breakpoints, and questions whether memory_remove_breakpoint should be changed to do this for all breakpoints. Possibly yes, though I'm not certain, hence this baby-steps patch. Tested on x86_64 Fedora 17, native and gdbserver. gdb/ 2014-04-23 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location): Tolerate errors if the breakpoint is set in a user-loaded objfile. (remove_breakpoint_1): Likewise. Also tolerate errors if the location is marked shlib_disabled. If the breakpoint is set in a user-loaded objfile is a GDB-side memory breakpoint, validate it before uninsertion. (disable_breakpoints_in_freed_objfile): Skip non-OBJF_USERLOADED objfiles. Don't clear the location's inserted flag. * mem-break.c (memory_validate_breakpoint): New function. * objfiles.c (userloaded_objfile_contains_address_p): New function. * objfiles.h (userloaded_objfile_contains_address_p): Declare. * target.h (memory_validate_breakpoint): New declaration. gdb/testsuite/ 2014-04-23 Pedro Alves <palves@redhat.com> * gdb.base/break-unload-file.c: New file. * gdb.base/break-unload-file.exp: New file. * gdb.base/sym-file-lib.c (baz): New function. * gdb.base/sym-file-loader.c (struct segment) <mapped_size>: New field. (load): Store the segment's mapped size. (unload): New function. (unload_shlib): New function. * gdb.base/sym-file-loader.h (unload_shlib): New declaration. * gdb.base/sym-file-main.c (main): Unload, and reload the library, set a breakpoint at baz, and call it. * gdb.base/sym-file.exp: New tests for stale breakpoint instructions.
2014-04-23 00:19:19 +02:00
# 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/>. */
# Test that "file" doesn't leave stale breakpoints planted in the
# target.
standard_testfile
"$ gdb PROGRAM" vs "(gdb) file PROGRAM" difference; warn on failure to remove breakpoint. Turns out there's a difference between loading the program with "gdb PROGRAM", vs loading it with "(gdb) file PROGRAM". The latter results in the objfile ending up with OBJF_USERLOADED set, while not with the former. (That difference seems bogus, but still that's not the point of this patch. We can revisit that afterwards.) The new code that suppresses breakpoint removal errors for add-symbol-file objects ends up being too greedy: /* In some cases, we might not be able to remove a breakpoint in a shared library that has already been removed, but we have not yet processed the shlib unload event. Similarly for an unloaded add-symbol-file object - the user might not yet have had the chance to remove-symbol-file it. shlib_disabled will be set if the library/object has already been removed, but the breakpoint hasn't been uninserted yet, e.g., after "nosharedlibrary" or "remove-symbol-file" with breakpoints always-inserted mode. */ if (val && (bl->loc_type == bp_loc_software_breakpoint && (bl->shlib_disabled || solib_name_from_address (bl->pspace, bl->address) || userloaded_objfile_contains_address_p (bl->pspace, bl->address)))) val = 0; as it turns out that OBJF_USERLOADED can be set for objfiles loaded by some other means not add-symbol-file. In this case, symbol-file (or "file", which is really just "exec-file"+"symbol-file"). Recall that add-symbol-file is documented as: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ And it's the "dynamically loaded" aspect that the breakpoint.c code cares about. So make add-symbol-file set OBJF_SHARED on its objfiles too, and tweak the breakpoint.c code to look for OBJF_SHARED instead of OBJF_USERLOADED. This restores back the missing breakpoint removal warning when we let sss-bp-on-user-bp-2.exp run on native GNU/Linux (https://sourceware.org/ml/gdb-patches/2014-06/msg00335.html): (gdb) PASS: gdb.base/sss-bp-on-user-bp-2.exp: define stepi_del_break stepi_del_break warning: Error removing breakpoint 3 (gdb) FAIL: gdb.base/sss-bp-on-user-bp-2.exp: stepi_del_break I say "restores" because this was GDB's behavior in 7.7 and earlier. And, likewise, "file" with no arguments only started turning breakpoints set in the main executable to "<pending>" with the remote-symbol-file patch (63644780). The old behavior is now restored, and we break-unload-file.exp test now exercizes both "gdb; file PROGRAM" and "gdb PROGRAM". gdb/ 2014-06-16 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location, remove_breakpoint_1): Adjust. (disable_breakpoints_in_freed_objfile): Skip objfiles that don't have OBJF_SHARED set. * objfiles.c (userloaded_objfile_contains_address_p): Rename to... (shared_objfile_contains_address_p): ... this. Check OBJF_SHARED instead of OBJF_USERLOADED. * objfiles.h (OBJF_SHARED): Update comment. (userloaded_objfile_contains_address_p): Rename to ... (shared_objfile_contains_address_p): ... this, and update comments. * symfile.c (add_symbol_file_command): Also set OBJF_SHARED in the new objfile. (remove_symbol_file_command): Skip objfiles that don't have OBJF_SHARED set. gdb/testsuite/ 2014-06-16 Pedro Alves <palves@redhat.com> * gdb.base/break-main-file-remove-fail.c: New file. * gdb.base/break-main-file-remove-fail.exp: New file. * gdb.base/break-unload-file.exp: Use build_executable instead of prepare_for_testing. (test_break): New parameter "initial_load". Handle it. (top level): Add initial_load cmdline/file axis.
2014-06-16 16:38:13 +02:00
if {[build_executable "failed to prepare" $testfile $srcfile debug]} {
Stale breakpoint instructions, spurious SIGTRAPS. Without the code portion of the patch, we get these failures: FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue FAIL: gdb.base/break-unload-file.exp: always-inserted on: hbreak: continue FAIL: gdb.base/sym-file.exp: stale bkpts: continue to breakpoint: end here They all looks like random SIGTRAPs: continue Continuing. Program received signal SIGTRAP, Trace/breakpoint trap. 0x0000000000400541 in foo () at ../../../src/gdb/testsuite/gdb.base/break-unload-file.c:21 21 } (gdb) FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue (This is a regression caused by the remove-symbol-file command series.) break-unload-file.exp is about having breakpoints inserted, and then doing "file". I caught this while writing a test that does "file PROGRAM", while PROGRAM was already loaded, which internally does "file" first, because I wanted to force a breakpoint_re_set, but the test is more explicit in case GDB ever optimizes out that re-set. The problem is that unloading the file with "file" ends up in disable_breakpoints_in_freed_objfile, which marks all breakpoint locations of the objfile as both shlib_disabled, _and_ clears the inserted flag, without actually removing the breakpoints from the inferior. Now, usually, in all-stop, breakpoints will already be removed from the inferior before the user can issue the "file" command, but, with non-stop, or breakpoints always-inserted on mode, breakpoints stay inserted even while the user has the prompt. In the latter case, then, if we let the program continue, and it executes the address where we had previously set the breakpoint, it'll actually execute the breakpoint instruction that we left behind... Now, one issue is that the intent of disable_breakpoints_in_freed_objfile is really to handle the unloading of OBJF_USERLOADED objfiles. These are objfiles that were added with add-symbol-file and that are removed with remove-symbol-file. "add-symbol-file"'s docs in the manual clearly say these commands are used to let GDB know about dynamically loaded code: You would use this command when @var{filename} has been dynamically loaded (by some other means) into the program that is running. Similarly, the online help says: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. So it makes sense to, like when shared libraries are unloaded through the generic solib machinery, mark the breakpoint locations as shlib_disabled. But, the "file" command is not about dynamically loaded code, it's about the main program. So the patch makes disable_breakpoints_in_freed_objfile skip all objfiles but OBJF_USERLOADED ones, thus skipping the main objfile. Then, the reason that disable_breakpoints_in_freed_objfile was clearing the inserted flag isn't clear, but likely to avoid breakpoint removal errors, assuming remove-symbol-file was called after the dynamic object was already unmapped from the inferior. In that case, it'd okay to simply clear the inserted flag, but not so if the user for example does remove-symbol-file to remove the library because he made a mistake in the library's address, and wants to re-do add-symbol-file with the correct address. To address all that, I propose an alternative implementation, that handles both cases. The patch includes changes to sym-file.exp to cover them. This implementation leaves the inserted flag alone, and handles breakpoint insertion/removal failure gracefully when the locations are in OBJF_USERLOADED objfiles, just like we handle insertion/removal failure gracefully for locations in shared libraries. To try to make sure we aren't patching back stale shadow memory contents into the inferior, in case the program mapped a different library at the same address where we had the breakpoint, without the user having had a chance of remove-symbol-file'ing before, this adds a new memory_validate_breakpoint function that checks if the breakpoint instruction is still in memory. ppc_linux_memory_remove_breakpoint does this unconditionally for all memory breakpoints, and questions whether memory_remove_breakpoint should be changed to do this for all breakpoints. Possibly yes, though I'm not certain, hence this baby-steps patch. Tested on x86_64 Fedora 17, native and gdbserver. gdb/ 2014-04-23 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location): Tolerate errors if the breakpoint is set in a user-loaded objfile. (remove_breakpoint_1): Likewise. Also tolerate errors if the location is marked shlib_disabled. If the breakpoint is set in a user-loaded objfile is a GDB-side memory breakpoint, validate it before uninsertion. (disable_breakpoints_in_freed_objfile): Skip non-OBJF_USERLOADED objfiles. Don't clear the location's inserted flag. * mem-break.c (memory_validate_breakpoint): New function. * objfiles.c (userloaded_objfile_contains_address_p): New function. * objfiles.h (userloaded_objfile_contains_address_p): Declare. * target.h (memory_validate_breakpoint): New declaration. gdb/testsuite/ 2014-04-23 Pedro Alves <palves@redhat.com> * gdb.base/break-unload-file.c: New file. * gdb.base/break-unload-file.exp: New file. * gdb.base/sym-file-lib.c (baz): New function. * gdb.base/sym-file-loader.c (struct segment) <mapped_size>: New field. (load): Store the segment's mapped size. (unload): New function. (unload_shlib): New function. * gdb.base/sym-file-loader.h (unload_shlib): New declaration. * gdb.base/sym-file-main.c (main): Unload, and reload the library, set a breakpoint at baz, and call it. * gdb.base/sym-file.exp: New tests for stale breakpoint instructions.
2014-04-23 00:19:19 +02:00
return -1
}
"$ gdb PROGRAM" vs "(gdb) file PROGRAM" difference; warn on failure to remove breakpoint. Turns out there's a difference between loading the program with "gdb PROGRAM", vs loading it with "(gdb) file PROGRAM". The latter results in the objfile ending up with OBJF_USERLOADED set, while not with the former. (That difference seems bogus, but still that's not the point of this patch. We can revisit that afterwards.) The new code that suppresses breakpoint removal errors for add-symbol-file objects ends up being too greedy: /* In some cases, we might not be able to remove a breakpoint in a shared library that has already been removed, but we have not yet processed the shlib unload event. Similarly for an unloaded add-symbol-file object - the user might not yet have had the chance to remove-symbol-file it. shlib_disabled will be set if the library/object has already been removed, but the breakpoint hasn't been uninserted yet, e.g., after "nosharedlibrary" or "remove-symbol-file" with breakpoints always-inserted mode. */ if (val && (bl->loc_type == bp_loc_software_breakpoint && (bl->shlib_disabled || solib_name_from_address (bl->pspace, bl->address) || userloaded_objfile_contains_address_p (bl->pspace, bl->address)))) val = 0; as it turns out that OBJF_USERLOADED can be set for objfiles loaded by some other means not add-symbol-file. In this case, symbol-file (or "file", which is really just "exec-file"+"symbol-file"). Recall that add-symbol-file is documented as: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ And it's the "dynamically loaded" aspect that the breakpoint.c code cares about. So make add-symbol-file set OBJF_SHARED on its objfiles too, and tweak the breakpoint.c code to look for OBJF_SHARED instead of OBJF_USERLOADED. This restores back the missing breakpoint removal warning when we let sss-bp-on-user-bp-2.exp run on native GNU/Linux (https://sourceware.org/ml/gdb-patches/2014-06/msg00335.html): (gdb) PASS: gdb.base/sss-bp-on-user-bp-2.exp: define stepi_del_break stepi_del_break warning: Error removing breakpoint 3 (gdb) FAIL: gdb.base/sss-bp-on-user-bp-2.exp: stepi_del_break I say "restores" because this was GDB's behavior in 7.7 and earlier. And, likewise, "file" with no arguments only started turning breakpoints set in the main executable to "<pending>" with the remote-symbol-file patch (63644780). The old behavior is now restored, and we break-unload-file.exp test now exercizes both "gdb; file PROGRAM" and "gdb PROGRAM". gdb/ 2014-06-16 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location, remove_breakpoint_1): Adjust. (disable_breakpoints_in_freed_objfile): Skip objfiles that don't have OBJF_SHARED set. * objfiles.c (userloaded_objfile_contains_address_p): Rename to... (shared_objfile_contains_address_p): ... this. Check OBJF_SHARED instead of OBJF_USERLOADED. * objfiles.h (OBJF_SHARED): Update comment. (userloaded_objfile_contains_address_p): Rename to ... (shared_objfile_contains_address_p): ... this, and update comments. * symfile.c (add_symbol_file_command): Also set OBJF_SHARED in the new objfile. (remove_symbol_file_command): Skip objfiles that don't have OBJF_SHARED set. gdb/testsuite/ 2014-06-16 Pedro Alves <palves@redhat.com> * gdb.base/break-main-file-remove-fail.c: New file. * gdb.base/break-main-file-remove-fail.exp: New file. * gdb.base/break-unload-file.exp: Use build_executable instead of prepare_for_testing. (test_break): New parameter "initial_load". Handle it. (top level): Add initial_load cmdline/file axis.
2014-06-16 16:38:13 +02:00
# Run the test proper. INITIAL_LOAD determines whether the program is
# initially loaded by the "file" command or by passing it to GDB on
# the command line. ALWAYS_INSERT determines whether always-inserted
# mode is on/off. BREAK_COMMAND is the break command being tested.
Stale breakpoint instructions, spurious SIGTRAPS. Without the code portion of the patch, we get these failures: FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue FAIL: gdb.base/break-unload-file.exp: always-inserted on: hbreak: continue FAIL: gdb.base/sym-file.exp: stale bkpts: continue to breakpoint: end here They all looks like random SIGTRAPs: continue Continuing. Program received signal SIGTRAP, Trace/breakpoint trap. 0x0000000000400541 in foo () at ../../../src/gdb/testsuite/gdb.base/break-unload-file.c:21 21 } (gdb) FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue (This is a regression caused by the remove-symbol-file command series.) break-unload-file.exp is about having breakpoints inserted, and then doing "file". I caught this while writing a test that does "file PROGRAM", while PROGRAM was already loaded, which internally does "file" first, because I wanted to force a breakpoint_re_set, but the test is more explicit in case GDB ever optimizes out that re-set. The problem is that unloading the file with "file" ends up in disable_breakpoints_in_freed_objfile, which marks all breakpoint locations of the objfile as both shlib_disabled, _and_ clears the inserted flag, without actually removing the breakpoints from the inferior. Now, usually, in all-stop, breakpoints will already be removed from the inferior before the user can issue the "file" command, but, with non-stop, or breakpoints always-inserted on mode, breakpoints stay inserted even while the user has the prompt. In the latter case, then, if we let the program continue, and it executes the address where we had previously set the breakpoint, it'll actually execute the breakpoint instruction that we left behind... Now, one issue is that the intent of disable_breakpoints_in_freed_objfile is really to handle the unloading of OBJF_USERLOADED objfiles. These are objfiles that were added with add-symbol-file and that are removed with remove-symbol-file. "add-symbol-file"'s docs in the manual clearly say these commands are used to let GDB know about dynamically loaded code: You would use this command when @var{filename} has been dynamically loaded (by some other means) into the program that is running. Similarly, the online help says: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. So it makes sense to, like when shared libraries are unloaded through the generic solib machinery, mark the breakpoint locations as shlib_disabled. But, the "file" command is not about dynamically loaded code, it's about the main program. So the patch makes disable_breakpoints_in_freed_objfile skip all objfiles but OBJF_USERLOADED ones, thus skipping the main objfile. Then, the reason that disable_breakpoints_in_freed_objfile was clearing the inserted flag isn't clear, but likely to avoid breakpoint removal errors, assuming remove-symbol-file was called after the dynamic object was already unmapped from the inferior. In that case, it'd okay to simply clear the inserted flag, but not so if the user for example does remove-symbol-file to remove the library because he made a mistake in the library's address, and wants to re-do add-symbol-file with the correct address. To address all that, I propose an alternative implementation, that handles both cases. The patch includes changes to sym-file.exp to cover them. This implementation leaves the inserted flag alone, and handles breakpoint insertion/removal failure gracefully when the locations are in OBJF_USERLOADED objfiles, just like we handle insertion/removal failure gracefully for locations in shared libraries. To try to make sure we aren't patching back stale shadow memory contents into the inferior, in case the program mapped a different library at the same address where we had the breakpoint, without the user having had a chance of remove-symbol-file'ing before, this adds a new memory_validate_breakpoint function that checks if the breakpoint instruction is still in memory. ppc_linux_memory_remove_breakpoint does this unconditionally for all memory breakpoints, and questions whether memory_remove_breakpoint should be changed to do this for all breakpoints. Possibly yes, though I'm not certain, hence this baby-steps patch. Tested on x86_64 Fedora 17, native and gdbserver. gdb/ 2014-04-23 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location): Tolerate errors if the breakpoint is set in a user-loaded objfile. (remove_breakpoint_1): Likewise. Also tolerate errors if the location is marked shlib_disabled. If the breakpoint is set in a user-loaded objfile is a GDB-side memory breakpoint, validate it before uninsertion. (disable_breakpoints_in_freed_objfile): Skip non-OBJF_USERLOADED objfiles. Don't clear the location's inserted flag. * mem-break.c (memory_validate_breakpoint): New function. * objfiles.c (userloaded_objfile_contains_address_p): New function. * objfiles.h (userloaded_objfile_contains_address_p): Declare. * target.h (memory_validate_breakpoint): New declaration. gdb/testsuite/ 2014-04-23 Pedro Alves <palves@redhat.com> * gdb.base/break-unload-file.c: New file. * gdb.base/break-unload-file.exp: New file. * gdb.base/sym-file-lib.c (baz): New function. * gdb.base/sym-file-loader.c (struct segment) <mapped_size>: New field. (load): Store the segment's mapped size. (unload): New function. (unload_shlib): New function. * gdb.base/sym-file-loader.h (unload_shlib): New declaration. * gdb.base/sym-file-main.c (main): Unload, and reload the library, set a breakpoint at baz, and call it. * gdb.base/sym-file.exp: New tests for stale breakpoint instructions.
2014-04-23 00:19:19 +02:00
#
"$ gdb PROGRAM" vs "(gdb) file PROGRAM" difference; warn on failure to remove breakpoint. Turns out there's a difference between loading the program with "gdb PROGRAM", vs loading it with "(gdb) file PROGRAM". The latter results in the objfile ending up with OBJF_USERLOADED set, while not with the former. (That difference seems bogus, but still that's not the point of this patch. We can revisit that afterwards.) The new code that suppresses breakpoint removal errors for add-symbol-file objects ends up being too greedy: /* In some cases, we might not be able to remove a breakpoint in a shared library that has already been removed, but we have not yet processed the shlib unload event. Similarly for an unloaded add-symbol-file object - the user might not yet have had the chance to remove-symbol-file it. shlib_disabled will be set if the library/object has already been removed, but the breakpoint hasn't been uninserted yet, e.g., after "nosharedlibrary" or "remove-symbol-file" with breakpoints always-inserted mode. */ if (val && (bl->loc_type == bp_loc_software_breakpoint && (bl->shlib_disabled || solib_name_from_address (bl->pspace, bl->address) || userloaded_objfile_contains_address_p (bl->pspace, bl->address)))) val = 0; as it turns out that OBJF_USERLOADED can be set for objfiles loaded by some other means not add-symbol-file. In this case, symbol-file (or "file", which is really just "exec-file"+"symbol-file"). Recall that add-symbol-file is documented as: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ And it's the "dynamically loaded" aspect that the breakpoint.c code cares about. So make add-symbol-file set OBJF_SHARED on its objfiles too, and tweak the breakpoint.c code to look for OBJF_SHARED instead of OBJF_USERLOADED. This restores back the missing breakpoint removal warning when we let sss-bp-on-user-bp-2.exp run on native GNU/Linux (https://sourceware.org/ml/gdb-patches/2014-06/msg00335.html): (gdb) PASS: gdb.base/sss-bp-on-user-bp-2.exp: define stepi_del_break stepi_del_break warning: Error removing breakpoint 3 (gdb) FAIL: gdb.base/sss-bp-on-user-bp-2.exp: stepi_del_break I say "restores" because this was GDB's behavior in 7.7 and earlier. And, likewise, "file" with no arguments only started turning breakpoints set in the main executable to "<pending>" with the remote-symbol-file patch (63644780). The old behavior is now restored, and we break-unload-file.exp test now exercizes both "gdb; file PROGRAM" and "gdb PROGRAM". gdb/ 2014-06-16 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location, remove_breakpoint_1): Adjust. (disable_breakpoints_in_freed_objfile): Skip objfiles that don't have OBJF_SHARED set. * objfiles.c (userloaded_objfile_contains_address_p): Rename to... (shared_objfile_contains_address_p): ... this. Check OBJF_SHARED instead of OBJF_USERLOADED. * objfiles.h (OBJF_SHARED): Update comment. (userloaded_objfile_contains_address_p): Rename to ... (shared_objfile_contains_address_p): ... this, and update comments. * symfile.c (add_symbol_file_command): Also set OBJF_SHARED in the new objfile. (remove_symbol_file_command): Skip objfiles that don't have OBJF_SHARED set. gdb/testsuite/ 2014-06-16 Pedro Alves <palves@redhat.com> * gdb.base/break-main-file-remove-fail.c: New file. * gdb.base/break-main-file-remove-fail.exp: New file. * gdb.base/break-unload-file.exp: Use build_executable instead of prepare_for_testing. (test_break): New parameter "initial_load". Handle it. (top level): Add initial_load cmdline/file axis.
2014-06-16 16:38:13 +02:00
proc test_break { initial_load always_inserted break_command } {
global srcdir subdir binfile
global gdb_prompt hex
global GDBFLAGS
append prefix "$initial_load: "
append prefix "always-inserted $always_inserted: "
append prefix "$break_command"
with_test_prefix "$prefix" {
gdb_exit
set saved_gdbflags $GDBFLAGS
# See "used to behave differently" further below.
if { $initial_load == "file" } {
gdb_start
gdb_file_cmd $binfile
} else {
global last_loaded_file
# gdb_file_cmd sets this. This is what gdb_reload
# implementations use as binary.
set last_loaded_file $binfile
set GDBFLAGS "$GDBFLAGS $binfile"
gdb_start
}
Stale breakpoint instructions, spurious SIGTRAPS. Without the code portion of the patch, we get these failures: FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue FAIL: gdb.base/break-unload-file.exp: always-inserted on: hbreak: continue FAIL: gdb.base/sym-file.exp: stale bkpts: continue to breakpoint: end here They all looks like random SIGTRAPs: continue Continuing. Program received signal SIGTRAP, Trace/breakpoint trap. 0x0000000000400541 in foo () at ../../../src/gdb/testsuite/gdb.base/break-unload-file.c:21 21 } (gdb) FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue (This is a regression caused by the remove-symbol-file command series.) break-unload-file.exp is about having breakpoints inserted, and then doing "file". I caught this while writing a test that does "file PROGRAM", while PROGRAM was already loaded, which internally does "file" first, because I wanted to force a breakpoint_re_set, but the test is more explicit in case GDB ever optimizes out that re-set. The problem is that unloading the file with "file" ends up in disable_breakpoints_in_freed_objfile, which marks all breakpoint locations of the objfile as both shlib_disabled, _and_ clears the inserted flag, without actually removing the breakpoints from the inferior. Now, usually, in all-stop, breakpoints will already be removed from the inferior before the user can issue the "file" command, but, with non-stop, or breakpoints always-inserted on mode, breakpoints stay inserted even while the user has the prompt. In the latter case, then, if we let the program continue, and it executes the address where we had previously set the breakpoint, it'll actually execute the breakpoint instruction that we left behind... Now, one issue is that the intent of disable_breakpoints_in_freed_objfile is really to handle the unloading of OBJF_USERLOADED objfiles. These are objfiles that were added with add-symbol-file and that are removed with remove-symbol-file. "add-symbol-file"'s docs in the manual clearly say these commands are used to let GDB know about dynamically loaded code: You would use this command when @var{filename} has been dynamically loaded (by some other means) into the program that is running. Similarly, the online help says: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. So it makes sense to, like when shared libraries are unloaded through the generic solib machinery, mark the breakpoint locations as shlib_disabled. But, the "file" command is not about dynamically loaded code, it's about the main program. So the patch makes disable_breakpoints_in_freed_objfile skip all objfiles but OBJF_USERLOADED ones, thus skipping the main objfile. Then, the reason that disable_breakpoints_in_freed_objfile was clearing the inserted flag isn't clear, but likely to avoid breakpoint removal errors, assuming remove-symbol-file was called after the dynamic object was already unmapped from the inferior. In that case, it'd okay to simply clear the inserted flag, but not so if the user for example does remove-symbol-file to remove the library because he made a mistake in the library's address, and wants to re-do add-symbol-file with the correct address. To address all that, I propose an alternative implementation, that handles both cases. The patch includes changes to sym-file.exp to cover them. This implementation leaves the inserted flag alone, and handles breakpoint insertion/removal failure gracefully when the locations are in OBJF_USERLOADED objfiles, just like we handle insertion/removal failure gracefully for locations in shared libraries. To try to make sure we aren't patching back stale shadow memory contents into the inferior, in case the program mapped a different library at the same address where we had the breakpoint, without the user having had a chance of remove-symbol-file'ing before, this adds a new memory_validate_breakpoint function that checks if the breakpoint instruction is still in memory. ppc_linux_memory_remove_breakpoint does this unconditionally for all memory breakpoints, and questions whether memory_remove_breakpoint should be changed to do this for all breakpoints. Possibly yes, though I'm not certain, hence this baby-steps patch. Tested on x86_64 Fedora 17, native and gdbserver. gdb/ 2014-04-23 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location): Tolerate errors if the breakpoint is set in a user-loaded objfile. (remove_breakpoint_1): Likewise. Also tolerate errors if the location is marked shlib_disabled. If the breakpoint is set in a user-loaded objfile is a GDB-side memory breakpoint, validate it before uninsertion. (disable_breakpoints_in_freed_objfile): Skip non-OBJF_USERLOADED objfiles. Don't clear the location's inserted flag. * mem-break.c (memory_validate_breakpoint): New function. * objfiles.c (userloaded_objfile_contains_address_p): New function. * objfiles.h (userloaded_objfile_contains_address_p): Declare. * target.h (memory_validate_breakpoint): New declaration. gdb/testsuite/ 2014-04-23 Pedro Alves <palves@redhat.com> * gdb.base/break-unload-file.c: New file. * gdb.base/break-unload-file.exp: New file. * gdb.base/sym-file-lib.c (baz): New function. * gdb.base/sym-file-loader.c (struct segment) <mapped_size>: New field. (load): Store the segment's mapped size. (unload): New function. (unload_shlib): New function. * gdb.base/sym-file-loader.h (unload_shlib): New declaration. * gdb.base/sym-file-main.c (main): Unload, and reload the library, set a breakpoint at baz, and call it. * gdb.base/sym-file.exp: New tests for stale breakpoint instructions.
2014-04-23 00:19:19 +02:00
"$ gdb PROGRAM" vs "(gdb) file PROGRAM" difference; warn on failure to remove breakpoint. Turns out there's a difference between loading the program with "gdb PROGRAM", vs loading it with "(gdb) file PROGRAM". The latter results in the objfile ending up with OBJF_USERLOADED set, while not with the former. (That difference seems bogus, but still that's not the point of this patch. We can revisit that afterwards.) The new code that suppresses breakpoint removal errors for add-symbol-file objects ends up being too greedy: /* In some cases, we might not be able to remove a breakpoint in a shared library that has already been removed, but we have not yet processed the shlib unload event. Similarly for an unloaded add-symbol-file object - the user might not yet have had the chance to remove-symbol-file it. shlib_disabled will be set if the library/object has already been removed, but the breakpoint hasn't been uninserted yet, e.g., after "nosharedlibrary" or "remove-symbol-file" with breakpoints always-inserted mode. */ if (val && (bl->loc_type == bp_loc_software_breakpoint && (bl->shlib_disabled || solib_name_from_address (bl->pspace, bl->address) || userloaded_objfile_contains_address_p (bl->pspace, bl->address)))) val = 0; as it turns out that OBJF_USERLOADED can be set for objfiles loaded by some other means not add-symbol-file. In this case, symbol-file (or "file", which is really just "exec-file"+"symbol-file"). Recall that add-symbol-file is documented as: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ And it's the "dynamically loaded" aspect that the breakpoint.c code cares about. So make add-symbol-file set OBJF_SHARED on its objfiles too, and tweak the breakpoint.c code to look for OBJF_SHARED instead of OBJF_USERLOADED. This restores back the missing breakpoint removal warning when we let sss-bp-on-user-bp-2.exp run on native GNU/Linux (https://sourceware.org/ml/gdb-patches/2014-06/msg00335.html): (gdb) PASS: gdb.base/sss-bp-on-user-bp-2.exp: define stepi_del_break stepi_del_break warning: Error removing breakpoint 3 (gdb) FAIL: gdb.base/sss-bp-on-user-bp-2.exp: stepi_del_break I say "restores" because this was GDB's behavior in 7.7 and earlier. And, likewise, "file" with no arguments only started turning breakpoints set in the main executable to "<pending>" with the remote-symbol-file patch (63644780). The old behavior is now restored, and we break-unload-file.exp test now exercizes both "gdb; file PROGRAM" and "gdb PROGRAM". gdb/ 2014-06-16 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location, remove_breakpoint_1): Adjust. (disable_breakpoints_in_freed_objfile): Skip objfiles that don't have OBJF_SHARED set. * objfiles.c (userloaded_objfile_contains_address_p): Rename to... (shared_objfile_contains_address_p): ... this. Check OBJF_SHARED instead of OBJF_USERLOADED. * objfiles.h (OBJF_SHARED): Update comment. (userloaded_objfile_contains_address_p): Rename to ... (shared_objfile_contains_address_p): ... this, and update comments. * symfile.c (add_symbol_file_command): Also set OBJF_SHARED in the new objfile. (remove_symbol_file_command): Skip objfiles that don't have OBJF_SHARED set. gdb/testsuite/ 2014-06-16 Pedro Alves <palves@redhat.com> * gdb.base/break-main-file-remove-fail.c: New file. * gdb.base/break-main-file-remove-fail.exp: New file. * gdb.base/break-unload-file.exp: Use build_executable instead of prepare_for_testing. (test_break): New parameter "initial_load". Handle it. (top level): Add initial_load cmdline/file axis.
2014-06-16 16:38:13 +02:00
gdb_reinitialize_dir $srcdir/$subdir
gdb_reload
set GDBFLAGS $saved_gdbflags
Stale breakpoint instructions, spurious SIGTRAPS. Without the code portion of the patch, we get these failures: FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue FAIL: gdb.base/break-unload-file.exp: always-inserted on: hbreak: continue FAIL: gdb.base/sym-file.exp: stale bkpts: continue to breakpoint: end here They all looks like random SIGTRAPs: continue Continuing. Program received signal SIGTRAP, Trace/breakpoint trap. 0x0000000000400541 in foo () at ../../../src/gdb/testsuite/gdb.base/break-unload-file.c:21 21 } (gdb) FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue (This is a regression caused by the remove-symbol-file command series.) break-unload-file.exp is about having breakpoints inserted, and then doing "file". I caught this while writing a test that does "file PROGRAM", while PROGRAM was already loaded, which internally does "file" first, because I wanted to force a breakpoint_re_set, but the test is more explicit in case GDB ever optimizes out that re-set. The problem is that unloading the file with "file" ends up in disable_breakpoints_in_freed_objfile, which marks all breakpoint locations of the objfile as both shlib_disabled, _and_ clears the inserted flag, without actually removing the breakpoints from the inferior. Now, usually, in all-stop, breakpoints will already be removed from the inferior before the user can issue the "file" command, but, with non-stop, or breakpoints always-inserted on mode, breakpoints stay inserted even while the user has the prompt. In the latter case, then, if we let the program continue, and it executes the address where we had previously set the breakpoint, it'll actually execute the breakpoint instruction that we left behind... Now, one issue is that the intent of disable_breakpoints_in_freed_objfile is really to handle the unloading of OBJF_USERLOADED objfiles. These are objfiles that were added with add-symbol-file and that are removed with remove-symbol-file. "add-symbol-file"'s docs in the manual clearly say these commands are used to let GDB know about dynamically loaded code: You would use this command when @var{filename} has been dynamically loaded (by some other means) into the program that is running. Similarly, the online help says: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. So it makes sense to, like when shared libraries are unloaded through the generic solib machinery, mark the breakpoint locations as shlib_disabled. But, the "file" command is not about dynamically loaded code, it's about the main program. So the patch makes disable_breakpoints_in_freed_objfile skip all objfiles but OBJF_USERLOADED ones, thus skipping the main objfile. Then, the reason that disable_breakpoints_in_freed_objfile was clearing the inserted flag isn't clear, but likely to avoid breakpoint removal errors, assuming remove-symbol-file was called after the dynamic object was already unmapped from the inferior. In that case, it'd okay to simply clear the inserted flag, but not so if the user for example does remove-symbol-file to remove the library because he made a mistake in the library's address, and wants to re-do add-symbol-file with the correct address. To address all that, I propose an alternative implementation, that handles both cases. The patch includes changes to sym-file.exp to cover them. This implementation leaves the inserted flag alone, and handles breakpoint insertion/removal failure gracefully when the locations are in OBJF_USERLOADED objfiles, just like we handle insertion/removal failure gracefully for locations in shared libraries. To try to make sure we aren't patching back stale shadow memory contents into the inferior, in case the program mapped a different library at the same address where we had the breakpoint, without the user having had a chance of remove-symbol-file'ing before, this adds a new memory_validate_breakpoint function that checks if the breakpoint instruction is still in memory. ppc_linux_memory_remove_breakpoint does this unconditionally for all memory breakpoints, and questions whether memory_remove_breakpoint should be changed to do this for all breakpoints. Possibly yes, though I'm not certain, hence this baby-steps patch. Tested on x86_64 Fedora 17, native and gdbserver. gdb/ 2014-04-23 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location): Tolerate errors if the breakpoint is set in a user-loaded objfile. (remove_breakpoint_1): Likewise. Also tolerate errors if the location is marked shlib_disabled. If the breakpoint is set in a user-loaded objfile is a GDB-side memory breakpoint, validate it before uninsertion. (disable_breakpoints_in_freed_objfile): Skip non-OBJF_USERLOADED objfiles. Don't clear the location's inserted flag. * mem-break.c (memory_validate_breakpoint): New function. * objfiles.c (userloaded_objfile_contains_address_p): New function. * objfiles.h (userloaded_objfile_contains_address_p): Declare. * target.h (memory_validate_breakpoint): New declaration. gdb/testsuite/ 2014-04-23 Pedro Alves <palves@redhat.com> * gdb.base/break-unload-file.c: New file. * gdb.base/break-unload-file.exp: New file. * gdb.base/sym-file-lib.c (baz): New function. * gdb.base/sym-file-loader.c (struct segment) <mapped_size>: New field. (load): Store the segment's mapped size. (unload): New function. (unload_shlib): New function. * gdb.base/sym-file-loader.h (unload_shlib): New declaration. * gdb.base/sym-file-main.c (main): Unload, and reload the library, set a breakpoint at baz, and call it. * gdb.base/sym-file.exp: New tests for stale breakpoint instructions.
2014-04-23 00:19:19 +02:00
if ![runto_main] then {
Fix test names starting with uppercase output by basic functions The following patch is based on the previous patch i sent and handles cases of test names that start with an uppercase letter. Test names should start with lowercase unless it starts with the name of a technology, architecture, ISA etc. This first patch addresses cases of test names output explicitly via xfail, kfail, kpass, fail, pass, unsupported, untested and also names set with the pattern "set test" and "set testname". gdb/testsuite/ChangeLog: 2016-12-01 Luis Machado <lgustavo@codesourcery.com> Fix test names starting with uppercase throughout all the files below. * gdb.ada/array_return.exp * gdb.ada/catch_ex.exp * gdb.ada/info_exc.exp * gdb.ada/mi_catch_ex.exp * gdb.ada/mi_dyn_arr.exp * gdb.ada/mi_ex_cond.exp * gdb.ada/mi_exc_info.exp * gdb.ada/mi_interface.exp * gdb.ada/mi_task_arg.exp * gdb.ada/mi_task_info.exp * gdb.ada/mi_var_array.exp * gdb.arch/alpha-step.exp * gdb.arch/amd64-disp-step.exp * gdb.arch/arm-disp-step.exp * gdb.arch/disp-step-insn-reloc.exp * gdb.arch/e500-prologue.exp * gdb.arch/ftrace-insn-reloc.exp * gdb.arch/gdb1558.exp * gdb.arch/i386-bp_permanent.exp * gdb.arch/i386-disp-step.exp * gdb.arch/i386-float.exp * gdb.arch/i386-gnu-cfi.exp * gdb.arch/ia64-breakpoint-shadow.exp * gdb.arch/mips16-thunks.exp * gdb.arch/pa-nullify.exp * gdb.arch/powerpc-aix-prologue.exp * gdb.arch/powerpc-power.exp * gdb.arch/ppc-dfp.exp * gdb.arch/s390-tdbregs.exp * gdb.arch/spu-info.exp * gdb.arch/spu-ls.exp * gdb.arch/thumb-bx-pc.exp * gdb.base/advance.exp * gdb.base/annota-input-while-running.exp * gdb.base/arrayidx.exp * gdb.base/asmlabel.exp * gdb.base/async.exp * gdb.base/attach-wait-input.exp * gdb.base/auto-connect-native-target.exp * gdb.base/batch-preserve-term-settings.exp * gdb.base/bfp-test.exp * gdb.base/bigcore.exp * gdb.base/bp-permanent.exp * gdb.base/break-always.exp * gdb.base/break-fun-addr.exp * gdb.base/break-idempotent.exp * gdb.base/break-main-file-remove-fail.exp * gdb.base/break-probes.exp * gdb.base/break-unload-file.exp * gdb.base/break.exp * gdb.base/call-ar-st.exp * gdb.base/call-rt-st.exp * gdb.base/call-sc.exp * gdb.base/call-signal-resume.exp * gdb.base/call-strs.exp * gdb.base/callexit.exp * gdb.base/callfuncs.exp * gdb.base/catch-gdb-caused-signals.exp * gdb.base/catch-signal-siginfo-cond.exp * gdb.base/catch-syscall.exp * gdb.base/compare-sections.exp * gdb.base/cond-eval-mode.exp * gdb.base/condbreak-call-false.exp * gdb.base/consecutive-step-over.exp * gdb.base/cursal.exp * gdb.base/disabled-location.exp * gdb.base/disasm-end-cu.exp * gdb.base/display.exp * gdb.base/double-prompt-target-event-error.exp * gdb.base/dprintf-bp-same-addr.exp * gdb.base/dprintf-detach.exp * gdb.base/dprintf-next.exp * gdb.base/dprintf-non-stop.exp * gdb.base/dprintf-pending.exp * gdb.base/dso2dso.exp * gdb.base/ending-run.exp * gdb.base/enum_cond.exp * gdb.base/examine-backward.exp * gdb.base/exe-lock.exp * gdb.base/exec-invalid-sysroot.exp * gdb.base/execl-update-breakpoints.exp * gdb.base/execution-termios.exp * gdb.base/fileio.exp * gdb.base/fixsection.exp * gdb.base/foll-exec-mode.exp * gdb.base/foll-exec.exp * gdb.base/fork-running-state.exp * gdb.base/frame-args.exp * gdb.base/fullpath-expand.exp * gdb.base/func-ptr.exp * gdb.base/gcore-relro-pie.exp * gdb.base/gdb1090.exp * gdb.base/gdb1555.exp * gdb.base/global-var-nested-by-dso.exp * gdb.base/gnu-ifunc.exp * gdb.base/hbreak-in-shr-unsupported.exp * gdb.base/hbreak-unmapped.exp * gdb.base/hook-stop.exp * gdb.base/infcall-input.exp * gdb.base/info-fun.exp * gdb.base/info-shared.exp * gdb.base/interrupt-noterm.exp * gdb.base/jit-so.exp * gdb.base/jit.exp * gdb.base/line-symtabs.exp * gdb.base/list.exp * gdb.base/longjmp.exp * gdb.base/macscp.exp * gdb.base/max-value-size.exp * gdb.base/nodebug.exp * gdb.base/nofield.exp * gdb.base/overlays.exp * gdb.base/paginate-after-ctrl-c-running.exp * gdb.base/paginate-bg-execution.exp * gdb.base/paginate-inferior-exit.exp * gdb.base/pending.exp * gdb.base/pr11022.exp * gdb.base/printcmds.exp * gdb.base/ptr-typedef.exp * gdb.base/ptype.exp * gdb.base/randomize.exp * gdb.base/range-stepping.exp * gdb.base/realname-expand.exp * gdb.base/relativedebug.exp * gdb.base/remote.exp * gdb.base/savedregs.exp * gdb.base/sepdebug.exp * gdb.base/set-noassign.exp * gdb.base/shlib-call.exp * gdb.base/shreloc.exp * gdb.base/sigaltstack.exp * gdb.base/sigbpt.exp * gdb.base/siginfo-addr.exp * gdb.base/siginfo-obj.exp * gdb.base/siginfo-thread.exp * gdb.base/signest.exp * gdb.base/signull.exp * gdb.base/sigrepeat.exp * gdb.base/skip.exp * gdb.base/so-impl-ld.exp * gdb.base/solib-corrupted.exp * gdb.base/solib-disc.exp * gdb.base/solib-display.exp * gdb.base/solib-overlap.exp * gdb.base/solib-search.exp * gdb.base/solib-symbol.exp * gdb.base/source-execution.exp * gdb.base/sss-bp-on-user-bp-2.exp * gdb.base/sss-bp-on-user-bp.exp * gdb.base/stack-checking.exp * gdb.base/stale-infcall.exp * gdb.base/step-break.exp * gdb.base/step-line.exp * gdb.base/step-over-exit.exp * gdb.base/step-test.exp * gdb.base/structs.exp * gdb.base/sym-file.exp * gdb.base/symtab-search-order.exp * gdb.base/term.exp * gdb.base/type-opaque.exp * gdb.base/unload.exp * gdb.base/until-nodebug.exp * gdb.base/until.exp * gdb.base/unwindonsignal.exp * gdb.base/watch-cond.exp * gdb.base/watch-non-mem.exp * gdb.base/watch_thread_num.exp * gdb.base/watchpoint-reuse-slot.exp * gdb.base/watchpoint-solib.exp * gdb.base/watchpoint.exp * gdb.btrace/dlopen.exp * gdb.cell/arch.exp * gdb.cell/break.exp * gdb.cell/bt.exp * gdb.cell/core.exp * gdb.cell/data.exp * gdb.cell/dwarfaddr.exp * gdb.cell/ea-cache.exp * gdb.cell/ea-standalone.exp * gdb.cell/ea-test.exp * gdb.cell/f-regs.exp * gdb.cell/fork.exp * gdb.cell/gcore.exp * gdb.cell/mem-access.exp * gdb.cell/ptype.exp * gdb.cell/registers.exp * gdb.cell/sizeof.exp * gdb.cell/solib-symbol.exp * gdb.cell/solib.exp * gdb.compile/compile-tls.exp * gdb.cp/exception.exp * gdb.cp/gdb2495.exp * gdb.cp/local.exp * gdb.cp/mb-inline.exp * gdb.cp/mb-templates.exp * gdb.cp/pr10687.exp * gdb.cp/pr9167.exp * gdb.cp/scope-err.exp * gdb.cp/templates.exp * gdb.cp/virtfunc.exp * gdb.dwarf2/dw2-dir-file-name.exp * gdb.dwarf2/dw2-single-line-discriminators.exp * gdb.fortran/complex.exp * gdb.fortran/library-module.exp * gdb.guile/guile.exp * gdb.guile/scm-cmd.exp * gdb.guile/scm-frame-inline.exp * gdb.guile/scm-objfile.exp * gdb.guile/scm-pretty-print.exp * gdb.guile/scm-symbol.exp * gdb.guile/scm-type.exp * gdb.guile/scm-value.exp * gdb.linespec/keywords.exp * gdb.linespec/ls-errs.exp * gdb.linespec/macro-relative.exp * gdb.linespec/thread.exp * gdb.mi/mi-breakpoint-changed.exp * gdb.mi/mi-dprintf-pending.exp * gdb.mi/mi-fullname-deleted.exp * gdb.mi/mi-logging.exp * gdb.mi/mi-pending.exp * gdb.mi/mi-solib.exp * gdb.mi/new-ui-mi-sync.exp * gdb.mi/user-selected-context-sync.exp * gdb.multi/dummy-frame-restore.exp * gdb.multi/multi-arch-exec.exp * gdb.multi/remove-inferiors.exp * gdb.multi/watchpoint-multi-exit.exp * gdb.opt/solib-intra-step.exp * gdb.perf/backtrace.exp * gdb.perf/single-step.exp * gdb.perf/skip-command.exp * gdb.perf/skip-prologue.exp * gdb.perf/solib.exp * gdb.python/lib-types.exp * gdb.python/py-as-string.exp * gdb.python/py-bad-printers.exp * gdb.python/py-block.exp * gdb.python/py-breakpoint.exp * gdb.python/py-cmd.exp * gdb.python/py-events.exp * gdb.python/py-evthreads.exp * gdb.python/py-finish-breakpoint.exp * gdb.python/py-finish-breakpoint2.exp * gdb.python/py-frame-inline.exp * gdb.python/py-frame.exp * gdb.python/py-inferior.exp * gdb.python/py-infthread.exp * gdb.python/py-mi.exp * gdb.python/py-objfile.exp * gdb.python/py-pp-maint.exp * gdb.python/py-pp-registration.exp * gdb.python/py-prettyprint.exp * gdb.python/py-recurse-unwind.exp * gdb.python/py-shared.exp * gdb.python/py-symbol.exp * gdb.python/py-symtab.exp * gdb.python/py-template.exp * gdb.python/py-type.exp * gdb.python/py-unwind-maint.exp * gdb.python/py-unwind.exp * gdb.python/py-value.exp * gdb.python/python.exp * gdb.reverse/finish-reverse-bkpt.exp * gdb.reverse/insn-reverse.exp * gdb.reverse/next-reverse-bkpt-over-sr.exp * gdb.reverse/solib-precsave.exp * gdb.reverse/solib-reverse.exp * gdb.stabs/gdb11479.exp * gdb.stabs/weird.exp * gdb.threads/fork-child-threads.exp * gdb.threads/fork-plus-threads.exp * gdb.threads/fork-thread-pending.exp * gdb.threads/forking-threads-plus-breakpoint.exp * gdb.threads/hand-call-in-threads.exp * gdb.threads/interrupted-hand-call.exp * gdb.threads/linux-dp.exp * gdb.threads/local-watch-wrong-thread.exp * gdb.threads/next-while-other-thread-longjmps.exp * gdb.threads/non-ldr-exit.exp * gdb.threads/pending-step.exp * gdb.threads/print-threads.exp * gdb.threads/process-dies-while-detaching.exp * gdb.threads/process-dies-while-handling-bp.exp * gdb.threads/pthreads.exp * gdb.threads/queue-signal.exp * gdb.threads/reconnect-signal.exp * gdb.threads/signal-command-handle-nopass.exp * gdb.threads/signal-command-multiple-signals-pending.exp * gdb.threads/signal-delivered-right-thread.exp * gdb.threads/signal-sigtrap.exp * gdb.threads/sigthread.exp * gdb.threads/staticthreads.exp * gdb.threads/stepi-random-signal.exp * gdb.threads/thread-unwindonsignal.exp * gdb.threads/thread_check.exp * gdb.threads/thread_events.exp * gdb.threads/tid-reuse.exp * gdb.threads/tls-nodebug.exp * gdb.threads/tls-shared.exp * gdb.threads/tls-so_extern.exp * gdb.threads/tls.exp * gdb.threads/wp-replication.exp * gdb.trace/actions-changed.exp * gdb.trace/actions.exp * gdb.trace/backtrace.exp * gdb.trace/change-loc.exp * gdb.trace/collection.exp * gdb.trace/deltrace.exp * gdb.trace/disconnected-tracing.exp * gdb.trace/entry-values.exp * gdb.trace/ftrace-lock.exp * gdb.trace/ftrace.exp * gdb.trace/infotrace.exp * gdb.trace/mi-trace-frame-collected.exp * gdb.trace/mi-trace-unavailable.exp * gdb.trace/mi-traceframe-changed.exp * gdb.trace/mi-tracepoint-changed.exp * gdb.trace/mi-tsv-changed.exp * gdb.trace/no-attach-trace.exp * gdb.trace/packetlen.exp * gdb.trace/passc-dyn.exp * gdb.trace/passcount.exp * gdb.trace/pending.exp * gdb.trace/pr16508.exp * gdb.trace/qtro.exp * gdb.trace/range-stepping.exp * gdb.trace/read-memory.exp * gdb.trace/report.exp * gdb.trace/save-trace.exp * gdb.trace/signal.exp * gdb.trace/stap-trace.exp * gdb.trace/status-stop.exp * gdb.trace/strace.exp * gdb.trace/tfile.exp * gdb.trace/tfind.exp * gdb.trace/trace-break.exp * gdb.trace/trace-condition.exp * gdb.trace/trace-enable-disable.exp * gdb.trace/trace-mt.exp * gdb.trace/tracecmd.exp * gdb.trace/tracefile-pseudo-reg.exp * gdb.trace/tspeed.exp * gdb.trace/tstatus.exp * gdb.trace/tsv.exp * gdb.trace/unavailable.exp * gdb.trace/while-dyn.exp * gdb.trace/while-stepping.exp * lib/gdb-guile.exp * lib/gdb.exp * lib/mi-support.exp * lib/pascal.exp * lib/perftest.exp * lib/prelink-support.exp * lib/selftest-support.exp
2016-12-01 21:40:05 +01:00
fail "can't run to main"
Stale breakpoint instructions, spurious SIGTRAPS. Without the code portion of the patch, we get these failures: FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue FAIL: gdb.base/break-unload-file.exp: always-inserted on: hbreak: continue FAIL: gdb.base/sym-file.exp: stale bkpts: continue to breakpoint: end here They all looks like random SIGTRAPs: continue Continuing. Program received signal SIGTRAP, Trace/breakpoint trap. 0x0000000000400541 in foo () at ../../../src/gdb/testsuite/gdb.base/break-unload-file.c:21 21 } (gdb) FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue (This is a regression caused by the remove-symbol-file command series.) break-unload-file.exp is about having breakpoints inserted, and then doing "file". I caught this while writing a test that does "file PROGRAM", while PROGRAM was already loaded, which internally does "file" first, because I wanted to force a breakpoint_re_set, but the test is more explicit in case GDB ever optimizes out that re-set. The problem is that unloading the file with "file" ends up in disable_breakpoints_in_freed_objfile, which marks all breakpoint locations of the objfile as both shlib_disabled, _and_ clears the inserted flag, without actually removing the breakpoints from the inferior. Now, usually, in all-stop, breakpoints will already be removed from the inferior before the user can issue the "file" command, but, with non-stop, or breakpoints always-inserted on mode, breakpoints stay inserted even while the user has the prompt. In the latter case, then, if we let the program continue, and it executes the address where we had previously set the breakpoint, it'll actually execute the breakpoint instruction that we left behind... Now, one issue is that the intent of disable_breakpoints_in_freed_objfile is really to handle the unloading of OBJF_USERLOADED objfiles. These are objfiles that were added with add-symbol-file and that are removed with remove-symbol-file. "add-symbol-file"'s docs in the manual clearly say these commands are used to let GDB know about dynamically loaded code: You would use this command when @var{filename} has been dynamically loaded (by some other means) into the program that is running. Similarly, the online help says: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. So it makes sense to, like when shared libraries are unloaded through the generic solib machinery, mark the breakpoint locations as shlib_disabled. But, the "file" command is not about dynamically loaded code, it's about the main program. So the patch makes disable_breakpoints_in_freed_objfile skip all objfiles but OBJF_USERLOADED ones, thus skipping the main objfile. Then, the reason that disable_breakpoints_in_freed_objfile was clearing the inserted flag isn't clear, but likely to avoid breakpoint removal errors, assuming remove-symbol-file was called after the dynamic object was already unmapped from the inferior. In that case, it'd okay to simply clear the inserted flag, but not so if the user for example does remove-symbol-file to remove the library because he made a mistake in the library's address, and wants to re-do add-symbol-file with the correct address. To address all that, I propose an alternative implementation, that handles both cases. The patch includes changes to sym-file.exp to cover them. This implementation leaves the inserted flag alone, and handles breakpoint insertion/removal failure gracefully when the locations are in OBJF_USERLOADED objfiles, just like we handle insertion/removal failure gracefully for locations in shared libraries. To try to make sure we aren't patching back stale shadow memory contents into the inferior, in case the program mapped a different library at the same address where we had the breakpoint, without the user having had a chance of remove-symbol-file'ing before, this adds a new memory_validate_breakpoint function that checks if the breakpoint instruction is still in memory. ppc_linux_memory_remove_breakpoint does this unconditionally for all memory breakpoints, and questions whether memory_remove_breakpoint should be changed to do this for all breakpoints. Possibly yes, though I'm not certain, hence this baby-steps patch. Tested on x86_64 Fedora 17, native and gdbserver. gdb/ 2014-04-23 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location): Tolerate errors if the breakpoint is set in a user-loaded objfile. (remove_breakpoint_1): Likewise. Also tolerate errors if the location is marked shlib_disabled. If the breakpoint is set in a user-loaded objfile is a GDB-side memory breakpoint, validate it before uninsertion. (disable_breakpoints_in_freed_objfile): Skip non-OBJF_USERLOADED objfiles. Don't clear the location's inserted flag. * mem-break.c (memory_validate_breakpoint): New function. * objfiles.c (userloaded_objfile_contains_address_p): New function. * objfiles.h (userloaded_objfile_contains_address_p): Declare. * target.h (memory_validate_breakpoint): New declaration. gdb/testsuite/ 2014-04-23 Pedro Alves <palves@redhat.com> * gdb.base/break-unload-file.c: New file. * gdb.base/break-unload-file.exp: New file. * gdb.base/sym-file-lib.c (baz): New function. * gdb.base/sym-file-loader.c (struct segment) <mapped_size>: New field. (load): Store the segment's mapped size. (unload): New function. (unload_shlib): New function. * gdb.base/sym-file-loader.h (unload_shlib): New declaration. * gdb.base/sym-file-main.c (main): Unload, and reload the library, set a breakpoint at baz, and call it. * gdb.base/sym-file.exp: New tests for stale breakpoint instructions.
2014-04-23 00:19:19 +02:00
return
}
delete_breakpoints
gdb_test_no_output "set breakpoint always-inserted $always_inserted"
set test "$break_command foo"
gdb_test_multiple "$break_command foo" $test {
-re "No hardware breakpoint support in the target.*$gdb_prompt $" {
unsupported $test
return
}
-re "Hardware breakpoints used exceeds limit.*$gdb_prompt $" {
unsupported $test
return
}
-re "Cannot insert hardware breakpoint.*$gdb_prompt $" {
unsupported $test
return
}
-re ".*reakpoint .* at .*$gdb_prompt $" {
pass $test
}
}
# The breakpoint shouldn't be pending now.
gdb_test "info break" "y.*$hex.*in foo at.*" \
"breakpoint is not pending"
# Remove the file, while the breakpoint above is inserted in a
# function in the main objfile. GDB used to have a bug where
# it would mark the breakpoint as uninserted, but actually
# would leave it inserted in the target.
set test "file"
gdb_test_multiple "file" $test {
-re "Are you sure you want to change the file. .*y or n. $" {
send_gdb "y\n"
exp_continue
}
-re "Discard symbol table from `.*'? .y or n. $" {
send_gdb "y\n"
exp_continue
}
-re "No symbol file now\\.\r\n$gdb_prompt $" {
pass $test
}
}
"$ gdb PROGRAM" vs "(gdb) file PROGRAM" difference; warn on failure to remove breakpoint. Turns out there's a difference between loading the program with "gdb PROGRAM", vs loading it with "(gdb) file PROGRAM". The latter results in the objfile ending up with OBJF_USERLOADED set, while not with the former. (That difference seems bogus, but still that's not the point of this patch. We can revisit that afterwards.) The new code that suppresses breakpoint removal errors for add-symbol-file objects ends up being too greedy: /* In some cases, we might not be able to remove a breakpoint in a shared library that has already been removed, but we have not yet processed the shlib unload event. Similarly for an unloaded add-symbol-file object - the user might not yet have had the chance to remove-symbol-file it. shlib_disabled will be set if the library/object has already been removed, but the breakpoint hasn't been uninserted yet, e.g., after "nosharedlibrary" or "remove-symbol-file" with breakpoints always-inserted mode. */ if (val && (bl->loc_type == bp_loc_software_breakpoint && (bl->shlib_disabled || solib_name_from_address (bl->pspace, bl->address) || userloaded_objfile_contains_address_p (bl->pspace, bl->address)))) val = 0; as it turns out that OBJF_USERLOADED can be set for objfiles loaded by some other means not add-symbol-file. In this case, symbol-file (or "file", which is really just "exec-file"+"symbol-file"). Recall that add-symbol-file is documented as: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ And it's the "dynamically loaded" aspect that the breakpoint.c code cares about. So make add-symbol-file set OBJF_SHARED on its objfiles too, and tweak the breakpoint.c code to look for OBJF_SHARED instead of OBJF_USERLOADED. This restores back the missing breakpoint removal warning when we let sss-bp-on-user-bp-2.exp run on native GNU/Linux (https://sourceware.org/ml/gdb-patches/2014-06/msg00335.html): (gdb) PASS: gdb.base/sss-bp-on-user-bp-2.exp: define stepi_del_break stepi_del_break warning: Error removing breakpoint 3 (gdb) FAIL: gdb.base/sss-bp-on-user-bp-2.exp: stepi_del_break I say "restores" because this was GDB's behavior in 7.7 and earlier. And, likewise, "file" with no arguments only started turning breakpoints set in the main executable to "<pending>" with the remote-symbol-file patch (63644780). The old behavior is now restored, and we break-unload-file.exp test now exercizes both "gdb; file PROGRAM" and "gdb PROGRAM". gdb/ 2014-06-16 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location, remove_breakpoint_1): Adjust. (disable_breakpoints_in_freed_objfile): Skip objfiles that don't have OBJF_SHARED set. * objfiles.c (userloaded_objfile_contains_address_p): Rename to... (shared_objfile_contains_address_p): ... this. Check OBJF_SHARED instead of OBJF_USERLOADED. * objfiles.h (OBJF_SHARED): Update comment. (userloaded_objfile_contains_address_p): Rename to ... (shared_objfile_contains_address_p): ... this, and update comments. * symfile.c (add_symbol_file_command): Also set OBJF_SHARED in the new objfile. (remove_symbol_file_command): Skip objfiles that don't have OBJF_SHARED set. gdb/testsuite/ 2014-06-16 Pedro Alves <palves@redhat.com> * gdb.base/break-main-file-remove-fail.c: New file. * gdb.base/break-main-file-remove-fail.exp: New file. * gdb.base/break-unload-file.exp: Use build_executable instead of prepare_for_testing. (test_break): New parameter "initial_load". Handle it. (top level): Add initial_load cmdline/file axis.
2014-06-16 16:38:13 +02:00
# This test used to behave differently depending on whether
# the program was first loaded through "file PROGRAM" or "gdb
# PROGRAM".
set ws "\[ \t\]"
gdb_test "info break" "breakpoint${ws}+keep${ws}+n${ws}+$hex${ws}*" \
"breakpoint is disabled"
Stale breakpoint instructions, spurious SIGTRAPS. Without the code portion of the patch, we get these failures: FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue FAIL: gdb.base/break-unload-file.exp: always-inserted on: hbreak: continue FAIL: gdb.base/sym-file.exp: stale bkpts: continue to breakpoint: end here They all looks like random SIGTRAPs: continue Continuing. Program received signal SIGTRAP, Trace/breakpoint trap. 0x0000000000400541 in foo () at ../../../src/gdb/testsuite/gdb.base/break-unload-file.c:21 21 } (gdb) FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue (This is a regression caused by the remove-symbol-file command series.) break-unload-file.exp is about having breakpoints inserted, and then doing "file". I caught this while writing a test that does "file PROGRAM", while PROGRAM was already loaded, which internally does "file" first, because I wanted to force a breakpoint_re_set, but the test is more explicit in case GDB ever optimizes out that re-set. The problem is that unloading the file with "file" ends up in disable_breakpoints_in_freed_objfile, which marks all breakpoint locations of the objfile as both shlib_disabled, _and_ clears the inserted flag, without actually removing the breakpoints from the inferior. Now, usually, in all-stop, breakpoints will already be removed from the inferior before the user can issue the "file" command, but, with non-stop, or breakpoints always-inserted on mode, breakpoints stay inserted even while the user has the prompt. In the latter case, then, if we let the program continue, and it executes the address where we had previously set the breakpoint, it'll actually execute the breakpoint instruction that we left behind... Now, one issue is that the intent of disable_breakpoints_in_freed_objfile is really to handle the unloading of OBJF_USERLOADED objfiles. These are objfiles that were added with add-symbol-file and that are removed with remove-symbol-file. "add-symbol-file"'s docs in the manual clearly say these commands are used to let GDB know about dynamically loaded code: You would use this command when @var{filename} has been dynamically loaded (by some other means) into the program that is running. Similarly, the online help says: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. So it makes sense to, like when shared libraries are unloaded through the generic solib machinery, mark the breakpoint locations as shlib_disabled. But, the "file" command is not about dynamically loaded code, it's about the main program. So the patch makes disable_breakpoints_in_freed_objfile skip all objfiles but OBJF_USERLOADED ones, thus skipping the main objfile. Then, the reason that disable_breakpoints_in_freed_objfile was clearing the inserted flag isn't clear, but likely to avoid breakpoint removal errors, assuming remove-symbol-file was called after the dynamic object was already unmapped from the inferior. In that case, it'd okay to simply clear the inserted flag, but not so if the user for example does remove-symbol-file to remove the library because he made a mistake in the library's address, and wants to re-do add-symbol-file with the correct address. To address all that, I propose an alternative implementation, that handles both cases. The patch includes changes to sym-file.exp to cover them. This implementation leaves the inserted flag alone, and handles breakpoint insertion/removal failure gracefully when the locations are in OBJF_USERLOADED objfiles, just like we handle insertion/removal failure gracefully for locations in shared libraries. To try to make sure we aren't patching back stale shadow memory contents into the inferior, in case the program mapped a different library at the same address where we had the breakpoint, without the user having had a chance of remove-symbol-file'ing before, this adds a new memory_validate_breakpoint function that checks if the breakpoint instruction is still in memory. ppc_linux_memory_remove_breakpoint does this unconditionally for all memory breakpoints, and questions whether memory_remove_breakpoint should be changed to do this for all breakpoints. Possibly yes, though I'm not certain, hence this baby-steps patch. Tested on x86_64 Fedora 17, native and gdbserver. gdb/ 2014-04-23 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location): Tolerate errors if the breakpoint is set in a user-loaded objfile. (remove_breakpoint_1): Likewise. Also tolerate errors if the location is marked shlib_disabled. If the breakpoint is set in a user-loaded objfile is a GDB-side memory breakpoint, validate it before uninsertion. (disable_breakpoints_in_freed_objfile): Skip non-OBJF_USERLOADED objfiles. Don't clear the location's inserted flag. * mem-break.c (memory_validate_breakpoint): New function. * objfiles.c (userloaded_objfile_contains_address_p): New function. * objfiles.h (userloaded_objfile_contains_address_p): Declare. * target.h (memory_validate_breakpoint): New declaration. gdb/testsuite/ 2014-04-23 Pedro Alves <palves@redhat.com> * gdb.base/break-unload-file.c: New file. * gdb.base/break-unload-file.exp: New file. * gdb.base/sym-file-lib.c (baz): New function. * gdb.base/sym-file-loader.c (struct segment) <mapped_size>: New field. (load): Store the segment's mapped size. (unload): New function. (unload_shlib): New function. * gdb.base/sym-file-loader.h (unload_shlib): New declaration. * gdb.base/sym-file-main.c (main): Unload, and reload the library, set a breakpoint at baz, and call it. * gdb.base/sym-file.exp: New tests for stale breakpoint instructions.
2014-04-23 00:19:19 +02:00
# Now delete the breakpoint from GDB's tables, to make sure
# GDB doesn't reinsert it, masking the bug (with the bug, on
# re-insert, GDB would fill the shadow buffer with a
# breakpoint instruction). Avoid delete_breakpoints as that
# doesn't record a pass/fail.
gdb_test "delete" "" "delete all breakpoints" \
"Delete all breakpoints.*y or n.*$" "y"
# Re-add symbols back.
set test "file \$binfile"
gdb_test_multiple "file $binfile" $test {
-re "Are you sure you want to change the file. .*y or n. $" {
send_gdb "y\n"
exp_continue
}
-re "Reading symbols from.*done.*$gdb_prompt $" {
pass $test
}
}
# Run to another function now. With the bug, GDB would trip
# on a spurious trap at foo.
gdb_test "b bar" ".*reakpoint .* at .*"
gdb_test "continue" "Breakpoint .*, bar .*"
}
}
"$ gdb PROGRAM" vs "(gdb) file PROGRAM" difference; warn on failure to remove breakpoint. Turns out there's a difference between loading the program with "gdb PROGRAM", vs loading it with "(gdb) file PROGRAM". The latter results in the objfile ending up with OBJF_USERLOADED set, while not with the former. (That difference seems bogus, but still that's not the point of this patch. We can revisit that afterwards.) The new code that suppresses breakpoint removal errors for add-symbol-file objects ends up being too greedy: /* In some cases, we might not be able to remove a breakpoint in a shared library that has already been removed, but we have not yet processed the shlib unload event. Similarly for an unloaded add-symbol-file object - the user might not yet have had the chance to remove-symbol-file it. shlib_disabled will be set if the library/object has already been removed, but the breakpoint hasn't been uninserted yet, e.g., after "nosharedlibrary" or "remove-symbol-file" with breakpoints always-inserted mode. */ if (val && (bl->loc_type == bp_loc_software_breakpoint && (bl->shlib_disabled || solib_name_from_address (bl->pspace, bl->address) || userloaded_objfile_contains_address_p (bl->pspace, bl->address)))) val = 0; as it turns out that OBJF_USERLOADED can be set for objfiles loaded by some other means not add-symbol-file. In this case, symbol-file (or "file", which is really just "exec-file"+"symbol-file"). Recall that add-symbol-file is documented as: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ And it's the "dynamically loaded" aspect that the breakpoint.c code cares about. So make add-symbol-file set OBJF_SHARED on its objfiles too, and tweak the breakpoint.c code to look for OBJF_SHARED instead of OBJF_USERLOADED. This restores back the missing breakpoint removal warning when we let sss-bp-on-user-bp-2.exp run on native GNU/Linux (https://sourceware.org/ml/gdb-patches/2014-06/msg00335.html): (gdb) PASS: gdb.base/sss-bp-on-user-bp-2.exp: define stepi_del_break stepi_del_break warning: Error removing breakpoint 3 (gdb) FAIL: gdb.base/sss-bp-on-user-bp-2.exp: stepi_del_break I say "restores" because this was GDB's behavior in 7.7 and earlier. And, likewise, "file" with no arguments only started turning breakpoints set in the main executable to "<pending>" with the remote-symbol-file patch (63644780). The old behavior is now restored, and we break-unload-file.exp test now exercizes both "gdb; file PROGRAM" and "gdb PROGRAM". gdb/ 2014-06-16 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location, remove_breakpoint_1): Adjust. (disable_breakpoints_in_freed_objfile): Skip objfiles that don't have OBJF_SHARED set. * objfiles.c (userloaded_objfile_contains_address_p): Rename to... (shared_objfile_contains_address_p): ... this. Check OBJF_SHARED instead of OBJF_USERLOADED. * objfiles.h (OBJF_SHARED): Update comment. (userloaded_objfile_contains_address_p): Rename to ... (shared_objfile_contains_address_p): ... this, and update comments. * symfile.c (add_symbol_file_command): Also set OBJF_SHARED in the new objfile. (remove_symbol_file_command): Skip objfiles that don't have OBJF_SHARED set. gdb/testsuite/ 2014-06-16 Pedro Alves <palves@redhat.com> * gdb.base/break-main-file-remove-fail.c: New file. * gdb.base/break-main-file-remove-fail.exp: New file. * gdb.base/break-unload-file.exp: Use build_executable instead of prepare_for_testing. (test_break): New parameter "initial_load". Handle it. (top level): Add initial_load cmdline/file axis.
2014-06-16 16:38:13 +02:00
foreach initial_load { "cmdline" "file" } {
# While it doesn't trigger the original bug this is a regression
# test for, test with breakpoint always-inserted off for extra
# coverage.
foreach always_inserted { "off" "on" } {
test_break $initial_load $always_inserted "break"
if {![skip_hw_breakpoint_tests]} {
test_break $initial_load $always_inserted "hbreak"
}
Stale breakpoint instructions, spurious SIGTRAPS. Without the code portion of the patch, we get these failures: FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue FAIL: gdb.base/break-unload-file.exp: always-inserted on: hbreak: continue FAIL: gdb.base/sym-file.exp: stale bkpts: continue to breakpoint: end here They all looks like random SIGTRAPs: continue Continuing. Program received signal SIGTRAP, Trace/breakpoint trap. 0x0000000000400541 in foo () at ../../../src/gdb/testsuite/gdb.base/break-unload-file.c:21 21 } (gdb) FAIL: gdb.base/break-unload-file.exp: always-inserted on: break: continue (This is a regression caused by the remove-symbol-file command series.) break-unload-file.exp is about having breakpoints inserted, and then doing "file". I caught this while writing a test that does "file PROGRAM", while PROGRAM was already loaded, which internally does "file" first, because I wanted to force a breakpoint_re_set, but the test is more explicit in case GDB ever optimizes out that re-set. The problem is that unloading the file with "file" ends up in disable_breakpoints_in_freed_objfile, which marks all breakpoint locations of the objfile as both shlib_disabled, _and_ clears the inserted flag, without actually removing the breakpoints from the inferior. Now, usually, in all-stop, breakpoints will already be removed from the inferior before the user can issue the "file" command, but, with non-stop, or breakpoints always-inserted on mode, breakpoints stay inserted even while the user has the prompt. In the latter case, then, if we let the program continue, and it executes the address where we had previously set the breakpoint, it'll actually execute the breakpoint instruction that we left behind... Now, one issue is that the intent of disable_breakpoints_in_freed_objfile is really to handle the unloading of OBJF_USERLOADED objfiles. These are objfiles that were added with add-symbol-file and that are removed with remove-symbol-file. "add-symbol-file"'s docs in the manual clearly say these commands are used to let GDB know about dynamically loaded code: You would use this command when @var{filename} has been dynamically loaded (by some other means) into the program that is running. Similarly, the online help says: (gdb) help add-symbol-file Load symbols from FILE, assuming FILE has been dynamically loaded. So it makes sense to, like when shared libraries are unloaded through the generic solib machinery, mark the breakpoint locations as shlib_disabled. But, the "file" command is not about dynamically loaded code, it's about the main program. So the patch makes disable_breakpoints_in_freed_objfile skip all objfiles but OBJF_USERLOADED ones, thus skipping the main objfile. Then, the reason that disable_breakpoints_in_freed_objfile was clearing the inserted flag isn't clear, but likely to avoid breakpoint removal errors, assuming remove-symbol-file was called after the dynamic object was already unmapped from the inferior. In that case, it'd okay to simply clear the inserted flag, but not so if the user for example does remove-symbol-file to remove the library because he made a mistake in the library's address, and wants to re-do add-symbol-file with the correct address. To address all that, I propose an alternative implementation, that handles both cases. The patch includes changes to sym-file.exp to cover them. This implementation leaves the inserted flag alone, and handles breakpoint insertion/removal failure gracefully when the locations are in OBJF_USERLOADED objfiles, just like we handle insertion/removal failure gracefully for locations in shared libraries. To try to make sure we aren't patching back stale shadow memory contents into the inferior, in case the program mapped a different library at the same address where we had the breakpoint, without the user having had a chance of remove-symbol-file'ing before, this adds a new memory_validate_breakpoint function that checks if the breakpoint instruction is still in memory. ppc_linux_memory_remove_breakpoint does this unconditionally for all memory breakpoints, and questions whether memory_remove_breakpoint should be changed to do this for all breakpoints. Possibly yes, though I'm not certain, hence this baby-steps patch. Tested on x86_64 Fedora 17, native and gdbserver. gdb/ 2014-04-23 Pedro Alves <palves@redhat.com> * breakpoint.c (insert_bp_location): Tolerate errors if the breakpoint is set in a user-loaded objfile. (remove_breakpoint_1): Likewise. Also tolerate errors if the location is marked shlib_disabled. If the breakpoint is set in a user-loaded objfile is a GDB-side memory breakpoint, validate it before uninsertion. (disable_breakpoints_in_freed_objfile): Skip non-OBJF_USERLOADED objfiles. Don't clear the location's inserted flag. * mem-break.c (memory_validate_breakpoint): New function. * objfiles.c (userloaded_objfile_contains_address_p): New function. * objfiles.h (userloaded_objfile_contains_address_p): Declare. * target.h (memory_validate_breakpoint): New declaration. gdb/testsuite/ 2014-04-23 Pedro Alves <palves@redhat.com> * gdb.base/break-unload-file.c: New file. * gdb.base/break-unload-file.exp: New file. * gdb.base/sym-file-lib.c (baz): New function. * gdb.base/sym-file-loader.c (struct segment) <mapped_size>: New field. (load): Store the segment's mapped size. (unload): New function. (unload_shlib): New function. * gdb.base/sym-file-loader.h (unload_shlib): New declaration. * gdb.base/sym-file-main.c (main): Unload, and reload the library, set a breakpoint at baz, and call it. * gdb.base/sym-file.exp: New tests for stale breakpoint instructions.
2014-04-23 00:19:19 +02:00
}
}