* dwarf2read.c (read_cutu_die_from_dwo): Fix function comment.
Remove unused local comp_dir_attr. Assert exactly one of
stub_comp_unit_die, stub_comp_dir is non-NULL.
Expand a bit the comments to answer some questions I had when looking
at why a target of mine would not have some default methods set.
gdb/ChangeLog:
* target.h (complete_target_initialization, add_target):
Add comment.
So that all native targets inherit a single "superclass".
Target methods that are set to or do the same as inf-child.c's are
removed.
Tested by cross building on Fedora 17.
gdb/
2014-03-07 Pedro Alves <palves@redhat.com>
* go32-nat.c: Include inf-child.h.
(go32_ops): Delete global.
(go32_close, go32_detach, go32_prepare_to_store, go32_can_run):
Delete methods.
(go32_create_inferior): Push the passed in target pointer instead
of referencing go32_ops.
(init_go32_ops): Delete function. Moved parts to _initialize_go32_nat.
(go32_target): New function, based on init_go32_ops, but inherit
inf_child_target.
(_initialize_go32_nat): Use go32_target. Move parts of
init_go32_ops here.
Some updates where needed after the minimal symbol handling got changed
a little. This patch makes those changes.
gdb/ChangeLog:
* sol-thread.c: #include "symtab.h", "minsym.h" and "objfiles.h".
(ps_pglobal_lookup): Use BMSYMBOL_VALUE_ADDRESS instead of
SYMBOL_VALUE_ADDRESS.
(info_cb): MSYMBOL_PRINT_NAME instead of SYMBOL_PRINT_NAME.
This patch is to remove parameter optional_p as it is always true,
in order to simplify get_tracepoint_by_number.
'optional_p' was added by this change,
1999-11-18 Tom Tromey <tromey@cygnus.com>
* tracepoint.h (get_tracepoint_by_number): Updated
declaration.
* tracepoint.c (trace_pass_command): Better error message.
Fixed logic when `all' not specified.
(get_tracepoint_by_number): Added `optional_p' argument. Fixed
all callers.
but after this patch,
FYI: remove `static's from cli-utils.c
https://sourceware.org/ml/gdb-patches/2011-03/msg00636.html
'optional_p' passed to get_tracepoint_by_number become always true.
gdb:
2014-03-06 Yao Qi <yao@codesourcery.com>
* breakpoint.c (get_tracepoint_by_number): Remove argument
optional_p. All callers updated. Adjust comments. Update
output message.
* breakpoint.h (get_tracepoint_by_number): Update declaration.
In GDB mainline, the error message for goto-bookmark
isn't perfect.
(gdb) goto-bookmark 1.1
goto-bookmark: no bookmark found for ''.
This patch tweaks the error message by checking the return value of
get_number. With patch applied, it becomes:
(gdb) goto-bookmark 1.1
goto-bookmark: invalid bookmark number '1.1'.
gdb:
2014-03-06 Yao Qi <yao@codesourcery.com>
* reverse.c (goto_bookmark_command): Add local 'p'. Emit error
early if get_number returns zero. Use 'p' instead of 'args'.
Hi,
GDB prints two warnings in one single line, as below:
(gdb) p 1.2
$1 = 1.2
(gdb) enable $1.2
History value must have integer type.Bad breakpoint number '$1'
This patch adds '\n' at the end of message.
gdb:
2014-03-06 Yao Qi <yao@codesourcery.com>
* cli/cli-utils.c (get_number_trailer): Add '\n' at the end of
message.
This patch fixes PR16508, which is about MI "-trace-find frame-number 0"
behaves differently from CLI "tfind 0". In CLI, we check both
status->running and status->filename, but in MI, we only check
status->running, which looks wrong to me. This patch moves the code
of checking to a new function check_trace_running, and use it in
both CLI and MI.
This patch also adds a test case pr16508.exp, which fails without this
fix, and passes with the fix applied.
FAIL: gdb.trace/pr16508.exp: interpreter-exec mi "-trace-find frame-number 0"
gdb:
2014-03-06 Yao Qi <yao@codesourcery.com>
PR breakpoints/16508
* tracepoint.c (check_trace_running): New function.
(trace_find_command): Move code to check_trace_running and
call check_trace_running.
(trace_find_pc_command): Likewise.
(trace_find_tracepoint_command): Likewise.
(trace_find_line_command): Likewise.
(trace_find_range_command): Likewise.
* tracepoint.h (check_trace_running): Likewise.
* mi/mi-main.c (mi_cmd_trace_find): Call check_trace_running.
gdb/testsuite:
2014-03-06 Yao Qi <yao@codesourcery.com>
* gdb.trace/pr16508.exp: New file.
This patch is to change the default implementation of to_traceframe_info
from 'return NULL' to tcomplain, which is intended. If new target
supports tracepoint, this method should be implemented, otherwise,
an error is thrown.
gdb:
2014-03-06 Yao Qi <yao@codesourcery.com>
* target.h (struct target_ops) <to_traceframe_info>: Use
TARGET_DEFAULT_NORETURN (tcomplain ()).
* target-delegates.c: Regenerated.
In non-stop mode, or rather, breakpoints always-inserted mode, the
code cache can easily end up with stale breakpoint instructions:
All it takes is filling a cache line when breakpoints already exist in
that memory region, and then delete the breakpoint.
Vis. (from the new test):
(gdb) set breakpoint always-inserted on
(gdb) b 23
Breakpoint 2 at 0x400540: file ../../../src/gdb/testsuite/gdb.base/breakpoint-shadow.c, line 23.
(gdb) b 24
Breakpoint 3 at 0x400547: file ../../../src/gdb/testsuite/gdb.base/breakpoint-shadow.c, line 24.
disass main
Dump of assembler code for function main:
0x000000000040053c <+0>: push %rbp
0x000000000040053d <+1>: mov %rsp,%rbp
=> 0x0000000000400540 <+4>: movl $0x1,-0x4(%rbp)
0x0000000000400547 <+11>: movl $0x2,-0x4(%rbp)
0x000000000040054e <+18>: mov $0x0,%eax
0x0000000000400553 <+23>: pop %rbp
0x0000000000400554 <+24>: retq
End of assembler dump.
So far so good. Now flush the code cache:
(gdb) set code-cache off
(gdb) set code-cache on
Requesting a disassembly works as expected, breakpoint shadowing is
applied:
(gdb) disass main
Dump of assembler code for function main:
0x000000000040053c <+0>: push %rbp
0x000000000040053d <+1>: mov %rsp,%rbp
=> 0x0000000000400540 <+4>: movl $0x1,-0x4(%rbp)
0x0000000000400547 <+11>: movl $0x2,-0x4(%rbp)
0x000000000040054e <+18>: mov $0x0,%eax
0x0000000000400553 <+23>: pop %rbp
0x0000000000400554 <+24>: retq
End of assembler dump.
However, now delete the breakpoints:
(gdb) delete
Delete all breakpoints? (y or n) y
And disassembly shows the old breakpoint instructions:
(gdb) disass main
Dump of assembler code for function main:
0x000000000040053c <+0>: push %rbp
0x000000000040053d <+1>: mov %rsp,%rbp
=> 0x0000000000400540 <+4>: int3
0x0000000000400541 <+5>: rex.RB cld
0x0000000000400543 <+7>: add %eax,(%rax)
0x0000000000400545 <+9>: add %al,(%rax)
0x0000000000400547 <+11>: int3
0x0000000000400548 <+12>: rex.RB cld
0x000000000040054a <+14>: add (%rax),%al
0x000000000040054c <+16>: add %al,(%rax)
0x000000000040054e <+18>: mov $0x0,%eax
0x0000000000400553 <+23>: pop %rbp
0x0000000000400554 <+24>: retq
End of assembler dump.
Those breakpoint instructions are no longer installed in target memory
they're stale in the code cache. Easily confirmed by just disabling
the code cache:
(gdb) set code-cache off
(gdb) disass main
Dump of assembler code for function main:
0x000000000040053c <+0>: push %rbp
0x000000000040053d <+1>: mov %rsp,%rbp
=> 0x0000000000400540 <+4>: movl $0x1,-0x4(%rbp)
0x0000000000400547 <+11>: movl $0x2,-0x4(%rbp)
0x000000000040054e <+18>: mov $0x0,%eax
0x0000000000400553 <+23>: pop %rbp
0x0000000000400554 <+24>: retq
End of assembler dump.
I stumbled upon this when writing a patch to infrun.c, that made
handle_inferior_event & co fill in the cache before breakpoints were
removed from the target. Recall that wait_for_inferior flushes the
dcache for every event. So in that case, always-inserted mode was not
necessary to trigger this. It's just a convenient way to expose the
issue.
The dcache works at the raw memory level. We need to update it
whenever memory is written, no matter what kind of target memory
object was originally passed down by the caller. The issue is that
the dcache update code isn't reached when a caller explicitly writes
raw memory. Breakpoint insertion/removal is one such case --
mem-break.c uses target_write_read_memory/target_write_raw_memory.
The fix is to move the dcache update code from memory_xfer_partial_1
to raw_memory_xfer_partial so that it's always reachable.
When we do that, we can actually simplify a series of things.
memory_xfer_partial_1 no longer needs to handle writes for any kind of
memory object, and therefore dcache_xfer_memory no longer needs to
handle writes either. So the latter (dcache_xfer_memory) and its
callees can be simplified to only care about reads. While we're
touching dcache_xfer_memory's prototype, might as well rename it to
reflect that fact that it only handles reads, and make it follow the
new target_xfer_status/xfered_len style. This made me notice that
dcache_xfer_memory loses the real error status if a memory read fails:
we could have failed to read due to TARGET_XFER_E_UNAVAILABLE, for
instance, but we always return TARGET_XFER_E_IO, hence the FIXME note.
I felt that fixing that fell out of the scope of this patch.
Currently dcache_xfer_memory handles the case of a write failing. The
whole cache line is invalidated when that happens. However,
dcache_update, the sole mechanism for handling writes that will remain
after the patch, does not presently handle that scenario. That's a
bug. The patch makes it handle that, by passing down the
target_xfer_status status from the caller, so that it can better
decide what to do itself. While I was changing the function's
prototype, I constified the myaddr parameter, getting rid of the need
for the cast as seen in its existing caller.
Tested on x86_64 Fedora 17, native and gdbserver.
gdb/
2014-03-05 Pedro Alves <palves@redhat.com>
PR gdb/16575
* dcache.c (dcache_poke_byte): Constify ptr parameter. Return
void. Update comment.
(dcache_xfer_memory): Delete.
(dcache_read_memory_partial): New, based on the read bits of
dcache_xfer_memory.
(dcache_update): Add status parameter. Use ULONGEST for len, and
adjust. Discard cache lines if the reason for the update was
error.
* dcache.h (dcache_xfer_memory): Delete declaration.
(dcache_read_memory_partial): New declaration.
(dcache_update): Update prototype.
* target.c (raw_memory_xfer_partial): Update the dcache here.
(memory_xfer_partial_1): Don't handle dcache writes here.
gdb/testsuite/
2014-03-05 Pedro Alves <palves@redhat.com>
PR gdb/16575
* gdb.base/breakpoint-shadow.exp (compare_disassembly): New
procedure.
(top level): Adjust to use it. Add tests that exercise breakpoint
interaction with the code-cache.
This patch moves the probe data from the objfile to the per-BFD
object. This lets the probes be shared between different inferiors
(and different objfiles when dlmopen is in use, should gdb ever handle
that).
2014-03-03 Tom Tromey <tromey@redhat.com>
* elfread.c (probe_key): Change to bfd_data.
(elf_get_probes, probe_key_free, _initialize_elfread): Probes are
now per-BFD, not per-objfile.
* stap-probe.c (stap_probe_destroy): Update comment.
(handle_stap_probe): Allocate on the per-BFD obstack.
This changes the probes to be independent of the program space.
After this, when a probe's address is needed, it is determined by
applying offsets at the point of use.
This introduces a bound_probe object, similar to bound minimal
symbols. Objects of this type are used when it's necessary to pass a
probe and its corresponding objfile.
This removes the backlink from probe to objfile, which was primarily
used to fetch the architecture to use.
This adds a get_probe_address function which calls a probe method to
compute the probe's relocated address. Similarly, it adds an objfile
parameter to the semaphore methods so they can do the relocation
properly as well.
2014-03-03 Tom Tromey <tromey@redhat.com>
* break-catch-throw.c (fetch_probe_arguments): Use bound probes.
* breakpoint.c (create_longjmp_master_breakpoint): Use
get_probe_address.
(add_location_to_breakpoint, bkpt_probe_insert_location)
(bkpt_probe_remove_location): Update.
* breakpoint.h (struct bp_location) <probe>: Now a bound_probe.
* elfread.c (elf_symfile_relocate_probe): Remove.
(elf_probe_fns): Update.
(insert_exception_resume_breakpoint): Change type of "probe"
parameter to bound_probe.
(check_exception_resume): Update.
* objfiles.c (objfile_relocate1): Don't relocate probes.
* probe.c (bound_probe_s): New typedef.
(parse_probes): Use get_probe_address. Set sal's objfile.
(find_probe_by_pc): Return a bound_probe.
(collect_probes): Return a VEC(bound_probe_s).
(compare_probes): Update.
(gen_ui_out_table_header_info): Change type of "probes"
parameter. Update.
(info_probes_for_ops): Update.
(get_probe_address): New function.
(probe_safe_evaluate_at_pc): Update.
* probe.h (struct probe_ops) <get_probe_address>: New field.
<set_semaphore, clear_semaphore>: Add objfile parameter.
(struct probe) <objfile>: Remove field.
<arch>: New field.
<address>: Update comment.
(struct bound_probe): New.
(find_probe_by_pc): Return a bound_probe.
(get_probe_address): Declare.
* solib-svr4.c (struct probe_and_action) <address>: New field.
(hash_probe_and_action, equal_probe_and_action): Update.
(register_solib_event_probe): Add address parameter.
(solib_event_probe_at): Update.
(svr4_create_probe_breakpoints): Add objfile parameter. Use
get_probe_address.
* stap-probe.c (struct stap_probe) <sem_addr>: Update comment.
(stap_get_probe_address): New function.
(stap_can_evaluate_probe_arguments, compute_probe_arg)
(compile_probe_arg): Update.
(stap_set_semaphore, stap_clear_semaphore): Compute semaphore's
address.
(handle_stap_probe): Don't relocate the probe.
(stap_relocate): Remove.
(stap_gen_info_probes_table_values): Update.
(stap_probe_ops): Remove stap_relocate.
* symfile-debug.c (debug_sym_relocate_probe): Remove.
(debug_sym_probe_fns): Update.
* symfile.h (struct sym_probe_fns) <sym_relocate_probe>: Remove.
* symtab.c (init_sal): Use memset.
* symtab.h (struct symtab_and_line) <objfile>: New field.
* tracepoint.c (start_tracing, stop_tracing): Update.
This fixes up a few mildly erroneous comments in probe.h.
2014-03-03 Tom Tromey <tromey@redhat.com>
* probe.h (parse_probes, find_probe_by_pc)
(find_probes_in_objfile): Fix comments.
OpenBSD 5.2 and later have a proper threads implementation based on
kernel threads. Debugging support is provided through additional
ptrace(2) requests, so this diff extends the generic code in
inf-ptrace.c with OpenBSD-specific code to discover additional threads.
gdb/ChangeLog:
* obsd-nat.h: New file.
* obsd-nat.c: New file.
* Makefile.in (HFILES_NO_SRCDIR): Add obsd-nat.h.
(ALLDEPFILES): Add obsd-nat.c.
This patch constifies ui_out_impl in struct ui_out, and various
instances of ui_out_impl.
This removes a couple of FIXME comments (near cli_ui_out_impl and
mi_ui_out_impl) that did not make sense to me.
Tested by rebuilding.
2014-02-28 Tom Tromey <tromey@redhat.com>
* cli-out.c (cli_ui_out_impl): Now const. Remove comment.
* cli-out.h (cli_ui_out_impl): Now const.
* mi/mi-out.c (mi_ui_out_impl): Now const. Remove comment.
* ui-out.c (struct ui_out) <impl>: Now const.
(default_ui_out_impl): Now const.
(ui_out_new): Make 'impl' parameter const.
* ui-out.h (ui_out_new): Update.
GCC 4.2.1 complains about first_l_name may be used uninitialized, and my brain
agrees.
gdb/ChangeLog:
* solib-svr4.c (svr4_read_so_list): Initialize first_l_name to 0.
runtest gdb.base/corefile.exp
==23174== ERROR: AddressSanitizer: heap-use-after-free on address 0x604400008c88 at pc 0x68f0be bp 0x7fffae9d7490 sp
0x7fffae9d7480
READ of size 8 at 0x604400008c88 thread T0
#0 0x68f0bd in svr4_read_so_list (/home/jkratoch/redhat/gdb-clean/gdb/gdb+0x68f0bd)
#1 0x68f64e in svr4_current_sos_direct (/home/jkratoch/redhat/gdb-clean/gdb/gdb+0x68f64e)
#2 0x68f757 in svr4_current_sos (/home/jkratoch/redhat/gdb-clean/gdb/gdb+0x68f757)
#3 0xcebbff in update_solib_list (/home/jkratoch/redhat/gdb-clean/gdb/gdb+0xcebbff)
0x604400008c88 is located 8 bytes inside of 1104-byte region [0x604400008c80,0x6044000090d0)
freed by thread T0 here:
#0 0x7f52677500f9 (/lib64/libasan.so.0+0x160f9)
#1 0xd2c68a in xfree (/home/jkratoch/redhat/gdb-clean/gdb/gdb+0xd2c68a)
#2 0xceb364 in free_so (/home/jkratoch/redhat/gdb-clean/gdb/gdb+0xceb364)
#3 0xca59f8 in do_free_so (/home/jkratoch/redhat/gdb-clean/gdb/gdb+0xca59f8)
#4 0x93432a in do_my_cleanups (/home/jkratoch/redhat/gdb-clean/gdb/gdb+0x93432a)
#5 0x934406 in do_cleanups (/home/jkratoch/redhat/gdb-clean/gdb/gdb+0x934406)
#6 0x68efa9 in svr4_read_so_list (/home/jkratoch/redhat/gdb-clean/gdb/gdb+0x68efa9)
I did not notice it during my review in:
Re: [PATCH v2] Skip vDSO when reading SO list (PR 8882)
https://sourceware.org/ml/gdb-patches/2013-09/msg00888.html
gdb/
2014-02-27 Jan Kratochvil <jan.kratochvil@redhat.com>
Additional PR 8882 fix.
* solib-svr4.c (svr4_read_so_list): Change first to first_l_name.
Message-ID: <20140226220918.GA10431@host2.jankratochvil.net>
Just a small optimization. No need to block/unblock signals if we're
not going to call sigsuspend.
gdb/
2014-02-27 Pedro Alves <palves@redhat.com>
* nat/linux-waitpid.c (my_waitpid): Only block signals if WNOHANG
isn't set.
GDBserver currently hangs forever in waitpid if the leader thread
exits before other threads, or if all resumed threads exit - e.g.,
next over a thread exit with sched-locking on. This is exposed by
leader-exit.exp. leader-exit.exp is part of a series of tests for a
set of related problems. See
<http://www.sourceware.org/ml/gdb-patches/2011-10/msg00704.html>:
"
To recap, on the Linux kernel, ptrace/waitpid don't allow reaping the
leader thread until all other threads in the group are reaped. When
the leader exits, it goes zombie, but waitpid will not return an exit
status until the other threads are gone. This is presently exercised
by the gdb.threads/leader-exit.exp test. The fix for that test, in
linux-nat.c:wait_lwp, handles the case where we see the leader gone
when we're stopping all threads to report an event to some other
thread to the core.
(...)
The latter bit about not blocking if there no resumed threads in the
process also applies to some other thread exiting, not just the main
thread. E.g., this test starts a thread, and runs to a breakpoint in
that thread:
...
(gdb) c
Continuing.
[New Thread 0x7ffff75a4700 (LWP 23397)]
[Switching to Thread 0x7ffff75a4700 (LWP 23397)]
Breakpoint 2, thread_a (arg=0x0) at ../../../src/gdb/testsuite/gdb.threads/no-unwaited-for-left.c:28
28 return 0; /* break-here */
(gdb) info threads
* 2 Thread 0x7ffff75a4700 (LWP 23397) thread_a (arg=0x0) at ../../../src/gdb/testsuite/gdb.threads/no-unwaited-for-left.c:28
1 Thread 0x7ffff7fcb720 (LWP 23391) 0x00007ffff7bc606d in pthread_join (threadid=140737343276800, thread_return=0x0) at pthread_join.c:89
The thread will exit as soon as we resume it. But if we only resume
that thread, leaving the rest of the threads stopped:
(gdb) set scheduler-locking on
(gdb) c
Continuing.
^C^C^C^C^C^C^C^C
"
This patch fixes the issues by implementing TARGET_WAITKIND_NO_RESUMED
on GDBserver, similarly to what the patch above did for native
Linux GDB.
gdb.threads/leader-exit.exp now passes.
gdb.threads/no-unwaited-for-left.exp now at least errors out instead
of hanging:
continue
Continuing.
warning: Remote failure reply: E.No unwaited-for children left.
[Thread 15454] #1 stopped.
0x00000034cf408e60 in pthread_join (threadid=140737353922368, thread_return=0x0) at pthread_join.c:93
93 lll_wait_tid (pd->tid);
(gdb) FAIL: gdb.threads/no-unwaited-for-left.exp: continue stops when the main thread exits
The gdb.threads/non-ldr-exc-*.exp tests are skipped because GDBserver
unfortunately doesn't support fork/exec yet, but I'm confident this
fixes the related issues.
I'm leaving modeling TARGET_WAITKIND_NO_RESUMED in the RSP for a
separate pass.
(BTW, in case of error in response to a vCont, it would be better for
GDB to query the target for the current thread, or re-select one,
instead of assuming current inferior_ptid is still the selected
thread.)
This implementation is a little different from GDB's, because I'm
avoiding bringing in more of this broken use of waitpid(PID) into
GDBserver. Specifically, this avoids waitpid(PID) when stopping all
threads. There's really no need for wait_for_sigstop to wait for each
LWP in turn. Instead, with some refactoring, we make it reuse
linux_wait_for_event.
gdb/gdbserver/
2014-02-27 Pedro Alves <palves@redhat.com>
PR 12702
* inferiors.h (A_I_NEXT, ALL_INFERIORS_TYPE, ALL_PROCESSES): New
macros.
* linux-low.c (delete_lwp, handle_extended_wait): Add debug
output.
(last_thread_of_process_p): Take a PID argument instead of a
thread pointer.
(linux_wait_for_lwp): Delete.
(num_lwps, check_zombie_leaders, not_stopped_callback): New
functions.
(linux_low_filter_event): New function, party factored out from
linux_wait_for_event.
(linux_wait_for_event): Rename to ...
(linux_wait_for_event_filtered): ... this. Add new filter ptid
argument. Partly rewrite. Always use waitpid(-1, WNOHANG) and
sigsuspend. Check for zombie leaders.
(linux_wait_for_event): Reimplement as wrapper around
linux_wait_for_event_filtered.
(linux_wait_1): Handle TARGET_WAITKIND_NO_RESUMED. Assume that if
a normal or signal exit is seen, it's the whole process exiting.
(wait_for_sigstop): No longer a for_each_inferior callback.
Rewrite on top of linux_wait_for_event_filtered.
(stop_all_lwps): Call wait_for_sigstop directly.
* server.c (resume, handle_target_event): Handle
TARGET_WAITKIND_NO_RESUMED.
So that gdbserver's Linux backend can use it too.
gdb/
2014-02-27 Pedro Alves <palves@redhat.com>
PR 12702
* linux-nat.c (status_to_str): Moved to nat/linux-waitpid.c.
* nat/linux-waitpid.c: Include string.h.
(status_to_str): Moved here and made extern.
* nat/linux-waitpid.h (status_to_str): New declaration.
Necessary to fix parsing auxv entries from core files on systems that use
the layout specified by ELF instead of the incompatible variant used by Linux.
gdb/Changelog:
* gdbarch.sh (auxv_parse): New.
* gdbarch.h: Regenerated.
* gdbarch.c: Regenerated.
* auxv.c (target_auxv_parse): Call gdbarch_parse_auxv if provided.
This is the GDBserver counterpart of a change we recently made in
GDB to only rely on get_image_name to determine its name.
This simplification, in turn, allows us to remove a fair amount of
functions and globals which now become unused.
gdb/gdbserver/ChangeLog:
* win32-low.c (psapi_get_dll_name,
* win32_CreateToolhelp32Snapshot): Delete.
(win32_CreateToolhelp32Snapshot, win32_Module32First)
(win32_Module32Next, load_toolhelp, toolhelp_get_dll_name):
Delete.
(handle_load_dll): Add function description.
Remove code using psapi_get_dll_name and toolhelp_get_dll_name.
This patch is a small cleanup that moves the magic 0x1000 offset
to apply to a DLL's base address inside the win32_add_one_solib
function, rather than delegate that reponsibility to its callers.
gdb/gdbserver/ChangeLog:
* win32-low.c (win32_add_one_solib): Add 0x1000 to load_addr.
Add comment.
(win32_add_all_dlls): Remove 0x1000 offset applied to DLL
base address when calling win32_add_one_solib.
(handle_load_dll): Delete local variable load_addr.
Remove 0x1000 offset applied to DLL base address when calling
win32_add_one_solib.
(handle_unload_dll): Add comment.
This GDBserver patch mirrors a change made in GDB wich aims at
simplifying DLL handling during the inferior initialization
(process creation during the "run", or during an "attach").
Instead of processing each DLL load event, which is sometimes
incomplete, we ignore these events until the inferior has completed
its startup phase, and then just iterate over all DLLs via
EnumProcessModules.
As a side-effect, it fixes a small bug where win32_ensure_ntdll_loaded
was missing a 0x1000 offset in the DLL base address. This problem
should only be visible on the 64bit version of Windows 8.1, since
this is the only platform where win32_ensure_ntdll_loaded is actually
needed.
gdb/gdbserver/ChangeLog:
* win32-low.c (win32_add_all_dlls): Renames
win32_ensure_ntdll_loaded. Rewrite function documentation.
Adjust implementation to always load all DLLs.
Add 0x1000 offset to DLL base address when calling
win32_add_one_solib.
(child_initialization_done): New static global.
(do_initial_child_stuff): Set child_initialization_done to
zero during child initialization, and 1 after. Replace call
to win32_ensure_ntdll_loaded by call to win32_add_all_dlls.
Add comment.
(match_dll_by_basename, dll_is_loaded_by_basename): Delete.
(handle_unload_dll): Add function documentation.
(get_child_debug_event): Ignore load and unload DLL events
during child initialization.
Starting with DWARF version 4, the description of the DW_AT_high_pc
attribute was amended to say:
if it is of class constant, the value is an unsigned integer offset
which when added to the low PC gives the address of the first
location past the last instruction associated with the entity.
A change was made in Apr 27th, 2012 to reflect that change:
| commit 91da14142c
| Author: Mark Wielaard <mjw@redhat.com>
| Date: Fri Apr 27 18:55:19 2012 +0000
|
| * dwarf2read.c (dwarf2_get_pc_bounds): Check DW_AT_high_pc form to
| see whether it is an address or a constant offset from DW_AT_low_pc.
| (dwarf2_record_block_ranges): Likewise.
| (read_partial_die): Likewise.
Unfortunately, this new interpretation is now used regardless of
the CU's DWARF version. It turns out that one of WindRiver's compilers
(FTR: Diabdata 4.4) is generating DWARF version 2 info with
DW_AT_high_pc attributes improperly using the data4 form. Because of
that, we miscompute all high PCs incorrectly. This leads to a lot of
symtabs having overlapping ranges, which in turn causes havoc in
pc-to-symtab-and-line translations.
One visible effect is when inserting a breakpoint on a given function:
(gdb) b world
Breakpoint 1 at 0x4005c4
The source location of the breakpoint is missing. The output should be:
(gdb) b world
Breakpoint 1 at 0x4005c8: file dw2-rel-hi-pc-world.c, line 24.
What happens in this case is that the pc-to-SAL translation first
starts be trying to find the symtab associated to our PC using
each symtab's ranges. Because of the high_pc miscomputation,
many symtabs end up matching, and the heuristic trying to select
the most probable one unfortunately returns one that is unrelated
(it really had no change in this case to do any better). Once we
have the wrong symtab, the start searching the associated linetable,
where the addresses are correct, thus finding no match, and therefore
no SAL.
This patch is an attempt at handling the situation as gracefully
as we can, without guarantees. It introduces a new function
"attr_value_as_address" which uses the correct accessor for getting
the value of a given attribute. It then adjust the code throughout
this unit to use this function instead of assuming that addresses always
have the DW_FORM_addr format.
It also fixes the original issue of miscomputing the high_pc
by limiting the new interpretation of constant form DW_AT_high_pc
attributes to units using DWARF version 4 or later.
gdb/ChangeLog:
* dwarf2read.c (attr_value_as_address): New function.
(dwarf2_find_base_address, read_call_site_scope): Use
attr_value_as_address in place of DW_ADDR.
(dwarf2_get_pc_bounds): Use attr_value_as_address to get
the low and high addresses. Slight rework of the handling
of the high pc being a constant form, and limit it to
DWARF verson 4 or higher.
(dwarf2_record_block_ranges): Likewise.
(read_partial_die): Likewise.
(new_symbol_full): Use attr_value_as_address in place of DW_ADDR.
gdb/testsuite/ChangeLog:
* gdb.dwarf2/dw2-abs-hi-pc-hello-dbg.S: New file.
* gdb.dwarf2/dw2-abs-hi-pc-hello.c: New file.
* gdb.dwarf2/dw2-abs-hi-pc-world-dbg.S: New file.
* gdb.dwarf2/dw2-abs-hi-pc-world.c: New file.
* gdb.dwarf2/dw2-abs-hi-pc.c: New file.
* gdb.dwarf2/dw2-abs-hi-pc.exp: New file.
Tested on x86_64-linux.
Right now the "file" command will discard the exec_bfd and then
possibly open a new one.
If this ends up reopening the same file, it can cause needless work by
gdb -- destroying all the per-BFD data just to re-read it again.
This patch changes the code to hold a reference to the old exec_bfd
while opening the new one.
The possible downside of this is a higher peak memory use.
2014-02-26 Tom Tromey <tromey@redhat.com>
* exec.c (exec_file_attach): Hold a reference to exec_bfd.
If minimal symbols have already been read into a per-BFD object, then
a symbol reader can skip re-reading them. This changes the ELF reader
to do so.
We only skip the work if the file is ELF+DWARF. If it has stabs or
mdebug sections, then I think extra information is computed during the
minsym creation pass; and so we must still repeat it. Eventually even
this will go away, once all symbol types have switched to being
progspace-independent. In the meantime this has no negative effect --
it is just a missing optimization for a small set of users.
This change also required a somewhat non-obvious change to the OBJSTAT
accounting code. If a symbol reader skips re-reading minimal symbols,
then the corresponding OBJSTAT will not be updated. This leads to a
test failure in gdb.base/maint.exp.
To fix this, I've moved the needed stat field out of objfile and into
the per-BFD object.
2014-02-26 Tom Tromey <tromey@redhat.com>
* elfread.c (elf_read_minimal_symbols): Return early if
minimal symbols have already been read. Add "ei" parameter.
(elf_symfile_read): Call elf_read_minimal_symbols earlier.
* minsyms.c (prim_record_minimal_symbol_full): Update.
* objfiles.h (struct objstats) <n_minsyms>: Move...
(struct objfile_per_bfd_storage) <n_minsyms>: ... here.
* symmisc.c (print_objfile_statistics): Update.
This is just a simple refactoring in elfread.c to split out the
minsym-reading code into its own function.
2014-02-26 Tom Tromey <tromey@redhat.com>
* elfread.c (elf_read_minimal_symbols): New function, from
elf_symfile_read.
(elf_symfile_read): Call it.
Now that minimal symbols are independent of the program space, we can
move them to the per-BFD object. This lets us save memory in the
multi-inferior case; and, once the symbol readers are updated, time.
The other prerequisite for this move is that all the objects referred
to by the minimal symbols have a lifetime at least as long as the
per-BFD object. I think this is satisfied partially by this patch
(moving the copied names there) and partially by earlier patches
moving the demangled name hash.
This patch contains a bit of logic to avoid creating new minimal
symbols if they have already been read for a given BFD. This allows
us to avoid trying to update all the symbol readers for this
condition. At first glance this may seem like a hack, but some symbol
readers mix psym and minsym reading, and would require logic just like
this regardless -- and it is simpler and less error-prone to just do
the work in a central spot.
2014-02-26 Tom Tromey <tromey@redhat.com>
* minsyms.c (lookup_minimal_symbol, iterate_over_minimal_symbols)
(lookup_minimal_symbol_text, lookup_minimal_symbol_by_pc_name)
(lookup_minimal_symbol_solib_trampoline)
(lookup_minimal_symbol_by_pc_section_1)
(lookup_minimal_symbol_and_objfile): Update.
(prim_record_minimal_symbol_full): Use the per-BFD obstack.
Don't allocate a minimal symbol if minsyms have already been read.
(build_minimal_symbol_hash_tables): Update.
(install_minimal_symbols): Do nothing if minsyms already read.
Use the per-BFD obstack.
(terminate_minimal_symbol_table): Use the per-BFD obstack.
* objfiles.c (allocate_objfile): Call
terminate_minimal_symbol_table later.
(have_minimal_symbols): Update.
* objfiles.h (struct objfile_per_bfd_storage) <msymbols,
minimal_symbol_count, msymbol_hash, msymbol_demangled_hash>:
Move from struct objfile.
<minsyms_read>: New field.
(struct objfile) <msymbols, minimal_symbol_count,
msymbol_hash, msymbol_demangled_hash>: Move.
(ALL_OBJFILE_MSYMBOLS): Update.
* symfile.c (read_symbols): Set minsyms_read.
(reread_symbols): Update.
* symmisc.c (dump_objfile, dump_msymbols): Update.
Doug Evans
Joel Brobecker
Pedro Alves
Markus Metzger
Yao Qi
Mike Frysinger
Tom Tromey
Doug Evans
Mark Kettenis
Siva Chandra
Jan Kratochvil
Hui Zhu
Ludovic Courtès