On Mon, 2020-08-17 at 09:38 +0100, Steven Price wrote:

On 15/08/2020 07:20, Chinwen Chang wrote:

quoted

smaps_rollup will try to grab mmap_lock and go through the whole vma
list until it finishes the iterating. When encountering large processes,
the mmap_lock will be held for a longer time, which may block other
write requests like mmap and munmap from progressing smoothly.

There are upcoming mmap_lock optimizations like range-based locks, but
the lock applied to smaps_rollup would be the coarse type, which doesn't
avoid the occurrence of unpleasant contention.

To solve aforementioned issue, we add a check which detects whether
anyone wants to grab mmap_lock for write attempts.

Change since v1:
- If current VMA is freed after dropping the lock, it will return
- incomplete result. To fix this issue, refine the code flow as
- suggested by Steve. [1]

Change since v2:
- When getting back the mmap lock, the address where you stopped last
- time could now be in the middle of a vma. Add one more check to handle
- this case as suggested by Michel. [2]

[1] https://lore.kernel.org/lkml/bf40676e-b14b-44cd-75ce-419c70194783@arm.com/ (local)
[2] https://lore.kernel.org/lkml/CANN689FtCsC71cjAjs0GPspOhgo_HRj+diWsoU1wr98YPktgWg@mail.gmail.com/ (local)

Signed-off-by: Chinwen Chang <redacted>
CC: Steven Price <steven.price@arm.com>
CC: Michel Lespinasse <redacted>

Reviewed-by: Steven Price <steven.price@arm.com>

quoted

---
  fs/proc/task_mmu.c | 73 +++++++++++++++++++++++++++++++++++++++++++++++++++---
  1 file changed, 70 insertions(+), 3 deletions(-)

diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 76e623a..945904e 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c

@@ -846,7 +846,7 @@ static int show_smaps_rollup(struct seq_file *m, void *v)
  	struct mem_size_stats mss;
  	struct mm_struct *mm;
  	struct vm_area_struct *vma;
-	unsigned long last_vma_end = 0;
+	unsigned long last_vma_end = 0, last_stopped = 0;
  	int ret = 0;
  
  	priv->task = get_proc_task(priv->inode);

@@ -867,9 +867,76 @@ static int show_smaps_rollup(struct seq_file *m, void *v)
  
  	hold_task_mempolicy(priv);
  
-	for (vma = priv->mm->mmap; vma; vma = vma->vm_next) {
-		smap_gather_stats(vma, &mss, 0);
+	for (vma = priv->mm->mmap; vma;) {
+		smap_gather_stats(vma, &mss, last_stopped);
+		last_stopped = 0;
  		last_vma_end = vma->vm_end;
+
+		/*
+		 * Release mmap_lock temporarily if someone wants to
+		 * access it for write request.
+		 */
+		if (mmap_lock_is_contended(mm)) {
+			mmap_read_unlock(mm);
+			ret = mmap_read_lock_killable(mm);
+			if (ret) {
+				release_task_mempolicy(priv);
+				goto out_put_mm;
+			}
+
+			/*
+			 * After dropping the lock, there are four cases to
+			 * consider. See the following example for explanation.
+			 *
+			 *   +------+------+-----------+
+			 *   | VMA1 | VMA2 | VMA3      |
+			 *   +------+------+-----------+
+			 *   |      |      |           |
+			 *  4k     8k     16k         400k
+			 *
+			 * Suppose we drop the lock after reading VMA2 due to
+			 * contention, then we get:
+			 *
+			 *	last_vma_end = 16k
+			 *
+			 * 1) VMA2 is freed, but VMA3 exists:
+			 *
+			 *    find_vma(mm, 16k - 1) will return VMA3.
+			 *    In this case, just continue from VMA3.
+			 *
+			 * 2) VMA2 still exists:
+			 *
+			 *    find_vma(mm, 16k - 1) will return VMA2.
+			 *    Iterate the loop like the original one.
+			 *
+			 * 3) No more VMAs can be found:
+			 *
+			 *    find_vma(mm, 16k - 1) will return NULL.
+			 *    No more things to do, just break.
+			 *
+			 * 4) (last_vma_end - 1) is the middle of a vma (VMA'):
+			 *
+			 *    find_vma(mm, 16k - 1) will return VMA' whose range
+			 *    contains last_vma_end.
+			 *    Iterate VMA' from last_vma_end.
+			 */
+			vma = find_vma(mm, last_vma_end - 1);
+			/* Case 3 above */
+			if (!vma)
+				break;
+
+			/* Case 1 above */
+			if (vma->vm_start >= last_vma_end)
+				continue;
+
+			/* Case 4 above */
+			if (vma->vm_end > last_vma_end) {
+				last_stopped = last_vma_end;
+				continue;

Note that instead of having last_stopped, you could replace the above 
with a direct call:

   smap_gather_stats(vma, &mss, last_vma_end);

I'm not sure which is cleaner though. last_stopped is a bit messy (it's 
easily confused with last_vma_end), but having just the one call site 
for smap_gather_stats() is nice too.

Steve

Hi Steve,

I think your idea is better. Let me try refactoring for further reviews.
Thanks for your kind suggestion:)

Chinwen

quoted

+			}
+		}
+		/* Case 2 above */
+		vma = vma->vm_next;
  	}
  
  	show_vma_header_prefix(m, priv->mm->mmap->vm_start,

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel