On Mon, 17 Mar 2008, Roland Mainz wrote:

> Keith M Wesolowski wrote:
>> On Sat, Mar 15, 2008 at 04:47:25PM +0100, Roland Mainz wrote:
>>> Does anyone know tools to track down problems caused by uninitalised
>>> global or stack variables except dbx's "check -access" functionality
>>> (which doesn't work in B84) which work in Solaris 11 ?
>>
>> gcc -Wuninitialized and lint in "enhanced mode" both do this today.
>
> "gcc" doesn't show any problems...
> ... how cam I enable "lint"'s enhanched mode ? The normal "lint" pass in
> OS/Net doesn't show any problems in the related code segments...

-Nlevel=1,2,3,4

it actually works only all of the messages of the normal mode
have been fixed (or switched off).

>>> BTW: Did anyone thought about adding compiler options yet to initalise
>>> "uninitalised" variables with a fixed byte sequence (e.g.
>>> -xinituninitalised=0xdeadbeef will fill all uninitalised memory
>>> belonging to variables on the stack with the (repeated) sequence
>>> 0xdeadbeef ; using -xinituninitalised=0xff will write 0xff repeatedly
>>> into such memory etc.) to make this kind of bug hunt more "predictable"
>>> ?
>>
>> Interesting idea.  Still, it could greatly reduce performance so you
>> wouldn't likely want to ship such binaries.
>
> The primary purpose of the option is debugging, e.g. something to get
> data in the case when other tools fail or are unavailable (like dbx's
> "check -access" mode). IMO it would be nice to have more tools to corner
> such bugs in an edge, e.g. I'd like to have fallback options if
> something goes wrong with tools like "dbx" or "lint" (in this specific
> case I already ran out of options and now I am manually editing source
> code and add lots of |foo=0FEDEADFE| to the code...).
>
> ----
>
> Bye,
> Roland
>
> --
>  __ .  . __
> (o.\ \/ /.o) roland.mainz at nrubsig.org
>  \__\/\/__/  MPEG specialist, C&&JAVA&&Sun&&Unix programmer
>  /O /==\ O\  TEL +49 641 7950090
> (;O/ \/ \O;)
> _______________________________________________
> tools-compilers mailing list
> tools-compilers at opensolaris.org
>


Reply via email to