Re: [PATCH v3] mm: fix unexpected changes to {failslab|fail_page_alloc}.attr

From: Qi Zheng
Date: Fri Nov 18 2022 - 20:43:07 EST




On 2022/11/19 05:42, Andrew Morton wrote:
On Fri, 18 Nov 2022 18:00:11 +0800 Qi Zheng <zhengqi.arch@xxxxxxxxxxxxx> wrote:

When we specify __GFP_NOWARN, we only expect that no warnings
will be issued for current caller. But in the __should_failslab()
and __should_fail_alloc_page(), the local GFP flags alter the
global {failslab|fail_page_alloc}.attr, which is persistent and
shared by all tasks. This is not what we expected, let's fix it.

Cc: stable@xxxxxxxxxxxxxxx
Fixes: 3f913fc5f974 ("mm: fix missing handler for __GFP_NOWARN")
Reported-by: Dmitry Vyukov <dvyukov@xxxxxxxxxx>
Signed-off-by: Qi Zheng <zhengqi.arch@xxxxxxxxxxxxx>
Reviewed-by: Akinobu Mita <akinobu.mita@xxxxxxxxx>

...

-bool should_fail(struct fault_attr *attr, ssize_t size)
+bool should_fail_ex(struct fault_attr *attr, ssize_t size, int flags)
{
bool stack_checked = false;
@@ -152,13 +149,20 @@ bool should_fail(struct fault_attr *attr, ssize_t size)
return false;
fail:
- fail_dump(attr);
+ if (!(flags & FAULT_NOWARN))
+ fail_dump(attr);
if (atomic_read(&attr->times) != -1)
atomic_dec_not_zero(&attr->times);
return true;
}
+EXPORT_SYMBOL_GPL(should_fail_ex);

I don't see a need to export this?

Yes, my initial thought was that there might be a driver using this
function, but there really isn't one yet.

And I see you've helped remove this, thanks a lot. :)




--
Thanks,
Qi