On Fri, Sep 18, 2026 at 05:52:31PM +0100, Catalin Marinas wrote:
TBH, I'm tempted not to support this configuration at all, just fail
gracefully if the realm page size doesn't match the host's. However, if
that's a real world configuration,
realm != host is going to be fairly common.
host > realm is probably not that common, I would guess.
I wonder whether could we leverage the swiotlb mechanism but with a
separate io_tlb_mem as a shared coco allocator with the right
alignment and use this pool when force_dma_unencrypted() instead of
specific alloc_pages() with adjusted order.
I propose we fix the far too low level set memory decrypted apis into
proper allocators, with APIs that can handle a round up. Then see
where we end up after that.
If we can hide 90% of it in an allocator then lets see if we can open
code the rest.
Jason