[PATCH v3 3/3] btrace, vdso: add vdso target sections

[PATCH v3 3/3] btrace, vdso: add vdso target sections

[Markus Metzger]

When loading symbols for the vdso, also add its sections to target_sections.

This fixes an issue with record btrace where vdso instructions could not be
disassembled during replay.

2014-05-21  Markus Metzger  <markus.t.metzger@intel.com>

	* symfile-mem.c (symbol_file_add_from_memory): Add BFD sections.

testsuite/
	* gdb.btrace/vdso.c: New.
	* gdb.btrace/vdso.exp: New.


---
 gdb/symfile-mem.c                 | 18 +++++++++++++
 gdb/testsuite/gdb.btrace/vdso.c   | 30 ++++++++++++++++++++++
 gdb/testsuite/gdb.btrace/vdso.exp | 53 +++++++++++++++++++++++++++++++++++++++
 3 files changed, 101 insertions(+)
 create mode 100644 gdb/testsuite/gdb.btrace/vdso.c
 create mode 100644 gdb/testsuite/gdb.btrace/vdso.exp

diff --git a/gdb/symfile-mem.c b/gdb/symfile-mem.c
index 3f09c4d..b29421e 100644
--- a/gdb/symfile-mem.c
+++ b/gdb/symfile-mem.c
@@ -92,6 +92,7 @@ symbol_file_add_from_memory (struct bfd *templ, CORE_ADDR addr,
   struct section_addr_info *sai;
   unsigned int i;
   struct cleanup *cleanup;
+  struct target_section *sections, *sections_end, *tsec;
 
   if (bfd_get_flavour (templ) != bfd_target_elf_flavour)
     error (_("add-symbol-file-from-memory not supported for this target"));
@@ -131,6 +132,23 @@ symbol_file_add_from_memory (struct bfd *templ, CORE_ADDR addr,
 				   from_tty ? SYMFILE_VERBOSE : 0,
                                    sai, OBJF_SHARED, NULL);
 
+  sections = NULL;
+  sections_end = NULL;
+
+  if (build_section_table (nbfd, &sections, &sections_end) == 0)
+    {
+      make_cleanup (xfree, sections);
+
+      /* Adjust the target section addresses by the load address.  */
+      for (tsec = sections; tsec != sections_end; ++tsec)
+	{
+	  tsec->addr += loadbase;
+	  tsec->endaddr += loadbase;
+	}
+
+      add_target_sections (&nbfd, sections, sections_end);
+    }
+
   /* This might change our ideas about frames already looked at.  */
   reinit_frame_cache ();
 
diff --git a/gdb/testsuite/gdb.btrace/vdso.c b/gdb/testsuite/gdb.btrace/vdso.c
new file mode 100644
index 0000000..e16a3c6
--- /dev/null
+++ b/gdb/testsuite/gdb.btrace/vdso.c
@@ -0,0 +1,30 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+   Copyright 2014 Free Software Foundation, Inc.
+
+   Contributed by Intel Corp. <markus.t.metzger@intel.com>
+
+   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/>.  */
+
+#include <sys/time.h>
+
+int
+main (void)
+{
+  struct timeval tv;
+
+  gettimeofday (&tv, 0); /* main.1.  */
+
+  return 0; /* main.2.  */
+}
diff --git a/gdb/testsuite/gdb.btrace/vdso.exp b/gdb/testsuite/gdb.btrace/vdso.exp
new file mode 100644
index 0000000..34a5bf0
--- /dev/null
+++ b/gdb/testsuite/gdb.btrace/vdso.exp
@@ -0,0 +1,53 @@
+# This testcase is part of GDB, the GNU debugger.
+#
+# Copyright 2014 Free Software Foundation, Inc.
+#
+# Contributed by Intel Corp. <markus.t.metzger@intel.com>
+#
+# 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 we can access the vdso memory during replay for stepping.
+
+# check for btrace support
+if { [skip_btrace_tests] } { return -1 }
+
+# start inferior
+standard_testfile
+if [prepare_for_testing $testfile.exp $testfile $srcfile] {
+    return -1
+}
+if ![runto_main] {
+    return -1
+}
+
+# capture the disassembly of gettimeofday while live debugging
+set live_gettimeofday [capture_command_output "disassemble gettimeofday" ""]
+
+# trace the test code
+gdb_test_no_output "record btrace"
+gdb_test "next" "main\.2.*"
+
+with_test_prefix "replay" {
+	# capture the disassembly of gettimeofday while replaying
+	gdb_test "record goto begin" "main\.1.*"
+	set replay_gettimeofday [capture_command_output "disassemble gettimeofday" ""]
+
+	# the two disassemblys must be identical
+	if ![string compare $live_gettimeofday $replay_gettimeofday]  {
+		pass "disassemble gettimeofday"
+	} else {
+		fail "disassemble gettimeofday"
+	}
+}
-- 
1.8.3.1

[PATCH v3 1/3] btrace: control memory access during replay

[Markus Metzger]

The btrace record target does not trace data.  We therefore do not allow
accessing read-write memory during replay.

In some cases, this might be useful to advanced users, though, who we assume
to know what they are doing.

Add a set|show command pair to turn this memory access restriction off.

CC: Eli Zaretskii <eliz@gnu.org>

2014-05-21  Markus Metzger  <markus.t.metzger@intel.com>

	* record-btrace.c (record_btrace_allow_memory_access): Remove.
	(replay_memory_access_read_only, replay_memory_access_read_write)
	(replay_memory_access_types, replay_memory_access)
	(set_record_btrace_cmdlist, show_record_btrace_cmdlist)
	(cmd_set_record_btrace, cmd_show_record_btrace)
	(cmd_show_replay_memory_access): New.
	(record_btrace_xfer_partial, record_btrace_insert_breakpoint)
	(record_btrace_remove_breakpoint): Replace
	record_btrace_allow_memory_access with replay_memory_access.
	(_initialize_record_btrace): Add commands.
	* NEWS: Announce it.

testsuite/
	* gdb.btrace/data.exp: Test it.

doc/
	* gdb.texinfo (Process Record and Replay): Document it.


---
 gdb/NEWS                          |  4 ++
 gdb/doc/gdb.texinfo               | 22 ++++++++++
 gdb/record-btrace.c               | 88 ++++++++++++++++++++++++++++++++++-----
 gdb/testsuite/gdb.btrace/data.exp | 10 ++++-
 4 files changed, 112 insertions(+), 12 deletions(-)

diff --git a/gdb/NEWS b/gdb/NEWS
index b23c8a0..62e0adc 100644
--- a/gdb/NEWS
+++ b/gdb/NEWS
@@ -61,6 +61,10 @@ maint ada show ignore-descriptive-types
   the user manual for more details on descriptive types and the intended
   usage of this option.
 
+set record btrace replay-memory-access (read-only|read-write)
+show record btrace replay-memory-access
+  Control what memory accesses are allowed during replay.
+
 * New features in the GDB remote stub, GDBserver
 
   ** New option --debug-format=option1[,option2,...] allows one to add
diff --git a/gdb/doc/gdb.texinfo b/gdb/doc/gdb.texinfo
index 6092ff4..b9e7964 100644
--- a/gdb/doc/gdb.texinfo
+++ b/gdb/doc/gdb.texinfo
@@ -6400,6 +6400,28 @@ results.
 @item show record full memory-query
 Show the current setting of @code{memory-query}.
 
+@kindex set record btrace
+The @code{btrace} record target does not trace data.  As a
+convenience, when replaying, @value{GDBN} reads read-only memory off
+the live program directly, assuming that the addresses of the
+read-only areas don't change.  This for example makes it possible to
+disassemble code while replaying, but not to print variables.
+In some cases, being able to inspect variables might be useful.
+You can use the following command for that:
+
+@item set record btrace replay-memory-access
+Control the behavior of the @code{btrace} recording method when
+accessing memory during replay.  If @code{read-only} (the default),
+@value{GDBN} will only allow accesses to read-only memory.
+If @code{read-write}, @value{GDBN} will allow accesses to read-only
+and to read-write memory.  Beware that the accessed memory corresponds
+to the live target and not necessarily to the current replay
+position.
+
+@kindex show record btrace
+@item show record btrace replay-memory-access
+Show the current setting of @code{replay-memory-access}.
+
 @kindex info record
 @item info record
 Show various statistics about the recording depending on the recording
diff --git a/gdb/record-btrace.c b/gdb/record-btrace.c
index d768225..f43436d 100644
--- a/gdb/record-btrace.c
+++ b/gdb/record-btrace.c
@@ -42,8 +42,22 @@ static struct target_ops record_btrace_ops;
 /* A new thread observer enabling branch tracing for the new thread.  */
 static struct observer *record_btrace_thread_observer;
 
-/* Temporarily allow memory accesses.  */
-static int record_btrace_allow_memory_access;
+/* Memory access types used in set/show record btrace replay-memory-access.  */
+static const char replay_memory_access_read_only[] = "read-only";
+static const char replay_memory_access_read_write[] = "read-write";
+static const char *const replay_memory_access_types[] =
+{
+  replay_memory_access_read_only,
+  replay_memory_access_read_write,
+  NULL
+};
+
+/* The currently allowed replay memory access type.  */
+static const char *replay_memory_access = replay_memory_access_read_only;
+
+/* Command lists for "set/show record btrace".  */
+static struct cmd_list_element *set_record_btrace_cmdlist;
+static struct cmd_list_element *show_record_btrace_cmdlist;
 
 /* Print a record-btrace debug message.  Use do ... while (0) to avoid
    ambiguities when used in if statements.  */
@@ -815,7 +829,8 @@ record_btrace_xfer_partial (struct target_ops *ops, enum target_object object,
   struct target_ops *t;
 
   /* Filter out requests that don't make sense during replay.  */
-  if (!record_btrace_allow_memory_access && record_btrace_is_replaying (ops))
+  if (replay_memory_access == replay_memory_access_read_only
+      && record_btrace_is_replaying (ops))
     {
       switch (object)
 	{
@@ -869,18 +884,19 @@ record_btrace_insert_breakpoint (struct target_ops *ops,
 				 struct bp_target_info *bp_tgt)
 {
   volatile struct gdb_exception except;
-  int old, ret;
+  const char *old;
+  int ret;
 
   /* Inserting breakpoints requires accessing memory.  Allow it for the
      duration of this function.  */
-  old = record_btrace_allow_memory_access;
-  record_btrace_allow_memory_access = 1;
+  old = replay_memory_access;
+  replay_memory_access = replay_memory_access_read_write;
 
   ret = 0;
   TRY_CATCH (except, RETURN_MASK_ALL)
     ret = ops->beneath->to_insert_breakpoint (ops->beneath, gdbarch, bp_tgt);
 
-  record_btrace_allow_memory_access = old;
+  replay_memory_access = old;
 
   if (except.reason < 0)
     throw_exception (except);
@@ -896,18 +912,19 @@ record_btrace_remove_breakpoint (struct target_ops *ops,
 				 struct bp_target_info *bp_tgt)
 {
   volatile struct gdb_exception except;
-  int old, ret;
+  const char *old;
+  int ret;
 
   /* Removing breakpoints requires accessing memory.  Allow it for the
      duration of this function.  */
-  old = record_btrace_allow_memory_access;
-  record_btrace_allow_memory_access = 1;
+  old = replay_memory_access;
+  replay_memory_access = replay_memory_access_read_write;
 
   ret = 0;
   TRY_CATCH (except, RETURN_MASK_ALL)
     ret = ops->beneath->to_remove_breakpoint (ops->beneath, gdbarch, bp_tgt);
 
-  record_btrace_allow_memory_access = old;
+  replay_memory_access = old;
 
   if (except.reason < 0)
     throw_exception (except);
@@ -1938,6 +1955,32 @@ cmd_record_btrace_start (char *args, int from_tty)
   execute_command ("target record-btrace", from_tty);
 }
 
+/* The "set record btrace" command.  */
+
+static void
+cmd_set_record_btrace (char *args, int from_tty)
+{
+  cmd_show_list (set_record_btrace_cmdlist, from_tty, "");
+}
+
+/* The "show record btrace" command.  */
+
+static void
+cmd_show_record_btrace (char *args, int from_tty)
+{
+  cmd_show_list (show_record_btrace_cmdlist, from_tty, "");
+}
+
+/* The "show record btrace replay-memory-access" command.  */
+
+static void
+cmd_show_replay_memory_access (struct ui_file *file, int from_tty,
+			       struct cmd_list_element *c, const char *value)
+{
+  fprintf_filtered (gdb_stdout, _("Replay memory access is %s.\n"),
+		    replay_memory_access);
+}
+
 void _initialize_record_btrace (void);
 
 /* Initialize btrace commands.  */
@@ -1950,6 +1993,29 @@ _initialize_record_btrace (void)
 	   &record_cmdlist);
   add_alias_cmd ("b", "btrace", class_obscure, 1, &record_cmdlist);
 
+  add_prefix_cmd ("btrace", class_support, cmd_set_record_btrace,
+		  _("Set record options"), &set_record_btrace_cmdlist,
+		  "set record btrace ", 0, &set_record_cmdlist);
+
+  add_prefix_cmd ("btrace", class_support, cmd_show_record_btrace,
+		  _("Show record options"), &show_record_btrace_cmdlist,
+		  "show record btrace ", 0, &show_record_cmdlist);
+
+  add_setshow_enum_cmd ("replay-memory-access", no_class,
+			replay_memory_access_types, &replay_memory_access, _("\
+Set what memory accesses are allowed during replay."), _("\
+Show what memory accesses are allowed during replay."),
+			   _("Default is READ-ONLY.\n\n\
+The btrace record target does not trace data.\n\
+The memory therefore corresponds to the live target and not \
+to the current replay position.\n\n\
+When READ-ONLY, allow accesses to read-only memory during replay.\n\
+When READ-WRITE, allow accesses to read-only and read-write memory during \
+replay."),
+			   NULL, cmd_show_replay_memory_access,
+			   &set_record_btrace_cmdlist,
+			   &show_record_btrace_cmdlist);
+
   init_record_btrace_ops ();
   add_target (&record_btrace_ops);
 
diff --git a/gdb/testsuite/gdb.btrace/data.exp b/gdb/testsuite/gdb.btrace/data.exp
index 64c5443..91d8d8b 100644
--- a/gdb/testsuite/gdb.btrace/data.exp
+++ b/gdb/testsuite/gdb.btrace/data.exp
@@ -40,6 +40,14 @@ gdb_test "reverse-step" ".*test\.4.*"
 gdb_test "print glob" "unavailable\[^\\\r\\\n\]*"
 gdb_test "print loc" "unavailable\[^\\\r\\\n\]*"
 
+# we can read memory if we explicitly allow it.
+gdb_test_no_output "set record btrace replay-memory-access read-write"
+gdb_test "print glob" "1"
+
+# we can't if we don't explicitly allow it.
+gdb_test_no_output "set record btrace replay-memory-access read-only"
+gdb_test "print glob" "unavailable\[^\\\r\\\n\]*"
+
 # stop replaying and try again
-gdb_test "record goto end"
+gdb_test "record goto end" ".*main\.3.*"
 gdb_test "print glob" "1"
-- 
1.8.3.1

[PATCH v3 2/3] test, gcore: move capture_command_output into lib/gdb.exp

[Markus Metzger]

Allow gcore's capture_command_output function to be used by other tests.

2014-05-21  Markus Metzger  <markus.t.metzger@intel.com>

testsuite/
	* gdb.base/gcore.exp (capture_command_output): Move ...
	* lib/gdb.exp (capture_command_output): ... here.


---
 gdb/testsuite/gdb.base/gcore.exp | 13 -------------
 gdb/testsuite/lib/gdb.exp        | 14 ++++++++++++++
 2 files changed, 14 insertions(+), 13 deletions(-)

diff --git a/gdb/testsuite/gdb.base/gcore.exp b/gdb/testsuite/gdb.base/gcore.exp
index 99743a0..c28a9b3 100644
--- a/gdb/testsuite/gdb.base/gcore.exp
+++ b/gdb/testsuite/gdb.base/gcore.exp
@@ -41,19 +41,6 @@ if { ! [ runto_main ] } then {
     return -1
 }
 
-proc capture_command_output { command prefix } {
-    global gdb_prompt
-    global expect_out
-
-    set output_string ""
-    gdb_test_multiple "$command" "capture_command_output for $command" {
-	-re "${command}\[\r\n\]+${prefix}(.*)\[\r\n\]+$gdb_prompt $" {
-	    set output_string $expect_out(1,string)
-	}
-    }
-    return $output_string
-}
-
 gdb_test "break terminal_func" "Breakpoint .* at .*${srcfile}, line .*" \
 	"set breakpoint at terminal_func"
 
diff --git a/gdb/testsuite/lib/gdb.exp b/gdb/testsuite/lib/gdb.exp
index 0fe6c31..22a43b6 100644
--- a/gdb/testsuite/lib/gdb.exp
+++ b/gdb/testsuite/lib/gdb.exp
@@ -4772,5 +4772,19 @@ proc parse_args { argset } {
     # number of items expected to be passed into the procedure...
 }
 
+# Capture the output of COMMAND in a string ignoring PREFIX; return that string.
+proc capture_command_output { command prefix } {
+    global gdb_prompt
+    global expect_out
+
+    set output_string ""
+    gdb_test_multiple "$command" "capture_command_output for $command" {
+	-re "${command}\[\r\n\]+${prefix}(.*)\[\r\n\]+$gdb_prompt $" {
+	    set output_string $expect_out(1,string)
+	}
+    }
+    return $output_string
+}
+
 # Always load compatibility stuff.
 load_lib future.exp
-- 
1.8.3.1

Re: [PATCH v3 1/3] btrace: control memory access during replay

[Pedro Alves]

On 05/21/2014 08:01 AM, Markus Metzger wrote:
> The btrace record target does not trace data.  We therefore do not allow
> accessing read-write memory during replay.
> 
> In some cases, this might be useful to advanced users, though, who we assume
> to know what they are doing.
> 
> Add a set|show command pair to turn this memory access restriction off.
> 

Code bits are OK.

-- 
Pedro Alves

Re: [PATCH v3 2/3] test, gcore: move capture_command_output into lib/gdb.exp

[Pedro Alves]

On 05/21/2014 08:01 AM, Markus Metzger wrote:
> Allow gcore's capture_command_output function to be used by other tests.
> 
> 2014-05-21  Markus Metzger  <markus.t.metzger@intel.com>
> 
> testsuite/
> 	* gdb.base/gcore.exp (capture_command_output): Move ...
> 	* lib/gdb.exp (capture_command_output): ... here.

OK.

-- 
Pedro Alves

Re: [PATCH v3 3/3] btrace, vdso: add vdso target sections

[Pedro Alves]

On 05/21/2014 08:01 AM, Markus Metzger wrote:

> 2014-05-21  Markus Metzger  <markus.t.metzger@intel.com>
> 
> 	* symfile-mem.c (symbol_file_add_from_memory): Add BFD sections.
> 
> testsuite/
> 	* gdb.btrace/vdso.c: New.
> 	* gdb.btrace/vdso.exp: New.

OK.

> +	# the two disassemblys must be identical

"disassemblies", I think.

-- 
Pedro Alves

Re: [PATCH v3 3/3] btrace, vdso: add vdso target sections

[Pedro Alves]

Hi Markus,

On 05/21/2014 08:01 AM, Markus Metzger wrote:
> When loading symbols for the vdso, also add its sections to target_sections.
> 
> This fixes an issue with record btrace where vdso instructions could not be
> disassembled during replay.
> 
> 2014-05-21  Markus Metzger  <markus.t.metzger@intel.com>
> 
> 	* symfile-mem.c (symbol_file_add_from_memory): Add BFD sections.

Unfortunately this caused regressions in break-interp.exp:

 Running ../../../src/gdb/testsuite/gdb.base/break-interp.exp ...
 ERROR: Process no longer exists

That is GDB crashing.

589fdceb99db9868e34bc976b7e234dbfaf327e2 is the first bad commit
commit 589fdceb99db9868e34bc976b7e234dbfaf327e2
Author: Markus Metzger <markus.t.metzger@intel.com>
Date:   Wed Apr 2 09:54:11 2014 +0200

    btrace, vdso: add vdso target sections

    When loading symbols for the vdso, also add its sections to target_sections.

    This fixes an issue with record btrace where vdso instructions could not be
    disassembled during replay.

        * symfile-mem.c (symbol_file_add_from_memory): Add BFD sections.

    testsuite/
        * gdb.btrace/vdso.c: New.
        * gdb.btrace/vdso.exp: New.


Core was generated by `/home/pedro/gdb/mygit/build/gdb/testsuite/../../gdb/gdb -nw -nx -data-directory'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0  0x0000000000647bce in exec_set_section_address (filename=0x27bb1f0 "/home/pedro/gdb/mygit/build/gdb/testsuite/gdb.base/break-interp-LDprelinkNOdebugNO", index=0,
    address=139930618982856) at ../../src/gdb/exec.c:919
919           if (filename_cmp (filename, p->the_bfd_section->owner->filename) == 0
Setting up the environment for debugging gdb.
Breakpoint 1 at 0x7127ee: file ../../src/gdb/utils.c, line 806.
Breakpoint 2 at 0x4fca7c: file ../../src/gdb/cli/cli-cmds.c, line 219.
(top-gdb) bt
#0  0x0000000000647bce in exec_set_section_address (filename=0x27bb1f0 "/home/pedro/gdb/mygit/build/gdb/testsuite/gdb.base/break-interp-LDprelinkNOdebugNO", index=0,
    address=139930618982856) at ../../src/gdb/exec.c:919
#1  0x000000000064b61d in objfile_relocate1 (objfile=0x27baf70, new_offsets=0x7fff536105a0) at ../../src/gdb/objfiles.c:819
#2  0x000000000064b673 in objfile_relocate (objfile=0x27baf70, new_offsets=0x7fff536105a0) at ../../src/gdb/objfiles.c:843
#3  0x000000000049eeb2 in svr4_relocate_main_executable () at ../../src/gdb/solib-svr4.c:2883
#4  0x000000000049ef55 in svr4_solib_create_inferior_hook (from_tty=0) at ../../src/gdb/solib-svr4.c:2927
#5  0x0000000000730b0e in solib_create_inferior_hook (from_tty=0) at ../../src/gdb/solib.c:1211
#6  0x00000000005e5345 in post_create_inferior (target=0xd02c00 <current_target>, from_tty=0) at ../../src/gdb/infcmd.c:443
#7  0x00000000005e5744 in run_command_1 (args=0x0, from_tty=1, tbreak_at_main=0) at ../../src/gdb/infcmd.c:628
#8  0x00000000005e57a0 in run_command (args=0x0, from_tty=1) at ../../src/gdb/infcmd.c:642
#9  0x00000000004f68b6 in do_cfunc (c=0x261b9e0, args=0x0, from_tty=1) at ../../src/gdb/cli/cli-decode.c:107
#10 0x00000000004f99aa in cmd_func (cmd=0x261b9e0, args=0x0, from_tty=1) at ../../src/gdb/cli/cli-decode.c:1886
#11 0x000000000070f7f2 in execute_command (p=0x250f393 "", from_tty=1) at ../../src/gdb/top.c:461
#12 0x0000000000610f0f in command_handler (command=0x250f390 "") at ../../src/gdb/event-top.c:438
#13 0x00000000006114c8 in command_line_handler (rl=0x2801b70 "run") at ../../src/gdb/event-top.c:635
#14 0x000000000077177a in rl_callback_read_char () at ../../src/readline/callback.c:220
#15 0x0000000000610a3d in rl_callback_read_char_wrapper (client_data=0x0) at ../../src/gdb/event-top.c:167
#16 0x0000000000610e21 in stdin_event_handler (error=0, client_data=0x0) at ../../src/gdb/event-top.c:378
#17 0x000000000060fa07 in handle_file_event (data=...) at ../../src/gdb/event-loop.c:768
#18 0x000000000060eee9 in process_event () at ../../src/gdb/event-loop.c:342
#19 0x000000000060ef8b in gdb_do_one_event () at ../../src/gdb/event-loop.c:394
#20 0x000000000060f000 in start_event_loop () at ../../src/gdb/event-loop.c:431
#21 0x0000000000610a6f in cli_command_loop (data=0x0) at ../../src/gdb/event-top.c:182
#22 0x0000000000606f09 in current_interp_command_loop () at ../../src/gdb/interps.c:350
#23 0x00000000006080ad in captured_command_loop (data=0x0) at ../../src/gdb/main.c:301
#24 0x0000000000604223 in catch_errors (func=0x608092 <captured_command_loop>, func_args=0x0, errstring=0x8e7001 "", mask=RETURN_MASK_ALL)
    at ../../src/gdb/exceptions.c:524
#25 0x00000000006094b3 in captured_main (data=0x7fff53610d50) at ../../src/gdb/main.c:1097
#26 0x0000000000604223 in catch_errors (func=0x608345 <captured_main>, func_args=0x7fff53610d50, errstring=0x8e7001 "", mask=RETURN_MASK_ALL)
    at ../../src/gdb/exceptions.c:524
#27 0x00000000006094dc in gdb_main (args=0x7fff53610d50) at ../../src/gdb/main.c:1105
#28 0x000000000045dbd7 in main (argc=5, argv=0x7fff53610e58) at ../../src/gdb/gdb.c:33

-- 
Pedro Alves

Re: [PATCH v3 3/3] btrace, vdso: add vdso target sections

[Pedro Alves]

On 05/29/2014 05:54 PM, Pedro Alves wrote:
> Hi Markus,
> 
> On 05/21/2014 08:01 AM, Markus Metzger wrote:
>> When loading symbols for the vdso, also add its sections to target_sections.
>>
>> This fixes an issue with record btrace where vdso instructions could not be
>> disassembled during replay.
>>
>> 2014-05-21  Markus Metzger  <markus.t.metzger@intel.com>
>>
>> 	* symfile-mem.c (symbol_file_add_from_memory): Add BFD sections.
> 
> Unfortunately this caused regressions in break-interp.exp:
> 
>  Running ../../../src/gdb/testsuite/gdb.base/break-interp.exp ...
>  ERROR: Process no longer exists
> 
> That is GDB crashing.
> 
> 589fdceb99db9868e34bc976b7e234dbfaf327e2 is the first bad commit
> commit 589fdceb99db9868e34bc976b7e234dbfaf327e2
> Author: Markus Metzger <markus.t.metzger@intel.com>
> Date:   Wed Apr 2 09:54:11 2014 +0200
> 
>     btrace, vdso: add vdso target sections
> 
>     When loading symbols for the vdso, also add its sections to target_sections.
> 
>     This fixes an issue with record btrace where vdso instructions could not be
>     disassembled during replay.
> 
>         * symfile-mem.c (symbol_file_add_from_memory): Add BFD sections.
> 
>     testsuite/
>         * gdb.btrace/vdso.c: New.
>         * gdb.btrace/vdso.exp: New.
> 
> 
> Core was generated by `/home/pedro/gdb/mygit/build/gdb/testsuite/../../gdb/gdb -nw -nx -data-directory'.
> Program terminated with signal SIGSEGV, Segmentation fault.
> #0  0x0000000000647bce in exec_set_section_address (filename=0x27bb1f0 "/home/pedro/gdb/mygit/build/gdb/testsuite/gdb.base/break-interp-LDprelinkNOdebugNO", index=0,
>     address=139930618982856) at ../../src/gdb/exec.c:919
> 919           if (filename_cmp (filename, p->the_bfd_section->owner->filename) == 0

'p', a target section, points at garbage.

Hmm, wait, nothing is removing the vdso symbols target sections, like ever,
not even when the process dies, isn't it?  I guess that's it.

-- 
Pedro Alves

RE: [PATCH v3 3/3] btrace, vdso: add vdso target sections

[Metzger, Markus T]

Thanks for reporting this crash.  I'll look into it.  I thought I ran all tests
for each patch in the vdso series.

It seems btrace got broken, as well.  I now get an assertion: "
to_execution_direction must be implemented ...".  The trivial solution
of providing a to_execution_direction method that just returns
execution_direction does not seem to work - or there's another
bug underneath.

Regards,
Markus.

> -----Original Message-----
> From: Pedro Alves [mailto:palves@redhat.com]
> Sent: Thursday, May 29, 2014 7:04 PM
> To: Metzger, Markus T
> Cc: gdb-patches@sourceware.org
> Subject: Re: [PATCH v3 3/3] btrace, vdso: add vdso target sections
> 
> On 05/29/2014 05:54 PM, Pedro Alves wrote:
> > Hi Markus,
> >
> > On 05/21/2014 08:01 AM, Markus Metzger wrote:
> >> When loading symbols for the vdso, also add its sections to
> target_sections.
> >>
> >> This fixes an issue with record btrace where vdso instructions could not be
> >> disassembled during replay.
> >>
> >> 2014-05-21  Markus Metzger  <markus.t.metzger@intel.com>
> >>
> >> 	* symfile-mem.c (symbol_file_add_from_memory): Add BFD
> sections.
> >
> > Unfortunately this caused regressions in break-interp.exp:
> >
> >  Running ../../../src/gdb/testsuite/gdb.base/break-interp.exp ...
> >  ERROR: Process no longer exists
> >
> > That is GDB crashing.
> >
> > 589fdceb99db9868e34bc976b7e234dbfaf327e2 is the first bad commit
> > commit 589fdceb99db9868e34bc976b7e234dbfaf327e2
> > Author: Markus Metzger <markus.t.metzger@intel.com>
> > Date:   Wed Apr 2 09:54:11 2014 +0200
> >
> >     btrace, vdso: add vdso target sections
> >
> >     When loading symbols for the vdso, also add its sections to
> target_sections.
> >
> >     This fixes an issue with record btrace where vdso instructions could not
> be
> >     disassembled during replay.
> >
> >         * symfile-mem.c (symbol_file_add_from_memory): Add BFD sections.
> >
> >     testsuite/
> >         * gdb.btrace/vdso.c: New.
> >         * gdb.btrace/vdso.exp: New.
> >
> >
> > Core was generated by
> `/home/pedro/gdb/mygit/build/gdb/testsuite/../../gdb/gdb -nw -nx -data-
> directory'.
> > Program terminated with signal SIGSEGV, Segmentation fault.
> > #0  0x0000000000647bce in exec_set_section_address
> (filename=0x27bb1f0
> "/home/pedro/gdb/mygit/build/gdb/testsuite/gdb.base/break-interp-
> LDprelinkNOdebugNO", index=0,
> >     address=139930618982856) at ../../src/gdb/exec.c:919
> > 919           if (filename_cmp (filename, p->the_bfd_section->owner-
> >filename) == 0
> 
> 'p', a target section, points at garbage.
> 
> Hmm, wait, nothing is removing the vdso symbols target sections, like ever,
> not even when the process dies, isn't it?  I guess that's it.
> 
> --
> Pedro Alves

Intel GmbH
Dornacher Strasse 1
85622 Feldkirchen/Muenchen, Deutschland
Sitz der Gesellschaft: Feldkirchen bei Muenchen
Geschaeftsfuehrer: Christian Lamprechter, Hannes Schwaderer, Douglas Lusk
Registergericht: Muenchen HRB 47456
Ust.-IdNr./VAT Registration No.: DE129385895
Citibank Frankfurt a.M. (BLZ 502 109 00) 600119052

RE: [PATCH v3 3/3] btrace, vdso: add vdso target sections

[Metzger, Markus T]

> -----Original Message-----
> From: Metzger, Markus T
> Sent: Monday, June 2, 2014 10:21 AM
> To: Pedro Alves
> Cc: gdb-patches@sourceware.org
> Subject: RE: [PATCH v3 3/3] btrace, vdso: add vdso target sections
> 
> Thanks for reporting this crash.  I'll look into it.  I thought I ran all tests
> for each patch in the vdso series.
> 
> It seems btrace got broken, as well.  I now get an assertion: "
> to_execution_direction must be implemented ...".  The trivial solution
> of providing a to_execution_direction method that just returns
> execution_direction does not seem to work - or there's another
> bug underneath.

Looks like btrace got broken when async was made default and btrace
did not support it.  I took the async bits from record-full and now at
least the btrace tests are running again.  I'll send the patch in a separate
email.

Regards,
Markus.


> 
> Regards,
> Markus.
> 
> > -----Original Message-----
> > From: Pedro Alves [mailto:palves@redhat.com]
> > Sent: Thursday, May 29, 2014 7:04 PM
> > To: Metzger, Markus T
> > Cc: gdb-patches@sourceware.org
> > Subject: Re: [PATCH v3 3/3] btrace, vdso: add vdso target sections
> >
> > On 05/29/2014 05:54 PM, Pedro Alves wrote:
> > > Hi Markus,
> > >
> > > On 05/21/2014 08:01 AM, Markus Metzger wrote:
> > >> When loading symbols for the vdso, also add its sections to
> > target_sections.
> > >>
> > >> This fixes an issue with record btrace where vdso instructions could not
> be
> > >> disassembled during replay.
> > >>
> > >> 2014-05-21  Markus Metzger  <markus.t.metzger@intel.com>
> > >>
> > >> 	* symfile-mem.c (symbol_file_add_from_memory): Add BFD
> > sections.
> > >
> > > Unfortunately this caused regressions in break-interp.exp:
> > >
> > >  Running ../../../src/gdb/testsuite/gdb.base/break-interp.exp ...
> > >  ERROR: Process no longer exists
> > >
> > > That is GDB crashing.
> > >
> > > 589fdceb99db9868e34bc976b7e234dbfaf327e2 is the first bad commit
> > > commit 589fdceb99db9868e34bc976b7e234dbfaf327e2
> > > Author: Markus Metzger <markus.t.metzger@intel.com>
> > > Date:   Wed Apr 2 09:54:11 2014 +0200
> > >
> > >     btrace, vdso: add vdso target sections
> > >
> > >     When loading symbols for the vdso, also add its sections to
> > target_sections.
> > >
> > >     This fixes an issue with record btrace where vdso instructions could not
> > be
> > >     disassembled during replay.
> > >
> > >         * symfile-mem.c (symbol_file_add_from_memory): Add BFD
> sections.
> > >
> > >     testsuite/
> > >         * gdb.btrace/vdso.c: New.
> > >         * gdb.btrace/vdso.exp: New.
> > >
> > >
> > > Core was generated by
> > `/home/pedro/gdb/mygit/build/gdb/testsuite/../../gdb/gdb -nw -nx -data-
> > directory'.
> > > Program terminated with signal SIGSEGV, Segmentation fault.
> > > #0  0x0000000000647bce in exec_set_section_address
> > (filename=0x27bb1f0
> > "/home/pedro/gdb/mygit/build/gdb/testsuite/gdb.base/break-interp-
> > LDprelinkNOdebugNO", index=0,
> > >     address=139930618982856) at ../../src/gdb/exec.c:919
> > > 919           if (filename_cmp (filename, p->the_bfd_section->owner-
> > >filename) == 0
> >
> > 'p', a target section, points at garbage.
> >
> > Hmm, wait, nothing is removing the vdso symbols target sections, like ever,
> > not even when the process dies, isn't it?  I guess that's it.
> >
> > --
> > Pedro Alves

Intel GmbH
Dornacher Strasse 1
85622 Feldkirchen/Muenchen, Deutschland
Sitz der Gesellschaft: Feldkirchen bei Muenchen
Geschaeftsfuehrer: Christian Lamprechter, Hannes Schwaderer, Douglas Lusk
Registergericht: Muenchen HRB 47456
Ust.-IdNr./VAT Registration No.: DE129385895
Citibank Frankfurt a.M. (BLZ 502 109 00) 600119052

RE: [PATCH v3 3/3] btrace, vdso: add vdso target sections

[Metzger, Markus T]

> -----Original Message-----
> From: Pedro Alves [mailto:palves@redhat.com]
> Sent: Thursday, May 29, 2014 7:04 PM


> Hmm, wait, nothing is removing the vdso symbols target sections, like ever,
> not even when the process dies, isn't it?  I guess that's it.

Yes, that seems to be it.

Target sections are typically owned by an so_list object and are removed
when the so_list object is removed.

Symfile target sections are owned by an objfile object and are removed
in a free_objfile observer in symfile.c.  That observer only removes target
sections for OBJF_USERLOADED objfiles, though.

The vdso target sections are owned by the vdso BFD and are never removed.

When I change symfile_add_from_memory to create target sections like this:

	add_target_sections_of_objfile (objf);

and further drop the restriction to only remove OBJF_USERLOADED objfiles in
symfile_free_objfile, the test passes.

diff --git a/gdb/symfile-mem.c b/gdb/symfile-mem.c
index b29421e..ef48f7d 100644
--- a/gdb/symfile-mem.c
+++ b/gdb/symfile-mem.c
@@ -92,7 +92,6 @@ symbol_file_add_from_memory (struct bfd *templ, CORE_ADDR addr,
   struct section_addr_info *sai;
   unsigned int i;
   struct cleanup *cleanup;
-  struct target_section *sections, *sections_end, *tsec;
 
   if (bfd_get_flavour (templ) != bfd_target_elf_flavour)
     error (_("add-symbol-file-from-memory not supported for this target"));
@@ -132,22 +131,7 @@ symbol_file_add_from_memory (struct bfd *templ, CORE_ADDR addr,
                                   from_tty ? SYMFILE_VERBOSE : 0,
                                    sai, OBJF_SHARED, NULL);
 
-  sections = NULL;
-  sections_end = NULL;
-
-  if (build_section_table (nbfd, &sections, &sections_end) == 0)
-    {
-      make_cleanup (xfree, sections);
-
-      /* Adjust the target section addresses by the load address.  */
-      for (tsec = sections; tsec != sections_end; ++tsec)
-       {
-         tsec->addr += loadbase;
-         tsec->endaddr += loadbase;
-       }
-
-      add_target_sections (&nbfd, sections, sections_end);
-    }
+  add_target_sections_of_objfile (objf);
 
   /* This might change our ideas about frames already looked at.  */
   reinit_frame_cache ();
diff --git a/gdb/symfile.c b/gdb/symfile.c
index 64a83c6..caa0722 100644
--- a/gdb/symfile.c
+++ b/gdb/symfile.c
@@ -3893,7 +3893,7 @@ static void
 symfile_free_objfile (struct objfile *objfile)
 {
   /* Remove the target sections of user-added objfiles.  */
-  if (objfile != 0 && objfile->flags & OBJF_USERLOADED)
+  if (objfile != NULL)
     remove_target_sections ((void *) objfile);
 }

Alternatively, I could add another OBJF_ flag and another free_objfile observer
to also remove the symfile-from-memory target sections.

Regards,
Markus.

Intel GmbH
Dornacher Strasse 1
85622 Feldkirchen/Muenchen, Deutschland
Sitz der Gesellschaft: Feldkirchen bei Muenchen
Geschaeftsfuehrer: Christian Lamprechter, Hannes Schwaderer, Douglas Lusk
Registergericht: Muenchen HRB 47456
Ust.-IdNr./VAT Registration No.: DE129385895
Citibank Frankfurt a.M. (BLZ 502 109 00) 600119052

RE: [PATCH v3 3/3] btrace, vdso: add vdso target sections

[Metzger, Markus T]

Hi Pedro,

I sent this as a patch:
https://sourceware.org/ml/gdb-patches/2014-06/msg00148.html

There's also another patch series open that fixes an assertion when
generating a core file while replaying:
https://sourceware.org/ml/gdb-patches/2014-05/msg00537.html
https://sourceware.org/ml/gdb-patches/2014-05/msg00539.html
https://sourceware.org/ml/gdb-patches/2014-05/msg00538.html

regards,
markus.

> -----Original Message-----
> From: Metzger, Markus T
> Sent: Monday, June 2, 2014 2:41 PM
> To: Pedro Alves
> Cc: gdb-patches@sourceware.org
> Subject: RE: [PATCH v3 3/3] btrace, vdso: add vdso target sections
> 
> > -----Original Message-----
> > From: Pedro Alves [mailto:palves@redhat.com]
> > Sent: Thursday, May 29, 2014 7:04 PM
> 
> 
> > Hmm, wait, nothing is removing the vdso symbols target sections, like ever,
> > not even when the process dies, isn't it?  I guess that's it.
> 
> Yes, that seems to be it.
> 
> Target sections are typically owned by an so_list object and are removed
> when the so_list object is removed.
> 
> Symfile target sections are owned by an objfile object and are removed
> in a free_objfile observer in symfile.c.  That observer only removes target
> sections for OBJF_USERLOADED objfiles, though.
> 
> The vdso target sections are owned by the vdso BFD and are never removed.
> 
> When I change symfile_add_from_memory to create target sections like
> this:
> 
> 	add_target_sections_of_objfile (objf);
> 
> and further drop the restriction to only remove OBJF_USERLOADED objfiles
> in
> symfile_free_objfile, the test passes.
> 
> diff --git a/gdb/symfile-mem.c b/gdb/symfile-mem.c
> index b29421e..ef48f7d 100644
> --- a/gdb/symfile-mem.c
> +++ b/gdb/symfile-mem.c
> @@ -92,7 +92,6 @@ symbol_file_add_from_memory (struct bfd *templ,
> CORE_ADDR addr,
>    struct section_addr_info *sai;
>    unsigned int i;
>    struct cleanup *cleanup;
> -  struct target_section *sections, *sections_end, *tsec;
> 
>    if (bfd_get_flavour (templ) != bfd_target_elf_flavour)
>      error (_("add-symbol-file-from-memory not supported for this target"));
> @@ -132,22 +131,7 @@ symbol_file_add_from_memory (struct bfd *templ,
> CORE_ADDR addr,
>                                    from_tty ? SYMFILE_VERBOSE : 0,
>                                     sai, OBJF_SHARED, NULL);
> 
> -  sections = NULL;
> -  sections_end = NULL;
> -
> -  if (build_section_table (nbfd, &sections, &sections_end) == 0)
> -    {
> -      make_cleanup (xfree, sections);
> -
> -      /* Adjust the target section addresses by the load address.  */
> -      for (tsec = sections; tsec != sections_end; ++tsec)
> -       {
> -         tsec->addr += loadbase;
> -         tsec->endaddr += loadbase;
> -       }
> -
> -      add_target_sections (&nbfd, sections, sections_end);
> -    }
> +  add_target_sections_of_objfile (objf);
> 
>    /* This might change our ideas about frames already looked at.  */
>    reinit_frame_cache ();
> diff --git a/gdb/symfile.c b/gdb/symfile.c
> index 64a83c6..caa0722 100644
> --- a/gdb/symfile.c
> +++ b/gdb/symfile.c
> @@ -3893,7 +3893,7 @@ static void
>  symfile_free_objfile (struct objfile *objfile)
>  {
>    /* Remove the target sections of user-added objfiles.  */
> -  if (objfile != 0 && objfile->flags & OBJF_USERLOADED)
> +  if (objfile != NULL)
>      remove_target_sections ((void *) objfile);
>  }
> 
> Alternatively, I could add another OBJF_ flag and another free_objfile
> observer
> to also remove the symfile-from-memory target sections.
> 
> Regards,
> Markus.

Intel GmbH
Dornacher Strasse 1
85622 Feldkirchen/Muenchen, Deutschland
Sitz der Gesellschaft: Feldkirchen bei Muenchen
Geschaeftsfuehrer: Christian Lamprechter, Hannes Schwaderer, Douglas Lusk
Registergericht: Muenchen HRB 47456
Ust.-IdNr./VAT Registration No.: DE129385895
Citibank Frankfurt a.M. (BLZ 502 109 00) 600119052