public inbox for archer-commits@sourceware.org
help / color / mirror / Atom feed
From: tromey@sourceware.org
To: archer-commits@sourceware.org
Subject: [SCM] tromey/cleanup-checker: introduce scoped cleanups
Date: Mon, 13 May 2013 17:45:00 -0000 [thread overview]
Message-ID: <20130513174506.23441.qmail@sourceware.org> (raw)
The branch, tromey/cleanup-checker has been updated
discards 9b1fe3dce529c03c56b26b92eadbac1f866ce955 (commit)
discards 28bbcf20b7b4a0f597237b56a7eb8793c65edbed (commit)
discards b742ab349892cce15df6d761190d17eb27a23976 (commit)
discards 72b491c4ccd7996cf471b85b346616f207e69760 (commit)
discards f1ed049596b4deed84518830462a7ffbd371979c (commit)
discards 09983ec5522012abad22c2484ede6d58c346cf26 (commit)
discards c76ac86917cfd5f8d783730da496285ac7d956d2 (commit)
discards cdcd1ee9c3501362c7071ec6b0e818d5ef775809 (commit)
discards bb421a989e7dc3696315823b8918092f0ee174c8 (commit)
discards 6d00e10eddf7b9b5594aedb29d2e2135eb116800 (commit)
discards 09f8d24b18b1f4beea87459625e0ce2a6bf03992 (commit)
discards a0b282774d0e5948ea2655ac2ec496cacea7a9ba (commit)
discards 29862d1dbc5bb0ffb1057fe4775efc0d0c83bdf0 (commit)
discards 27ac4476e9e4f7a238ca9f86ec955ef6d2d21c02 (commit)
discards 9de1593e2a53e2414bfb06ccd238a2a4adc148be (commit)
discards 22b53119e9c8e26a32ff9300d698725e29cd123c (commit)
discards 5bd523b8ded5668a3e8ad94f2951300d4eebcecf (commit)
discards 8151eef178cd96ab3106cccaa3afc8ece5b69d71 (commit)
discards 1ebfefe7d65c7532c9e13534b43915d6c0f3661e (commit)
discards f839ed581f8468feb9294637cf8e8367de36080c (commit)
discards 0c04640cda1b579c51d4e8025be9202a374f040b (commit)
discards 0567c8d5e10110e8c6590c0b1ca9bcfad6500f49 (commit)
discards 76ff0a4abfd2d95bee57e14b151c26f4ad42e80c (commit)
discards f5a385fdac093cd96ac660b38a2a9ba199b98044 (commit)
discards ab1481fe59388b02c51c672114261bcb40261ca7 (commit)
discards 4904a5c8b2929216849840bc807a85ea30300176 (commit)
discards 568e3554958cb42488bc51f5e2bd8b5cd817803c (commit)
discards 9950c2e4c9891c1b716d6c6d32107ddd7e97c5a5 (commit)
via 212f7dae72349005357393d8c3b80e88c2e2c46b (commit)
via ac28d0b42cadec26606c351c4dc0da64b4ad7cc9 (commit)
via e13ea91044ecb6b7e11ef6a0948a6344ffd74c7b (commit)
via b5934aa76217694c13b6942ceb8cdc1347b3fe90 (commit)
via b908d1f6d6bb9abe29ae0d51b50590ef2b05b2a8 (commit)
via 62c38fe0b1ff8bf404a80e3fdcaf7674447342d5 (commit)
via 68aa964f224cd625de302c7a92cf04002ced6011 (commit)
via 3126887c2a009b84551f6928146274eb257d299c (commit)
via 1b6aab822d8534f9aa988c4a6decebbbc7cc9682 (commit)
via cd87ddaa4a8f00efdda393d1e46c97026061c60f (commit)
via 0e4714eb136cb739e765b48bf92577f4d97a845f (commit)
via 0045f22ad9b72875d5ac8cea97490ae0d0cd5e9b (commit)
via afc715666e5fa331f966e3039b0c235155ce1277 (commit)
via d8eab1fc5ef85749c7323128f719c9f7f6d62a66 (commit)
via 2d3e0116baaa0352c711db1cbb2149a40c84fd48 (commit)
via f3940903f9784069c2b7291830fd1a5d05bc595e (commit)
via d2a26c8f60250b73c0af1dc69afd7e24d07709f5 (commit)
via 19ad7176bb73d31dcf94ec0c73739b2d3d906719 (commit)
via 41aceedec390eee91fc7ea6cb64ab83c88ae7f2a (commit)
via 581961da08a63cfee0d973a33ef7faa3c87b143f (commit)
via 84e0e8372df7f377bb9352ea0412ce8412459cc1 (commit)
via 4325d09461e89376402d4c594c34d4d51ee54eb2 (commit)
via a737657e27e1595d5b04bdbbd772be1b7a746696 (commit)
via cd06d044eb69a3e50650744f950179eeed15e3b6 (commit)
via 8aaa4bff619497e49e8a61a54df0b7f82779bdaa (commit)
via 3b154a472ffac9cea9c09221ece6af9e77b31625 (commit)
via 4fa96e4172912d0b76468717cc3571bed32dd2cd (commit)
via 7c03920f78b234c426fdb1307a8035b268d21562 (commit)
from 9b1fe3dce529c03c56b26b92eadbac1f866ce955 (commit)
Those revisions listed above that are new to this repository have
not appeared on any other notification email.
- Log -----------------------------------------------------------------
commit 212f7dae72349005357393d8c3b80e88c2e2c46b
Author: Tom Tromey <tromey@redhat.com>
Date: Thu May 9 11:44:45 2013 -0600
introduce scoped cleanups
This adds scoped cleanups to gdb. A scoped cleanup is a more
efficient, and checkable, alternative to the common idiom:
cleanup = make_cleanup (null_cleanup, NULL);
...
do_cleanups (cleanup);
... where the cleanup is always run when exiting the scope.
The efficiency comes from stack allocating the scoped cleanup. This
is probably minor. However, I've noticed myself sometimes avoiding
null cleanups on this basis, and it is nice to erase that bias.
The checking comes from using a GCC extension when available. This
check ensures that the scoped cleanup was either run or discarded when
the scope exits.
commit ac28d0b42cadec26606c351c4dc0da64b4ad7cc9
Author: Tom Tromey <tromey@redhat.com>
Date: Wed May 8 14:57:11 2013 -0600
fix up xml-support.c
xml-support.c has a function that returns a cleanup via an out parameter.
This changes this function to be a normal cleanup constructor --
returning the cleanup directly and returning the other result via an
out parameter.
This is sort of a hack, but it lets the checker work here.
I also noticed that gdb_xml_create_parser_and_cleanup is not really
needed, so I removed it.
* xml-support.c (gdb_xml_create_parser_and_cleanup_1): Return
a cleanup. Remove 'old_chain' argument. Add 'parser_result'
argument.
(gdb_xml_create_parser_and_cleanup): Remove.
(gdb_xml_parse_quick): Use gdb_xml_create_parser_and_cleanup_1.
(xml_process_xincludes): Update.
* xml-support.h (gdb_xml_create_parser_and_cleanup): Don't
declare.
commit e13ea91044ecb6b7e11ef6a0948a6344ffd74c7b
Author: Tom Tromey <tromey@redhat.com>
Date: Wed May 8 14:52:26 2013 -0600
fix compile_rx_or_error
compile_rx_or_error looks like a constructor, but it can return NULL.
This patch changes it to remove the NULL return, making it work
like any other cleanup constructor.
This is a stylistic patch but I think it is also better for code to
follow the normal conventions.
* probe.c (collect_probes): Check arguments for NULL before
calling compile_rx_or_error.
* utils.c (compile_rx_or_error): Require 'rx' to be non-NULL.
Remove NULL return.
commit b5934aa76217694c13b6942ceb8cdc1347b3fe90
Author: Tom Tromey <tromey@redhat.com>
Date: Tue May 7 11:57:54 2013 -0600
some fixes to infrun.c
This fixes some of the problems in infrun.c that the checker reported.
I filed the remaining problems as bugs.
This patch is purely stylistic.
* infrun.c (handle_vfork_child_exec_or_exit)
(fetch_inferior_event, adjust_pc_after_break): Introduce
an outer null cleanup.
commit b908d1f6d6bb9abe29ae0d51b50590ef2b05b2a8
Author: Tom Tromey <tromey@redhat.com>
Date: Tue May 7 11:36:46 2013 -0600
fix breakpoint_1
This is a stylistic change to make breakpoint_1 analyzable.
The issue here is maybe fixable in the analyzer with some work.
* breakpoint.c (breakpoint_1): Introduce an outer null cleanup.
commit 62c38fe0b1ff8bf404a80e3fdcaf7674447342d5
Author: Tom Tromey <tromey@redhat.com>
Date: Tue May 7 11:34:59 2013 -0600
introduce dangling_cleanup attribute and change source to use it
Some functions leak cleanups on purpose. However, these functions
aren't "constructors" in the cleanup-checker sense, as they do not
return the cleanup.
This patch adds an attribute that can be used to mark these functions.
This avoids a bunch of false reports from the checker.
This patch is really a huge hack. It is basically giving up on
cleanly checking the use of cleanups. You can see this most easily by
noticing that I elected not to have this property propagate through
the call chain.
Still, I figured I'd see what people thought of this patch. Without
it the checker is harder to use, as you must know which functions are
ignorable.
Maybe some of these are fixable with a bit more thought.
* ada-lang.c (old_renaming_is_invisible): Use DANGLING_CLEANUP.
* breakpoint.c (breakpoint_re_set_default): Use DANGLING_CLEANUP.
* c-exp.y (operator_stoken): Use DANGLING_CLEANUP.
* coffread.c (coff_locate_sections): Use DANGLING_CLEANUP.
* contrib/cleanup_check.py (leaks_cleanup): New global.
(note_dangling_cleanup, leaves_dangling_cleanup): New functions.
(CleanupChecker.compute_master): Add comment. Check
leaves_dangling_cleanup.
(CleanupChecker.check_cleanups): Don't mention special constructors
to the user.
(register_attributes): New function. Register it with gcc.
* defs.h (DANGLING_CLEANUP): New define.
* dwarf2read.c (read_cutu_die_from_dwo, handle_DW_AT_stmt_list)
(setup_type_unit_groups, dwarf2_add_field, dwarf2_add_typedef)
(dwarf2_add_member_fn, psymtab_include_file_name)
(dwarf_decode_lines): Use DANGLING_CLEANUP.
* linux-nat.c (linux_child_pid_to_exec_file): Use DANGLING_CLEANUP.
* python/py-prettyprint.c (push_dummy_python_frame): Use
DANGLING_CLEANUP.
* stabsread.c (read_member_functions, read_struct_fields)
(read_baseclasses): Use DANGLING_CLEANUP.
* utils.h (gdb_bfd_errmsg): Use DANGLING_CLEANUP.
commit 68aa964f224cd625de302c7a92cf04002ced6011
Author: Tom Tromey <tromey@redhat.com>
Date: Tue May 7 10:55:56 2013 -0600
fix mi-cmd-var.c
This is a stylistic change in mi-cmd-var.c that adds outer cleanups
where needed by the checker.
* mi/mi-cmd-var.c (mi_cmd_var_list_children): Add an outer
null cleanup.
(mi_cmd_var_update, varobj_update_one): Likewise.
commit 3126887c2a009b84551f6928146274eb257d299c
Author: Tom Tromey <tromey@redhat.com>
Date: Tue May 7 10:54:48 2013 -0600
fix cli-script.c
read_command_lines_1 had some (IMNSHO) spaghetti-ish code for cleanup
handling. This makes the code much simpler to understand, by
introducing an outer cleanup.
This is another case where a stylistic change for the checker is also
nice for the reader.
* cli/cli-script.c (read_command_lines_1): Use a null cleanup
for 'old_chain'. Do not check 'head' before processing
cleanups.
commit 1b6aab822d8534f9aa988c4a6decebbbc7cc9682
Author: Tom Tromey <tromey@redhat.com>
Date: Tue May 7 09:23:12 2013 -0600
fix mi-cmd-stack.c
mi-cmd-stack.d had a conditional cleanup, "cleanup_tuple" that
confused the checker. However, there was no need for this, since it
was only used via do_cleanups at the end of the function, just before
another call to do_cleanups.
So, while this is a stylistic patch for the checker, I also consider
it a generic improvement for readers of the code.
* mi/mi-cmd-stack.c (list_arg_or_local): Remove
"cleanup_tuple".
commit cd87ddaa4a8f00efdda393d1e46c97026061c60f
Author: Tom Tromey <tromey@redhat.com>
Date: Tue May 7 09:03:52 2013 -0600
fix dbxread.c
This is a stylistic change to make some code in dbxread.c analyzable
by the checker.
* dbxread.c (dbx_read_symtab): Declare 'back_to' in a more
inner scope. Unconditionally call do_cleanups.
commit 0e4714eb136cb739e765b48bf92577f4d97a845f
Author: Tom Tromey <tromey@redhat.com>
Date: Tue May 7 09:01:51 2013 -0600
fix source.c
find_and_open_source can leak a cleanup.
* source.c (find_and_open_source): Call do_cleanups.
commit 0045f22ad9b72875d5ac8cea97490ae0d0cd5e9b
Author: Tom Tromey <tromey@redhat.com>
Date: Tue May 7 08:53:19 2013 -0600
fix linux-thread-db.c
This is a stylistic change to make it so the checker can analyze a
function in linux-thread-db.c.
* linux-thread-db.c (thread_db_load_search): Unconditionally
call do_cleanups.
commit afc715666e5fa331f966e3039b0c235155ce1277
Author: Tom Tromey <tromey@redhat.com>
Date: Mon May 6 13:47:54 2013 -0600
fix in solib-aix.c
solib_aix_bfd_open has an early "return" that doesn't run cleanups.
This fixes the problem by dropping the null_cleanup and using a later
cleanup as the master cleanup for the function.
* solib-aix.c (solib_aix_bfd_open): Don't use a null cleanup
for 'cleanup'; instead use a later one.
commit d8eab1fc5ef85749c7323128f719c9f7f6d62a66
Author: Tom Tromey <tromey@redhat.com>
Date: Thu Mar 7 07:43:40 2013 -0700
use explicit returns to avoid checker confusion
The checker does not understand the idiom
if (except.reason < 0) {
do_cleanups (whatever);
GDB_PY_HANDLE_EXCEPTION (except);
}
because it doesn't realize that the nested 'if' actually has the same
condition.
This fixes instances of this to be more explicit.
* python/py-breakpoint.c (bppy_get_commands): Use
explicit, unconditional return.
* python/py-frame.c (frapy_read_var): Likewise.
* python/python.c (gdbpy_decode_line): Likewise.
commit 2d3e0116baaa0352c711db1cbb2149a40c84fd48
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Mar 5 14:43:27 2013 -0700
fix cp-namespace.c
cp_lookup_symbol_imports_or_template could return without
running cleanups.
* cp-namespace.c (cp_lookup_symbol_imports_or_template): Call
do_cleanups on all return paths.
commit f3940903f9784069c2b7291830fd1a5d05bc595e
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Mar 5 14:36:21 2013 -0700
fix top.c
execute_command can leak a cleanup along one return path.
* top.c (execute_command): Discard 'cleanup_if_error' cleanups.
commit d2a26c8f60250b73c0af1dc69afd7e24d07709f5
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Mar 5 14:33:18 2013 -0700
fix one bug in stabsread.c
Some code in stabsread.c can return without running cleanups.
* stabsread.c (read_struct_type): Call do_cleanups along
all return paths.
commit 19ad7176bb73d31dcf94ec0c73739b2d3d906719
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Mar 5 14:31:14 2013 -0700
fix mipsread.c
Some code in mipsread.c could leak cleanups along some return paths.
* mipsread.c (read_alphacoff_dynamic_symtab): Call do_cleanups
along all return paths.
commit 41aceedec390eee91fc7ea6cb64ab83c88ae7f2a
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Mar 5 14:22:40 2013 -0700
fix one bug in symfile.c
find_separate_debug_file could leak a cleanup along some return paths.
* symfile.c (find_separate_debug_file): Call do_cleanups
along all return paths.
commit 581961da08a63cfee0d973a33ef7faa3c87b143f
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Mar 5 14:21:16 2013 -0700
fix symtab.c
search_symbols had some bad code resulting in a cleanup being both
discarded and run.
* symtab.c (search_symbols): Introduce a null cleanup for
'retval_chain'.
commit 84e0e8372df7f377bb9352ea0412ce8412459cc1
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Mar 5 14:13:35 2013 -0700
fix py-value.c
Some code in py-value.c could exit a loop without running some
cleanups made in the loop.
* python/py-value.c (valpy_binop): Call do_cleanups before
exiting loop.
commit 4325d09461e89376402d4c594c34d4d51ee54eb2
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Mar 5 14:11:25 2013 -0700
fix py-prettyprint.c
print_children, in py-prettyprint.c, could call do_cleanups twice on
the same cleanup.
* python/py-prettyprint.c (print_children): Remove extra
do_cleanups call.
commit a737657e27e1595d5b04bdbbd772be1b7a746696
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Mar 5 14:10:29 2013 -0700
fix py-frame.c
A couple return paths in frapy_read_var were missing do_cleanups calls.
* python/py-frame.c (frapy_read_var): Call do_cleanups along
all return paths.
commit cd06d044eb69a3e50650744f950179eeed15e3b6
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Mar 5 14:08:39 2013 -0700
fix py-breakpoint.c
One return path in bppy_get_commands was missing a do_cleanups call.
* python/py-breakpoint.c (bppy_get_commands): Call do_cleanups
along all return paths.
commit 8aaa4bff619497e49e8a61a54df0b7f82779bdaa
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Mar 5 14:05:33 2013 -0700
simplify cli-logging.c for analysis
This is another stylistic patch. It changes cli-logging.c to be
analyzable by the checker, again following the method of adding an
outer cleanup and unconditionally calling do_cleanups.
* cli/cli-logging.c (set_logging_redirect): Unconditionally
call do_cleanups.
commit 3b154a472ffac9cea9c09221ece6af9e77b31625
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Mar 5 14:04:14 2013 -0700
fix varobj.c
c_value_of_root is missing a call to do_cleanups at one return.
This fixes the problem by removing that return and letting control
fall through.
* varobj.c (c_value_of_root): Call do_cleanups along all
return paths.
commit 4fa96e4172912d0b76468717cc3571bed32dd2cd
Author: Tom Tromey <tromey@redhat.com>
Date: Tue Mar 5 14:00:20 2013 -0700
make a cleanup unconditionally in tracepoint.c
This is another cosmetic patch. It introduces an "outer" cleanup in
trace_dump_command and arranges to unconditionally call do_cleanups.
This lets the checker analyze the function.
* tracepoint.c (trace_dump_command): Unconditionally call
do_cleanups.
commit 7c03920f78b234c426fdb1307a8035b268d21562
Author: Tom Tromey <tromey@redhat.com>
Date: Mon Mar 4 14:08:50 2013 -0700
fix two buglets in breakpoint.c
First, output_thread_groups leaks a cleanup along one return path.
Second, parse_cmd_to_aexpr could return without running its cleanups,
if there was an exception in a TRY_CATCH.
* breakpoint.c (output_thread_groups, parse_cmd_to_aexpr): Call
do_cleanups earlier.
-----------------------------------------------------------------------
Summary of changes:
gdb/breakpoint.c | 10 ++++------
gdb/utils.c | 2 ++
2 files changed, 6 insertions(+), 6 deletions(-)
First 500 lines of diff:
diff --git a/gdb/breakpoint.c b/gdb/breakpoint.c
index a9f1921..47539f1 100644
--- a/gdb/breakpoint.c
+++ b/gdb/breakpoint.c
@@ -5805,8 +5805,7 @@ output_thread_groups (struct ui_out *uiout,
VEC(int) *inf_num,
int mi_only)
{
- struct cleanup *back_to = make_cleanup_ui_out_list_begin_end (uiout,
- field_name);
+ struct cleanup *back_to;
int is_mi = ui_out_is_mi_like_p (uiout);
int inf;
int i;
@@ -5814,10 +5813,9 @@ output_thread_groups (struct ui_out *uiout,
/* For backward compatibility, don't display inferiors in CLI unless
there are several. Always display them for MI. */
if (!is_mi && mi_only)
- {
- do_cleanups (back_to);
- return;
- }
+ return;
+
+ back_to = make_cleanup_ui_out_list_begin_end (uiout, field_name);
for (i = 0; VEC_iterate (int, inf_num, i, inf); ++i)
{
diff --git a/gdb/utils.c b/gdb/utils.c
index fa54e10..18ee9bb 100644
--- a/gdb/utils.c
+++ b/gdb/utils.c
@@ -1135,6 +1135,8 @@ compile_rx_or_error (regex_t *pattern, const char *rx, const char *message)
{
int code;
+ gdb_assert (rx != NULL);
+
code = regcomp (pattern, rx, REG_NOSUB);
if (code != 0)
{
hooks/post-receive
--
Repository for Project Archer.
next reply other threads:[~2013-05-13 17:45 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-05-13 17:45 tromey [this message]
-- strict thread matches above, loose matches on Subject: below --
2013-05-14 20:42 tromey
2013-05-09 17:56 tromey
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=20130513174506.23441.qmail@sourceware.org \
--to=tromey@sourceware.org \
--cc=archer-commits@sourceware.org \
/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).