The patch titled
Subject: scatterlist: fix sg_phys() masking
has been removed from the -mm tree. Its filename was
scatterlist-fix-sg_phys-masking.patch
This patch was dropped because an alternative patch was merged
------------------------------------------------------
From: Dan Williams <[email protected]>
Subject: scatterlist: fix sg_phys() masking
commit db0fa0cb0157 "scatterlist: use sg_phys()" did replacements of the
form:
phys_addr_t phys = page_to_phys(sg_page(s));
phys_addr_t phys = sg_phys(s) & PAGE_MASK;
However, this breaks platforms where sizeof(phys_addr_t) > sizeof(unsigned
long). Since PAGE_MASK is an unsigned long this inadvertently masks the
high bits returned by sg_phys(). Convert to PHYSICAL_PAGE_MASK in these
cases which will do the proper sign extension.
As caught by the kbuild robot, a generic fallback definition of
PHYSICAL_PAGE_MASK is needed for several archs.
Signed-off-by: Dan Williams <[email protected]>
Reported-by: Vitaly Lavrov <[email protected]>
Cc: Jens Axboe <[email protected]>
Cc: Joerg Roedel <[email protected]>
Cc: Christoph Hellwig <[email protected]>
Cc: Russell King <[email protected]>
Cc: David Woodhouse <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
---
arch/arm/mm/dma-mapping.c | 2 +-
drivers/iommu/intel-iommu.c | 2 +-
drivers/staging/android/ion/ion_chunk_heap.c | 4 ++--
include/linux/mm.h | 12 ++++++++++++
4 files changed, 16 insertions(+), 4 deletions(-)
diff -puN arch/arm/mm/dma-mapping.c~scatterlist-fix-sg_phys-masking
arch/arm/mm/dma-mapping.c
--- a/arch/arm/mm/dma-mapping.c~scatterlist-fix-sg_phys-masking
+++ a/arch/arm/mm/dma-mapping.c
@@ -1521,7 +1521,7 @@ static int __map_sg_chunk(struct device
return -ENOMEM;
for (count = 0, s = sg; count < (size >> PAGE_SHIFT); s = sg_next(s)) {
- phys_addr_t phys = sg_phys(s) & PAGE_MASK;
+ phys_addr_t phys = sg_phys(s) & PHYSICAL_PAGE_MASK;
unsigned int len = PAGE_ALIGN(s->offset + s->length);
if (!is_coherent &&
diff -puN drivers/iommu/intel-iommu.c~scatterlist-fix-sg_phys-masking
drivers/iommu/intel-iommu.c
--- a/drivers/iommu/intel-iommu.c~scatterlist-fix-sg_phys-masking
+++ a/drivers/iommu/intel-iommu.c
@@ -2159,7 +2159,7 @@ static int __domain_mapping(struct dmar_
sg_res = aligned_nrpages(sg->offset, sg->length);
sg->dma_address = ((dma_addr_t)iov_pfn <<
VTD_PAGE_SHIFT) + sg->offset;
sg->dma_length = sg->length;
- pteval = (sg_phys(sg) & PAGE_MASK) | prot;
+ pteval = (sg_phys(sg) & PHYSICAL_PAGE_MASK) | prot;
phys_pfn = pteval >> VTD_PAGE_SHIFT;
}
diff -puN
drivers/staging/android/ion/ion_chunk_heap.c~scatterlist-fix-sg_phys-masking
drivers/staging/android/ion/ion_chunk_heap.c
---
a/drivers/staging/android/ion/ion_chunk_heap.c~scatterlist-fix-sg_phys-masking
+++ a/drivers/staging/android/ion/ion_chunk_heap.c
@@ -81,7 +81,7 @@ static int ion_chunk_heap_allocate(struc
err:
sg = table->sgl;
for (i -= 1; i >= 0; i--) {
- gen_pool_free(chunk_heap->pool, sg_phys(sg) & PAGE_MASK,
+ gen_pool_free(chunk_heap->pool, sg_phys(sg) &
PHYSICAL_PAGE_MASK,
sg->length);
sg = sg_next(sg);
}
@@ -109,7 +109,7 @@ static void ion_chunk_heap_free(struct i
DMA_BIDIRECTIONAL);
for_each_sg(table->sgl, sg, table->nents, i) {
- gen_pool_free(chunk_heap->pool, sg_phys(sg) & PAGE_MASK,
+ gen_pool_free(chunk_heap->pool, sg_phys(sg) &
PHYSICAL_PAGE_MASK,
sg->length);
}
chunk_heap->allocated -= allocated_size;
diff -puN include/linux/mm.h~scatterlist-fix-sg_phys-masking include/linux/mm.h
--- a/include/linux/mm.h~scatterlist-fix-sg_phys-masking
+++ a/include/linux/mm.h
@@ -90,6 +90,18 @@ extern int overcommit_kbytes_handler(str
/* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE
*/
#define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)addr, PAGE_SIZE)
+#ifndef PHYSICAL_PAGE_MASK
+/*
+ * Cast *PAGE_MASK to a signed type so that it is sign-extended if
+ * virtual addresses are 32-bits but physical addresses are larger (ie,
+ * 32-bit PAE).
+ *
+ * An arch may redefine this to mask out values outside the max
+ * address-width of the cpu.
+ */
+#define PHYSICAL_PAGE_MASK ((signed long) PAGE_MASK)
+#endif
+
/*
* Linux kernel virtual memory manager primitives.
* The idea being to have a "virtual" mm in the same way
_
Patches currently in -mm which might be from [email protected] are
pmem-dax-clean-up-clear_pmem.patch
dax-increase-granularity-of-dax_clear_blocks-operations.patch
dax-guarantee-page-aligned-results-from-bdev_direct_access.patch
dax-fix-lifetime-of-in-kernel-dax-mappings-with-dax_map_atomic.patch
dax-fix-lifetime-of-in-kernel-dax-mappings-with-dax_map_atomic-v3.patch
um-kill-pfn_t.patch
kvm-rename-pfn_t-to-kvm_pfn_t.patch
mm-dax-pmem-introduce-pfn_t.patch
mm-dax-pmem-introduce-pfn_t-v3.patch
mm-introduce-find_dev_pagemap.patch
x86-mm-introduce-vmem_altmap-to-augment-vmemmap_populate.patch
libnvdimm-pfn-pmem-allocate-memmap-array-in-persistent-memory.patch
avr32-convert-to-asm-generic-memory_modelh.patch
hugetlb-fix-compile-error-on-tile.patch
frv-fix-compiler-warning-from-definition-of-__pmd.patch
x86-mm-introduce-_page_devmap.patch
mm-dax-gpu-convert-vm_insert_mixed-to-pfn_t.patch
mm-dax-convert-vmf_insert_pfn_pmd-to-pfn_t.patch
list-introduce-list_del_poison.patch
libnvdimm-pmem-move-request_queue-allocation-earlier-in-probe.patch
mm-dax-pmem-introduce-getput_dev_pagemap-for-dax-gup.patch
mm-dax-dax-pmd-vs-thp-pmd-vs-hugetlbfs-pmd.patch
mm-x86-get_user_pages-for-dax-mappings.patch
dax-provide-diagnostics-for-pmd-mapping-failures.patch
dax-re-enable-dax-pmd-mappings.patch
--
To unsubscribe from this list: send the line "unsubscribe stable" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
