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]. [ 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 + * will be wasted. + */ +#define IOVA_MAG_SIZE 127 #define MAX_GLOBAL_MAGS 32 /* magazines per bin */ struct iova_magazine { -- 2.27.0
