[PATCH] mm: use PFN_PHYS() in phys_mem_access_prot()

[Ivan Ezequiel Rodriguez <ivanrwcm25@gmail.com>]

From: Iván Ezequiel Rodriguez <ivanrwcm25@gmail.com>

phys_mem_access_prot() stores "pfn << PAGE_SHIFT" in a phys_addr_t.
The shift is evaluated as unsigned long, so on a 32-bit kernel with
CONFIG_PHYS_ADDR_T_64BIT an address above 4 GiB is truncated before
uncached_access() compares it with __pa(high_memory).

mmap_mem_prepare() already shifts in phys_addr_t, so the mapping is
accepted and then left cacheable. PFN_PHYS() casts before the shift.

On a MIPS32 XPA Malta guest, mmap of /dev/mem at 0xf00000000 without
O_DSYNC used the boot CCA. The same mapping with O_DSYNC was uncached.
With PFN_PHYS() both are uncached.

Fixes: 7e6735c3578e ("/dev/mem: use phys_addr_t for physical addresses")
Cc: stable@vger.kernel.org
Signed-off-by: Iván Ezequiel Rodriguez <ivanrwcm25@gmail.com>
---
 mm/char-mem.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/char-mem.c b/mm/char-mem.c
index c5761ddedbd7..53ce0034a4c1 100644
--- a/mm/char-mem.c
+++ b/mm/char-mem.c
@@ -272,7 +272,7 @@ static pgprot_t phys_mem_access_prot(struct file *file, unsigned long pfn,
 				     unsigned long size, pgprot_t vma_prot)
 {
 #ifdef pgprot_noncached
-	phys_addr_t offset = pfn << PAGE_SHIFT;
+	phys_addr_t offset = PFN_PHYS(pfn);
 
 	if (uncached_access(file, offset))
 		return pgprot_noncached(vma_prot);
-- 
2.43.0