From: jose.marchesi@oracle.com (Jose E. Marchesi)
To: gdb-patches@sourceware.org
Subject: Re: [PATCH] hardware watchpoints turned off, inferior not yet started
Date: Thu, 07 Nov 2013 12:21:00 -0000 [thread overview]
Message-ID: <87fvr8xwo8.fsf@oracle.com> (raw)
[Following https://sourceware.org/ml/gdb-patches/2013-10/msg00563.html]
Hi.
The patch posted above introduced a regression in sparc64-*-linux-gnu,
and possibly in other targets not supporting hardware watchpoints at
all.
Today I was surprised to find this in a sparc64-*-linux-gnu target:
$ gdb foo
... intro text ...
(gdb) watch global_var
Hardware watchpoint 1: global_var
The cause for this regression is this code in update_watchpoint
(breakpoint.c):
+ if (!target_has_execution)
{
/* Without execution, memory can't change. No use to try and
set watchpoint locations. The watchpoint will be reset when
the target gains execution, through breakpoint_re_set. */
+ if (!can_use_hw_watchpoints)
+ {
+ if (b->base.ops->works_in_software_mode (&b->base))
+ b->base.type = bp_watchpoint;
+ else
+ error (_("Software read/access watchpoints not supported."));
+ }
The watchpoint must be downgraded to a software watchpoint if the target
does not support hw watchpoints, even if can_use_hw_watchpoints is 1.
The following patch fixes this and also adds an additional test to
testsuite/watchpoints.exp to catch this problem.
2013-11-07 Jose E. Marchesi <jose.marchesi@oracle.com>
* breakpoint.c (update_watchpoint): Downgrade hw watchpoints to sw
watchpoints in targets not supporting them, when the inferior is
not running.
2013-11-07 Jose E. Marchesi <jose.marchesi@oracle.com>
* gdb.base/watchpoints.exp: Add a test to ensure that a watchpoint
is not created as a hw watchpoint in targets not supporting them,
even if can-use-hw-watchpoints is 1 when the inferior is not
running.
diff --git a/gdb/breakpoint.c b/gdb/breakpoint.c
index ffe73fd..597e6f9 100644
--- a/gdb/breakpoint.c
+++ b/gdb/breakpoint.c
@@ -1800,7 +1800,10 @@ update_watchpoint (struct watchpoint *b, int reparse)
/* Without execution, memory can't change. No use to try and
set watchpoint locations. The watchpoint will be reset when
the target gains execution, through breakpoint_re_set. */
- if (!can_use_hw_watchpoints)
+ int i = hw_breakpoint_used_count ();
+ int target_supports_hw_watchpoints =
+ target_can_use_hardware_watchpoint (bp_hardware_watchpoint, i + 1, 0);
+ if (!target_supports_hw_watchpoints || !can_use_hw_watchpoints)
{
if (b->base.ops->works_in_software_mode (&b->base))
b->base.type = bp_watchpoint;
--- a/gdb/testsuite/gdb.base/watchpoints.exp
+++ b/gdb/testsuite/gdb.base/watchpoints.exp
@@ -30,6 +30,15 @@ if {[prepare_for_testing $testfile.exp $testfile $srcfile debug]} {
}
with_test_prefix "before inferior start" {
+ if [target_info exists gdb,no_hardware_watchpoints] {
+ # Ensure that a watchpoint is not created as a hw watchpoint
+ # even if can-use-hw-watchpoints is 1 when the inferior is not
+ # running.
+ gdb_test_no_output "set can-use-hw-watchpoints 1"
+ gdb_test "watch ival1" "Watchpoint \[0-9\]+: ival1" \
+ "create sw watchpoint"
+ }
+
# Ensure that if we turn off hardware watchpoints and set a watch point
# before starting the inferior the watchpoint created will not be a
# hardware watchpoint.
next reply other threads:[~2013-11-07 11:43 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-11-07 12:21 Jose E. Marchesi [this message]
2013-11-08 12:03 ` Pedro Alves
-- strict thread matches above, loose matches on Subject: below --
2013-10-16 9:39 Andrew Burgess
2013-10-16 12:32 ` Pedro Alves
2013-10-16 15:43 ` Andrew Burgess
2013-10-16 16:28 ` Pedro Alves
2013-10-18 9:14 ` Andrew Burgess
2013-10-18 14:37 ` Andrew Burgess
2013-10-18 15:33 ` Pedro Alves
2013-10-18 16:26 ` Andrew Burgess
2013-10-18 15:26 ` Pedro Alves
2013-10-18 15:43 ` Andrew Burgess
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=87fvr8xwo8.fsf@oracle.com \
--to=jose.marchesi@oracle.com \
--cc=gdb-patches@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