TTool merge requestshttps://gitlab.telecom-paris.fr/mbe-tools/TTool/-/merge_requests2022-09-30T05:03:04Zhttps://gitlab.telecom-paris.fr/mbe-tools/TTool/-/merge_requests/442Update Trace and Timeline HTML2022-09-30T05:03:04ZJawher JerrayUpdate Trace and Timeline HTMLThis merge request fixes a bug in the graphic trace of buses, also it extends the update of commit 6e1d4454 to other functions: HW2HTML() and schedule2HTML() for CPU, Tasks and Buses.This merge request fixes a bug in the graphic trace of buses, also it extends the update of commit 6e1d4454 to other functions: HW2HTML() and schedule2HTML() for CPU, Tasks and Buses.Jawher JerrayJawher Jerrayhttps://gitlab.telecom-paris.fr/mbe-tools/TTool/-/merge_requests/477[WIP] simulation code generation fixes2023-11-29T14:36:15ZMarvin Häuser[WIP] simulation code generation fixesRunning DSE on "SmartCardProtocol" currently crashes all simulation binaries on AARCH64 macOS with Xcode Clang. I identified multiple culprits, such as undefined behaviour when traversing a variadic argument list, or dereferencing a NULL...Running DSE on "SmartCardProtocol" currently crashes all simulation binaries on AARCH64 macOS with Xcode Clang. I identified multiple culprits, such as undefined behaviour when traversing a variadic argument list, or dereferencing a NULL pointer. To further harden the code against such bugs, I propose enabling the typical "extra" warnings, as well as making most warnings fatal. This already uncovered further issues, such as uninitialized return values. Please refer for the individual commits for details.
This remains a WIP draft for three reasons:
- There are open questions for commits marked as [WIP]:
- https://gitlab.telecom-paris.fr/mbe-tools/TTool/-/commit/6f1d3dd6e69e742daab3d0c7eaaf8f58ac461c81: This is supposed to be NULL in real-world runs?
- https://gitlab.telecom-paris.fr/mbe-tools/TTool/-/commit/9f51e068890eedc5ed978295a29f1c9895ba7def: The types for these are unsigned, so the condition is always true. Is it realistic for the subtraction done at initialization to underflow and must this be detected instead? Should the code maybe use overflow-safe arithmetics?
- https://gitlab.telecom-paris.fr/mbe-tools/TTool/-/commit/f62c886cd9fc34abfda76ebe66cb29c957335177: I took best guesses here at what is intended to fall through and what might be a bug, but I have no definite answers.
- I tested the changes only with "SmartCardProtocol", other code generation samples may emit errors.
- Project-specific code used by end-user projects may start emitting errors as well. Maybe this should be suppressible via an environment variable or some custom Makefile with a property? I am not sure which route to take.