On Fri, Dec 21, 2018 at 12:03:00PM +0800, kbuild test robot wrote:
> tree: git://git.cmpxchg.org/linux-mmotm.git master
> head: 98c1d1d6a1d1553512e5db8c07a149c41e7c2f84
> commit: 47931f365e6eaac24d1653e3ce00f69e76187c08 [310/355] mm: treewide: remove unused address argument from pte_alloc functions
> config: x86_64-rhel-7.2-clear (attached as .config)
> compiler: gcc-7 (Debian 7.3.0-1) 7.3.0
> reproduce:
> git checkout 47931f365e6eaac24d1653e3ce00f69e76187c08
> # save the attached .config to linux build tree
> make ARCH=x86_64
>
> Note: the mmotm/master HEAD 98c1d1d6a1d1553512e5db8c07a149c41e7c2f84 builds fine.
> It only hurts bisectibility.
>
> All errors (new ones prefixed by >>):
>
> mm/memory.c: In function '__do_fault':
> >> mm/memory.c:3007:23: error: too many arguments to function 'pte_alloc_one'
> vmf->prealloc_pte = pte_alloc_one(vmf->vma->vm_mm, vmf->address);
> ^~~~~~~~~~~~~
> In file included from arch/x86/include/asm/mmu_context.h:12:0,
> from mm/memory.c:74:
> arch/x86/include/asm/pgalloc.h:51:18: note: declared here
> extern pgtable_t pte_alloc_one(struct mm_struct *);
> ^~~~~~~~~~~~~
>
> vim +/pte_alloc_one +3007 mm/memory.c
>
> ^1da177e4 Linus Torvalds 2005-04-16 2991
> 9a95f3cf7 Paul Cassella 2014-08-06 2992 /*
> 9a95f3cf7 Paul Cassella 2014-08-06 2993 * The mmap_sem must have been held on entry, and may have been
> 9a95f3cf7 Paul Cassella 2014-08-06 2994 * released depending on flags and vma->vm_ops->fault() return value.
> 9a95f3cf7 Paul Cassella 2014-08-06 2995 * See filemap_fault() and __lock_page_retry().
> 9a95f3cf7 Paul Cassella 2014-08-06 2996 */
> 2b7403035 Souptick Joarder 2018-08-23 2997 static vm_fault_t __do_fault(struct vm_fault *vmf)
> 7eae74af3 Kirill A. Shutemov 2014-04-03 2998 {
> 82b0f8c39 Jan Kara 2016-12-14 2999 struct vm_area_struct *vma = vmf->vma;
> 2b7403035 Souptick Joarder 2018-08-23 3000 vm_fault_t ret;
> 7eae74af3 Kirill A. Shutemov 2014-04-03 3001
> d85ec7561 Michal Hocko 2018-12-19 3002 /*
> d85ec7561 Michal Hocko 2018-12-19 3003 * Preallocate pte before we take page_lock because this might lead to
> d85ec7561 Michal Hocko 2018-12-19 3004 * deadlocks for memcg reclaim which waits for pages under writeback.
> d85ec7561 Michal Hocko 2018-12-19 3005 */
> d85ec7561 Michal Hocko 2018-12-19 3006 if (pmd_none(*vmf->pmd) && !vmf->prealloc_pte) {
> d85ec7561 Michal Hocko 2018-12-19 @3007 vmf->prealloc_pte = pte_alloc_one(vmf->vma->vm_mm, vmf->address);
Taking a look at linux-next, this has already been fixed so I believe the
report is based on an older kernel.
thanks,
- Joel
