Incorrect Initialization Method

[zkh8227]

In the following code, the variable name is initialized to its own indeterminate value with char *name = name. Even though name will be reassigned later, what is the intention behind such an initialization?

https://github.com/landley/toybox/blob/master/toys/other/vmstat.c#L52

[landley]

It silences gcc's spurious "may be used uninitialized" warnings without generating extra code (the NOP assignment is optimized out but silences the spurious warning). GCC is full of false positive warnings because it can't reliably do dependency tracking across two or more variables. The gcc developers have been insane about this for literally decades, ala: https://stackoverflow.com/questions/5080848/disable-gcc-may-be-used-uninitialized-on-a-particular-variable It's similar to the (void) typecasts on function return values where the compiler complains you're ignoring the return value of a write to the watchdog timer which, if it did fail, would mean the system was about to reboot so what exactly are you supposed to do about it? (Real example, toys/other/watchdog.c line 34. Note that Ubuntu locally patched its version of gcc to ignore the explicit suppression and still warn, because ubuntu.) Most these were switched to using the QUIET macro in lib/portability.h although that generates wasted bytes of code (either writing a value guaranteed to be overwritten, or writing a value guaranteed never to be used). But at least in that case llvm isn't penalized for gcc being stupid.

[zkh8227]

Initialization is sometimes useful.I've noticed that you've fixed the initialization code in the following cases: for example, when it's necessary to check whether fscanf has successfully read the command variable, an initial value for command is required at this time.
For other cases, the approach like name=name is indeed fine.

*command=0;
https://github.com/landley/toybox/blob/master/scripts/mkflags.c#L181
char *p = 0;
https://github.com/landley/toybox/blob/master/toys/other/vmstat.c#L52