Compact-unwind encodings are more limited than DWARF frame descriptions. For functions whose frame layout cannot be described by a compact unwind encoding, the encoding for the function will specify "use DWARF", and the corresponding unwind-info record will use the low bits of the encoding to point to the FDE for the function. We test this with a frame-pointer=none function, since these frame layouts always triger a fall-back to DWARF on arm64.
27 KiB
27 KiB