Fix gdb.reverse/i386-sse-reverse.exp against gdbserver.
Pedro Alves
palves@redhat.com
Tue Jan 17 22:27:00 GMT 2012
I see:
continue
Continuing.
Cannot remove breakpoints because program is no longer writable.
Further execution is probably impossible.
Breakpoint 2, sse_test () at ../../../src/gdb/testsuite/gdb.reverse/i386-sse-reverse.c:73
73 } /* end sse_test */
(gdb) FAIL: gdb.reverse/i386-sse-reverse.exp: continue to end of sse4_test
The does does "record; b 73; c; reverse-step n times ; continue".
The breakpoint is inserted last while replaying (for the continue), which means
it wasn't inserted in the inferior/gdbserver and when "continue" reaches the
end of the record list (no longer replaying), we try to remove that said breakpoint.
Since we are no longer replaying, we try to remove it from gdbserver, which errors
out for not knowing about any such breakpoint.
This fixes it by teaching the record target to keep track of breakpoints inserted,
and keeping track if it has forwarded the breakpoint to the target beneath.
Tested on x86_64 / fedora 16. Will check in in a bit.
I think this is a necessary (but not sufficient) step to support breakpoint
always-inserted mode in the record target, btw.
gdb/
2012-01-17 Pedro Alves <palves@redhat.com>
* record.c (struct record_breakpoint, record_breakpoint_p)
(record_breakpoints): New.
(record_insert_breakpoint, record_remove_breakpoint): Manage
record breakpoints list. Only remove breakpoints from the
inferior if they had been inserted there in the first place.
---
gdb/record.c | 80 +++++++++++++++++++++++++++++++++++++++++++++++++++-------
1 files changed, 70 insertions(+), 10 deletions(-)
diff --git a/gdb/record.c b/gdb/record.c
index e396262..eb3e8e8 100644
--- a/gdb/record.c
+++ b/gdb/record.c
@@ -1718,24 +1718,68 @@ record_xfer_partial (struct target_ops *ops, enum target_object object,
offset, len);
}
-/* Behavior is conditional on RECORD_IS_REPLAY.
- We will not actually insert or remove breakpoints when replaying,
- nor when recording. */
+/* This structure represents a breakpoint inserted while the record
+ target is active. We use this to know when to install/remove
+ breakpoints in/from the target beneath. For example, a breakpoint
+ may be inserted while recording, but removed when not replaying nor
+ recording. In that case, the breakpoint had not been inserted on
+ the target beneath, so we should not try to remove it there. */
+
+struct record_breakpoint
+{
+ /* The address and address space the breakpoint was set at. */
+ struct address_space *address_space;
+ CORE_ADDR addr;
+
+ /* True when the breakpoint has been also installed in the target
+ beneath. This will be false for breakpoints set during replay or
+ when recording. */
+ int in_target_beneath;
+};
+
+typedef struct record_breakpoint *record_breakpoint_p;
+DEF_VEC_P(record_breakpoint_p);
+
+/* The list of breakpoints inserted while the record target is
+ active. */
+VEC(record_breakpoint_p) *record_breakpoints = NULL;
+
+/* Behavior is conditional on RECORD_IS_REPLAY. We will not actually
+ insert or remove breakpoints in the real target when replaying, nor
+ when recording. */
static int
record_insert_breakpoint (struct gdbarch *gdbarch,
struct bp_target_info *bp_tgt)
{
+ struct record_breakpoint *bp;
+ int in_target_beneath = 0;
+
if (!RECORD_IS_REPLAY)
{
+ /* When recording, we currently always single-step, so we didn't
+ really need to install regular breakpoints in the inferior.
+ However, we do have to insert software single-step
+ breakpoints, in case the target can't hardware step. To keep
+ things single, we always insert. */
struct cleanup *old_cleanups = record_gdb_operation_disable_set ();
int ret = record_beneath_to_insert_breakpoint (gdbarch, bp_tgt);
do_cleanups (old_cleanups);
- return ret;
+ if (ret != 0)
+ return ret;
+
+ in_target_beneath = 1;
}
+ bp = XNEW (struct record_breakpoint);
+
+ bp->addr = bp_tgt->placed_address;
+ bp->address_space = bp_tgt->placed_address_space;
+ bp->in_target_beneath = in_target_beneath;
+ VEC_safe_push (record_breakpoint_p, record_breakpoints, bp);
+
return 0;
}
@@ -1745,17 +1789,33 @@ static int
record_remove_breakpoint (struct gdbarch *gdbarch,
struct bp_target_info *bp_tgt)
{
- if (!RECORD_IS_REPLAY)
+ struct record_breakpoint *bp;
+ int ix;
+
+ for (ix = 0;
+ VEC_iterate (record_breakpoint_p, record_breakpoints, ix, bp);
+ ++ix)
{
- struct cleanup *old_cleanups = record_gdb_operation_disable_set ();
- int ret = record_beneath_to_remove_breakpoint (gdbarch, bp_tgt);
+ if (bp->addr == bp_tgt->placed_address
+ && bp->address_space == bp_tgt->placed_address_space)
+ {
+ if (bp->in_target_beneath)
+ {
+ struct cleanup *old_cleanups = record_gdb_operation_disable_set ();
+ int ret = record_beneath_to_remove_breakpoint (gdbarch, bp_tgt);
- do_cleanups (old_cleanups);
+ do_cleanups (old_cleanups);
- return ret;
+ if (ret != 0)
+ return ret;
+ }
+
+ VEC_unordered_remove (record_breakpoint_p, record_breakpoints, ix);
+ return 0;
+ }
}
- return 0;
+ gdb_assert_not_reached ("removing unknown breakpoint");
}
/* "to_can_execute_reverse" method for process record target. */
More information about the Gdb-patches
mailing list