This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
[PATCH 0/2 v2] Add options to skip unavailable locals
- From: Yao Qi <yao at codesourcery dot com>
- To: <gdb-patches at sourceware dot org>
- Date: Sun, 25 Aug 2013 11:42:01 +0800
- Subject: [PATCH 0/2 v2] Add options to skip unavailable locals
- References: <51FA557F dot 5 at redhat dot com>
Hi,
In the review to '[PATCH 2/2] Add options to skip unavailable locals',
Pedro pointed out two things here
<http://sourceware.org/ml/gdb-patches/2013-08/msg00019.html>
On 08/01/2013 08:33 PM, Pedro Alves wrote:>> @@ -295,18 +293,39 @@ mi_cmd_stack_list_args (char *command, char **argv, int argc)
>> > struct ui_out *uiout = current_uiout;
>> > int raw_arg = 0;
>> > enum py_bt_status result = PY_BT_ERROR;
>> >+ int skip_unavailable = 0;
>> >+ int oind = 0;
>> >
>> >- if (argc > 0)
>> >- raw_arg = parse_no_frames_option (argv[0]);
>> >+ /* We can't use mi_getopt here, because the number of options is not
>> >+ determined. */
> Hmm. Isn't that easy to fix though? We'd just need an mi_getopt
> variant that doesn't error out when it sees an unknown option, but
> instead returns that option's position (similarly to getopt). It's
> then the caller's responsibility to parse the rest of the option
> string.
I add a variant mi_getopt_silent in patch 1/2 and use it here.
>> >@@ -597,7 +626,6 @@ list_args_or_locals (enum what_to_list what, enum print_values values,
>> > if (print_me)
>> > {
>> > struct symbol *sym2;
>> >- struct frame_arg arg, entryarg;
>> >
>> > if (SYMBOL_IS_ARGUMENT (sym))
>> > sym2 = lookup_symbol (SYMBOL_LINKAGE_NAME (sym),
>> >@@ -607,33 +635,56 @@ list_args_or_locals (enum what_to_list what, enum print_values values,
>> > sym2 = sym;
>> > gdb_assert (sym2 != NULL);
>> >
>> >- memset (&arg, 0, sizeof (arg));
>> >- arg.sym = sym2;
>> >- arg.entry_kind = print_entry_values_no;
>> >- memset (&entryarg, 0, sizeof (entryarg));
>> >- entryarg.sym = sym2;
>> >- entryarg.entry_kind = print_entry_values_no;
>> >-
>> >- switch (values)
>> >+ /* Need to read the value before being able to determine
>> >+ whether its unavailable. */
>> >+ if (values == PRINT_ALL_VALUES
>> >+ || values == PRINT_SIMPLE_VALUES
>> >+ || skip_unavailable)
>> >+ val = read_var_value (sym2, fi);
>> >+
>> >+ if (skip_unavailable
>> >+ && (value_entirely_unavailable (val)
>> >+ /* A scalar object that does not have all bits
>> >+ available is also considered unavailable,
>> >+ because all bits contribute to its
>> >+ representation. */
>> >+ || (val_print_scalar_type_p (value_type (val))
>> >+ && !value_bytes_available (val,
>> >+ value_embedded_offset (val),
>> >+ TYPE_LENGTH (value_type (val))))))
>> >+ ;
>> >+ else
> I don't think this has been updated right for entry values. With
> entry values, we now have_two_ values to account for. I think
> we need to do this once for each of the regular arg and the
> entry arg?
>
Patch 2/2 is updated to handle entry values by adding a parameter
'skip_unavailable' in function list_arg_or_local.
Other comments on changelog and doc are addressed too. Regression
tested on x86_64-linux both native and gdbserver.
*** BLURB HERE ***
Yao Qi (2):
Use mi_getopt_silent
Add options to skip unavailable locals
gdb/NEWS | 4 +
gdb/doc/gdb.texinfo | 17 +++-
gdb/mi/mi-cmd-stack.c | 127 +++++++++++++++-------
gdb/mi/mi-getopt.c | 35 +++++-
gdb/mi/mi-getopt.h | 8 +-
gdb/testsuite/gdb.trace/entry-values.exp | 5 +
gdb/testsuite/gdb.trace/mi-trace-unavailable.exp | 9 ++
gdb/valprint.c | 8 +-
gdb/valprint.h | 2 +
gdb/value.c | 20 ++++
gdb/value.h | 4 +
11 files changed, 185 insertions(+), 54 deletions(-)
--
1.7.7.6