[Lldb-commits] [lldb] [lldb] Fix stale L1 memory cache read after memory write (PR #208347)

[Felipe de Azevedo Piovezan via lldb-commits]

https://github.com/felipepiovezan commented:

So I'm still trying to understand, isn't the algorithm still quadratic?

This is the kind of pattern I am worried about:

insert(addr: 100, size: 16KB)

and then we do something like a multi memory read of 1000 pointers, so 1000 8-byte addresses, 8 bytes apart each  (so something like reading 8 bytes from each of those addresses:  16000, 1616, 1632, 1648, etc...).

Every time we insert one new 8-byte range in the cache, we are going to scan all previously scanned ranges because max size = 16KB.

https://github.com/llvm/llvm-project/pull/208347