Kernel developers seem to be excited to discuss the benefits of a possible rust code approval in the Linux kernel. Linus Torvalds himself made some initial comments on the subject before the Linux Plumbers 2020 conference, in which the matter is discussed in detail.
The mailing list thread, when discussing Greg Kroah-Hartman’s earlier comments on the prospect of rust for the kernel, mentioned that one of the conditions sought is that it is effectively disabled by default until sufficient testing has been done.
Linus Torvalds, however, interfered with his own opinion. Linus commented that he wants it to be activated effectively by default to ensure that there are widespread tests and not isolated use where developers can do “crazy”
No, please make it an automatic configuration option “Is rust available”. This is exactly how we make the compiler versions and check at configuration time whether compiler flags are available.
See init / Kconfig for things like
def_bool $ (success, $ (LD) -v | head -n 1 | grep -q LLD)
and the grate support should be similar. Something like
def_bool $ (success, $ (RUST) ..sometest ..)
because I don’t want us to be in a situation where new rust support isn’t even tested on builds by default.
On the contrary. I want the first grate driver (or whatever) to be introduced in such a simple format that mistakes are obvious and easy.
The worst situation is that a (small) group of people starts testing their very special situation and doing bad and crazy things because “nobody else cares, it’s hidden”.
At least he didn’t shoot down the idea of rust support in the kernel. We’ll see where the direction of Rust language support within the Linux kernel after LPC2020 goes.
The virtual Linux Plumbers Conference takes place from August 24th to 28th. The annual meeting of key Linux kernel developers was scheduled to take place in Halifax, Canada until the corona virus forced them to void these plans when they held their first online LPC event.