Re: [PATCH] fix error handling in load_module()

[Tejun Heo <tj@kernel.org>]

Hello, Rusty.

Rusty Russell wrote:
>> My reverse engineering of the secret, undocumented percpu_modfree()
>> indicates that its mad inventor intended that percpu_modfree(NULL) be a
>> valid thing to do.
> 
> Yes, percpu_modfree() should handle NULL.  If it doesn't, that's the bug.

Hmm... It seems the report was against pre-2.6.32-rcX window.  ppc is
still using the original percpu_modfree() and I don't think the it
ever supported NULL free.  In the current linus tree, the triggering
BUG() would be at line 500 which is inside percpu_modfree() which
checks @freeme matches any allocated address and if not triggers
BUG().

I think something like the following should fix it.

 kernel/module.c |    3 +++
 1 file changed, 3 insertions(+)

diff --git a/kernel/module.c b/kernel/module.c
index 05ce49c..eed1f9e 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -490,6 +490,9 @@ static void percpu_modfree(void *freeme)
 	void *ptr = __per_cpu_start + block_size(pcpu_size[0]);
 	int cpu;

+	if (unlikely(!freeme))
+		return;
+
 	/* First entry is core kernel percpu data. */
 	for (i = 1; i < pcpu_num_used; ptr += block_size(pcpu_size[i]), i++) {
 		if (ptr == freeme) {

-- 
tejun