On 8/4/21 9:57 AM, Sebastian Andrzej Siewior wrote: > On 2021-08-04 09:49:48 [-0600], Jens Axboe wrote: >> >> That'd work on non-RT as well, but it makes it worse on non-RT as well with >> the irq enable/disable dance. While that's not the end of the world, would >> be nice to have a solution that doesn't sacrifice anything, yet doesn't >> make RT unhappy. > > There were plans to make local_irq_disable() mostly a nop (similar to > what ppc64 does). But I have no idea what happened. I hope that work > gets picked up. So what do you propose in the interim? As far as io_uring is concerned, it's not a _huge_ deal to do the IRQ dance, but it does bother me that we're making things slightly worse for the mainline kernel just to make the out-of-tree patches happy. -- Jens Axboe
