On 2022/12/29 6:17, Andrew Morton wrote:
> On Mon, 5 Dec 2022 11:41:05 +0800 Wupeng Ma <mawupeng1@xxxxxxxxxx> wrote:
>
>> While testing mlock, we have a problem if the len of mlock is ULONG_MAX.
>> The return value of mlock is zero. But nothing will be locked since the
>> len in do_mlock overflows to zero due to the following code in mlock:
>>
>> len = PAGE_ALIGN(len + (offset_in_page(start)));
>>
>> The same problem happens in munlock.
>>
>> Since TASK_SIZE is the maximum user space address. The start or len of
>> mlock shouldn't be bigger than this. Function access_ok can be used to
>> check this issue, so return -EINVAL if bigger.
>
> What happens if userspace uses a value somewhat smaller than ULONG_MAX?
>
> mlock(addr, ULONG_MAX - 1000000);
>
> ?
>
> Because if the above works successfully and if it no longer works
> successfully with this patchset then that could be a
> backward-compatibility problem.
For mlock:
mlock(addr, ULONG_MAX - 1000000) will ret -1 and the errno is EINVAL(22) due to
the following call trace.
do_mlock
apply_vma_lock_flags
end = start + len;
if (end < start)
return -EINVAL;
Just like you said, we need to keep backward-compatibility. Maybe we can only catch and fix
the overflowing scenarios since they are absolutely wrong. here is the diff:
diff --git a/mm/mlock.c b/mm/mlock.c
index 7032f6dd0ce1..fd5e857ab245 100644
--- a/mm/mlock.c
+++ b/mm/mlock.c
@@ -569,6 +569,7 @@ static __must_check int do_mlock(unsigned long start, size_t len, vm_flags_t fla
unsigned long locked;
unsigned long lock_limit;
int error = -ENOMEM;
+ size_t old_len = len;
start = untagged_addr(start);
@@ -578,6 +579,9 @@ static __must_check int do_mlock(unsigned long start, size_t len, vm_flags_t fla
len = PAGE_ALIGN(len + (offset_in_page(start)));
start &= PAGE_MASK;
+ if (old_len != 0 && len == 0)
+ return -EINVAL;
+
lock_limit = rlimit(RLIMIT_MEMLOCK);
lock_limit >>= PAGE_SHIFT;
locked = len >> PAGE_SHIFT;
>
