On 7/26/19 4:19 PM, Walter Wu wrote: > On Fri, 2019-07-26 at 15:52 +0300, Andrey Ryabinin wrote: >> >> On 7/26/19 3:28 PM, Walter Wu wrote: >>> On Fri, 2019-07-26 at 15:00 +0300, Andrey Ryabinin wrote: >>>> >>> >>>>> >>>>> >>>>> I remember that there are already the lists which you concern. Maybe we >>>>> can try to solve those problems one by one. >>>>> >>>>> 1. deadlock issue? cause by kmalloc() after kfree()? >>>> >>>> smp_call_on_cpu() >>> >>>>> 2. decrease allocation fail, to modify GFP_NOWAIT flag to GFP_KERNEL? >>>> >>>> No, this is not gonna work. Ideally we shouldn't have any allocations there. >>>> It's not reliable and it hurts performance. >>>> >>> I dont know this meaning, we need create a qobject and put into >>> quarantine, so may need to call kmem_cache_alloc(), would you agree this >>> action? >>> >> >> How is this any different from what you have now? > > I originally thought you already agreed the free-list(tag-based > quarantine) after fix those issue. If no allocation there, If no allocation there, than it must be somewhere else. We known exactly the amount of memory we need, so it's possible to preallocate it in advance. > i think maybe > only move generic quarantine into tag-based kasan, but its memory > consumption is more bigger our patch. what do you think? >
