On Monday 13 June 2005 2:22 am, Michael Zapf wrote:
> Hi David,
>
> > Send a patch and I'll sign off on it. It'd be a bit annoying to spin
> >for that long (cpu doing nothing else!) waiting for slow hardware, but
> >it's clearly more annoying to get those reset errors. In fact, make it
> >more like 750 usec, if 600 is that marginal. This might be a good fix
> >to put into 2.6.12-final.
>
> sorry, I don't have any experience with providing a proper patch.
It's easy, see Documentation/SubmittingPatches. Here's what I put
together for this one, and I'll forward it in hopes that it can
squeak in before 2.6.12-final.
- Dave
As reported by Michael Zapf <[EMAIL PROTECTED]>, some EHCI
controllers seem to take forever to finish port resets and produce
"port N reset error -110" type errors. Spinning a bit longer helps.
Signed-off-by: David Brownell <[EMAIL PROTECTED]>
--- g26.orig/drivers/usb/host/ehci-hub.c 2005-06-06 12:25:46.000000000 -0700
+++ g26/drivers/usb/host/ehci-hub.c 2005-06-13 06:58:13.000000000 -0700
@@ -439,9 +439,12 @@ static int ehci_hub_control (
/* force reset to complete */
writel (temp & ~PORT_RESET,
&ehci->regs->port_status [wIndex]);
+ /* REVISIT: some hardware needs 550+ usec to clear
+ * this bit; seems too long to spin routinely...
+ */
retval = handshake (
&ehci->regs->port_status [wIndex],
- PORT_RESET, 0, 500);
+ PORT_RESET, 0, 750);
if (retval != 0) {
ehci_err (ehci, "port %d reset error %d\n",
wIndex + 1, retval);
