On So, 2019-05-19 at 22:56 -0700, Christoph Hellwig wrote: > Folks, you can't just pass arbitary GFP_ flags to dma allocation > routines, beause very often they are not just wrappers around > the page allocator. > > So no, you can't just fine grained control the flags, but the > existing code is just as buggy. > > Please switch to use memalloc_noio_save() instead. Thinking about this again, we have a problem. We introduced memalloc_noio_save() in 3.10 . Hence the code should have been correct in v4.14. Which means that either 6518202970c1 "(mm/cma: remove unsupported gfp_mask parameter from cma_alloc()" is buggy, or the original issue with a delay of 2 seconds still exist. Do we need to do something? Regards Oliver
