Thread (27 messages) 27 messages, 3 authors, 2018-12-15

Re: [PATCH 3/5] dma-direct: refine dma_direct_alloc zone selection

From: Robin Murphy <robin.murphy@arm.com>
Date: 2018-09-27 14:30:26
Also in: linux-iommu, lkml

On 20/09/18 19:52, Christoph Hellwig wrote:
quoted hunk ↗ jump to hunk
We need to take the DMA offset and encryption bit into account when
selecting a zone.  User the opportunity to factor out the zone
selection into a helper for reuse.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
  kernel/dma/direct.c | 31 +++++++++++++++++++++----------
  1 file changed, 21 insertions(+), 10 deletions(-)
diff --git a/kernel/dma/direct.c b/kernel/dma/direct.c
index 81b73a5bba54..3c404e33d946 100644
--- a/kernel/dma/direct.c
+++ b/kernel/dma/direct.c
@@ -68,6 +68,22 @@ u64 dma_direct_get_required_mask(struct device *dev)
  	return (1ULL << (fls64(max_dma) - 1)) * 2 - 1;
  }
  
+static gfp_t __dma_direct_optimal_gfp_mask(struct device *dev, u64 dma_mask,
+		u64 *phys_mask)
+{
+	if (force_dma_unencrypted())
+		*phys_mask = __dma_to_phys(dev, dma_mask);
+	else
+		*phys_mask = dma_to_phys(dev, dma_mask);
Maybe make phys_to_dma_direct() take u64 instead of phys_addr_t so we 
can reuse it here?
+	/* GFP_DMA32 and GFP_DMA are no ops without the corresponding zones: */
+	if (*phys_mask <= DMA_BIT_MASK(ARCH_ZONE_DMA_BITS))
+		return GFP_DMA;
If we don't have ZONE_DMA it would in theory be a tiny bit better to 
fall back to GFP_DMA32 instead of returning 0 here, but I'm not sure 
it's worth the bother.
quoted hunk ↗ jump to hunk
+	if (*phys_mask <= DMA_BIT_MASK(32))
+		return GFP_DMA32;
+	return 0;
+}
+
  static bool dma_coherent_ok(struct device *dev, phys_addr_t phys, size_t size)
  {
  	return phys_to_dma_direct(dev, phys) + size - 1 <=
@@ -80,17 +96,13 @@ void *dma_direct_alloc_pages(struct device *dev, size_t size,
  	unsigned int count = PAGE_ALIGN(size) >> PAGE_SHIFT;
  	int page_order = get_order(size);
  	struct page *page = NULL;
+	u64 phys_mask;
  	void *ret;
  
  	/* we always manually zero the memory once we are done: */
  	gfp &= ~__GFP_ZERO;
-
-	/* GFP_DMA32 and GFP_DMA are no ops without the corresponding zones: */
-	if (dev->coherent_dma_mask <= DMA_BIT_MASK(ARCH_ZONE_DMA_BITS))
-		gfp |= GFP_DMA;
-	if (dev->coherent_dma_mask <= DMA_BIT_MASK(32) && !(gfp & GFP_DMA))
-		gfp |= GFP_DMA32;
-
+	gfp |= __dma_direct_optimal_gfp_mask(dev, dev->coherent_dma_mask,
+			&phys_mask);
  again:
  	/* CMA can be used only in the context which permits sleeping */
  	if (gfpflags_allow_blocking(gfp)) {
@@ -109,15 +121,14 @@ void *dma_direct_alloc_pages(struct device *dev, size_t size,
  		page = NULL;
  
  		if (IS_ENABLED(CONFIG_ZONE_DMA32) &&
-		    dev->coherent_dma_mask < DMA_BIT_MASK(64) &&
+		    phys_mask < DMA_BIT_MASK(64) &&
  		    !(gfp & (GFP_DMA32 | GFP_DMA))) {
  			gfp |= GFP_DMA32;
  			goto again;
Ah right, in that situation we should probably end up here anyway, so 
that's good enough - definitely not worth any more #ifdeffery above. 
Nits aside,

Reviewed-by: Robin Murphy <robin.murphy@arm.com>
  		}
  
  		if (IS_ENABLED(CONFIG_ZONE_DMA) &&
-		    dev->coherent_dma_mask < DMA_BIT_MASK(32) &&
-		    !(gfp & GFP_DMA)) {
+		    phys_mask < DMA_BIT_MASK(32) && !(gfp & GFP_DMA)) {
  			gfp = (gfp & ~GFP_DMA32) | GFP_DMA;
  			goto again;
  		}
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help