https://github.com/mozilla-servo/libwapcaplet/commit/1c3343639571d65bc0c45cab8253dabe33aace9e introduced locking in libwapcaplet (in October 2012, by Brian Anderson). Does anyone remember why this was necessary?
rust_lwc_lock calls pthread_mutex_lock lwc_string_ref calls rust_lwc_lock rust_lwc_string_ref calls lwc_string_ref from_lwc_string calls rust_lwc_string_ref NetSurf CSS callbacks (node_has_id, node_has_class) call from_lwc_string NetSurf CSS callbacks (node_has_id, node_has_class) can be called millions of times even for a relatively simple page (GitHub source code view) So we call pthread_mutex_lock millions of times to layout a page. We measured that this accounts for 10%(!) of CSS selector matching time. I am curious if locking is unnecessary, or failing that, if locking can be done higher up in less fine grained way instead of doing millions of times. While this all may be unnecessary after we move to the new style system, I still want to understand what's going on since it seems plausible to me that we may encounter analogous problem even in the new style system. Thanks! _______________________________________________ dev-servo mailing list dev-servo@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-servo
