Lighttpd

mod_webdav crash with keep-alive

mod_webdav allocated its own handler context in lighttpd 1.4.43, and cleaned it up in connection_reset hook, but did not properly reset con->plugin_ctx[p->id] = NULL. lighttpd 1.4.44 moved the plugin cleanup check from the connection_reset hook to connection_close hook, leading to possibility that mod_webdav handler_ctx would be cleaned up multiple times if there were other keep-alive requests (not handled by mod_webdav) on the same socket connection, or simply an HTTP/1.1 without Connection: close.