The function target_type::deassert_reset() is called for every
target after reset is deasserted. If the target fails to get
halted, we log a warning and issue a halt request for the target
itself.
Current code calls the generic target_halt() that:
- extends the halt to all the targets in the SMP node;
- waits for targets to halt.
This breaks the logic of running target_type::deassert_reset()
per target. Plus, waiting for targets to halt delays the call of
target_type::deassert_reset() for the next targets.
Replace the call to target_halt() with the aarch64 specific
function to halt the single target. Pass the parameter HALT_LAZY
to prevent the wait for target halted.
Similar solution is already implemented for cortex_a.
Change-Id: I446dc03cb91524c6d388db485bc2388177af77b6
Signed-off-by: Antonio Borneo <borneo.antonio@gmail.com>
Reviewed-on: https://review.openocd.org/c/openocd/+/6947
Tested-by: jenkins
if (target->state != TARGET_HALTED) {
LOG_WARNING("%s: ran after reset and before halt ...",
target_name(target));
- retval = target_halt(target);
- if (retval != ERROR_OK)
- return retval;
+ if (target_was_examined(target)) {
+ retval = aarch64_halt_one(target, HALT_LAZY);
+ if (retval != ERROR_OK)
+ return retval;
+ } else {
+ target->state = TARGET_UNKNOWN;
+ }
}
}