On Wed, Nov 18, 2015 at 04:34:30PM +0100, Vlastimil Babka wrote: > On 11/09/2015 08:23 AM, Joonsoo Kim wrote: > > CMA allocation should be guaranteed to succeed by definition, but, > > unfortunately, it would be failed sometimes. It is hard to track down > > the problem, because it is related to page reference manipulation and > > we don't have any facility to analyze it. > > Reminds me of the PeterZ's VM_PINNED patchset. What happened to it? > https://lwn.net/Articles/600502/ > > > This patch adds tracepoints to track down page reference manipulation. > > With it, we can find exact reason of failure and can fix the problem. > > Following is an example of tracepoint output. > > > > <...>-9018 [004] 92.678375: page_ref_set: pfn=0x17ac9 flags=0x0 count=1 mapcount=0 mapping=(nil) mt=4 val=1 > > <...>-9018 [004] 92.678378: kernel_stack: > > => get_page_from_freelist (ffffffff81176659) > > => __alloc_pages_nodemask (ffffffff81176d22) > > => alloc_pages_vma (ffffffff811bf675) > > => handle_mm_fault (ffffffff8119e693) > > => __do_page_fault (ffffffff810631ea) > > => trace_do_page_fault (ffffffff81063543) > > => do_async_page_fault (ffffffff8105c40a) > > => async_page_fault (ffffffff817581d8) > > [snip] > > <...>-9018 [004] 92.678379: page_ref_mod: pfn=0x17ac9 flags=0x40048 count=2 mapcount=1 mapping=0xffff880015a78dc1 mt=4 val=1 > > [snip] > > ... > > ... > > <...>-9131 [001] 93.174468: test_pages_isolated: start_pfn=0x17800 end_pfn=0x17c00 fin_pfn=0x17ac9 ret=fail > > [snip] > > <...>-9018 [004] 93.174843: page_ref_mod_and_test: pfn=0x17ac9 flags=0x40068 count=0 mapcount=0 mapping=0xffff880015a78dc1 mt=4 val=-1 ret=1 > > => release_pages (ffffffff8117c9e4) > > => free_pages_and_swap_cache (ffffffff811b0697) > > => tlb_flush_mmu_free (ffffffff81199616) > > => tlb_finish_mmu (ffffffff8119a62c) > > => exit_mmap (ffffffff811a53f7) > > => mmput (ffffffff81073f47) > > => do_exit (ffffffff810794e9) > > => do_group_exit (ffffffff81079def) > > => SyS_exit_group (ffffffff81079e74) > > => entry_SYSCALL_64_fastpath (ffffffff817560b6) > > > > This output shows that problem comes from exit path. In exit path, > > to improve performance, pages are not freed immediately. They are gathered > > and processed by batch. During this process, migration cannot be possible > > and CMA allocation is failed. This problem is hard to find without this > > page reference tracepoint facility. > > Yeah but when you realized it was this problem, what was the fix? Probably not > remove batching from exit path? Shouldn't CMA in this case just try waiting for > the pins to go away, which would eventually happen? And for long-term pins, > VM_PINNED would make sure the pages are migrated away from CMA pageblocks first? > > So I'm worried that this is quite nontrivial change for a very specific usecase. This patch is not to solve the problem but just to expose what is culprit. For using VM_PINNED, firstly, we should know where are long-term pins. There are obviously clear places which will be first target if we use VM_PINNED but somewhere are vague. For vague places, this patch will help finding there. Even we don't use VM_PINNED, this patch will expose current obstacle which can help to understand current problems. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>