Add library to cache decode instr info in analysis tools #7113
Assignees
Labels
Comments
abhinav92003
added a commit
that referenced
this issue
Dec 9, 2024
Adds a new library to cache information about decoded instructions. This can be used by analysis tools that need to decode the instr encodings in the trace. The library allows the tools to specify what information they need to cache. Refactors the invariant checker tool to use this library. Issue: #7113
abhinav92003
added a commit
that referenced
this issue
Dec 11, 2024
Moves the module mapper and related functionality into raw2trace_shared. This is in preparation for further changes that include the module mapper logic in a new instr_decode_cache_t. To avoid having to pull in the whole drmemtrace_raw2trace implementation into the instr decode cache library, which will end up including it in the invariant checker and everything that depends on, we separate out the module mapper into raw2trace_shared which is intended for such cases. Issue: #7113
abhinav92003
added a commit
that referenced
this issue
Dec 11, 2024
This is in preparation for further changes that include the module mapper logic in a new library (drmemtrace_instr_decode_cache) that provides an instr decode cache (instr_decode_cache_t). To avoid having to pull in the whole drmemtrace_raw2trace implementation into drmemtrace_instr_decode_cache, which will end up including it in the invariant checker and everything that depends on, we separate out the module mapper into raw2trace_shared which is intended for such cases. Issue: #7113
abhinav92003
added a commit
that referenced
this issue
Dec 12, 2024
…7123) Moves the module mapper and related functionality into raw2trace_shared. This is in preparation for further changes that include the module mapper logic in a new library (drmemtrace_instr_decode_cache) that provides an instr decode cache (instr_decode_cache_t). To avoid having to pull in the whole drmemtrace_raw2trace implementation into drmemtrace_instr_decode_cache, which will end up including it in the invariant checker and everything that depends on, we separate out the module mapper into raw2trace_shared which is intended for such cases. This is a pure code move, no changes were made to logic or style. Issue: #7113
abhinav92003
added a commit
that referenced
this issue
Dec 15, 2024
Moves the module read functionality into raw2trace_shared. Refactors opcode_mix and view to use the new function instead of creating a raw2trace_directory_t. This is in preparation for further changes that include the module read logic in a new library (drmemtrace_decode_cache) that provides an instr decode cache (decode_cache_t). To avoid having to pull in the whole drmemtrace_raw2trace implementation into drmemtrace_decode_cache, which will end up including it in the invariant checker and everything that depends on, we separate out the module read logic into raw2trace_shared which is intended for such cases. While the raw2trace and raw2trace_shared implementation usually is made to deal with only file streams, we make an exception here for read_module_file() which uses DR file APIs instead. This is likely fine because this function is not intended for use with code that wants file streams, and it is better than other alternatives: moving raw2trace-using functionality out of raw2trace_directory_t so we can use raw2trace_directory_t in decode_cache_t (this would still add zlib/lz4 etc unnecessary file libs to drmemtrace_decode_cache), create yet another raw2trace_shared_non_stream.cpp for shared code that uses DR file APIs, or duplicating the read_module_file impl in decode_cache_t. This also removes raw2trace_directory_t::initialize_module_file() as that functionality is now provided by read_module_file in raw2trace_shared. Issue: #7113
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Various analyzers require decoding instr encodings, e.g. opcode_mix and invariant_checker. Some want to remember the full decoded instr_t, whereas others want to record only specific information about the decoded instr. It would be nice to have a shared library to obviate the need for new analysis tools to re-implement the logic.
The text was updated successfully, but these errors were encountered: