Frank Hofmann wrote:
> On Wed, 27 Dec 2006, Garrett D'Amore wrote:
>
>> Cyril Plisko wrote:
>>> Howdy,
>>>
>>> While browsing sources under usr/src/uts I noticed that many
>>> modules are reimplementing some sort of debug printf-like
>>> functionality all over again. In many cases it looks like one
>>> framework copying the defines from the other changing only
>>> a prefix. Wouldn't it be beneficial for the kernel to have some
>>> sort of generic DPRINTF/dprintf functionality, available to all
>>> the components ? IMHO, besides overall cleanness it will
>>> also allow developers to concentrate on their specific tasks
>>> instead of re-inventing the wheel again and again.
>>>
>>> May be there is something similar on the way already ?
>>>
>>
>> It would be nice if we had some standard macros for this. The problem
>> is that developers want to turn on debugging in _their_ subsystem, and
>> nowhere else. Now the kernel could figure where the call came from, and
>> set up a table of kernel module -> debug tunable, which could be
>> helpful.
>>
>> Then you could use KPRINTF() calls, which evaluate to { } in a module
>> built without debug facilities.
>
> <personal opinion>
> Please no - it must be possible to runtime enable it. Even if the
> module that uses it is compiled DEBUG. I think modules, even DEBUG
> modules, that unconditionally spam the messages are very bad citizens.
I agree, but I think you're misunderstanding what I'm saying. What I'm
saying is if the module is built with DEBUG, then it enables a call into
the kernel facility, which checks against a module-specific debug value
kept by the kernel. This could reside in a "master table" somewhere.
>
> I don't mind if for speed reasons, the [DK]PRINTF() evaluates to
> nothing (read: is not compiled in) if DEBUG is not defined. But if it
> is, I still don't want to be spammed by every off-the-shelf module
> just because I'm using a debug kernel.
>
> Can we put a "debuglevel" into "struct modlinkage" or so, to allow
> selectively enabling these facilities per module ?
That might work, but I want an override in /etc/system or somesuch, so
that I can enable it at run time. (Its harder to get to modlinkage
structure members.)
A better way, IMO, would be to create a new global symbol, "_debug" or
somesuch, that is an integer providing the value. This could be
overridden using the "normal" methods, and would also avoid any
compatibility issues in extending modlinkage.
-- Garrett
> </personal opinion>
>
> FrankH.
>
>>
>> -- Garrett
>>
>> --
>> Garrett D'Amore, Principal Software Engineer
>> Tadpole Computer / Computing Technologies Division,
>> General Dynamics C4 Systems
>> http://www.tadpolecomputer.com/
>> Phone: 951 325-2134 Fax: 951 325-2191
>>
>> _______________________________________________
>> opensolaris-code mailing list
>> [email protected]
>> http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
>>
--
Garrett D'Amore, Principal Software Engineer
Tadpole Computer / Computing Technologies Division,
General Dynamics C4 Systems
http://www.tadpolecomputer.com/
Phone: 951 325-2134 Fax: 951 325-2191
_______________________________________________
opensolaris-code mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
