mm/memory.c: actually remap enough memory
authorGrazvydas Ignotas <notasas@gmail.com>
Thu, 12 Feb 2015 23:00:19 +0000 (15:00 -0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 18 Mar 2015 12:22:28 +0000 (13:22 +0100)
commit 9cb12d7b4ccaa976f97ce0c5fd0f1b6a83bc2a75 upstream.

For whatever reason, generic_access_phys() only remaps one page, but
actually allows to access arbitrary size.  It's quite easy to trigger
large reads, like printing out large structure with gdb, which leads to a
crash.  Fix it by remapping correct size.

Fixes: 28b2ee20c7cb ("access_process_vm device memory infrastructure")
Signed-off-by: Grazvydas Ignotas <notasas@gmail.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
mm/memory.c

index 8b2d75f61b324ef5795ccd61e62c64952b6ff84e..04232bb173f05a7d2604b470643fee10cdc074da 100644 (file)
@@ -4088,7 +4088,7 @@ int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
        if (follow_phys(vma, addr, write, &prot, &phys_addr))
                return -EINVAL;
 
-       maddr = ioremap_prot(phys_addr, PAGE_SIZE, prot);
+       maddr = ioremap_prot(phys_addr, PAGE_ALIGN(len + offset), prot);
        if (write)
                memcpy_toio(maddr + offset, buf, len);
        else