powerpc/mm: Fix false detection of read faults

To support detection of read faults with Radix execute-only memory, the
vma_is_accessible() check in access_error() (which checks for PROT_NONE)
was replaced with a check to see if VM_READ was missing, and if so,
returns true to assert the fault was caused by a bad read.

This is incorrect, as it ignores that both VM_WRITE and VM_EXEC imply
read on powerpc, as defined in protection_map[].  This causes mappings
containing VM_WRITE or VM_EXEC without VM_READ to misreport the cause of
page faults, since the MMU is still allowing reads.

Correct this by restoring the original vma_is_accessible() check for
PROT_NONE mappings, and adding a separate check for Radix PROT_EXEC-only
mappings.

Fixes: 395cac7 ("powerpc/mm: Support execute-only memory on the Radix MMU")
Reported-by: Michal Suchánek <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Tested-by: Benjamin Gray <[email protected]>
Signed-off-by: Russell Currey <[email protected]>
Signed-off-by: Michael Ellerman <[email protected]>
Link: https://msgid.link/[email protected]

Author: ruscur

arch/powerpc/mm/fault.c

@@ -271,11 +271,16 @@ static bool access_error(bool is_write, bool is_exec, struct vm_area_struct *vma
 	}
 
 	/*
-	 * Check for a read fault.  This could be caused by a read on an
-	 * inaccessible page (i.e. PROT_NONE), or a Radix MMU execute-only page.
+	 * VM_READ, VM_WRITE and VM_EXEC all imply read permissions, as
+	 * defined in protection_map[].  Read faults can only be caused by
+	 * a PROT_NONE mapping, or with a PROT_EXEC-only mapping on Radix.
 	 */
-	if (unlikely(!(vma->vm_flags & VM_READ)))
+	if (unlikely(!vma_is_accessible(vma)))
 		return true;
+
+	if (unlikely(radix_enabled() && ((vma->vm_flags & VM_ACCESS_FLAGS) == VM_EXEC)))
+		return true;
+
 	/*
 	 * We should ideally do the vma pkey access check here. But in the
 	 * fault path, handle_mm_fault() also does the same check. To avoid