Re: strace of io_uring events?
From: Jens Axboe
Date: Tue Jul 21 2020 - 13:30:55 EST
On 7/21/20 11:23 AM, Andy Lutomirski wrote:
> On Tue, Jul 21, 2020 at 8:31 AM Jens Axboe <axboe@xxxxxxxxx> wrote:
>>
>> On 7/21/20 9:27 AM, Andy Lutomirski wrote:
>>> On Fri, Jul 17, 2020 at 1:02 AM Stefano Garzarella <sgarzare@xxxxxxxxxx> wrote:
>>>>
>>>> On Thu, Jul 16, 2020 at 08:12:35AM -0700, Kees Cook wrote:
>>>>> On Thu, Jul 16, 2020 at 03:14:04PM +0200, Stefano Garzarella wrote:
>>>
>>>>> access (IIUC) is possible without actually calling any of the io_uring
>>>>> syscalls. Is that correct? A process would receive an fd (via SCM_RIGHTS,
>>>>> pidfd_getfd, or soon seccomp addfd), and then call mmap() on it to gain
>>>>> access to the SQ and CQ, and off it goes? (The only glitch I see is
>>>>> waking up the worker thread?)
>>>>
>>>> It is true only if the io_uring istance is created with SQPOLL flag (not the
>>>> default behaviour and it requires CAP_SYS_ADMIN). In this case the
>>>> kthread is created and you can also set an higher idle time for it, so
>>>> also the waking up syscall can be avoided.
>>>
>>> I stared at the io_uring code for a while, and I'm wondering if we're
>>> approaching this the wrong way. It seems to me that most of the
>>> complications here come from the fact that io_uring SQEs don't clearly
>>> belong to any particular security principle. (We have struct creds,
>>> but we don't really have a task or mm.) But I'm also not convinced
>>> that io_uring actually supports cross-mm submission except by accident
>>> -- as it stands, unless a user is very careful to only submit SQEs
>>> that don't use user pointers, the results will be unpredictable.
>>
>> How so?
>
> Unless I've missed something, either current->mm or sqo_mm will be
> used depending on which thread ends up doing the IO. (And there might
> be similar issues with threads.) Having the user memory references
> end up somewhere that is an implementation detail seems suboptimal.
current->mm is always used from the entering task - obviously if done
synchronously, but also if it needs to go async. The only exception is a
setup with SQPOLL, in which case ctx->sqo_mm is the task that set up the
ring. SQPOLL requires root privileges to setup, and there's no task
entering the io_uring at all necessarily. It'll just submit sqes with
the credentials that are registered with the ring.
>>> Perhaps we can get away with this:
>>>
>>> diff --git a/fs/io_uring.c b/fs/io_uring.c
>>> index 74bc4a04befa..92266f869174 100644
>>> --- a/fs/io_uring.c
>>> +++ b/fs/io_uring.c
>>> @@ -7660,6 +7660,20 @@ SYSCALL_DEFINE6(io_uring_enter, unsigned int,
>>> fd, u32, to_submit,
>>> if (!percpu_ref_tryget(&ctx->refs))
>>> goto out_fput;
>>>
>>> + if (unlikely(current->mm != ctx->sqo_mm)) {
>>> + /*
>>> + * The mm used to process SQEs will be current->mm or
>>> + * ctx->sqo_mm depending on which submission path is used.
>>> + * It's also unclear who is responsible for an SQE submitted
>>> + * out-of-process from a security and auditing perspective.
>>> + *
>>> + * Until a real usecase emerges and there are clear semantics
>>> + * for out-of-process submission, disallow it.
>>> + */
>>> + ret = -EACCES;
>>> + goto out;
>>> + }
>>> +
>>> /*
>>> * For SQ polling, the thread will do all submissions and completions.
>>> * Just return the requested submit count, and wake the thread if
>>
>> That'll break postgres that already uses this, also see:
>>
>> commit 73e08e711d9c1d79fae01daed4b0e1fee5f8a275
>> Author: Jens Axboe <axboe@xxxxxxxxx>
>> Date: Sun Jan 26 09:53:12 2020 -0700
>>
>> Revert "io_uring: only allow submit from owning task"
>>
>> So no, we can't do that.
>>
>
> Yikes, I missed that.
>
> Andres, how final is your Postgres branch? I'm wondering if we could
> get away with requiring a special flag when creating an io_uring to
> indicate that you intend to submit IO from outside the creating mm.
>
> Even if we can't make this change, we could plausibly get away with
> tying seccomp-style filtering to sqo_mm. IOW we'd look up a
> hypothetical sqo_mm->io_uring_filters to filter SQEs even when
> submitted from a different mm.
This is just one known use case, there may very well be others. Outside
of SQPOLL, which is special, I don't see a reason to restrict this.
Given that you may have a fuller understanding of it after the above
explanation, please clearly state what problem you're seeing that
warrants a change.
--
Jens Axboe