jasonmolenda wrote:

Yes I agree if there is a 64-bit WebAssembly, and we need a way to indicate an 
address scoped to a specific module instance, we'll need something very similar 
to the address space proposal to accommodate.  Is a 64-bit webasm something 
that exists today?  Or just something that we can see existing in the future?  
I'm mostly curious if module instance relative addresses still exist in a 
64-bit address space.

My reservation about adopting address_space:base10 before the address space PR 
lands is that they might change it.  It could go to address_space:base16 or 
as:base10, I don't know.  Maybe all it takes is a comment on that PR saying 
"we're about to adopt address_space:base10 in another context, that part of the 
proposal is locked down".

Do we need to coordinate with gdb or other webasm gdb stub implementations on 
these details?  They may try to solve the same problem for globals (or need to 
solve it for 64-bit address space webasm), not know we have a solution already 
in use, and implement their own approach.

If there's a possibility of needing address space specifiers for a 64-bit 
address space webasm, I have little reservations about using it to hold a 
module instance id here when looking up globals, even though it's not a module 
instance address space address, REALLY, it's a global variable index in a 
module instance, but the concept of "select a module instance with this number, 
then process the rest of this packet" is sensible.

https://github.com/llvm/llvm-project/pull/213176
_______________________________________________
lldb-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits

Reply via email to