Change Details

Some BPF programs compiled on s390 fail to load, because s390 arch-specific linux headers contain float and double types. At the moment there is no BTF_KIND for floats and doubles, so release version of LLVM ends up emitting type id 0 for them, which in-kernel verifier does not accept. This should also be the case for structs and enums with >=64k members. Fix by emitting byte arrays instead of real types. Another option would be to skip the offending types altogether, but this would be unnecessarily restrictive. Yet another option would be what pahole does - represent floats and doubles as ints, but that won't cover other "weird" cases like large structs.

Some BPF programs compiled on s390 fail to load, because s390 arch-specific linux headers contain float and double types. At the moment there is no BTF_KIND for floats and doubles, so the release version of LLVM ends up emitting type id 0 for them, which the in-kernel verifier does not accept. Introduce support for such types to libbpf by representing them using the new BTF_KIND_FLOAT.

Some BPF programs compiled on s390 fail to load, because s390 arch-specific linux headers contain float and double types. At the moment there is no BTF_KIND for floats and doubles, so release versionso the release version of LLVM ends up emitting type id 0 for them, which in-kernel verifierthe does not accept. This should also be the case for structs and enums with >=64k membersin-kernel verifier does not accept. Fix by emitting byte arrays instead of real types. Another option would be to skip the offending types altogether, but this would be unnecessarily restrictive. Yet another option would be what pahole does - represent floats and doubles as ints, but that won't cover otherIntroduce support for such types to libbpf by representing them using "weird" cases like large structsthe new BTF_KIND_FLOAT.