Re: [RFC] android: ion: How to properly clean caches for uncached allocations

From: Laura Abbott
Date: Thu Mar 08 2018 - 21:01:21 EST


On 03/08/2018 04:45 PM, Liam Mark wrote:
On Wed, 7 Mar 2018, Laura Abbott wrote:

On 02/28/2018 09:18 PM, Liam Mark wrote:
The issue:

Currently in ION if you allocate uncached memory it is possible that there
are still dirty lines in the cache. And often these dirty lines in the
cache are the zeros which were meant to clear out any sensitive kernel
data.

What this means is that if you allocate uncached memory from ION, and then
subsequently write to that buffer (using the uncached mapping you are
provided by ION) then the data you have written could be corrupted at some
point in the future if a dirty line is evicted from the cache.

Also this means there is a potential security issue. If an un-privileged
userspace user allocated uncached memory (for example from the system heap)
and then if they were to read from that buffer (through the un-cached
mapping they are provided by ION), and if some of the zeros which were
written to that memory are still in the cache then this un-privileged
userspace user could read potentially sensitive kernel data.

For the use case you are describing we don't actually need the
memory to be non-cached until it comes time to do the dma mapping.
Here's a proposal to shoot holes in:

- Before any dma_buf attach happens, all mmap mappings are cached
- At the time attach happens, we shoot down any existing userspace
mappings, do the dma_map with appropriate flags to clean the pages
and then allow remapping to userspace as uncached. Really this
looks like a variation on the old Ion faulting code which I removed
except it's for uncached buffers instead of cached buffers.


Thanks Laura, I will take a look to see if I can think of any concerns.

Initial thoughts.
- What about any kernel mappings (kmap/vmap) the client has made?


We could either synchronize with dma_buf_{begin,end}_cpu_access
or just disallow the mapping to happen if there's an outstanding
kmap or vmap. Is this an actual problem or only theoretical?

- I guess it would be tempting to only do this behavior for memory that
came from buddy (as opposed to the pool since it should be clean), but we
would need to be careful that no pages sneak into the pool without being
cleaned (example: client allocs then frees without ever call
dma_buf_attach).


You're welcome to try that optimization but I think we should
focus on the basics first. Honestly it might make sense just to
have a single pool at this point since the cost of syncing
is not happening on the allocation path.

Potential problems:
- I'm not 100% about the behavior here if the attaching device
is already dma_coherent. I also consider uncached mappings
enough of a device specific optimization that you shouldn't
do them unless you know it's needed.

I don't believe we want to allow uncached memory to be dma mapped by an
io-coherent device and this is something I would like to eventually block.

Since there is always a kernel cached mapping for ION uncached memory then
speculative access could still be putting lines in the cache, so when an
io-coherent device tries to read this uncached memory its snoop into the
cache could find one of these 'stale' clean cache lines and end up using
stale data.
Agree?


Sounds right.

- The locking/sequencing with userspace could be tricky
since userspace may not like us ripping mappings out from
underneath if it's trying to access.

Perhaps delay this work to the dma_map_attachment call since when the data
is dma mapped the CPU shouldn't be accessing it?

Or worst case perhaps fail all map attempts to uncached memory until the
memory has been dma mapped (and cleaned) at least once?


My concern was mostly concurrent userspace access on a buffer
that's being dma_mapped but that sounds racy to begin with.

I suggested disallowing mmap until dma_mapping before and I thought
that was not possible?

Thanks,
Laura

Thanks,
Liam

Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project