mm: abstract get_arg_page() stack expansion and mmap read lock

Right now fs/exec.c invokes expand_downwards(), an otherwise internal
implementation detail of the VMA logic in order to ensure that an arg page
can be obtained by get_user_pages_remote().

In order to be able to move the stack expansion logic into mm/vma.c to
make it available to userland testing we need to find an alternative
approach here.

We do so by providing the mmap_read_lock_maybe_expand() function which
also helpfully documents what get_arg_page() is doing here and adds an
additional check against VM_GROWSDOWN to make explicit that the stack
expansion logic is only invoked when the VMA is indeed a downward-growing
stack.

This allows expand_downwards() to become a static function.

Importantly, the VMA referenced by mmap_read_maybe_expand() must NOT be
currently user-visible in any way, that is place within an rmap or VMA
tree.  It must be a newly allocated VMA.

This is the case when exec invokes this function.

Link: https://lkml.kernel.org/r/5295d1c70c58e6aa63d14be68d4e1de9fa1c8e6d.1733248985.git.lorenzo.stoakes@oracle.com
Signed-off-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Jann Horn <jannh@google.com>
Cc: Kees Cook <kees@kernel.org>
Cc: Liam R. Howlett <Liam.Howlett@Oracle.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

fs/exec.c

@@ -205,18 +205,10 @@ static struct page *get_arg_page(struct linux_binprm *bprm, unsigned long pos,
 	/*
 	 * Avoid relying on expanding the stack down in GUP (which
 	 * does not work for STACK_GROWSUP anyway), and just do it
-	 * by hand ahead of time.
+	 * ahead of time.
 	 */
-	if (write && pos < vma->vm_start) {
-		mmap_write_lock(mm);
-		ret = expand_downwards(vma, pos);
-		if (unlikely(ret < 0)) {
-			mmap_write_unlock(mm);
-			return NULL;
-		}
-		mmap_write_downgrade(mm);
-	} else
-		mmap_read_lock(mm);
+	if (!mmap_read_lock_maybe_expand(mm, vma, pos, write))
+		return NULL;
 
 	/*
 	 * We are doing an exec().  'current' is the process

include/linux/mm.h

@@ -3324,6 +3324,8 @@ extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admi
 extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
 extern void exit_mmap(struct mm_struct *);
 int relocate_vma_down(struct vm_area_struct *vma, unsigned long shift);
+bool mmap_read_lock_maybe_expand(struct mm_struct *mm, struct vm_area_struct *vma,
+				 unsigned long addr, bool write);
 
 static inline int check_data_rlimit(unsigned long rlim,
 				    unsigned long new,
@@ -3437,9 +3439,6 @@ extern unsigned long stack_guard_gap;
 int expand_stack_locked(struct vm_area_struct *vma, unsigned long address);
 struct vm_area_struct *expand_stack(struct mm_struct * mm, unsigned long addr);
 
-/* CONFIG_STACK_GROWSUP still needs to grow downwards at some places */
-int expand_downwards(struct vm_area_struct *vma, unsigned long address);
-
 /* Look up the first VMA which satisfies  addr < vm_end,  NULL if none. */
 extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
 extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,

mm/mmap.c

@@ -1009,7 +1009,7 @@ static int expand_upwards(struct vm_area_struct *vma, unsigned long address)
  * vma is the first one with address < vma->vm_start.  Have to extend vma.
  * mmap_lock held for writing.
  */
-int expand_downwards(struct vm_area_struct *vma, unsigned long address)
+static int expand_downwards(struct vm_area_struct *vma, unsigned long address)
 {
 	struct mm_struct *mm = vma->vm_mm;
 	struct vm_area_struct *prev;
@@ -1940,3 +1940,55 @@ int relocate_vma_down(struct vm_area_struct *vma, unsigned long shift)
 	/* Shrink the vma to just the new range */
 	return vma_shrink(&vmi, vma, new_start, new_end, vma->vm_pgoff);
 }
+
+#ifdef CONFIG_MMU
+/*
+ * Obtain a read lock on mm->mmap_lock, if the specified address is below the
+ * start of the VMA, the intent is to perform a write, and it is a
+ * downward-growing stack, then attempt to expand the stack to contain it.
+ *
+ * This function is intended only for obtaining an argument page from an ELF
+ * image, and is almost certainly NOT what you want to use for any other
+ * purpose.
+ *
+ * IMPORTANT - VMA fields are accessed without an mmap lock being held, so the
+ * VMA referenced must not be linked in any user-visible tree, i.e. it must be a
+ * new VMA being mapped.
+ *
+ * The function assumes that addr is either contained within the VMA or below
+ * it, and makes no attempt to validate this value beyond that.
+ *
+ * Returns true if the read lock was obtained and a stack was perhaps expanded,
+ * false if the stack expansion failed.
+ *
+ * On stack expansion the function temporarily acquires an mmap write lock
+ * before downgrading it.
+ */
+bool mmap_read_lock_maybe_expand(struct mm_struct *mm,
+				 struct vm_area_struct *new_vma,
+				 unsigned long addr, bool write)
+{
+	if (!write || addr >= new_vma->vm_start) {
+		mmap_read_lock(mm);
+		return true;
+	}
+
+	if (!(new_vma->vm_flags & VM_GROWSDOWN))
+		return false;
+
+	mmap_write_lock(mm);
+	if (expand_downwards(new_vma, addr)) {
+		mmap_write_unlock(mm);
+		return false;
+	}
+
+	mmap_write_downgrade(mm);
+	return true;
+}
+#else
+bool mmap_read_lock_maybe_expand(struct mm_struct *mm, struct vm_area_struct *vma,
+				 unsigned long addr, bool write)
+{
+	return false;
+}
+#endif