On 10/19/22 16:16, Miaohe Lin wrote: > On 2022/10/19 7:36, Mike Kravetz wrote: > > The hugetlb vma_lock structure hangs off the vm_private_data pointer > > of sharable hugetlb vmas. The structure is vma specific and can not > > be shared between vmas. At fork and various other times, vmas are > > duplicated via vm_area_dup(). When this happens, the pointer in the > > newly created vma must be cleared and the structure reallocated. Two > > hugetlb specific routines deal with this hugetlb_dup_vma_private and > > hugetlb_vm_op_open. Both routines are called for newly created vmas. > > hugetlb_dup_vma_private would always clear the pointer and > > hugetlb_vm_op_open would allocate the new vms_lock structure. This did > > not work in the case of this calling sequence pointed out in [1]. > > move_vma > > copy_vma > > new_vma = vm_area_dup(vma); > > new_vma->vm_ops->open(new_vma); --> new_vma has its own vma lock. > > is_vm_hugetlb_page(vma) > > clear_vma_resv_huge_pages > > hugetlb_dup_vma_private --> vma->vm_private_data is set to NULL > > When clearing hugetlb_dup_vma_private we actually leak the associated > > vma_lock structure. > > > > The vma_lock structure contains a pointer to the associated vma. This > > information can be used in hugetlb_dup_vma_private and hugetlb_vm_op_open > > to ensure we only clear the vm_private_data of newly created (copied) > > vmas. In such cases, the vma->vma_lock->vma field will not point to the > > vma. > > > > Update hugetlb_dup_vma_private and hugetlb_vm_op_open to not clear > > vm_private_data if vma->vma_lock->vma == vma. Also, log a warning if > > hugetlb_vm_op_open ever encounters the case where vma_lock has already > > been correctly allocated for the vma. > > > > [1] https://lore.kernel.org/linux-mm/5154292a-4c55-28cd-0935-82441e512fc3@xxxxxxxxxx/ > > > > Fixes: 131a79b474e9 ("hugetlb: fix vma lock handling during split vma and range unmapping") > > Signed-off-by: Mike Kravetz <mike.kravetz@xxxxxxxxxx> > > --- > > mm/hugetlb.c | 31 ++++++++++++++++++++++++------- > > 1 file changed, 24 insertions(+), 7 deletions(-) > > > > diff --git a/mm/hugetlb.c b/mm/hugetlb.c > > index 02f781624fce..7f74cbff6619 100644 > > --- a/mm/hugetlb.c > > +++ b/mm/hugetlb.c > > @@ -1014,15 +1014,23 @@ void hugetlb_dup_vma_private(struct vm_area_struct *vma) > > VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma); > > /* > > * Clear vm_private_data > > + * - For shared mappings this is a per-vma semaphore that may be > > + * allocated in a subsequent call to hugetlb_vm_op_open. > > + * Before clearing, make sure pointer is not associated with vma > > + * as this will leak the structure. This is the case when called > > + * via clear_vma_resv_huge_pages() and hugetlb_vm_op_open has already > > + * been called to allocate a new structure. > > * - For MAP_PRIVATE mappings, this is the reserve map which does > > * not apply to children. Faults generated by the children are > > * not guaranteed to succeed, even if read-only. > > - * - For shared mappings this is a per-vma semaphore that may be > > - * allocated in a subsequent call to hugetlb_vm_op_open. > > */ > > - vma->vm_private_data = (void *)0; > > - if (!(vma->vm_flags & VM_MAYSHARE)) > > - return; > > + if (vma->vm_flags & VM_MAYSHARE) { > > + struct hugetlb_vma_lock *vma_lock = vma->vm_private_data; > > + > > + if (vma_lock && vma_lock->vma != vma) > > + vma->vm_private_data = NULL; > > + } else > > + vma->vm_private_data = NULL; > > } > > > > /* > > @@ -4601,6 +4609,7 @@ static void hugetlb_vm_op_open(struct vm_area_struct *vma) > > struct resv_map *resv = vma_resv_map(vma); > > > > /* > > + * HPAGE_RESV_OWNER indicates a private mapping. > > * This new VMA should share its siblings reservation map if present. > > * The VMA will only ever have a valid reservation map pointer where > > * it is being copied for another still existing VMA. As that VMA > > @@ -4616,10 +4625,18 @@ static void hugetlb_vm_op_open(struct vm_area_struct *vma) > > /* > > * vma_lock structure for sharable mappings is vma specific. > > * Clear old pointer (if copied via vm_area_dup) and create new. > > + * Before clearing, make sure vma_lock is not for this vma. > > */ > > if (vma->vm_flags & VM_MAYSHARE) { > > - vma->vm_private_data = NULL; > > - hugetlb_vma_lock_alloc(vma); > > + struct hugetlb_vma_lock *vma_lock = vma->vm_private_data; > > + > > + if (vma_lock) { > > Thanks Mike. It seems the case of "vma_lock == NULL" is missed, i.e. if vma->vm_private_data == NULL, > hugetlb_vm_op_open won't allocate a new vma lock? Thank you so much! Yes, you are correct. Your review comments have prevented numerous bugs and led to better code. I will send v2 shortly. -- Mike Kravetz
