Philip Craig

read: add support for custom address types

1

Add the `Reader::Address` associated type, which implements the `ReaderAddress` trait. This trait is implemented for u64 to provide the existing behaviour of supporting multiple address sizes, as well as for...

Simpler attribute value parsing

1

`Attribute::value` is okay for things like the `dwarfdump` example or the convert code, where we can treat all attributes the same without needing to interpret the values. However for many...

Is `read::RawRngListEntry::AddressOrOffsetPair` needed?

2

For DWARF 4, we currently allow range list entries to use a pair of addresses. However, the DWARF 4 standard (section 2.17.3) only talks about address offsets. We should check...

There might be some ideas for things that can be improved in: https://github.com/sola-st/wasm-type-prediction/blob/a69a0310324fb1a695cee6d97e9db61d638df0ef/implementation/2-extraction/src/dwarf/util.rs

When we added macro parsing support, this code should have been updated too: https://github.com/gimli-rs/gimli/blob/94682004e67689e9e1ee9ddf18e2be135f394c5c/src/read/dwarf.rs#L1105-L1107

`DW_FORM_exprloc ` was introduced in DWARF 4, so we shouldn't use it. We should check for similar problems with other forms

The easiest fix is to print an error in `readobj` if the tree level is too high. But it would be nice to include this check in `object::read::pe` somehow.

ELF symbols usually include a size, and Mach-O symbols can be delimited by LC_FUNCTION_STARTS and segment sizes. This will help with https://github.com/gimli-rs/addr2line/issues/324.

It probably makes sense for the unified read API to hide the default weak symbol, and instead adjust the symbol for the `coff::IMAGE_SYM_CLASS_WEAK_EXTERNAL` entry to include some of the information...