On 10/2/20 7:49 PM, Thomas Gleixner wrote: > On Fri, Oct 02 2020 at 17:38, Oleg Nesterov wrote: >> On 10/02, Thomas Gleixner wrote: >>> >>> I think it's fundamentaly wrong that we have several places and several >>> flags which handle task_work_run() instead of having exactly one place >>> and one flag. >> >> Damn yes, agreed. > > Actually there are TWO places, but they don't interfere: > > 1) exit to user > > 2) enter guest > > From the kernel POV they are pretty much the same as both are leaving > the kernel domain. But they have a few subtle different requirements > what has to be done or not. > > So any change to that logic needs to fixup both places, Right, I actually did spot that, but didn't include it in the initial series. I've split up the series a bit more, into functional bits. Should be easier to reason/discuss: https://git.kernel.dk/cgit/linux-block/log/?h=tif-task_work -- Jens Axboe
