Messages in this thread | | | From | Jiasheng Jiang <> | Subject | Re: Re: [PATCH v2] sched: Fix implicit type conversion | Date | Mon, 25 Oct 2021 13:29:41 +0000 |
| |
On Mon, Oct 25, 2021 at 12:43:55AM +0000, Peter Zijlstra wrote: >> The variable 'n' is defined as ULONG. However in the cpumask_next(), >> it is used as INT. >> That is vulnerable and may cause overflow. >> For example, if the value of 'n' is (2^31 - 1), then it can pass the
>That would need nr_cpu_ids to be that large. How's that going to happen?
First, maybe it is hard to exploit it now, but who knows the future. Second, the patch cost much less than the loss caused by the unexpected input. Third, it is universally accepted that the implicit type conversion is vulnerable. Therefore, it will set an example for others that having the good programming custom.
| |