On Mon, 23 Feb 2004, Stephen Hemminger wrote:

> The list macro's now do poisoning to detect attempted traversal of
> dead elements, but the UHCI wants to initialize them as empty.
> The problem with marking them as empty, is if there ever is a locking
> problem, then the code traversing the list can hit the initialized 
> element and end up in a endless circle...
> 
> Therefore poisoning is safer and faster.

Your rationale is correct, but I want to save this patch for later on.

Everything else being equal I would go ahead and submit it, but I'm 
planning to do some significant alterations to the internal workings of 
the driver.  I will be less worried about races from improper locking than 
about failing to remove an element from a list, and leaving the code the 
way it is will facilitate detecting such errors (which are initially more 
likely, IMHO).

Once those changes are in place and working stably, remind about this
patch (and of course making the comparable change for members of other
lists).

Alan Stern



-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to