Andrew Pogrebnoi
Andrew Pogrebnoi
Does anybody work on this? I can take on it.
@ianlancetaylor, I've just discovered the patch from @prattmic: https://go.dev/cl/363417. As far as I can tell, it only lacks `runtime·epollwait2` implementations for other than `amd64` platforms. Are there some reasons you...
Thanks, @prattmic. I'll prepare CL then and will do some benchmarking.
Oh, cool, thanks for the advice Austin!
I have a few questions. @prattmic, in your CL you use `_ENOSYS` to define if the syscall exists. But I cannot find its definition anywhere other than in the "golang.org.x/sys/unix"...
> For the syscall number constants, I would create new per-arch files in runtime/internal/syscall to define those constants. Eventually, we'll want to shift all of the constants out of sys_linux_xxx.s...
Ah, as far as I can see there is no generator for the runtime now. Hence no easy way to generate syscall defs. So I'll probably just add constants manually....
Hi, I'm again with questions. 1. I made functions for epoll_pwait and epoll_pwait2 calls in `runtime/netpoll_epoll.go` instead of assembler once, which looks like this: ```go var _zero uintptr func epollwait2(epfd...
Hi Michael, thank you for the feedback and explanations. > I think you should have at least 2 CLs: one (or more) moving things to `runtime/internal/syscall` with no behavior changes,...
To the comment of complexity - I agree, but we don't have much choice for this code to be used with frontend tools. The other way would be get back...