On 24/03/07, Pete Zaitcev <[EMAIL PROTECTED]> wrote: > On Sat, 24 Mar 2007 17:23:55 +0100, René van Paassen <[EMAIL PROTECTED]> > wrote: > > > I will change the del_timer to del_timer_sync, to remove the race where > > the timer´s activity is running while the timer is being removed. > > > > Does that sound reasonable? > > Yes, it's a good idea, however this is not quite what I was concerned > about. For a timer, it's possible to run on one CPU while a URB callback > is being delivered on another CPU. Even if Dmitry does everything > right and provides a perfect exclusion inside input_report(), you still > may end with input events interleaving without some kind of exclusion > in the driver. This is why I asked if aiptek can call input_report_abs() > with interrupts closed (which happens if you bracket them with > spin_lock_irqsave() in your timer routine).
Ah, but with the rest of my story I was trying to convey the idea that the timer activity does NOT run in parallel with the input thread. After entry of aiptek_irq, the timer is deleted, *before* any input_report_* calls. Then the data from the tablet is processed, leading to input_report_* calls, but the timer routine is *not* running at this time. Only after that the timer is started again. So, unless I have got the wrong model here, input_report_* calls from timer and aiptek_irq cannot occur in parallel anyhow, and a spin lock is not needed. René ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ linux-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
