From: Gary Benson <gbenson@redhat.com>
To: Tom Tromey <tromey@redhat.com>
Cc: gdb-patches@sourceware.org
Subject: [RFA 2/7 take 2] API for inhibiting section map updates
Date: Fri, 24 May 2013 07:47:00 -0000 [thread overview]
Message-ID: <20130524074737.GB4602@blade.nx> (raw)
In-Reply-To: <87mwrpohnv.fsf@fleche.redhat.com>
[-- Attachment #1: Type: text/plain, Size: 1028 bytes --]
Tom Tromey wrote:
> >>>>> "Gary" == Gary Benson <gbenson@redhat.com> writes:
>
> Gary> +void
> Gary> +inhibit_section_map_updates (void)
> Gary> +{
> Gary> + get_objfile_pspace_data (current_program_space)->inhibit_updates = 1;
> Gary> +}
> Gary> +
> Gary> +/* See comments in objfiles.h. */
> Gary> +
> Gary> +void
> Gary> +resume_section_map_updates (void)
> Gary> +{
> Gary> + get_objfile_pspace_data (current_program_space)->inhibit_updates = 0;
> Gary> +}
> Gary> +
> Gary> +/* See comments in objfiles.h. */
> Gary> +
> Gary> +void
> Gary> +resume_section_map_updates_cleanup (void *arg)
> Gary> +{
> Gary> + resume_section_map_updates ();
> Gary> }
>
> It seems like it might be slightly safer if these all took the
> program space as an argument. That way if something happens to
> change the current program space, then, e.g., the cleanup will
> still affect the program space for which it was made.
>
> What do you think of that?
That's an excellent idea. The attached patch implements this.
Cheers,
Gary
[-- Attachment #2: rtld-probes-2-inhibit-sm-updates.patch --]
[-- Type: text/plain, Size: 6026 bytes --]
2013-05-23 Gary Benson <gbenson@redhat.com>
* objfiles.h (inhibit_section_map_updates): New function
declaration.
(resume_section_map_updates): Likewise.
(resume_section_map_updates_cleanup): Likewise.
* objfiles.c (objfile_pspace_info): Removed field
"objfiles_changed_p". New fields "new_objfiles_available",
"section_map_dirty" and "inhibit_updates".
(allocate_objfile): Set new_objfiles_available.
(free_objfile): Set section_map_dirty.
(objfile_relocate1): Likewise.
(in_plt_section): Likewise.
(find_pc_section): Update the conditions under which the
section map will be updated.
(inhibit_section_map_updates): New function.
(resume_section_map_updates): Likewise.
(resume_section_map_updates_cleanup): Likewise.
diff --git a/gdb/objfiles.h b/gdb/objfiles.h
index 93149e2..adb1ef8 100644
--- a/gdb/objfiles.h
+++ b/gdb/objfiles.h
@@ -501,6 +501,22 @@ extern int in_plt_section (CORE_ADDR, char *);
modules. */
DECLARE_REGISTRY(objfile);
+/* In normal use, the section map will be rebuilt by find_pc_section
+ if objfiles have been added, removed or relocated since it was last
+ called. Calling inhibit_section_map_updates will inhibit this
+ behavior until resume_section_map_updates is called. If you call
+ inhibit_section_map_updates you must ensure that every call to
+ find_pc_section in the inhibited region relates to a section that
+ is already in the section map and has not since been removed or
+ relocated. */
+extern void inhibit_section_map_updates (struct program_space *pspace);
+
+/* Resume automatically rebuilding the section map as required. */
+extern void resume_section_map_updates (struct program_space *pspace);
+
+/* Version of the above suitable for use as a cleanup. */
+extern void resume_section_map_updates_cleanup (void *arg);
+
extern void default_iterate_over_objfiles_in_search_order
(struct gdbarch *gdbarch,
iterate_over_objfiles_in_search_order_cb_ftype *cb,
diff --git a/gdb/objfiles.c b/gdb/objfiles.c
index 3e49ea2..8842131 100644
--- a/gdb/objfiles.c
+++ b/gdb/objfiles.c
@@ -67,9 +67,18 @@ struct objfile *rt_common_objfile; /* For runtime common symbols */
struct objfile_pspace_info
{
- int objfiles_changed_p;
struct obj_section **sections;
int num_sections;
+
+ /* Nonzero if object files have been added since the section map
+ was last updated. */
+ int new_objfiles_available;
+
+ /* Nonzero if the section map MUST be updated before use. */
+ int section_map_dirty;
+
+ /* Nonzero if section map updates should be inhibited if possible. */
+ int inhibit_updates;
};
/* Per-program-space data key. */
@@ -317,7 +326,7 @@ allocate_objfile (bfd *abfd, int flags)
objfile->flags |= flags;
/* Rebuild section map next time we need it. */
- get_objfile_pspace_data (objfile->pspace)->objfiles_changed_p = 1;
+ get_objfile_pspace_data (objfile->pspace)->new_objfiles_available = 1;
return objfile;
}
@@ -646,7 +655,7 @@ free_objfile (struct objfile *objfile)
obstack_free (&objfile->objfile_obstack, 0);
/* Rebuild section map next time we need it. */
- get_objfile_pspace_data (objfile->pspace)->objfiles_changed_p = 1;
+ get_objfile_pspace_data (objfile->pspace)->section_map_dirty = 1;
xfree (objfile);
}
@@ -826,7 +835,7 @@ objfile_relocate1 (struct objfile *objfile,
}
/* Rebuild section map next time we need it. */
- get_objfile_pspace_data (objfile->pspace)->objfiles_changed_p = 1;
+ get_objfile_pspace_data (objfile->pspace)->section_map_dirty = 1;
/* Update the table in exec_ops, used to read memory. */
ALL_OBJFILE_OSECTIONS (objfile, s)
@@ -1291,11 +1300,14 @@ static void
update_section_map (struct program_space *pspace,
struct obj_section ***pmap, int *pmap_size)
{
+ struct objfile_pspace_info *pspace_info;
int alloc_size, map_size, i;
struct obj_section *s, **map;
struct objfile *objfile;
- gdb_assert (get_objfile_pspace_data (pspace)->objfiles_changed_p != 0);
+ pspace_info = get_objfile_pspace_data (current_program_space);
+ gdb_assert (pspace_info->section_map_dirty != 0
+ || pspace_info->new_objfiles_available != 0);
map = *pmap;
xfree (map);
@@ -1365,7 +1377,9 @@ find_pc_section (CORE_ADDR pc)
return s;
pspace_info = get_objfile_pspace_data (current_program_space);
- if (pspace_info->objfiles_changed_p != 0)
+ if (pspace_info->section_map_dirty
+ || (pspace_info->new_objfiles_available
+ && !pspace_info->inhibit_updates))
{
update_section_map (current_program_space,
&pspace_info->sections,
@@ -1373,7 +1387,8 @@ find_pc_section (CORE_ADDR pc)
/* Don't need updates to section map until objfiles are added,
removed or relocated. */
- pspace_info->objfiles_changed_p = 0;
+ pspace_info->new_objfiles_available = 0;
+ pspace_info->section_map_dirty = 0;
}
/* The C standard (ISO/IEC 9899:TC2) requires the BASE argument to
@@ -1414,14 +1429,38 @@ in_plt_section (CORE_ADDR pc, char *name)
}
\f
-/* Set objfiles_changed_p so section map will be rebuilt next time it
+/* Set section_map_dirty so section map will be rebuilt next time it
is used. Called by reread_symbols. */
void
objfiles_changed (void)
{
/* Rebuild section map next time we need it. */
- get_objfile_pspace_data (current_program_space)->objfiles_changed_p = 1;
+ get_objfile_pspace_data (current_program_space)->section_map_dirty = 1;
+}
+
+/* See comments in objfiles.h. */
+
+void
+inhibit_section_map_updates (struct program_space *pspace)
+{
+ get_objfile_pspace_data (pspace)->inhibit_updates = 1;
+}
+
+/* See comments in objfiles.h. */
+
+void
+resume_section_map_updates (struct program_space *pspace)
+{
+ get_objfile_pspace_data (pspace)->inhibit_updates = 0;
+}
+
+/* See comments in objfiles.h. */
+
+void
+resume_section_map_updates_cleanup (void *arg)
+{
+ resume_section_map_updates (arg);
}
/* The default implementation for the "iterate_over_objfiles_in_search_order"
next prev parent reply other threads:[~2013-05-24 7:47 UTC|newest]
Thread overview: 59+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-05-16 14:43 [RFA 0/7] Improved linker-debugger interface Gary Benson
2013-05-16 14:47 ` [RFA 1/7] Probes API convenience patch Gary Benson
2013-05-16 14:48 ` [RFA 4/7] GDB support for new gdbserver functionality Gary Benson
2013-05-16 14:48 ` [RFA 3/7] New " Gary Benson
2013-05-16 18:18 ` Tom Tromey
2013-05-24 7:46 ` [RFA 3/7 take 2] " Gary Benson
2013-05-25 21:05 ` Jan Kratochvil
2013-05-26 2:45 ` Eli Zaretskii
2013-05-29 18:50 ` Pedro Alves
2013-05-30 9:38 ` Gary Benson
2013-05-30 10:40 ` Pedro Alves
2013-05-30 10:54 ` Gary Benson
2013-05-30 16:31 ` Eli Zaretskii
2013-05-30 17:22 ` Gary Benson
2013-05-16 14:48 ` [RFA 2/7] API for inhibiting section map updates Gary Benson
2013-05-20 14:22 ` Tom Tromey
2013-05-24 7:47 ` Gary Benson [this message]
2013-05-24 14:18 ` [RFA 2/7 take 2] " Tom Tromey
2013-05-29 17:18 ` Pedro Alves
2013-05-30 9:12 ` Gary Benson
2013-05-16 14:55 ` [RFA 6/7] Linker-debugger interface tests by Jan Gary Benson
2013-05-29 19:06 ` Pedro Alves
2013-05-30 9:19 ` Gary Benson
2013-05-16 14:55 ` [RFA 5/7] Improved linker-debugger interface Gary Benson
2013-05-17 19:05 ` Jan Kratochvil
2013-05-24 8:30 ` [RFA 5/7 take 2] " Gary Benson
2013-05-25 21:05 ` Jan Kratochvil
2013-05-29 18:51 ` Pedro Alves
2013-05-30 10:43 ` [RFA 5/7 take 3] " Gary Benson
2013-05-30 17:18 ` Pedro Alves
2013-05-31 13:22 ` [RFA 5/7 take 4] " Gary Benson
2013-05-31 13:27 ` Pedro Alves
2013-06-03 10:31 ` Pedro Alves
2013-06-03 16:37 ` Jan Kratochvil
2013-06-03 17:28 ` Pedro Alves
2013-06-04 11:33 ` Gary Benson
2013-05-16 14:55 ` [RFA 7/7] Linker-debugger interface tests Gary Benson
2013-05-19 13:45 ` Jan Kratochvil
2013-05-19 16:43 ` Jan Kratochvil
2013-05-24 8:38 ` [RFA 7/7 take 2] " Gary Benson
2013-05-24 8:58 ` Jan Kratochvil
2013-05-24 14:05 ` [obv] Fix excessive backslashes in testsuite Gary Benson
2013-05-25 21:06 ` [RFA 7/7 take 2] Linker-debugger interface tests Jan Kratochvil
2013-05-16 17:33 ` [RFA 0/7] Improved linker-debugger interface Tom Tromey
2013-05-16 18:53 ` Gary Benson
2013-06-06 9:00 ` Gary Benson
2013-05-19 13:46 ` Jan Kratochvil
2013-05-29 19:08 ` Pedro Alves
2013-06-04 13:38 ` [commit] " Gary Benson
2013-06-25 21:04 ` Joel Brobecker
2013-06-25 22:03 ` Sergio Durigan Junior
2013-06-26 0:49 ` Joel Brobecker
2013-07-09 8:41 ` Gary Benson
[not found] ` <20130708104719.GA11176@blade.nx>
2013-07-09 14:45 ` Joel Brobecker
2013-06-26 15:38 ` Tom Tromey
2013-06-26 17:23 ` Sergio Durigan Junior
2013-06-26 19:15 ` Joel Brobecker
2013-06-27 23:33 ` Tom Tromey
2013-06-30 3:12 ` Sergio Durigan Junior
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=20130524074737.GB4602@blade.nx \
--to=gbenson@redhat.com \
--cc=gdb-patches@sourceware.org \
--cc=tromey@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