On Wed, 2018-11-28 at 15:11 -0800, Andrew Morton wrote:

On Tue, 27 Nov 2018 16:07:54 -0800 Rick Edgecombe [off-list ref]
wrote:

quoted

Change the module allocations to flush before freeing the pages.

...

--- a/arch/x86/kernel/module.c
+++ b/arch/x86/kernel/module.c

@@ -87,8 +87,8 @@ void *module_alloc(unsigned long size)
 	p = __vmalloc_node_range(size, MODULE_ALIGN,
 				    MODULES_VADDR + get_module_load_offset(),
 				    MODULES_END, GFP_KERNEL,
-				    PAGE_KERNEL_EXEC, 0, NUMA_NO_NODE,
-				    __builtin_return_address(0));
+				    PAGE_KERNEL_EXEC, VM_IMMEDIATE_UNMAP,
+				    NUMA_NO_NODE, __builtin_return_address(0));
 	if (p && (kasan_module_alloc(p, size) < 0)) {
 		vfree(p);
 		return NULL;

Should any other architectures do this?

I would think everything that has something like an NX bit and doesn't use the
default module_alloc implementation.

I could add the flag for every arch that defines PAGE_KERNEL_EXEC, but I don't
have a good way to test on all of those architectures.

Thanks,

Rick