Thales
Thales
As explained in the [Open Questions](https://github.com/ra-kete/dma-poc/blob/master/README.md#open-questions), we need a system of traits to mark safe buffers types for DMA. This issue is open for anyone to propose a trait system...
`Addr::send` will actually check if the Mailbox is full, but the future created by it uses a clone of `AddressSender`, and due to how `AddressSender::clone` is implemented, the newly cloned...
Instead of having: ```rust pub struct Node { next: AtomicPtr, pub(crate) data: UnsafeCell, } ``` We could just use one location to store both things, since we don't need to...
This is a nice building block to implement client certificate verification, it's even used internally for another verifier in rustls. Currently, `new` returns `Arc` instead of `Self`, creating unneeded indirection...
This make it possible to re-use this to write other serial drivers, like an async one... https://github.com/akiles/embassy/pull/75
https://github.com/hyperium/h2/blob/a8af2358647cd2a68f851dfe3228cc3116654223/src/server.rs#L1378-L1379 This is arguably a bad default, given that it will block any server pushes unless the client explicitly sends a non-zero `MAX_CONCURRENT_STREAMS` setting. I don't think we should limit...
Quoting @adamgreig > if the dcache was enabled for the region the descriptor (or buffer) is in, it would need cleaning at this point (and the buffers would need invalidating...
Since this implementation uses a FIFO, the tx queue (in PRX mode) will stall if the first item to be popped doesn't match the pipe address of the received packet....