On 4/5/18 8:45 AM, Paolo Valente wrote: > > >> Il giorno 05 apr 2018, alle ore 15:15, Sami Farin <hvtaifwkbgefbaei@xxxxxxxxx> ha scritto: >> >> I was using chacharand to fill 32 GB SD card (VFAT fs) (maybe 30 MiB/s) >> with random data, it froze halfway. There was 400 MiB Dirty data. >> After reboot the filling operation went OK when I used kyber scheduler. >> System is Fedora 27 on Core i5 2500K / 16 GiB. >> > > I'm afraid this crash is caused by a bug fixed for 4.16 [1]. In the > same thread [1], Oleksander (in CC) proposed to backport this and > other fixes and improvements to 4.15. But Jens (in CC) didn't accept, > because too general stuff was included in the batch. Maybe this bug > report could be the opportunity to reconsider that backport or part of > it? I never objected to back porting the single fix. What I did object to was a huge list of other fixes. I'm also fine with back porting a set of fixes, if they are all relevant to back port. I don't want a whole sale list of "these are all the changes we did to BFQ, let's back port them". -- Jens Axboe