{"id":37291,"date":"2022-10-10T16:57:50","date_gmt":"2022-10-10T20:57:50","guid":{"rendered":"https:\/\/mjtsai.com\/blog\/?p=37291"},"modified":"2022-10-17T15:52:17","modified_gmt":"2022-10-17T19:52:17","slug":"improving-firefox-responsiveness-on-macos","status":"publish","type":"post","link":"https:\/\/mjtsai.com\/blog\/2022\/10\/10\/improving-firefox-responsiveness-on-macos\/","title":{"rendered":"Improving Firefox Responsiveness on macOS"},"content":{"rendered":"

Gabriele Svelto<\/a> (tweet<\/a>, Hacker News<\/a>):<\/p>\n

Performance in some of our automated tests degraded by as much as 30%<\/a>. os_unfair_lock<\/code> might be better behaved than OSSpinLock<\/code>, but it sucked.<\/p>

As it turns out os_unfair_lock<\/code> doesn’t spin on contention, it makes the calling thread sleep right away when it finds a contended lock<\/a>.<\/p>

For the memory allocator this behavior was suboptimal and the performance regression unacceptable.<\/p>

[…]<\/p>

However, as I dug into Apple’s libraries and kernel, I noticed that some spin locks were indeed available, and they did the spinning in kernel-space where they could make a more informed choice with regards to load and scheduling. Those would have been an excellent choice for our use-case.<\/p>

So how do you use them? Well, it turns out they’re not documented. They rely on a non-public function and flags which I had to duplicate in Firefox.<\/p>

The function is os_unfair_lock_with_options()<\/code><\/a> and the options I used are OS_UNFAIR_LOCK_DATA_SYNCHRONIZATION<\/code> and OS_UNFAIR_LOCK_ADAPTIVE_SPIN<\/code>.<\/p><\/blockquote>\n\n

Jeff Johnson<\/a>:<\/p>\n

\n

This only works because Firefox is not in the Mac App Store.<\/p>\n<\/blockquote>\n\n

Previously:<\/p>\n