This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH v2] Catch exception in value_rtti_indirect_type
- From: Simon Marchi <simon dot marchi at ericsson dot com>
- To: Pedro Alves <palves at redhat dot com>, <gdb-patches at sourceware dot org>
- Cc: <simon dot marchi at polymtl dot ca>
- Date: Thu, 22 Jan 2015 15:37:18 -0500
- Subject: Re: [PATCH v2] Catch exception in value_rtti_indirect_type
- Authentication-results: sourceware.org; auth=none
- References: <1411174953-8930-1-git-send-email-simon dot marchi at ericsson dot com> <54809E38 dot 3080606 at redhat dot com> <54821670 dot 6020006 at ericsson dot com>
On 14-12-05 03:32 PM, Simon Marchi wrote:
> On 2014-12-04 12:47 PM, Pedro Alves wrote:
>> On 09/20/2014 02:02 AM, Simon Marchi wrote:
>>> In the situation described in bug 17416 [1],
>>
>> Could you paste that in the commit log please?
>
> Done.
>
>>> an exception thrown in
>>> value_ind can propagate too far and leave an half-built variable object,
>>> leading to a wrong state. This patch adds a TRY_CATCH to catch it and
>>> makes value_rtti_indirect_type return NULL in that case, meaning
>>> that the type of the pointed object could not be found.
>>>
>>> If you want, I can also integrate the test case provided in the bug
>>
>> Yes please.
>
> Done.
>
>>> description. I just don't know how to name it without giving it a
>>> ridiculously long name such as
>>> mi-var-list-children-with-print-object-on-and-a-null-pointer-to-a-structure-that-contains-a-pointer-to-a-structure.exp.
>>
>> Perhaps mi-var-list-children-invalid-grandchild.exp.
>
> Perfect, thanks.
>
>>>
>>> I tested the change on my machine, Ubuntu 14.10 x86-64.
>>>
>>> gdb/Changelog:
>>>
>>> * valops.c (value_rtti_indirect_type): Catch exception thrown by
>>> value_ind.
>>>
>>> [1] https://sourceware.org/bugzilla/show_bug.cgi?id=17416
>>> ---
>>> gdb/valops.c | 13 ++++++++++++-
>>> 1 file changed, 12 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/gdb/valops.c b/gdb/valops.c
>>> index e1decf0..c1a0c86 100644
>>> --- a/gdb/valops.c
>>> +++ b/gdb/valops.c
>>> @@ -3609,7 +3609,18 @@ value_rtti_indirect_type (struct value *v, int *full,
>>> if (TYPE_CODE (type) == TYPE_CODE_REF)
>>> target = coerce_ref (v);
>>> else if (TYPE_CODE (type) == TYPE_CODE_PTR)
>>> - target = value_ind (v);
>>> + {
>>> + volatile struct gdb_exception except;
>>> +
>>> + TRY_CATCH (except, RETURN_MASK_ERROR)
>>> + {
>>> + target = value_ind (v);
>>> + }
>>> + if (except.error == MEMORY_ERROR)
>>> + return NULL;
>>> + else if (except.error != GDB_NO_ERROR)
>>> + throw_exception (except);
>>
>> Indentation looks odd. But, could you write it like this instead ?
>>
>> volatile struct gdb_exception except;
>>
>> TRY_CATCH (except, RETURN_MASK_ERROR)
>> {
>> target = value_ind (v);
>> }
>> if (except.reason < 0)
>> {
>> if (except.error == MEMORY_ERROR)
>> {
>> /* Add comment here. */
>> return NULL;
>> }
>> throw_exception (except);
>> }
>
> Indeed it was odd. I took your suggestion. Here is the updated v2:
>
> From 4f3924e505d6343c430292e2261d146e66d8799a Mon Sep 17 00:00:00 2001
> From: Simon Marchi <simon.marchi@ericsson.com>
> Date: Fri, 19 Sep 2014 21:02:33 -0400
> Subject: [PATCH v2] Catch exception in value_rtti_indirect_type
>
> In the situation described in bug 17416 [1]:
>
> * "set print object" is on;
> * The variable object is a pointer to a struct, and it contains an
> invalid value (e.g. NULL, or random uninitialized value);
> * The variable object (struct) has a child which is also a pointer to a
> struct;
> * We try to use "-var-list-children".
>
> ... an exception thrown in value_ind can propagate too far and leave an
> half-built variable object, leading to a wrong state. This patch adds a
> TRY_CATCH to catch it and makes value_rtti_indirect_type return NULL in
> that case, meaning that the type of the pointed object could not be
> found.
>
> A test for the fix is also added.
>
> New in v2:
>
> * Added test.
> * Restructured "catch" code.
> * Added details about the bug in commit log.
>
> gdb/Changelog:
>
> * valops.c (value_rtti_indirect_type): Catch exception thrown by
> value_ind.
>
> gdb/testsuite/ChangeLog
>
> * gdb.mi/mi-var-list-children-invalid-grandchild.c: New file.
> * gdb.mi/mi-var-list-children-invalid-grandchild.exp: New file.
>
> [1] https://sourceware.org/bugzilla/show_bug.cgi?id=17416
> ---
> .../mi-var-list-children-invalid-grandchild.c | 46 ++++++++++++++++
> .../mi-var-list-children-invalid-grandchild.exp | 63 ++++++++++++++++++++++
> gdb/valops.c | 20 ++++++-
> 3 files changed, 128 insertions(+), 1 deletion(-)
> create mode 100644 gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.c
> create mode 100644 gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.exp
>
> diff --git a/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.c b/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.c
> new file mode 100644
> index 0000000..b67d875
> --- /dev/null
> +++ b/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.c
> @@ -0,0 +1,46 @@
> +/* 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/>. */
> +
> +struct inner
> +{
> + int a;
> +};
> +
> +struct outer
> +{
> + struct inner *inner;
> +};
> +
> +int main (void)
> +{
> + struct inner inner;
> + struct outer outer;
> + struct outer *p_outer;
> +
> + inner.a = 42;
> + outer.inner = &inner;
> +
> + /* We force p_outer to an invalid value, but this also happens naturally
> + * when a variable has not been initialized. */
> +
> + p_outer = 0;
> + /* p_outer set to invalid value */
> + p_outer = &outer;
> + /* p_outer set to valid value */
> +
> + return 0;
> +}
> diff --git a/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.exp b/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.exp
> new file mode 100644
> index 0000000..ed8460d
> --- /dev/null
> +++ b/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.exp
> @@ -0,0 +1,63 @@
> +# 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/>.
> +
> +# This is a regression test for bug 17416
> +# https://sourceware.org/bugzilla/show_bug.cgi?id=17416
> +
> +load_lib mi-support.exp
> +set MIFLAGS "-i=mi"
> +
> +gdb_exit
> +if [mi_gdb_start] {
> + continue
> +}
> +
> +standard_testfile
> +
> +if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug}] != "" } {
> + untested mi-var-list-children-invalid-grandchild.exp
> + return -1
> +}
> +
> +mi_delete_breakpoints
> +mi_gdb_reinitialize_dir $srcdir/$subdir
> +mi_gdb_load ${binfile}
> +
> +set line_invalid_pointer_value [gdb_get_line_number "p_outer set to invalid value"]
> +set line_valid_pointer_value [gdb_get_line_number "p_outer set to valid value"]
> +
> +# set print object on
> +mi_gdb_test "-interpreter-exec console \"set print object on\"" \
> + "(.*=cmd-param-changed,param=\"print object\",value=\"on\".*|)\\^done" \
> + "set print object on"
> +
> +mi_runto main
> +
> +mi_continue_to_line $line_invalid_pointer_value "continue to invalid pointer value"
> +
> +# Create variable object
> +mi_gdb_test "-var-create var1 * p_outer" "\\^done,name=\"var1\",numchild=\"1\",value=\"0x0\",type=\"struct outer \\*\",thread-id=\"1\",has_more=\"0\"" "create variable object"
> +
> +# List children while the pointer is uninitialized.
> +# Note: if you remove this -var-list-children, the second one will work.
> +mi_gdb_test "-var-list-children var1" "\\^done,numchild=\"1\",children=\\\[child={name=\"var1.inner\",exp=\"inner\",numchild=\"1\",type=\"struct inner \\*\",thread-id=\"1\"}\\\],has_more=\"0\"" "list children #1"
> +
> +mi_continue_to_line $line_valid_pointer_value "continue to valid pointer value"
> +
> +# Update and list children again
> +mi_gdb_test "-var-update var1" "\\^done,changelist=\\\[{name=\"var1\",in_scope=\"true\",type_changed=\"false\",has_more=\"0\"},{name=\"var1.inner\",in_scope=\"true\",type_changed=\"false\",has_more=\"0\"}\\\]" "update variable object"
> +
> +mi_gdb_test "-var-list-children var1" "\\^done,numchild=\"1\",children=\\\[child={name=\"var1.inner\",exp=\"inner\",numchild=\"1\",type=\"struct inner \\*\",thread-id=\"1\"}\\\],has_more=\"0\"" "list children #2"
> +
> diff --git a/gdb/valops.c b/gdb/valops.c
> index 4125fc0..d806f27 100644
> --- a/gdb/valops.c
> +++ b/gdb/valops.c
> @@ -3609,7 +3609,25 @@ value_rtti_indirect_type (struct value *v, int *full,
> if (TYPE_CODE (type) == TYPE_CODE_REF)
> target = coerce_ref (v);
> else if (TYPE_CODE (type) == TYPE_CODE_PTR)
> - target = value_ind (v);
> + {
> + volatile struct gdb_exception except;
> +
> + TRY_CATCH (except, RETURN_MASK_ERROR)
> + {
> + target = value_ind (v);
> + }
> + if (except.reason < 0)
> + {
> + if (except.error == MEMORY_ERROR)
> + {
> + /* value_ind threw a memory error. The pointer is NULL or
> + contains an uninitialized value: we can't determine any
> + type. */
> + return NULL;
> + }
> + throw_exception (except);
> + }
> + }
> else
> return NULL;
Ping.
(I realized I had forgotten about this patch while reviewing my opened bugs)