The problem is, VM_* flags are a limited resource. As we add support for new use cases to userfaultfd, there are new reasons why a userfault might be triggered, but we can't keep adding new VM_* flags.
So, introduce a new enum, to which we can add arbitrarily many reasons going forward. The intent is: 1. Page fault handlers will notice a userfaultfd registration (VM_UFFD_MISSING or VM_UFFD_WP). 2. They'll call handle_userfault() to resolve it, with the reason: page missing, write protect fault, or (in the future) minor fault, etc... Importantly, the possible reasons for triggering a userfault will no longer match 1:1 with VM_* flags; there can be > 1 reason to trigger a fault for a single VM_* flag. Signed-off-by: Axel Rasmussen <[email protected]> --- fs/userfaultfd.c | 21 +++++++++------------ include/linux/userfaultfd_k.h | 12 ++++++++++-- mm/huge_memory.c | 4 ++-- mm/hugetlb.c | 2 +- mm/memory.c | 8 ++++---- mm/shmem.c | 2 +- 6 files changed, 27 insertions(+), 22 deletions(-) diff --git a/fs/userfaultfd.c b/fs/userfaultfd.c index 1f4a34b1a1e7..8d663eae0266 100644 --- a/fs/userfaultfd.c +++ b/fs/userfaultfd.c @@ -190,7 +190,7 @@ static inline void msg_init(struct uffd_msg *msg) static inline struct uffd_msg userfault_msg(unsigned long address, unsigned int flags, - unsigned long reason, + enum uffd_trigger_reason reason, unsigned int features) { struct uffd_msg msg; @@ -206,7 +206,7 @@ static inline struct uffd_msg userfault_msg(unsigned long address, * a write fault. */ msg.arg.pagefault.flags |= UFFD_PAGEFAULT_FLAG_WRITE; - if (reason & VM_UFFD_WP) + if (reason == UFFD_REASON_WP) /* * If UFFD_FEATURE_PAGEFAULT_FLAG_WP was set in the * uffdio_api.features and UFFD_PAGEFAULT_FLAG_WP was @@ -229,7 +229,7 @@ static inline bool userfaultfd_huge_must_wait(struct userfaultfd_ctx *ctx, struct vm_area_struct *vma, unsigned long address, unsigned long flags, - unsigned long reason) + enum uffd_trigger_reason reason) { struct mm_struct *mm = ctx->mm; pte_t *ptep, pte; @@ -251,7 +251,7 @@ static inline bool userfaultfd_huge_must_wait(struct userfaultfd_ctx *ctx, */ if (huge_pte_none(pte)) ret = true; - if (!huge_pte_write(pte) && (reason & VM_UFFD_WP)) + if (!huge_pte_write(pte) && (reason == UFFD_REASON_WP)) ret = true; out: return ret; @@ -261,7 +261,7 @@ static inline bool userfaultfd_huge_must_wait(struct userfaultfd_ctx *ctx, struct vm_area_struct *vma, unsigned long address, unsigned long flags, - unsigned long reason) + enum uffd_trigger_reason reason) { return false; /* should never get here */ } @@ -277,7 +277,7 @@ static inline bool userfaultfd_huge_must_wait(struct userfaultfd_ctx *ctx, static inline bool userfaultfd_must_wait(struct userfaultfd_ctx *ctx, unsigned long address, unsigned long flags, - unsigned long reason) + enum uffd_trigger_reason reason) { struct mm_struct *mm = ctx->mm; pgd_t *pgd; @@ -316,7 +316,7 @@ static inline bool userfaultfd_must_wait(struct userfaultfd_ctx *ctx, goto out; if (pmd_trans_huge(_pmd)) { - if (!pmd_write(_pmd) && (reason & VM_UFFD_WP)) + if (!pmd_write(_pmd) && (reason == UFFD_REASON_WP)) ret = true; goto out; } @@ -332,7 +332,7 @@ static inline bool userfaultfd_must_wait(struct userfaultfd_ctx *ctx, */ if (pte_none(*pte)) ret = true; - if (!pte_write(*pte) && (reason & VM_UFFD_WP)) + if (!pte_write(*pte) && (reason == UFFD_REASON_WP)) ret = true; pte_unmap(pte); @@ -366,7 +366,7 @@ static inline long userfaultfd_get_blocking_state(unsigned int flags) * fatal_signal_pending()s, and the mmap_lock must be released before * returning it. */ -vm_fault_t handle_userfault(struct vm_fault *vmf, unsigned long reason) +vm_fault_t handle_userfault(struct vm_fault *vmf, enum uffd_trigger_reason reason) { struct mm_struct *mm = vmf->vma->vm_mm; struct userfaultfd_ctx *ctx; @@ -401,9 +401,6 @@ vm_fault_t handle_userfault(struct vm_fault *vmf, unsigned long reason) BUG_ON(ctx->mm != mm); - VM_BUG_ON(reason & ~(VM_UFFD_MISSING|VM_UFFD_WP)); - VM_BUG_ON(!(reason & VM_UFFD_MISSING) ^ !!(reason & VM_UFFD_WP)); - if (ctx->features & UFFD_FEATURE_SIGBUS) goto out; if ((vmf->flags & FAULT_FLAG_USER) == 0 && diff --git a/include/linux/userfaultfd_k.h b/include/linux/userfaultfd_k.h index c63ccdae3eab..cc1554e7162f 100644 --- a/include/linux/userfaultfd_k.h +++ b/include/linux/userfaultfd_k.h @@ -9,6 +9,14 @@ #ifndef _LINUX_USERFAULTFD_K_H #define _LINUX_USERFAULTFD_K_H +/* Denotes the reason why handle_userfault() is being triggered. */ +enum uffd_trigger_reason { + /* A page was missing. */ + UFFD_REASON_MISSING, + /* A write protect fault occurred. */ + UFFD_REASON_WP, +}; + #ifdef CONFIG_USERFAULTFD #include <linux/userfaultfd.h> /* linux/include/uapi/linux/userfaultfd.h */ @@ -32,7 +40,7 @@ extern int sysctl_unprivileged_userfaultfd; -extern vm_fault_t handle_userfault(struct vm_fault *vmf, unsigned long reason); +extern vm_fault_t handle_userfault(struct vm_fault *vmf, enum uffd_trigger_reason reason); extern ssize_t mcopy_atomic(struct mm_struct *dst_mm, unsigned long dst_start, unsigned long src_start, unsigned long len, @@ -111,7 +119,7 @@ extern void userfaultfd_unmap_complete(struct mm_struct *mm, /* mm helpers */ static inline vm_fault_t handle_userfault(struct vm_fault *vmf, - unsigned long reason) + enum uffd_trigger_reason reason) { return VM_FAULT_SIGBUS; } diff --git a/mm/huge_memory.c b/mm/huge_memory.c index 395c75111d33..1d740b43bcc5 100644 --- a/mm/huge_memory.c +++ b/mm/huge_memory.c @@ -629,7 +629,7 @@ static vm_fault_t __do_huge_pmd_anonymous_page(struct vm_fault *vmf, spin_unlock(vmf->ptl); put_page(page); pte_free(vma->vm_mm, pgtable); - ret2 = handle_userfault(vmf, VM_UFFD_MISSING); + ret2 = handle_userfault(vmf, UFFD_REASON_MISSING); VM_BUG_ON(ret2 & VM_FAULT_FALLBACK); return ret2; } @@ -748,7 +748,7 @@ vm_fault_t do_huge_pmd_anonymous_page(struct vm_fault *vmf) } else if (userfaultfd_missing(vma)) { spin_unlock(vmf->ptl); pte_free(vma->vm_mm, pgtable); - ret = handle_userfault(vmf, VM_UFFD_MISSING); + ret = handle_userfault(vmf, UFFD_REASON_MISSING); VM_BUG_ON(ret & VM_FAULT_FALLBACK); } else { set_huge_zero_page(pgtable, vma->vm_mm, vma, diff --git a/mm/hugetlb.c b/mm/hugetlb.c index 0d45a01a85f8..2a90e0b4bf47 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -4305,7 +4305,7 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm, hash = hugetlb_fault_mutex_hash(mapping, idx); mutex_unlock(&hugetlb_fault_mutex_table[hash]); i_mmap_unlock_read(mapping); - ret = handle_userfault(&vmf, VM_UFFD_MISSING); + ret = handle_userfault(&vmf, UFFD_REASON_MISSING); i_mmap_lock_read(mapping); mutex_lock(&hugetlb_fault_mutex_table[hash]); goto out; diff --git a/mm/memory.c b/mm/memory.c index bc4a41ec81aa..995a95826f4d 100644 --- a/mm/memory.c +++ b/mm/memory.c @@ -3100,7 +3100,7 @@ static vm_fault_t do_wp_page(struct vm_fault *vmf) if (userfaultfd_pte_wp(vma, *vmf->pte)) { pte_unmap_unlock(vmf->pte, vmf->ptl); - return handle_userfault(vmf, VM_UFFD_WP); + return handle_userfault(vmf, UFFD_REASON_WP); } vmf->page = vm_normal_page(vma, vmf->address, vmf->orig_pte); @@ -3535,7 +3535,7 @@ static vm_fault_t do_anonymous_page(struct vm_fault *vmf) /* Deliver the page fault to userland, check inside PT lock */ if (userfaultfd_missing(vma)) { pte_unmap_unlock(vmf->pte, vmf->ptl); - return handle_userfault(vmf, VM_UFFD_MISSING); + return handle_userfault(vmf, UFFD_REASON_MISSING); } goto setpte; } @@ -3577,7 +3577,7 @@ static vm_fault_t do_anonymous_page(struct vm_fault *vmf) if (userfaultfd_missing(vma)) { pte_unmap_unlock(vmf->pte, vmf->ptl); put_page(page); - return handle_userfault(vmf, VM_UFFD_MISSING); + return handle_userfault(vmf, UFFD_REASON_MISSING); } inc_mm_counter_fast(vma->vm_mm, MM_ANONPAGES); @@ -4195,7 +4195,7 @@ static inline vm_fault_t wp_huge_pmd(struct vm_fault *vmf, pmd_t orig_pmd) { if (vma_is_anonymous(vmf->vma)) { if (userfaultfd_huge_pmd_wp(vmf->vma, orig_pmd)) - return handle_userfault(vmf, VM_UFFD_WP); + return handle_userfault(vmf, UFFD_REASON_WP); return do_huge_pmd_wp_page(vmf, orig_pmd); } if (vmf->vma->vm_ops->huge_fault) { diff --git a/mm/shmem.c b/mm/shmem.c index 06c771d23127..e1e2513b4298 100644 --- a/mm/shmem.c +++ b/mm/shmem.c @@ -1849,7 +1849,7 @@ static int shmem_getpage_gfp(struct inode *inode, pgoff_t index, */ if (vma && userfaultfd_missing(vma)) { - *fault_type = handle_userfault(vmf, VM_UFFD_MISSING); + *fault_type = handle_userfault(vmf, UFFD_REASON_MISSING); return 0; } -- 2.30.0.478.g8a0d178c01-goog
