Re: kernel BUG at mm/page-writeback.c:2241 [ BUG_ON(PageWriteback(page); ]

[Matthew Wilcox <willy@infradead.org>]

On Thu, Oct 22, 2020 at 11:35:26AM -0400, Qian Cai wrote:
> On Thu, 2020-10-22 at 01:49 +0100, Matthew Wilcox wrote:
> > On Wed, Oct 21, 2020 at 08:30:18PM -0400, Qian Cai wrote:
> > > Today's linux-next starts to trigger this wondering if anyone has any clue.
> > 
> > I've seen that occasionally too.  I changed that BUG_ON to VM_BUG_ON_PAGE
> > to try to get a clue about it.  Good to know it's not the THP patches
> > since they aren't in linux-next.
> > 
> > I don't understand how it can happen.  We have the page locked, and then we
> > do:
> > 
> >                         if (PageWriteback(page)) {
> >                                 if (wbc->sync_mode != WB_SYNC_NONE)
> >                                         wait_on_page_writeback(page);
> >                                 else
> >                                         goto continue_unlock;
> >                         }
> > 
> >                         VM_BUG_ON_PAGE(PageWriteback(page), page);
> > 
> > Nobody should be able to put this page under writeback while we have it
> > locked ... right?  The page can be redirtied by the code that's supposed
> > to be writing it back, but I don't see how anyone can make PageWriteback
> > true while we're holding the page lock.
> 
> It happened again on today's linux-next:
> 
> [ 7613.579890][T55770] page:00000000a4b35e02 refcount:3 mapcount:0 mapping:00000000457ceb87 index:0x3e pfn:0x1cef4e
> [ 7613.590594][T55770] aops:xfs_address_space_operations ino:805d85a dentry name:"doio.f1.55762"
> [ 7613.599192][T55770] flags: 0xbfffc0000000bf(locked|waiters|referenced|uptodate|dirty|lru|active)
> [ 7613.608596][T55770] raw: 00bfffc0000000bf ffffea0005027d48 ffff88810eaec030 ffff888231f3a6a8
> [ 7613.617101][T55770] raw: 000000000000003e 0000000000000000 00000003ffffffff ffff888143724000
> [ 7613.625590][T55770] page dumped because: VM_BUG_ON_PAGE(PageWriteback(page))
> [ 7613.632695][T55770] page->mem_cgroup:ffff888143724000

Seems like it reproduces for you pretty quickly.  I have no luck ;-(

Can you add this?

+++ b/mm/page-writeback.c
@@ -2774,6 +2774,7 @@ int __test_set_page_writeback(struct page *page, bool keep_write)
        struct address_space *mapping = page_mapping(page);
        int ret, access_ret;
 
+       VM_BUG_ON_PAGE(!PageLocked(page), page);
        lock_page_memcg(page);
        if (mapping && mapping_use_writeback_tags(mapping)) {
                XA_STATE(xas, &mapping->i_pages, page_index(page));

This is the only place (afaict) that sets PageWriteback, so that will
tell us whether someone is setting Writeback without holding the lock,
or whether we're suffering from a spurious wakeup.