[PATCH 10/17] prmem: documentation

[Andy Lutomirski]

> On Oct 30, 2018, at 1:43 PM, Igor Stoppa <igor.stoppa at gmail.com> wrote:
> 
>> On 30/10/2018 21:20, Matthew Wilcox wrote:
>>> On Tue, Oct 30, 2018 at 12:28:41PM -0600, Tycho Andersen wrote:
>>>> On Tue, Oct 30, 2018 at 10:58:14AM -0700, Matthew Wilcox wrote:
>>>> On Tue, Oct 30, 2018 at 10:06:51AM -0700, Andy Lutomirski wrote:
>>>>>> On Oct 30, 2018, at 9:37 AM, Kees Cook <keescook at chromium.org> wrote:
>>>>> I support the addition of a rare-write mechanism to the upstream kernel.
>>>>> And I think that there is only one sane way to implement it: using an
>>>>> mm_struct. That mm_struct, just like any sane mm_struct, should only
>>>>> differ from init_mm in that it has extra mappings in the *user* region.
>>>> 
>>>> I'd like to understand this approach a little better.  In a syscall path,
>>>> we run with the user task's mm.  What you're proposing is that when we
>>>> want to modify rare data, we switch to rare_mm which contains a
>>>> writable mapping to all the kernel data which is rare-write.
>>>> 
>>>> So the API might look something like this:
>>>> 
>>>>    void *p = rare_alloc(...);    /* writable pointer */
>>>>    p->a = x;
>>>>    q = rare_protect(p);        /* read-only pointer */
> 
> With pools and memory allocated from vmap_areas, I was able to say
> 
> protect(pool)
> 
> and that would do a swipe on all the pages currently in use.
> In the SELinux policyDB, for example, one doesn't really want to individually protect each allocation.
> 
> The loading phase happens usually at boot, when the system can be assumed to be sane (one might even preload a bare-bone set of rules from initramfs and then replace it later on, with the full blown set).
> 
> There is no need to process each of these tens of thousands allocations and initialization as write-rare.
> 
> Would it be possible to do the same here?

I don’t see why not, although getting the API right will be a tad complicated.

> 
>>>> 
>>>> To subsequently modify q,
>>>> 
>>>>    p = rare_modify(q);
>>>>    q->a = y;
>>> 
>>> Do you mean
>>> 
>>>     p->a = y;
>>> 
>>> here? I assume the intent is that q isn't writable ever, but that's
>>> the one we have in the structure at rest.
>> Yes, that was my intent, thanks.
>> To handle the list case that Igor has pointed out, you might want to
>> do something like this:
>>    list_for_each_entry(x, &xs, entry) {
>>        struct foo *writable = rare_modify(entry);
> 
> Would this mapping be impossible to spoof by other cores?
> 

Indeed. Only the core with the special mm loaded could see it.

But I dislike allowing regular writes in the protected region. We really only need four write primitives:

1. Just write one value.  Call at any time (except NMI).

2. Just copy some bytes. Same as (1) but any number of bytes.

3,4: Same as 1 and 2 but must be called inside a special rare write region. This is purely an optimization.

Actually getting a modifiable pointer should be disallowed for two reasons:

1. Some architectures may want to use a special write-different-address-space operation. Heck, x86 could, too: make the actual offset be a secret and shove the offset into FSBASE or similar. Then %fs-prefixed writes would do the rare writes.

2. Alternatively, x86 could set the U bit. Then the actual writes would use the uaccess helpers, giving extra protection via SMAP.

We don’t really want a situation where an unchecked pointer in the rare write region completely defeats the mechanism.