Re: [dmidecode] Crash bug in dmidecde 2.5?

[Jean Delvare]

Hi Petter, 

> Compiling with -Wconversion and -Wsign-compare might give you some
> knowledge about possible problems.

I think we already have -Wsign-compare since it is implied by -W, which
dmidecode uses since the very beginning.

Just tried -Wconversions, it yields many many warnings. You are correct
that it would have triggered on the recent signedness issue, but it
raises so many warnings for non-faulty cases that the signal would
probably have been left unnoticed in all the noise anyway.

> Or, if you want to get lots of information about possible problems,
> use this:
> 
>   -W -Wall -Wcast-align -Wmissing-prototypes -Wstrict-prototypes \
>   -Wpointer-arith -Wreturn-type -Wconversion -Wsign-compare \
>   -Wcast-qual -Wmissing-declarations

Dmidecode already uses most of these. Missing ones were:
-Wmissing-prototypes
-Wreturn-type (implied by -Wall)
-Wconversion
-Wsign-compare (implied by -W)
-Wmissing-declarations

I can't get the difference between -Wmissing-prototypes and
-Wmissing-declarations, let alone the fact that the former is for C only
(doesn't change a thing in our case). I think I'll add
-Wmissing-prototypes, it cannot hurt.

Thanks a lot for the hints,
-- 
Jean Delvare