Cuong Manh Le
Cuong Manh Le
What should we do for `purego` implementation, since it could not access runtime hasher?
The problem is with inline static init, you can pass `-d=inlstaticinit=0` to workaround this issue.
> cc @cuonglm as [CL 421880](https://go-review.googlesource.com/c/go/+/421880) seems to be the first commit that causes: https://build.golang.org/log/b787fac76217446a7796bfe687e299e2313baf67 This is a real failed and was fixed at tip. The OOM killed seems different...
I think the function name should be `UnsafeSlice` instead of `Slice`. `Slice` is already taken as a `Kind`.
Does `reflect.SliceAt` do all the runtime validations that `unsafe.Slice` do? For example, should this code panic? ``` _ = SliceAt(TypeOf(0), unsafe.Pointer((*int)(nil)), 1) ```
> So it might help if someone wants to look into how we might fix that reporting issue. For this issue, could we just move the `check bce` pass after...
This seems to me a `cmd/cgo` issue, not the runtime. cgo does not recorgnize that `up(&x.self)` is a type conversion, so it emits the cgo check pointer call with `up(&x.self)`...
@ianlancetaylor We could probably record type alias only, something like `map[string]string{"up": "unsafe.Pointer"}`, then the `p.isType` check can correctly recognize and stripe the type conversion.