On Thu, 6 Mar 2025 14:45:05 +0100 Danilo Krummrich <dakr@xxxxxxxxxx> wrote: > On Thu, Mar 06, 2025 at 09:56:38PM +0900, FUJITA Tomonori wrote: >> On Tue, 4 Mar 2025 18:34:52 +0100 >> Danilo Krummrich <dakr@xxxxxxxxxx> wrote: >> >> > +Delay / Sleep abstractions >> > +-------------------------- >> > + >> > +Rust abstractions for the kernel's delay() and sleep() functions. >> > + >> > +There is some ongoing work from FUJITA Tomonori [1], which has not seen any updates >> > +since Oct. 24. >> > + >> > +| Complexity: Beginner >> > +| Link: https://lore.kernel.org/netdev/20241001112512.4861-2-fujita.tomonori@xxxxxxxxx/ [1] >> >> I posted v11 last month. >> >> https://lore.kernel.org/netdev/20250220070611.214262-1-fujita.tomonori@xxxxxxxxx/ > > Thanks for letting me know. > > I think I lost track of this because in v1 the series was named "add delay > abstraction (sleep functions)" and with v2 it was switched to "rust: Add IO > polling" and I was searching for subsequent patch series with the "delay" > keyword. I see. During the review process, I changed the subject due to the consensus that, in most cases, device drivers should use read_poll_timeout instead of calling the sleep function directly. > Anyways, AFAICS you ended up with adding fsleep(). I think nova-core will need > control over having a busy loop or actually re-schedule. I plan to add read_poll_timeout_atomic() with delay functions: https://lore.kernel.org/netdev/20250228.080550.354359820929821928.fujita.tomonori@xxxxxxxxx/ delay functions need Delta struct in the patchset so the patchset needs to be merged first.
