I am currently working on research in developing a transform/pass within CHERI LLVM to optimize capability deployment.
I want to use an external analysis tool (originally designed to optimize SafeStack) to optimize capability deployment and wanted to figure out how capabilities are deployed. I understand pure-capability mode makes all pointers capabilities but I also heard that there was an existing analysis step in purecap mode compilation that determines and optimizes capability deployment.
are there any documents detailing CHERI LLVM's additions to LLVM? In particular the analysis step or capability deployment in LLVM?
Thank you,Yuh K.
Hi Yuh,You are more likely to get satisfactory answers about CHERI-LLVM by asking on one of the CHERI channels; the Morello Forum is more appropriate for Morello-specific questions. Two good options:- The cl-cheri-discuss list: lists.cam.ac.uk/.../cl-cheri-discuss- The CHERI-CPU Slack: www.cl.cam.ac.uk/.../cheri-slack.htmlCheers,Kevin