Re: [PATCH] iommu/iova: change IOVA_MAG_SIZE to 127 to save memory

From: Robin Murphy
Date: Thu Jun 30 2022 - 06:07:06 EST


On 2022-06-30 10:37, John Garry wrote:
On 30/06/2022 10:02, Robin Murphy wrote:
On 2022-06-30 08:33, Feng Tang wrote:
kmalloc will round up the request size to power of 2, and current
iova_magazine's size is 1032 (1024+8) bytes, so each instance
allocated will get 2048 bytes from kmalloc, causing around 1KB
waste.

And in some exstreme case, the memory wasted can trigger OOM as
reported in 2019 on a crash kernel with 256 MB memory [1].

I don't think it really needs pointing out that excessive memory consumption can cause OOM. Especially not in the particularly silly context of a system with only 2MB of RAM per CPU - that's pretty much guaranteed to be doomed one way or another.

   [    4.319253] iommu: Adding device 0000:06:00.2 to group 5
   [    4.325869] iommu: Adding device 0000:20:01.0 to group 15
   [    4.332648] iommu: Adding device 0000:20:02.0 to group 16
   [    4.338946] swapper/0 invoked oom-killer: gfp_mask=0x6040c0(GFP_KERNEL|__GFP_COMP), nodemask=(null), order=0, oom_score_adj=0
   [    4.350251] swapper/0 cpuset=/ mems_allowed=0
   [    4.354618] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.19.57.mx64.282 #1
   [    4.355612] Hardware name: Dell Inc. PowerEdge R7425/08V001, BIOS 1.9.3 06/25/2019
   [    4.355612] Call Trace:
   [    4.355612]  dump_stack+0x46/0x5b
   [    4.355612]  dump_header+0x6b/0x289
   [    4.355612]  out_of_memory+0x470/0x4c0
   [    4.355612]  __alloc_pages_nodemask+0x970/0x1030
   [    4.355612]  cache_grow_begin+0x7d/0x520
   [    4.355612]  fallback_alloc+0x148/0x200
   [    4.355612]  kmem_cache_alloc_trace+0xac/0x1f0
   [    4.355612]  init_iova_domain+0x112/0x170
   [    4.355612]  amd_iommu_domain_alloc+0x138/0x1a0
   [    4.355612]  iommu_group_get_for_dev+0xc4/0x1a0
   [    4.355612]  amd_iommu_add_device+0x13a/0x610
   [    4.355612]  add_iommu_group+0x20/0x30
   [    4.355612]  bus_for_each_dev+0x76/0xc0
   [    4.355612]  bus_set_iommu+0xb6/0xf0
   [    4.355612]  amd_iommu_init_api+0x112/0x132
   [    4.355612]  state_next+0xfb1/0x1165
   [    4.355612]  amd_iommu_init+0x1f/0x67
   [    4.355612]  pci_iommu_init+0x16/0x3f
   ...
   [    4.670295] Unreclaimable slab info:
   ...
   [    4.857565] kmalloc-2048           59164KB      59164KB

Change IOVA_MAG_SIZE from 128 to 127 to make size of 'iova_magazine'
1024 bytes so that no memory will be wasted.

[1]. https://lkml.org/lkml/2019/8/12/266

Signed-off-by: Feng Tang <feng.tang@xxxxxxxxx>
---
  drivers/iommu/iova.c | 7 ++++++-
  1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/drivers/iommu/iova.c b/drivers/iommu/iova.c
index db77aa675145b..27634ddd9b904 100644
--- a/drivers/iommu/iova.c
+++ b/drivers/iommu/iova.c
@@ -614,7 +614,12 @@ EXPORT_SYMBOL_GPL(reserve_iova);
   * dynamic size tuning described in the paper.
   */
-#define IOVA_MAG_SIZE 128
+/*
+ * As kmalloc's buffer size is fixed to power of 2, 127 is chosen to
+ * assure size of 'iova_magzine' to be 1024 bytes, so that no memory

Typo: iova_magazine

+ * will be wasted.
+ */
+#define IOVA_MAG_SIZE 127

I do wonder if we will see some strange new behaviour since IOVA_FQ_SIZE % IOVA_MAG_SIZE != 0 now...

I doubt it - even if a flush queue does happen to be entirely full of equal-sized IOVAs, a CPU's loaded magazines also both being perfectly empty when it comes to dump a full fq seem further unlikely, so in practice I don't see this making any appreciable change to the likelihood of spilling back to the depot or not. In fact the smaller the magazines get, the less time would be spent flushing the depot back to the rbtree, where your interesting workload falls off the cliff and never catches back up with the fq timer, so at some point it might even improve (unless it's also already close to the point where smaller caches would bottleneck allocation)... might be interesting to experiment with a wider range of magazine sizes if you had the time and inclination.

Cheers,
Robin.



The change itself seems perfectly reasonable, though.

Acked-by: Robin Murphy <robin.murphy@xxxxxxx>