[PATCH 3.16 044/134] vfio/type1: Remove locked page accounting workqueue

[Ben Hutchings]

3.16.47-rc1 review patch.  If anyone has any objections, please let me know.

------------------

From: Alex Williamson <alex.williamson@xxxxxxxxxx>

commit 0cfef2b7410b64d7a430947e0b533314c4f97153 upstream.

If the mmap_sem is contented then the vfio type1 IOMMU backend will
defer locked page accounting updates to a workqueue task.  This has a
few problems and depending on which side the user tries to play, they
might be over-penalized for unmaps that haven't yet been accounted or
race the workqueue to enter more mappings than they're allowed.  The
original intent of this workqueue mechanism seems to be focused on
reducing latency through the ioctl, but we cannot do so at the cost
of correctness.  Remove this workqueue mechanism and update the
callers to allow for failure.  We can also now recheck the limit under
write lock to make sure we don't exceed it.

vfio_pin_pages_remote() also now necessarily includes an unwind path
which we can jump to directly if the consecutive page pinning finds
that we're exceeding the user's memory limits.  This avoids the
current lazy approach which does accounting and mapping up to the
fault, only to return an error on the next iteration to unwind the
entire vfio_dma.

Reviewed-by: Peter Xu <peterx@xxxxxxxxxx>
Reviewed-by: Kirti Wankhede <kwankhede@xxxxxxxxxx>
Signed-off-by: Alex Williamson <alex.williamson@xxxxxxxxxx>
[bwh: Backported to 3.16:
 - vfio_lock_acct() always operates on current->mm
 - Drop changes to vfio_{,un}pin_page_external() and
   vfio_iommu_unmap_unpin_reaccount()
 - Drop test of rsvd flag
 - Fix up the disable_hugepages case in vfio_pin_pages()
 - Use down_write() instead of down_write_killable()
 - Adjust context]
Signed-off-by: Ben Hutchings <ben@xxxxxxxxxxxxxxx>
---
--- a/drivers/vfio/vfio_iommu_type1.c
+++ b/drivers/vfio/vfio_iommu_type1.c
@@ -128,57 +128,37 @@ static void vfio_unlink_dma(struct vfio_
 	rb_erase(&old->node, &iommu->dma_list);
 }
 
-struct vwork {
-	struct mm_struct	*mm;
-	long			npage;
-	struct work_struct	work;
-};
-
-/* delayed decrement/increment for locked_vm */
-static void vfio_lock_acct_bg(struct work_struct *work)
+static int vfio_lock_acct(long npage, bool *lock_cap)
 {
-	struct vwork *vwork = container_of(work, struct vwork, work);
 	struct mm_struct *mm;
+	int ret;
 
-	mm = vwork->mm;
-	down_write(&mm->mmap_sem);
-	mm->locked_vm += vwork->npage;
-	up_write(&mm->mmap_sem);
-	mmput(mm);
-	kfree(vwork);
-}
+	if (!npage)
+		return 0;
 
-static void vfio_lock_acct(long npage)
-{
-	struct vwork *vwork;
-	struct mm_struct *mm;
+	mm = current->mm;
+	if (!mm)
+		return -ESRCH; /* process exited */
 
-	if (!current->mm || !npage)
-		return; /* process exited or nothing to do */
+	ret = 0;
+	down_write(&mm->mmap_sem);
+	if (npage > 0) {
+		if (lock_cap ? !*lock_cap : !capable(CAP_IPC_LOCK)) {
+			unsigned long limit;
 
-	if (down_write_trylock(&current->mm->mmap_sem)) {
-		current->mm->locked_vm += npage;
-		up_write(&current->mm->mmap_sem);
-		return;
-	}
+			limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT;
 
-	/*
-	 * Couldn't get mmap_sem lock, so must setup to update
-	 * mm->locked_vm later. If locked_vm were atomic, we
-	 * wouldn't need this silliness
-	 */
-	vwork = kmalloc(sizeof(struct vwork), GFP_KERNEL);
-	if (!vwork)
-		return;
-	mm = get_task_mm(current);
-	if (!mm) {
-		kfree(vwork);
-		return;
+			if (mm->locked_vm + npage > limit)
+				ret = -ENOMEM;
+		}
 	}
-	INIT_WORK(&vwork->work, vfio_lock_acct_bg);
-	vwork->mm = mm;
-	vwork->npage = npage;
-	schedule_work(&vwork->work);
+
+	if (!ret)
+		mm->locked_vm += npage;
+
+	up_write(&mm->mmap_sem);
+
+	return ret;
 }
 
 /*
@@ -260,7 +240,7 @@ static int vaddr_get_pfn(unsigned long v
 static long vfio_pin_pages(unsigned long vaddr, long npage,
 			   int prot, unsigned long *pfn_base)
 {
-	unsigned long limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT;
+	unsigned long pfn = 0, limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT;
 	bool lock_cap = capable(CAP_IPC_LOCK);
 	long ret, i;
 
@@ -282,14 +262,13 @@ static long vfio_pin_pages(unsigned long
 	}
 
 	if (unlikely(disable_hugepages)) {
-		vfio_lock_acct(1);
-		return 1;
+		ret = vfio_lock_acct(1, &lock_cap);
+		i = 1;
+		goto unpin_out;
 	}
 
 	/* Lock all the consecutive pages from pfn_base */
 	for (i = 1, vaddr += PAGE_SIZE; i < npage; i++, vaddr += PAGE_SIZE) {
-		unsigned long pfn = 0;
-
 		ret = vaddr_get_pfn(vaddr, prot, &pfn);
 		if (ret)
 			break;
@@ -303,11 +282,20 @@ static long vfio_pin_pages(unsigned long
 			put_pfn(pfn, prot);
 			pr_warn("%s: RLIMIT_MEMLOCK (%ld) exceeded\n",
 				__func__, limit << PAGE_SHIFT);
-			break;
+			ret = -ENOMEM;
+			goto unpin_out;
 		}
 	}
 
-	vfio_lock_acct(i);
+	ret = vfio_lock_acct(i, &lock_cap);
+
+unpin_out:
+	if (ret) {
+		for (pfn = *pfn_base ; i ; pfn++, i--)
+			put_pfn(pfn, prot);
+
+		return ret;
+	}
 
 	return i;
 }
@@ -322,7 +310,7 @@ static long vfio_unpin_pages(unsigned lo
 		unlocked += put_pfn(pfn++, prot);
 
 	if (do_accounting)
-		vfio_lock_acct(-unlocked);
+		vfio_lock_acct(-unlocked, NULL);
 
 	return unlocked;
 }
@@ -368,7 +356,7 @@ static void vfio_unmap_unpin(struct vfio
 		iova += unmapped;
 	}
 
-	vfio_lock_acct(-unlocked);
+	vfio_lock_acct(-unlocked, NULL);
 }
 
 static void vfio_remove_dma(struct vfio_iommu *iommu, struct vfio_dma *dma)