Re: bus_dmamap_sync() for USB

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Mon, Jun 23, 2008 at 05:41:53PM -0400, Michael Lorenz wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hello,
> 
> On Jun 23, 2008, at 12:20, Manuel Bouyer wrote:
> 
> >attached is a patch that adds bus_dmamap_sync() calls for uhci,  
> >ohci and ehci.
> >I've tested uhci and ehci on amd64 (both intel and add-on VIA ehci)  
> >and
> >sgimips (add-on VIA ehci). ohci has been tested on amd64 and macppc
> >(both with the build-in USB controller). All tests have been done with
> >USB keyboard/mouse, USB umass key and umodem. This patch, along  
> >with the
> >one in PR port-i386/38935, fix the race condition that cause
> >"host controller process error" and "host controller halted" errors  
> >when
> >umodem@uhci is under heavy use.
> >
> >Michael Lorenz reported issues with this on uhci/ehci in a sgimips O2,
> >but I couldn't reproduce it. It may be an issue with his add-on  
> >adapter.
> 
> I see the same problems with an ohci/ehci card equipped with a NEC  
> chip which works perfectly fine on macppc and sparc64, so I guess the  
> problem is sgimips-specific. Since you don't see it I better look for  
> local hacks in my source tree :/

Or some difference between CPUs. But from what you said, my R10000 should
cause more issues than your R5000, as your should only reorder writes
while mine can also reorder reads.

> The funny thing is that I only have problems with USB cards, an ex  
> and the onboard ahc's Just Work.

The way USB controllers (all of uhci, ohci and ehci) works makes it much more
sensitive to race conditions: there are linked lists of DMA descriptors,
which is managed by both the host and the contrller (both can add or
remove elements from the lists). If both the host and the controller don't
write list pointers to main memory, or read list pointers from main memory, 
in a very specific order, the list can become corrupted. I wouldn't be
surprised if there were some silicon bugs in this area, which would show up
only in some timing conditions. What makes things worse, and different from
an ethernet or disk controllers, is that the USB controller scan
the descriptor lists every microsecond, even if there's nothing to do.
ethernet (at last the ex) or disk controllers will do DMA only when they know
they have work to do.



-- 
Manuel Bouyer, LIP6, Universite Paris VI.           
Manuel.Bouyer%lip6.fr@localhost
     NetBSD: 26 ans d'experience feront toujours la difference
--