Seiya Nuta

> As of the current implementation, only the data region will be accessible. What is the memory region succeeded by? We do not have an overflow red-zone in malloc right?...

Thanks for sharing. I'll have a look at later! > overflow_redzone is commented out when I saw last It is commented out because the length of `data[]` is arbitrary. The...

FWIW, here's a tip: if I were you, I'd debug unexpected page faults occurred in vm server in following steps: 1. Print IP and the inaccessible memory address in `kernel/task.c:343`....

Might be not related to that bug, I noticed some pitfalls in your source code: - In kasan.h, `shadow` is declared as a global variable. It should be `extern`. -...

> Will the compiler also insert hook functions for shadow memory accesses? I checked the OP-TEE's implementation but it seems we don't have to care about that since they do...

It seems the tcpip server is not embedded into the OS image (because `[vm] lauching tcpip...` does not appear in the log) despite it is enabled in the build config....

I can't reproduce the problem. Could you share the build log (the output of `make V=1`)?

Could you try building with that config in a newly cloned repository? If the problem still appears, I'd like to know your development environment (especially `make --version`).

Hmm I still cannot reproduce this problem. Which operating system do you use? I couldn't reproduce it on macOS 10.15.7 and Ubuntu 20.04. Just in case: did you tried with...

Hi, try `clang --target=x86_64` to let it cross-compile. Also, you would be interested in the following options: ``` clang --target=x86_64 -mcmodel=large -fno-omit-frame-pointer -ffreestanding -fno-builtin -nostdlib -nostdinc \ -mno-red-zone -mno-mmx -mno-sse...