Meng-Hung Chen

There are some topics that we can delve into further: 1. Register Allocation The current register allocator has a simple mechanism that picking the candidate of the register and mapping...

Should we also rework the preprocessor? I think the current cfront with the compiler directives and macro expansion might cause some problems in your working.

Should we abjust the directory structure that making `src/ssa.c` focus on building the SSA and create another `src/optimizer.c` to include the existent `src/peephole.c` and the other optimizers ?

The self-hosting on Raspberry Pi 3 is broken since d9a989d also. After enter the commend `make out/shecc-stage2.elf`, I got: ```shell CC+LD out/inliner GEN out/libc.inc CC out/src/main.o LD out/shecc SHECC out/shecc-stage1.elf...

> @vacantron, I'm currently modifying the parser to achieve left-to-right evaluation. But if I start modifying the implementation from the parser, the changes to the codegen might not be necessary....

> In `src/ssa.c`, within the `insert_phi_insn` function, why we didn't link the `prev` pointer between the original instruction and the new phi-func? The property `prev` was added later and I...

> So should I keep the modifications in append_unwound_phi_insn for now? You can add a comment like `FIXME: ...` to indicate the workaround.

@nosba0957,I have opened the #137. You can redirect this pull request to that issue.

Maybe modifying the setting of the overcommit fix this problem? Try `echo 1 | sudo tee /proc/sys/vm/overcommit_memory`.

> [@howjmay](https://github.com/howjmay), can you confirm with the latest `master` branch? The executable has been added, however, the Aarch64 support of architecture test is currently in progress in #547 .