[pushed] Don't suppress errors inserting/removing hardware breakpoints in shared libraries.
Commit Message
As explained in
https://sourceware.org/ml/gdb-patches/2008-08/msg00361.html, after a
shared library was unloaded, we can no longer insert or remove
breakpoints into/from its (no longer present) code segment. That'll
fail with memory errors. However, that concern does not apply to
hardware breakpoints. By definition, hardware breakpoints are
implemented using a mechanism that is not dependent on being able to
modify the target's memory. Usually, by setting up CPU debug
registers. IOW, we should be able to set hw breakpoints in an
unmapped address. We don't seem to have a test that exercises that,
so this patch adds one.
I noticed the error supression because of a related issue -- the
target_insert_hw_breakpoint/target_remove_hw_breakpoint interfaces
don't really distinguish "not supported" from "error" return, and so
remote.c returns -1 in both cases. This results in hardware
breakpoints set in shared libraries silently ending up pending forever
even though the target doesn't actually support hw breakpoints.
(gdb) set breakpoint always-inserted on
(gdb) set remote Z-packet off
(gdb) info breakpoints
No breakpoints or watchpoints.
(gdb) hbreak shrfunc
Hardware assisted breakpoint 3 at 0x7ffff7dfb657: file ../../../src/gdb/testsuite/gdb.base/hbreak-in-shr-unsupported-shr.c, line 21.
(gdb) info break
Num Type Disp Enb Address What
3 hw breakpoint keep y <PENDING> shrfunc
After the patch we get the expected:
(gdb) hbreak shrfunc
Hardware assisted breakpoint 3 at 0x7ffff7dfb657: file ../../../src/gdb/testsuite/gdb.base/hbreak-in-shr-unsupported-shr.c, line 21.
Warning:
Cannot insert hardware breakpoint 3.
Could not insert hardware breakpoints:
You may have requested too many hardware breakpoints/watchpoints.
(gdb) info break
Num Type Disp Enb Address What
3 hw breakpoint keep y 0x00007ffff7dfb657 in shrfunc at ../../../src/gdb/testsuite/gdb.base/hbreak-in-shr-unsupported-shr.c:21
(HW breakpoints set in the main executable, when the target doesn't
support HW breakpoints always resulted in the latter output.)
We probably should improve the insert/remove interface to return a
different error code for unsupported. But I chose to fix the error
supression first, as it's a deeper and wider issue.
Tested on x86_64 Fedora 17, native and gdbserver.
gdb/
2014-04-23 Pedro Alves <palves@redhat.com>
* breakpoint.c (insert_bp_location, remove_breakpoint_1): If
the breakpoint is set in a shared library, only suppress
errors for software breakpoints, not hardware breakpoints.
gdb/testsuite/
2014-04-23 Pedro Alves <palves@redhat.com>
* gdb.base/hbreak-in-shr-unsupported-shr.c: New file.
* gdb.base/hbreak-in-shr-unsupported.c: New file.
* gdb.base/hbreak-in-shr-unsupported.exp: New file.
* gdb.base/hbreak-unmapped.c: New file.
* gdb.base/hbreak-unmapped.exp: New file.
* gdb.trace/qtro.exp (gdb_is_target_remote): Move ...
* lib/gdb.exp (gdb_is_target_remote): ... here.
---
gdb/ChangeLog | 6 ++
gdb/breakpoint.c | 14 +--
gdb/testsuite/ChangeLog | 10 ++
.../gdb.base/hbreak-in-shr-unsupported-shr.c | 22 ++++
gdb/testsuite/gdb.base/hbreak-in-shr-unsupported.c | 28 +++++
.../gdb.base/hbreak-in-shr-unsupported.exp | 120 +++++++++++++++++++++
gdb/testsuite/gdb.base/hbreak-unmapped.c | 22 ++++
gdb/testsuite/gdb.base/hbreak-unmapped.exp | 72 +++++++++++++
gdb/testsuite/gdb.trace/qtro.exp | 16 ---
gdb/testsuite/lib/gdb.exp | 19 ++++
10 files changed, 307 insertions(+), 22 deletions(-)
create mode 100644 gdb/testsuite/gdb.base/hbreak-in-shr-unsupported-shr.c
create mode 100644 gdb/testsuite/gdb.base/hbreak-in-shr-unsupported.c
create mode 100644 gdb/testsuite/gdb.base/hbreak-in-shr-unsupported.exp
create mode 100644 gdb/testsuite/gdb.base/hbreak-unmapped.c
create mode 100644 gdb/testsuite/gdb.base/hbreak-unmapped.exp
Comments
On 04/23/2014 10:08 PM, Pedro Alves wrote:
> +# Hardware breakpoints are implemented using a mechanism that is not
> +# dependent on being able to modify the target's memory, we should be
> +# able to set them even in unmapped memory areas.
> +gdb_test "hbreak *0" "Hardware assisted breakpoint \[0-9\]+ at 0x0"
> +
> +gdb_test "info break" "hw breakpoint.*y.*0x0\+\[ \t\]\+" \
> + "info break shows hw breakpoint"
These two tests fail on arm-none-eabi target,
hbreak *0^M
Hardware assisted breakpoint 3 at 0x0: file
/scratch/yqi/arm-none-eabi-lite/obj/cs3-2014.11-999999-arm-none-eabi-i686-pc-linux-gnu/generated/arm-vector.S,
line 25.^M
(gdb) FAIL: gdb.base/hbreak-unmapped.exp: hbreak *0
info break^M
Num Type Disp Enb Address What^M
3 hw breakpoint keep y 0x00000000
/scratch/yqi/arm-none-eabi-lite/obj/cs3-2014.11-999999-arm-none-eabi-i686-pc-linux-gnu/generated/arm-vector.S:25^M
(gdb) FAIL: gdb.base/hbreak-unmapped.exp: info break shows hw breakpoint
delete $bpnum
because the vector table base address is 0x0.
We can either relax the regex pattern or skip this test on bare metal
targets. However, after I read the commit log, I find
hbreak-unmapped.exp isn't necessary because we've already had
hbreak-in-shr-unsupported.exp. I don't see anything hbreak-unmapped.exp
can cover while hbreak-in-shr-unsupported.exp doesn't. I suggest we
1) remove hbreak-unmapped.exp and hbreak-unmapped.c
2) or skip hbreak-unmapped.exp on bare metal and uclinux targets,
3) or relax the pattern
What do you think?
On 06/03/2014 10:40 AM, Yao Qi wrote:
> On 04/23/2014 10:08 PM, Pedro Alves wrote:
>> +# Hardware breakpoints are implemented using a mechanism that is not
>> +# dependent on being able to modify the target's memory, we should be
>> +# able to set them even in unmapped memory areas.
>> +gdb_test "hbreak *0" "Hardware assisted breakpoint \[0-9\]+ at 0x0"
>> +
>> +gdb_test "info break" "hw breakpoint.*y.*0x0\+\[ \t\]\+" \
>> + "info break shows hw breakpoint"
>
> These two tests fail on arm-none-eabi target,
>
> hbreak *0^M
> Hardware assisted breakpoint 3 at 0x0: file
> /scratch/yqi/arm-none-eabi-lite/obj/cs3-2014.11-999999-arm-none-eabi-i686-pc-linux-gnu/generated/arm-vector.S,
> line 25.^M
> (gdb) FAIL: gdb.base/hbreak-unmapped.exp: hbreak *0
> info break^M
> Num Type Disp Enb Address What^M
> 3 hw breakpoint keep y 0x00000000
> /scratch/yqi/arm-none-eabi-lite/obj/cs3-2014.11-999999-arm-none-eabi-i686-pc-linux-gnu/generated/arm-vector.S:25^M
> (gdb) FAIL: gdb.base/hbreak-unmapped.exp: info break shows hw breakpoint
> delete $bpnum
>
> because the vector table base address is 0x0.
>
> We can either relax the regex pattern or skip this test on bare metal
> targets. However, after I read the commit log, I find
> hbreak-unmapped.exp isn't necessary because we've already had
> hbreak-in-shr-unsupported.exp. I don't see anything hbreak-unmapped.exp
> can cover while hbreak-in-shr-unsupported.exp doesn't. I suggest we
They're opposites. The real tests are not the "hbreak main" parts,
which is the tests probing target features, but what follows. The
former tests that inserting a hbreak in an unmapped region
doesn't error. While the latter tests that inserting a hbreak in
a mapped region when the target doesn't support hbreak at all
results in an error.
> 2) or skip hbreak-unmapped.exp on bare metal and uclinux targets,
Yes, but no need to match by target, we can do like e.g.,
gdb.base/signull.exp does, and check whether "x /b 0" manages
to read memory. If so, then skip the test. Any sane target with
an MMU will leave page 0 unmapped, so no need to look for some other
address.
@@ -1,3 +1,9 @@
+2014-04-23 Pedro Alves <palves@redhat.com>
+
+ * breakpoint.c (insert_bp_location, remove_breakpoint_1): If
+ the breakpoint is set in a shared library, only suppress
+ errors for software breakpoints, not hardware breakpoints.
+
2014-04-22 Pedro Alves <palves@redhat.com>
* infrun.c (schedlock_applies): New function, factored out from
@@ -2642,12 +2642,12 @@ insert_bp_location (struct bp_location *bl,
breakpoint in a shared library that has already been
removed, but we have not yet processed the shlib unload
event. Unfortunately, some targets that implement
- breakpoint insertion themselves (necessary if this is a
- HW breakpoint, but SW breakpoints likewise) can't tell
- why the breakpoint insertion failed (e.g., the remote
- target doesn't define error codes), so we must treat
- generic errors as memory errors. */
+ breakpoint insertion themselves can't tell why the
+ breakpoint insertion failed (e.g., the remote target
+ doesn't define error codes), so we must treat generic
+ errors as memory errors. */
if ((bp_err == GENERIC_ERROR || bp_err == MEMORY_ERROR)
+ && bl->loc_type == bp_loc_software_breakpoint
&& solib_name_from_address (bl->pspace, bl->address))
{
/* See also: disable_breakpoints_in_shlibs. */
@@ -3826,7 +3826,9 @@ remove_breakpoint_1 (struct bp_location *bl, insertion_state_t is)
/* 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. */
- if (val && solib_name_from_address (bl->pspace, bl->address))
+ if (val
+ && bl->loc_type == bp_loc_software_breakpoint
+ && solib_name_from_address (bl->pspace, bl->address))
val = 0;
if (val)
@@ -1,3 +1,13 @@
+2014-04-23 Pedro Alves <palves@redhat.com>
+
+ * gdb.base/hbreak-in-shr-unsupported-shr.c: New file.
+ * gdb.base/hbreak-in-shr-unsupported.c: New file.
+ * gdb.base/hbreak-in-shr-unsupported.exp: New file.
+ * gdb.base/hbreak-unmapped.c: New file.
+ * gdb.base/hbreak-unmapped.exp: New file.
+ * gdb.trace/qtro.exp (gdb_is_target_remote): Move ...
+ * lib/gdb.exp (gdb_is_target_remote): ... here.
+
2014-04-22 Pedro Alves <palves@redhat.com>
* gdb.base/consecutive-step-over.c: New file.
new file mode 100644
@@ -0,0 +1,22 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+ Copyright 2004-2014 Free Software Foundation, Inc.
+
+ 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/>. */
+
+int
+shrfunc (int x)
+{
+ return x * 10;
+}
new file mode 100644
@@ -0,0 +1,28 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+ Copyright 2004-2014 Free Software Foundation, Inc.
+
+ 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/>. */
+
+extern int shrfunc (int x);
+
+int
+main (void)
+{
+ int y;
+
+ y = shrfunc (1);
+
+ return 0;
+}
new file mode 100644
@@ -0,0 +1,120 @@
+# Copyright 2014 Free Software Foundation, Inc.
+
+# 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 trying to inserting a hw breakpoint in a shared library
+# when the target doesn't support hw breakpoints doesn't silently
+# error out without informing the user.
+
+if {[skip_shlib_tests]} {
+ return -1
+}
+
+set main_src hbreak-in-shr-unsupported.c
+set lib_src hbreak-in-shr-unsupported-shr.c
+standard_testfile ${main_src} ${lib_src}
+set lib_basename hbreak-in-shr-unsupported-shr.so
+set lib_so [standard_output_file ${lib_basename}]
+
+set lib_opts "debug"
+set exec_opts [list debug shlib=${lib_so}]
+
+if [get_compiler_info] {
+ return -1
+}
+
+if { [gdb_compile_shlib ${srcdir}/${subdir}/${lib_src} ${lib_so} $lib_opts] != ""
+ || [gdb_compile ${srcdir}/${subdir}/${main_src} ${binfile} executable $exec_opts] != ""} {
+ untested "Could not compile ${subdir}/$lib_src or ${subdir}/$srcfile."
+ return -1
+}
+
+clean_restart $binfile
+gdb_load_shlibs $lib_so
+
+if ![runto_main] then {
+ fail "Can't run to main"
+ return -1
+}
+
+set is_target_remote [gdb_is_target_remote]
+
+# Get main breakpoint out of the way.
+delete_breakpoints
+
+# Easier to test if GDB inserts breakpoints immediately.
+gdb_test_no_output "set breakpoint always-inserted on"
+
+# Force-disable Z1 packets, in case the target actually supports
+# these.
+if {$is_target_remote} {
+ gdb_test_no_output "set remote Z-packet off"
+}
+
+# Probe for hw breakpoints support. With Z packets disabled, this
+# should always fail with remote targets. For other targets, with no
+# way to force-disable hw breakpoints support, best we can do is skip
+# the remainder of the test if hardware breakpoint insertion in a
+# function in the main executable succeeds.
+set cant_insert_hbreak 0
+set supports_hbreak 0
+set test "probe hbreak support"
+gdb_test_multiple "hbreak main" $test {
+ -re "No hardware breakpoint support in the target.*$gdb_prompt $" {
+ pass $test
+ }
+ -re "Hardware breakpoints used exceeds limit.*$gdb_prompt $" {
+ pass $test
+ }
+ -re "Cannot insert hardware breakpoint.*$gdb_prompt $" {
+ set cant_insert_hbreak 1
+ set supports_hbreak 1
+ pass $test
+ }
+ -re "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*$gdb_prompt $" {
+ set supports_hbreak 1
+ if {$is_target_remote} {
+ # Z-packets have been force-disabled, so this shouldn't
+ # happen.
+ fail $test
+ } else {
+ pass $test
+ }
+ }
+}
+
+if {!$supports_hbreak} {
+ unsupported "no hbreak support"
+ return
+}
+
+if {!$cant_insert_hbreak} {
+ unsupported "can't disable hw breakpoint support"
+ return
+}
+
+# Get previous hw breakpoint out of the way.
+delete_breakpoints
+
+# Without target support, this should always complain. GDB used to
+# suppress the error if the breakpoint was set in a shared library.
+# While that makes sense for software breakpoints (the memory might be
+# unmapped), it doesn't for hardware breakpoints, as those by
+# definition are implemented using a mechanism that is not dependent
+# on being able to modify the target's memory.
+gdb_test "hbreak shrfunc" "Cannot insert hardware breakpoint.*" \
+ "hbreak shrfunc complains"
+
+gdb_test "info break" "hw breakpoint.*y.*$hex.*in shrfunc at.*" \
+ "breakpoint not pending"
new file mode 100644
@@ -0,0 +1,22 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+ Copyright 2014 Free Software Foundation, Inc.
+
+ 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/>. */
+
+int
+main (void)
+{
+ return 0;
+}
new file mode 100644
@@ -0,0 +1,72 @@
+# Copyright 2014 Free Software Foundation, Inc.
+
+# 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/>.
+
+if {[skip_hw_breakpoint_tests]} {
+ return 0
+}
+
+standard_testfile
+
+if {[prepare_for_testing "failed to prepare" ${testfile} ${srcfile}]} {
+ return -1
+}
+
+if ![runto_main] {
+ fail "Can't run to main"
+ return -1
+}
+
+delete_breakpoints
+
+# Test whether the target supports hardware breakpoints at all.
+
+set supports_hbreak 0
+set test "probe hardware breakpoint support"
+gdb_test_multiple "hbreak main" $test {
+ -re "No hardware breakpoint support in the target.*$gdb_prompt $" {
+ pass $test
+ }
+ -re "Hardware breakpoints used exceeds limit.*$gdb_prompt $" {
+ pass $test
+ }
+ -re "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*$gdb_prompt $" {
+ set supports_hbreak 1
+ pass $test
+ }
+}
+
+if {!$supports_hbreak} {
+ unsupported "hardware breakpoints"
+ return
+}
+
+delete_breakpoints
+
+# Force immediate breakpoint insertion.
+gdb_test_no_output "set breakpoint always-inserted on"
+
+# Hardware breakpoints are implemented using a mechanism that is not
+# dependent on being able to modify the target's memory, we should be
+# able to set them even in unmapped memory areas.
+gdb_test "hbreak *0" "Hardware assisted breakpoint \[0-9\]+ at 0x0"
+
+gdb_test "info break" "hw breakpoint.*y.*0x0\+\[ \t\]\+" \
+ "info break shows hw breakpoint"
+
+gdb_test_no_output "delete \$bpnum" "" "delete" \
+ "delete hw breakpoint"
+
+gdb_test "info break" "No breakpoints or watchpoints\." \
+ "info break shows no breakpoints"
@@ -35,22 +35,6 @@ if ![runto_main] {
# Check whether we're testing with the remote or extended-remote
# targets, and whether the target supports tracepoints.
-proc gdb_is_target_remote { } {
- global gdb_prompt
-
- set test "probe for target remote"
- gdb_test_multiple "maint print target-stack" $test {
- -re ".*emote serial target in gdb-specific protocol.*$gdb_prompt $" {
- pass $test
- return 1
- }
- -re "$gdb_prompt $" {
- pass $test
- }
- }
- return 0
-}
-
if ![gdb_is_target_remote] {
return -1
}
@@ -2454,6 +2454,25 @@ proc skip_libstdcxx_probe_tests {} {
return $ok
}
+# Check whether we're testing with the remote or extended-remote
+# targets.
+
+proc gdb_is_target_remote {} {
+ global gdb_prompt
+
+ set test "probe for target remote"
+ gdb_test_multiple "maint print target-stack" $test {
+ -re ".*emote serial target in gdb-specific protocol.*$gdb_prompt $" {
+ pass $test
+ return 1
+ }
+ -re "$gdb_prompt $" {
+ pass $test
+ }
+ }
+ return 0
+}
+
set compiler_info "unknown"
set gcc_compiled 0
set hp_cc_compiler 0