[PATCH 5/5] iommu/vt-d: don't disable preemption while accessing deferred_flush()

From: Sebastian Andrzej Siewior
Date: Fri Oct 21 2016 - 13:06:58 EST


get_cpu() disables preemption and returns the current CPU number. The
CPU number is later only used once while retrieving the address of the
local's CPU deferred_flush pointer.
We can instead use raw_cpu_ptr() while we remain preemptible. The worst
thing that can happen is that flush_unmaps_timeout() is invoked multiple
times: once by taskA after seeing HIGH_WATER_MARK and then preempted to
another CPU and then by taskB which saw HIGH_WATER_MARK on the same CPU
as taskA. It is also likely that ->size got from HIGH_WATER_MARK to 0
right after its read because another CPU invoked flush_unmaps_timeout()
for this CPU.
The access to flush_data is protected by a spinlock so even if we get
migrated to another CPU or preempted - the data structure is protected.

While at it, I marked deferred_flush static since I can't find a
reference to it outside of this file.

Cc: David Woodhouse <dwmw2@xxxxxxxxxxxxx>
Cc: Joerg Roedel <joro@xxxxxxxxxx>
Cc: iommu@xxxxxxxxxxxxxxxxxxxxxxxxxx
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>
---
drivers/iommu/intel-iommu.c | 8 ++------
1 file changed, 2 insertions(+), 6 deletions(-)

diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c
index a4407eabf0e6..0dad326a9483 100644
--- a/drivers/iommu/intel-iommu.c
+++ b/drivers/iommu/intel-iommu.c
@@ -479,7 +479,7 @@ struct deferred_flush_data {
struct deferred_flush_table *tables;
};

-DEFINE_PER_CPU(struct deferred_flush_data, deferred_flush);
+static DEFINE_PER_CPU(struct deferred_flush_data, deferred_flush);

/* bitmap for indexing intel_iommus */
static int g_num_of_iommus;
@@ -3673,10 +3673,8 @@ static void add_unmap(struct dmar_domain *dom, unsigned long iova_pfn,
struct intel_iommu *iommu;
struct deferred_flush_entry *entry;
struct deferred_flush_data *flush_data;
- unsigned int cpuid;

- cpuid = get_cpu();
- flush_data = per_cpu_ptr(&deferred_flush, cpuid);
+ flush_data = raw_cpu_ptr(&deferred_flush);

/* Flush all CPUs' entries to avoid deferring too much. If
* this becomes a bottleneck, can just flush us, and rely on
@@ -3709,8 +3707,6 @@ static void add_unmap(struct dmar_domain *dom, unsigned long iova_pfn,
}
flush_data->size++;
spin_unlock_irqrestore(&flush_data->lock, flags);
-
- put_cpu();
}

static void intel_unmap(struct device *dev, dma_addr_t dev_addr, size_t size)
--
2.9.3