Changqing-JING wrote: In compiler principle point of view, I understand the LTO concept. But my AOT compiler runs on embedded devices, which only have around 200-300k RAM. And the host code is flashed in Pflash already. Currently to link a import function to Wasm code, I just need to link the start address of the function. To do the machine code LTO, the AOT compiler need machine code decoder for each instruction set. In engineering point of view, it's too much overhead.
https://github.com/llvm/llvm-project/pull/200722 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
