Why are not all functions in syscalls.c/sysmem.c marked as weak?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content
‎2020-12-26 05:38 AM
Hello,
some functions like _read()/_write() are marked as weak, so they can be overriden by own implementations. But other functions, like _exit(), etc. are not marked as weak. Why is this the case?
I made a small experiment and changed some functions in syscalls.c, just to see what happens when re-generating the code from the CubeMX file. The changes are preserved, so the file is only generated once. So, doesn't this make the weak attribute of _read()/_write() useless? In fact, either all or none of the functions should be marked as weak. And the file should always be re-generated.
The same seems to be true for the linker script, it's not re-generated. However, re-generating it would destroy any changes made to the memory layout, etc. so it's obvious why it's not re-generated.
This is a bit confusing - what are the rules for (not) re-generating files?
Regards
- Labels:
-
STM32CubeIDE
-
STM32CubeMX
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content
‎2020-12-26 07:04 AM
CubeMX code generation is not meant to be infinitely configurable. The vast vast majority of users aren't going to change _exit, let alone even know about it.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content
‎2020-12-26 10:16 AM
> So, doesn't this make the weak attribute of _read()/_write() useless?
Consider if the user does not want to touch the syscalls.c and provides their own _read()/_write() in other .c file.
If the program does not use _read()/_write() at all, the weak functions in syscalls.c are ok.
Or, you can implement your _read()/_write() and others in syscalls.c (and remove weak) since it is won't be re-generated.
> This is a bit confusing - what are the rules for (not) re-generating files?
Maybe the CubeMX/CubeIDE should put a comment in all generated files to help us understand it.
Another variant - after re-generation, produce a manifest of all re-generated and auto-deleted files.
-- pa
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content
‎2020-12-28 01:14 AM
Hello TDK & Pavel,
I understand that CubeMX can't configure each possible case. but I think it should be possible to have those files always re-created and using BEGIN/END and weak definitions. Just to get rid of the confusion. If it's not easily possible, than Pavels suggestion regarding comments and/or an overview about modified files would be helpful.
Regards