From: "Marc-André Laperle" <marc-andre.laperle@ericsson.com>
To: Pedro Alves <palves@redhat.com>, <gdb-patches@sourceware.org>
Subject: Re: [PATCH 3/3] Add -file-list-shared-libraries MI command
Date: Wed, 04 Jan 2017 17:19:00 -0000 [thread overview]
Message-ID: <ddab3a27-8fd3-ba10-c7e1-27ea8b85efa5@ericsson.com> (raw)
In-Reply-To: <176aae27-f7c3-c563-141d-35eb1f69ee02@redhat.com>
Thanks for the comments! A few questions/suggestions below.
On 2016-11-23 08:06 AM, Pedro Alves wrote:
> Hi Marc-Andre,
>
> Great that you're tackling this. Thanks much for digging into GDB. :-)
> I like this. As for following GDB's code standards, it's almost perfect.
>
> But I have questions on the MI output, though. See below.
>
> On 09/12/2016 09:27 PM, Marc-Andre Laperle wrote:
>
>> diff --git a/gdb/doc/gdb.texinfo b/gdb/doc/gdb.texinfo
>> index d1a5e7c..c6b2133 100644
>> --- a/gdb/doc/gdb.texinfo
>> +++ b/gdb/doc/gdb.texinfo
>> @@ -31401,26 +31401,35 @@ The @value{GDBN} equivalent is @samp{info sources}.
>> (gdb)
>> @end smallexample
>>
>> -@ignore
>> @subheading The @code{-file-list-shared-libraries} Command
>> @findex -file-list-shared-libraries
>>
>> @subsubheading Synopsis
>>
>> @smallexample
>> - -file-list-shared-libraries
>> + -file-list-shared-libraries [ @var{regexp} ]
>> @end smallexample
>>
>> List the shared libraries in the program.
>> +With a regular expression @var{regexp}, only those libraries whose
>> +names match @var{regexp} are listed.
>>
>> @subsubheading @value{GDBN} Command
>>
>> The corresponding @value{GDBN} command is @samp{info shared}.
>>
>> @subsubheading Example
>> -N.A.
>> +@smallexample
>> +(gdb)
>> +-file-list-exec-source-files
>> +^done,shared-libraries=[
>> +@{from="0x72815989",to="0x728162c0",syms-read="1",name="/lib/libfoo.so"@},
>> +@{from="0x76ee48c0",to="0x76ee9160",syms-read="1",name="/lib/libbar.so"@}]
> I don't see where the documentation describes what the
> attributes of each list element are. Sorry if I missed it.
>
> I find it surprising that the attributes output don't match
> attributes output by =library-loaded ? I'd think they should
> match. That'd simplify the documentation too, as one place
> would refer to the other for attributes list.
>
> BTW, =library-loaded doesn't output from/to. ISTR that that
> was discussed and left out, because from/to assume there's
> a contiguous range to report, while that's not true on all
> targets (i.e., assumes a single segment). So it seems to
> me that that's a separate discussion/patch would better
> be addressed in both places.
Those attributes are from the "info shared" command which is what this
new MI command is inspired from. The from/to are shown in Eclipse and I
personally use it to know in which library a given stack frame is, in
absence of debug symbols. I could change the attributes to be just like
"=library-loaded" but it would be inconsistent with "info shared" and
will remove information that was available before. Perhaps the best way
forward would be to make all attributes the same as "=library-loaded"
but also add the from/to.
To make the MI more future proof, it could be a list of from/to segments
instead, but for now it will only one segment. What do you think?
>> +void
>> +mi_cmd_file_list_shared_libraries (char *command, char **argv, int argc)
>> +{
>> + struct ui_out *uiout = current_uiout;
>> + const char *pattern;
>> + struct so_list *so = NULL;
>> + struct gdbarch *gdbarch = target_gdbarch ();
>> +
>> + switch (argc)
>> + {
>> + case 0:
>> + pattern = NULL;
>> + break;
>> + case 1:
>> + pattern = argv[0];
>> + break;
>> + default:
>> + error (_("Usage: -file-list-shared-libraries [REGEXP]"));
>> + break;
> error doesn't return. No need for the break.
>
>> + }
>> +
>> + if (pattern != NULL)
>> + {
>> + char *re_err = re_comp (pattern);
>> +
>> + if (re_err != NULL)
>> + error (_("Invalid regexp: %s"), re_err);
>> + }
>> +
>> + update_solib_list (1);
>> +
>> + /* Print the table header. */
>> + ui_out_begin (uiout, ui_out_type_list, "shared-libraries");
>> +
>> + ALL_SO_LIBS (so)
>> + {
>> + if (so->so_name[0] == '\0')
>> + continue;
>> + if (pattern != NULL && !re_exec (so->so_name))
>> + continue;
>> +
>> + ui_out_begin (uiout, ui_out_type_tuple, NULL);
>> +
>> + if (so->addr_high != 0)
>> + {
>> + ui_out_field_core_addr (uiout, "from", gdbarch, so->addr_low);
>> + ui_out_field_core_addr (uiout, "to", gdbarch, so->addr_high);
>> + }
>> + else
>> + {
>> + ui_out_field_skip (uiout, "from");
>> + ui_out_field_skip (uiout, "to");
>> + }
>> +
>> + ui_out_field_int (uiout, "syms-read", so->symbols_loaded ? 1 : 0);
>> +
>> + ui_out_field_string (uiout, "name", so->so_name);
> So seems to me that the inner body of this loop would be better
> calling a function that is shared with =library-loaded.
>
>> +
>> --- a/gdb/solist.h
>> +++ b/gdb/solist.h
>> @@ -23,6 +23,11 @@
>> /* For domain_enum domain. */
>> #include "symtab.h"
>>
>> +#define ALL_SO_LIBS(so) \
>> + for (so = current_program_space->so_list; \
>> + so; \
> Write explicit 'so != NULL'.
>
>> + so = so->next)
>> +
>> /* Forward declaration for target specific link map information. This
>> struct is opaque to all but the target specific file. */
>> struct lm_info;
>> diff --git a/gdb/testsuite/gdb.mi/mi-solib.exp b/gdb/testsuite/gdb.mi/mi-solib.exp
>> index 2227987..4c40ba4 100644
>> --- a/gdb/testsuite/gdb.mi/mi-solib.exp
>> +++ b/gdb/testsuite/gdb.mi/mi-solib.exp
>> @@ -48,27 +48,47 @@ if { [gdb_compile_shlib ${srcfile_lib} ${binfile_lib} $lib_flags] != ""
>>
>> mi_delete_breakpoints
>> mi_gdb_reinitialize_dir $srcdir/$subdir
>> -mi_gdb_reinitialize_dir $srcdir/$subdir
>> mi_gdb_load ${binfile}
>>
>> mi_load_shlibs $binfile_lib
>>
>> -mi_gdb_test "777-gdb-set stop-on-solib-events 1" "777\\^done" \
>> - "set stop-on-solib-events"
>> +proc test_stop_on_solib_events {} {
> Looks like a good candidate for the the new proc_with_prefix.
>
>> + mi_gdb_test "777-gdb-set stop-on-solib-events 1" "777\\^done" \
>> + "set stop-on-solib-events"
>>
>> -# We use "run" rather than "-exec-run" here in order to test that CLI
>> -# commands still cause the correct MI output to be generated.
>> -mi_run_with_cli
>> + # We use "run" rather than "-exec-run" here in order to test that CLI
>> + # commands still cause the correct MI output to be generated.
>> + mi_run_with_cli
>>
>> -# Also test that the CLI solib event note is output.
>> -set test "CLI prints solib event"
>> -gdb_expect {
>> - -re "~\"Stopped due to shared library event \\(no libraries added or removed\\)\\\\n" {
>> - pass "$test"
>> - }
>> - timeout {
>> - fail "$test (timeout)"
>> + # Also test that the CLI solib event note is output.
>> + set test "CLI prints solib event"
>> + gdb_expect {
>> + -re "~\"Stopped due to shared library event \\(no libraries added or removed\\)\\\\n" {
>> + pass "$test"
>> + }
>> + timeout {
>> + fail "$test (timeout)"
>> + }
>> }
>> +
>> + mi_expect_stop solib-event .* .* .* .* .* "check for solib event"
>> +
>> + # Unset solib events to avoid interfering with other tests.
>> + mi_gdb_test "778-gdb-set stop-on-solib-events 0" "778\\^done" \
>> + "unset stop-on-solib-events"
>> +}
>> +
>> +proc test_file_list_shared_libraries {} {
>> + global libname
>> + global binfile
>> +
>> + mi_continue_to main
>> +
>> + mi_gdb_test "222-file-list-shared-libraries" \
>> + "222\\^done,shared-libraries=\\\[.*\{from=\".*\",to=\".*\",syms-read=\"1\",name=\".*${libname}.so\"\}.*]" \
>> + "Getting a list of shared libraries."
> Lowercase, no period at end, and use imperative:
>
> "get list of shared libraries"
>
> Thanks,
> Pedro Alves
>
next prev parent reply other threads:[~2017-01-04 17:19 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-09-12 20:28 [PATCH 1/3] Remove unused parameter in solib_add and update_solib_list Marc-Andre Laperle
2016-09-12 20:28 ` [PATCH 3/3] Add -file-list-shared-libraries MI command Marc-Andre Laperle
2016-10-14 21:20 ` Marc-André Laperle
2016-10-20 5:05 ` Simon Marchi
2016-11-23 13:06 ` Pedro Alves
2017-01-04 17:19 ` Marc-André Laperle [this message]
2017-01-12 16:15 ` Pedro Alves
2016-09-12 20:28 ` [PATCH 2/3] Add a better diagnostic message in mi_gdb_test Marc-Andre Laperle
2016-10-20 4:24 ` Simon Marchi
2016-11-23 13:03 ` Pedro Alves
2016-10-20 4:20 ` [PATCH 1/3] Remove unused parameter in solib_add and update_solib_list Simon Marchi
2016-11-23 13:03 ` Pedro Alves
2017-02-03 17:17 ` [PATCH v2 0/3] Shared libraries MI command Marc-Andre Laperle
2017-02-03 17:16 ` [PATCH v2 2/3] Add a better diagnostic message in mi_gdb_test Marc-Andre Laperle
2017-02-03 17:16 ` [PATCH v2 3/3] Add -file-list-shared-libraries MI command Marc-Andre Laperle
2017-02-06 12:40 ` Pedro Alves
2017-02-28 22:08 ` [Patch v3] " Marc-Andre Laperle
2017-03-01 15:50 ` Marc-André Laperle
2017-03-01 16:12 ` Eli Zaretskii
2017-03-01 16:38 ` [Patch v4] " Marc-Andre Laperle
2017-03-01 17:09 ` Eli Zaretskii
2017-03-02 14:56 ` [Patch v5] " Marc-Andre Laperle
2017-03-09 19:12 ` Marc-André Laperle
2017-03-17 16:55 ` Pedro Alves
2017-03-20 19:07 ` Marc-André Laperle
2017-02-03 17:16 ` [PATCH v2 1/3] Remove unused parameter in solib_add and update_solib_list Marc-Andre Laperle
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=ddab3a27-8fd3-ba10-c7e1-27ea8b85efa5@ericsson.com \
--to=marc-andre.laperle@ericsson.com \
--cc=gdb-patches@sourceware.org \
--cc=palves@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).