Skip to contentSkip to conversation

powerepc/book3s64/hash: Align start/end address correctly with bolt mapping

PATCH Applied Aneesh Kumar K.V 2 participants 2 messages Sep 7, 2020 – Sep 17, 2020 PowerPC

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 powerepc/book3s64/hash: Align start/end address correctly with bolt mapping

This ensures we don't do a partial mapping of memory. With nvdimm, when creating namespaces with size not aligned to 16MB, the kernel ends up partially mapping the pages. This can result in kernel adding multiple hash page table entries for the same range. A new namespace will result in create_section_mapping() with start and end overlapping an already existing bolted hash page table entry.

commit: 6acd7d5ef264 ("libnvdimm/namespace: Enforce memremap_compat_align()") made sure that we always create namespaces aligned to 16MB. But we can do better by avoiding mapping pages that are not aligned. This helps to catch access to these partially mapped pages early.

---
 arch/powerpc/mm/book3s64/hash_utils.c    | 12 +++++++++---
 arch/powerpc/mm/book3s64/radix_pgtable.c |  1 +
 2 files changed, 10 insertions(+), 3 deletions(-)
arch/powerpc/mm/book3s64/hash_utils.c +9 -3
index c663e7ba801f..7185bc43b24f 100644
@@ -260,8 +260,12 @@ int htab_bolt_mapping(unsigned long vstart, unsigned long vend,
 	DBG("htab_bolt_mapping(%lx..%lx -> %lx (%lx,%d,%d)\n",
 	    vstart, vend, pstart, prot, psize, ssize);
 
-	for (vaddr = vstart, paddr = pstart; vaddr < vend;
-	     vaddr += step, paddr += step) {
+	/* Carefully map only the possible range */
+	vaddr = ALIGN(vstart, step);
+	paddr = ALIGN(pstart, step);
+	vend  = ALIGN_DOWN(vend, step);
+
+	for (; vaddr < vend; vaddr += step, paddr += step) {
 		unsigned long hash, hpteg;
 		unsigned long vsid = get_kernel_vsid(vaddr, ssize);
 		unsigned long vpn  = hpt_vpn(vaddr, vsid, ssize);
@@ -343,7 +347,9 @@ int htab_remove_mapping(unsigned long vstart, unsigned long vend,
 	if (!mmu_hash_ops.hpte_removebolted)
 		return -ENODEV;
 
-	for (vaddr = vstart; vaddr < vend; vaddr += step) {
+	/* Unmap the full range specificied */
+	vaddr = ALIGN_DOWN(vstart, step);
+	for (;vaddr < vend; vaddr += step) {
 		rc = mmu_hash_ops.hpte_removebolted(vaddr, psize, ssize);
 		if (rc == -ENOENT) {
 			ret = -ENOENT;
arch/powerpc/mm/book3s64/radix_pgtable.c +1 -0
index d5f0c10d752a..5c8adeb8c955 100644
@@ -276,6 +276,7 @@ static int __meminit create_physical_mapping(unsigned long start,
 	int psize;
 
 	start = ALIGN(start, PAGE_SIZE);
+	end   = ALIGN_DOWN(end, PAGE_SIZE);
 	for (addr = start; addr < end; addr += mapping_size) {
 		unsigned long gap, previous_size;
 		int rc;
Signature · 1 line
2.26.2
#1 lore raw
Quoted 8 lines · Aneesh Kumar K.V

This ensures we don't do a partial mapping of memory. With nvdimm, when creating namespaces with size not aligned to 16MB, the kernel ends up partially mapping the pages. This can result in kernel adding multiple hash page table entries for the same range. A new namespace will result in create_section_mapping() with start and end overlapping an already existing bolted hash page table entry.

[...]

Applied to powerpc/next.

[1/1] powerepc/book3s64/hash: Align start/end address correctly with bolt mapping

      https://git.kernel.org/powerpc/c/79b123cdf9cf0d4a1620baa8c611962626323a08

cheers