Re: Question regarding SLAB corruption

[Keir Fraser <keir@xensource.com>]

[-- Attachment #1: Type: text/plain, Size: 847 bytes --]

On 9/7/07 13:07, "Lukas Hejtmanek" <xhejtman@ics.muni.cz> wrote:

> According to IB developers, it uses alloc_pages and then pci_map_sg which they
> state that is DMA API. So, is it correct approach?

Well, we currently assume that segments of a scatterlist (as passed to
pci_map_sg) do not cross page boundaries. It looks like this assumption is
broken (certainly for the infiniband driver!).

I've just checked in the attached patch to fix this issue. Please give it a
try in your dom0 tree.

> I did debug like this - so it prints physical and virtual addresses and they
> seems to be perfectly contiguous. Are the physical addresses real physical
> addresses or virtualized by Xen hypervisor?

You want to use virt_to_machine(). As you say, the addresses returned by
virt_to_phys() are virtual-physical, which is not what you want.

 -- Keir


[-- Attachment #2: map_sg.patch --]
[-- Type: application/octet-stream, Size: 2897 bytes --]

# HG changeset patch
# User kfraser@localhost.localdomain
# Date 1183985110 -3600
# Node ID 08cf42135056cbc07a6d790d4851e0e4b160f847
# Parent  f833757672a70ee43afd0bfbfaa22cec3b132445
x86: dma_map_sg() must handle multi-page segments.
Signed-off-by: Keir Fraser <keir@xensource.com>

diff -r f833757672a7 -r 08cf42135056 arch/i386/kernel/pci-dma-xen.c
--- a/arch/i386/kernel/pci-dma-xen.c	Sat Jul 07 11:44:16 2007 +0100
+++ b/arch/i386/kernel/pci-dma-xen.c	Mon Jul 09 13:45:10 2007 +0100
@@ -97,6 +97,9 @@ dma_map_sg(struct device *hwdev, struct 
 			BUG_ON(!sg[i].page);
 			IOMMU_BUG_ON(address_needs_mapping(
 				hwdev, sg[i].dma_address));
+			IOMMU_BUG_ON(range_straddles_page_boundary(
+				page_to_pseudophys(sg[i].page) + sg[i].offset,
+				sg[i].length));
 		}
 		rc = nents;
 	}
@@ -338,7 +341,7 @@ dma_map_single(struct device *dev, void 
 	} else {
 		dma = gnttab_dma_map_page(virt_to_page(ptr)) +
 		      offset_in_page(ptr);
-		IOMMU_BUG_ON(range_straddles_page_boundary(ptr, size));
+		IOMMU_BUG_ON(range_straddles_page_boundary(__pa(ptr), size));
 		IOMMU_BUG_ON(address_needs_mapping(dev, dma));
 	}
 
diff -r f833757672a7 -r 08cf42135056 arch/i386/kernel/swiotlb.c
--- a/arch/i386/kernel/swiotlb.c	Sat Jul 07 11:44:16 2007 +0100
+++ b/arch/i386/kernel/swiotlb.c	Mon Jul 09 13:45:10 2007 +0100
@@ -480,7 +480,7 @@ swiotlb_map_single(struct device *hwdev,
 	 * we can safely return the device addr and not worry about bounce
 	 * buffering it.
 	 */
-	if (!range_straddles_page_boundary(ptr, size) &&
+	if (!range_straddles_page_boundary(__pa(ptr), size) &&
 	    !address_needs_mapping(hwdev, dev_addr))
 		return dev_addr;
 
@@ -577,7 +577,9 @@ swiotlb_map_sg(struct device *hwdev, str
 	for (i = 0; i < nelems; i++, sg++) {
 		dev_addr = gnttab_dma_map_page(sg->page) + sg->offset;
 
-		if (address_needs_mapping(hwdev, dev_addr)) {
+		if (range_straddles_page_boundary(page_to_pseudophys(sg->page)
+						  + sg->offset, sg->length)
+		    || address_needs_mapping(hwdev, dev_addr)) {
 			gnttab_dma_unmap_page(dev_addr);
 			buffer.page   = sg->page;
 			buffer.offset = sg->offset;
diff -r f833757672a7 -r 08cf42135056 include/asm-i386/mach-xen/asm/dma-mapping.h
--- a/include/asm-i386/mach-xen/asm/dma-mapping.h	Sat Jul 07 11:44:16 2007 +0100
+++ b/include/asm-i386/mach-xen/asm/dma-mapping.h	Mon Jul 09 13:45:10 2007 +0100
@@ -23,11 +23,11 @@ address_needs_mapping(struct device *hwd
 }
 
 static inline int
-range_straddles_page_boundary(void *p, size_t size)
+range_straddles_page_boundary(paddr_t p, size_t size)
 {
 	extern unsigned long *contiguous_bitmap;
-	return (((((unsigned long)p & ~PAGE_MASK) + size) > PAGE_SIZE) &&
-		!test_bit(__pa(p) >> PAGE_SHIFT, contiguous_bitmap));
+	return ((((p & ~PAGE_MASK) + size) > PAGE_SIZE) &&
+		!test_bit(p >> PAGE_SHIFT, contiguous_bitmap));
 }
 
 #define dma_alloc_noncoherent(d, s, h, f) dma_alloc_coherent(d, s, h, f)

[-- Attachment #3: Type: text/plain, Size: 138 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel