Hello, I am writing to inquire about the availability of any "__sanitizer_cov_trace_basic_block" implementations in gcc. I somewhat learned from the following mail discussion ( https://gcc.gnu.org/pipermail/gcc-patches/2015-December/437137.html) that `__sanitizer_cov_trace_basic_block` is provided; however, I cannot find a implementation here. gcc also does not provide `-fsanitize-coverage=trace-bb` option for use.
For the user space we may add it to sanitizer_common/sanitizer_coverage_libcdep.cc, however for userspace the benefit of having __sanitizer_cov_trace_pc is not so big because we already have __sanitizer_cov_trace_basic_block. __sanitizer_cov_trace_basic_block has similar functionality but a bit more hairy implementation in the compiler and requires an initialization step done at module init, which in the kernel we don't have. Currently I want to record all the executed basic blocks and mapped them back to *GIMPLE basic blocks*. While `__sanitizer_cov_trace_pc` can help to log the program counter value, that's still not what I want. If I use `__sanitizer_cov_trace_pc`, I would envision mapping the `pc` value to assembly basic block info, and then map assembly program basic block back to GIMPLE basic block. While it could be doable (suppose I disable all optimization), it seems really clumsy. It seems to me `__sanitizer_cov_trace_basic_block` allows to pass some parameters and therefore I guess there is a way to directly pass the GIMPLE basic block id as the input of `__sanitizer_cov_trace_basic_block`, while I just cannot find a way of doing so for `__sanitizer_cov_trace_pc`. Am I missing anything here? Best, Shuai
