Fix calling ntop on tracepoint args

[danobi]

Commit f05b4cd ("Introduce Type::ctx") introduced a ctx type
for values that come from the bpf prog context. This subtly broke ntop()
because ntop was not yet aware of Type::ctx.

Fix this by relaxing semantic analyser check to include Type::ctx.
Similarly relax for codegen.

Checklist

• Language changes are updated in docs/reference_guide.md
• User-visible and non-trivial changes updated in CHANGELOG.md

[danobi]

I wonder if Type::ctx should be limited to just the ctx builtin. Reasoning is that a value coming from the prog context is more an attribute and not a type unto itself. For example, saddr_v6 can be both an array and from the prog ctx. It doesn't necessarily have to be one or the other.

@mmisono any thoughts?

[fbs]

I wonder if Type::ctx should be limited to just the ctx builtin. Reasoning is that a value coming from the prog context is more an attribute and not a type unto itself.

I'm working on replacing cast with a pointer and a record type. As ctx can currently be both and more which makes it confusing I'm going to remove ctx too and make it an attribute instead.

https://github.com/fbs/bpftrace/tree/pointer_type

[danobi]

Sounds good. Looking forward to seeing that merged :)

[mmisono]

Thanks for the fix.

I thought Type::ctx simplified the implementation but actually there are several corner cases and I've thought some rework is needed. Making ctx an attribute seems good.