[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-arm-kernel
Subject:    Re: Alignment trap handler: ai_usermode uninitialised
From:       Ralph Siemsen <ralphs () netwinder ! org>
Date:       2005-04-28 17:49:46
Message-ID: 4271223A.9000200 () netwinder ! org
[Download RAW message or body]

Peter Menzebach wrote:

> I would be fine, if the default value would be *any* other value than 0, 
> preferable 5. Is there any known reason, why it is 0?
> 
> It took me about the same time than Thilo to find out, what had happened.

There were lengthy debates about the appropriate default value on this 
list several years ago.  In the end it was agreed that only kernel 
should be fixedup by default; any userspace apps are faulty and should 
be corrected.  The reason is that a significant performance hit is taken 
by each fixup, so they should not occur silently.  The only sure way to 
notice there is a problem is for the task to get a fatal signal.

(Just printing messages to syslog or even the current console isn't good 
enough.  Users of X will never see it.  And in my case, I do all my 
builds on headless machines, connected via ssh, so I didn't notice the 
10,000's of alignment faults I was getting from gettext.)

What would be nice would be to have a new signal type, other than 
segfault.. but that's not easily done.

-Ralph

-------------------------------------------------------------------
Subscription options: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel
FAQ:       http://www.arm.linux.org.uk/armlinux/mlfaq.php
Etiquette: http://www.arm.linux.org.uk/armlinux/mletiquette.php
[prev in list] [next in list] [prev in thread] [next in thread]