[Lldb-commits] [lldb] [lldb] Fix a crash when using .dwp files and make type lookup reliable with the index cache (PR #79544)

[David Blaikie via lldb-commits]

dwblaikie wrote:

> > > Even if we want to have the skeleton compile unit be parsed first, we would need to know this from any accelerator table entry, wether it be from .debug_names or from our internal manual index.
> > 
> > 
> > True enough, but I think letting this become a lazy property post-construction is a bit more problematic as it might lead to more complicated invariants/situations further downstream. By making it a precondition that any split unit, on construction, has the association with the skeleton unit - there would be fewer "interesting" situations later where the lookup occurs when someone isn't expecting it. The split unit isn't useful without the skeleton for addresses, etc, so it's not like delaying the skeleton lookup provides better performance, etc.
> 
> It actually does provide better performance as we will often do the type lookup solely in the .dwp file and determine we don't need to actually parse and return the type because the context doesn't match. There is no need for the skeleton unit and the address tables and lines tables in the type lookup case. It allows us to traverse the DIE tree in the .dwp file without ever needing to possibly do an expensive lookup by DWO ID for DWARF4, when it actually isn't needed. We have an accessor to allow us to get the skeleton unit if and only if we need it, and yes, most times it will be filled in already. But when it isn't we can easily access it.

Huh, fair enough - bit surprising, but it works. Still worried about doing things a bit too lazily, more complicated states to deal with through the debugger, but fair enough. Thanks for walking me through it!

https://github.com/llvm/llvm-project/pull/79544