[PATCH] Introduce Tyr

[FUJITA Tomonori]

Hi,

On Sat, 28 Jun 2025 10:05:11 -0300
Daniel Almeida <daniel.almeida at collabora.com> wrote:

> Anyway, I thought that having a branch would be more tidy than listing them, as
> the branch shows in what order they're applied and etc. For example, the patch
> for read_poll_timeout was cherry-picked from Fujita's v12 series, and that was
> subsequently dropped in v13 until the rest of the series was merged on v15. I
> thought that referring to v12 of that series would be slightly confusing.
> 
> IOW: this should be appliable as soon as the dependencies themselves are
> merged. I am hoping that this can happen on the 6.17 merge window as the
> comments on most of them appear to be dying down so maybe the r-b's will start
> coming soon. It also gives a user to read_poll_timeout(), which may prompt Fujita
> to keep working on it.

I expect read_poll_timeout() to be merged during the 6.18 window.

To explain the situation, read_poll_timeout() depends on fsleep() and
might_sleep() abstractions.

I expect the former to be part of what Andreas is preparing to merge
for the 6.17 window, along with the patchset converting hrtimer to use
Instant and Delta.

Boqun has submitted the latter as a pull request to the tip tree for
inclusion in the 6.17 window.

Since the two features are being merged through different trees and I
don’t want to complicate the process, I’m planning to target
read_poll_timeout() for the 6.18 merge window.

If you're targeting this driver for 6.17, it might be safer to
implement a similar functionality like the nova driver did and replace
it later.


Thanks,