FreeChainXenon/binutils-gdb
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

2011-06-06 Pedro Alves <pedro@codesourcery.com>

Browse source 
gdb/
	* infcall.c (run_inferior_call): Don't mask async.  Instead force
	a synchronous wait, if the target can async.

	* target.h (struct target_ops): Delete to_async_mask.
	(target_async_mask): Delete.
	* target.c (update_current_target): Delete references to to_async_mask.
	* linux-nat.c (linux_nat_async_mask_value): Delete.
	(linux_nat_is_async_p, linux_nat_can_async_p): Remove references
	to linux_nat_async_mask_value.
	(linux_nat_async_mask): Delete.
	(linux_nat_async, linux_nat_close): Remove references to
	linux_nat_async_mask_value.
	* record.c (record_async_mask_value): Delete.
	(record_async): Remove references to record_async_mask_value.
	(record_async_mask): Delete.
	(record_can_async_p, record_is_async_p): Remove references to
	record_async_mask_value.
	(init_record_ops, init_record_core_ops): Remove references to
	record_async_mask.
	* remote.c (remote_async_mask_value): Delete.
	(init_remote_ops): Remove reference to remote_async_mask.
	(remote_can_async_p, remote_is_async_p): Remove references to
	remote_async_mask_value.
	(remote_async): Remove references to remote_async_mask_value.
	(remote_async_mask): Delete.

	* infrun.c (fetch_inferior_event): Don't claim registers changed
	if the current thread is already not executing.
This commit is contained in:
Pedro Alves 2011-06-06 12:47:07 +00:00
parent 3c0013bf3b
commit 3dd5b83d5b
8 changed files with 61 additions and 138 deletions
Download patch file Download diff file Expand all files Collapse all files

11
gdb/infrun.c
View file

@ -2739,7 +2739,16 @@ fetch_inferior_event (void *client_data)
status mechanism. */
overlay_cache_invalid = 1;
registers_changed ();
/* But don't do it if the current thread is already stopped (hence
this is either a delayed event that will result in
TARGET_WAITKIND_IGNORE, or it's an event for another thread (and
we always clear the register and frame caches when the user
switches threads anyway). If we didn't do this, a spurious
delayed event in all-stop mode would make the user lose the
selected frame. */
if (non_stop || is_executing (inferior_ptid))
registers_changed ();
make_cleanup_restore_integer (&execution_direction);
execution_direction = target_execution_direction ();