User Activity

  • Posted a comment on ticket #2634 on Small Device C Compiler suite

    Wow, this bug was introduced 10 years ago in [r4758].

  • Posted a comment on ticket #528 on Small Device C Compiler suite

    Currently SDCC does not even deduce that Ch_loc is always initialized / assigned after the while loop and thus it generates warning 84's. I would not in general want this warning to be suppressed by the substitution just because it is assigned only once somewhere in the code.

  • Modified a comment on ticket #2627 on Small Device C Compiler suite

    Fixed in revision [r9940]. Philipp

  • Posted a comment on ticket #528 on Small Device C Compiler suite

    Is Dhrystone meant for testing compiler efficiency or for testing the speed of the underlying hardware? If the latter, maybe the code was meant to be non-optimizable.

  • Modified a comment on ticket #519 on Small Device C Compiler suite

    The assembler does not have a .weak directive and the linker doesn't understand the concept either. I've never seen the .weak concept used by anything but GCC / binutils ld. And just like all other microcontroller toolchains SDCC has its own way of defining interrupt handlers. We allow the user to specify the name of the ISR instead of use a fixed one, because it might handle any kind of peripheral. Further when an interrupt is not used, there is no point in consuming code space for the vector and...

  • Posted a comment on ticket #519 on Small Device C Compiler suite

    The assembler does not have a .weak directive and the linker doesn't understand the concept either. I've never seen the .weak concept used by anything but GCC. And just like all other microcontroller toolchains SDCC has its own way of defining interrupt handlers. We allow the user to specify the name of the ISR instead of use a fixed one, because it might handle any kind of peripheral. Further when an interrupt is not used, there is no point in consuming code space for the vector and a dummy implementation....

  • Posted a comment on ticket #2668 on Small Device C Compiler suite

    That doesn't really matter. The problem is that labelRefCountChange has a side-effect even when labelRefCount returns false and prohibits the rule to be applied.

  • Posted a comment on ticket #2668 on Small Device C Compiler suite

    Are you sure this is right? I had the impression that the conditions were handled from left to right. This would mean that first labelRefCountChange(%1 -1) updates the refcount and then labelRefCount(%0 8) is checked.

View All

Personal Data

Username:
maartenbrock
Joined:
2003-10-16 08:59:06
Location:
Tilburg / Netherlands / CEST
Gender:
Male

Projects

Skills

  • No skills entered.

Personal Tools