On Fri, Oct 25, 2002, David Brownell <[EMAIL PROTECTED]> wrote:
> 
> > Don't you have to set urb->dev before calling usb_submit_urb() again?
> > That's necessary for bulk urbs.  Or do we have another difference in the
> > endpoint logic somethere?  :)
> 
> No, I removed the nulling of urb->dev ... the original reason for
> having that was to eliminate some confusion about whether the HCD
> had really unlinked an urb that could be automagically resubmitted.
> (Mostly usb-uhci as I recall, which oopsed in certain cases there
> and was rather convoluted in how it handled such logic ...)
> 
> Now there's no longer such confusion:  all callbacks mean the urb
> has been unlinked.  So no need to null it, and no need to create
> problems like oopsable races between clearing urb->dev in giveback()
> and accessing through a driver unlink.
> 
> I think Johannes and Oliver were both lobbying to get rid of that
> nulling, and with this patch it became fully safe to get rid of it.

I don't remember if I lobbied for it, but I definately think it's a good
idea to get rid of it.

It's unnecessary now, unprecedented in the kernel and just makes thing
more complicated.

JE



-------------------------------------------------------
This sf.net email is sponsored by: Influence the future 
of Java(TM) technology. Join the Java Community 
Process(SM) (JCP(SM)) program now. 
http://ads.sourceforge.net/cgi-bin/redirect.pl?sunm0004en
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to