On Thu, Dec 05, 2024 at 05:12:47PM -0500, Peter Xu wrote:
> In case if I missed it, a runnable patch would work to clarify.
Ohhh no need now, I see what you meant.
But then you'll really need to comment p->sem, with something like:
- /* sem where to wait for more work */
+ /* sem where to wait for more work. If there's no any work, it means
+ * a local sync. */
QemuSemaphore sem;
Do you like it? I definitely don't.. because it's confusing why p->sem can
imply a sync request if we already have pending_sync. IMHO it's cleaner
when we have pending_sync, use it for all kinds of syncs.
--
Peter Xu
