Re: [linux-usb-devel] Re: kernel crash during shutdown, patch

Bernd Porr Tue, 03 Feb 2004 16:42:21 -0800

Were you able to test that "interrupt transfers don't
queue" case?

Sorry no.

That it would.  How are you forcing the controller into that
particular rude shutdown path?

Don't know. It's my aura. I attract chaos. ;-)


Better would be to just remove the HCD_IS_RUNNING test I highlighted,
which will mean that aborting a running controller takes a smidgeon
longer than it should ... in just the last frame.  Try that and see
how it behaves.

It behaves good:

Feb 3 23:28:34 snoopy kernel: ehci_hcd 0000:00:10.3: shutdown urb c807f940 pipe 40408180 ep1in-intr Feb 3 23:28:34 snoopy kernel: usb usb4: unregistering interface 4-0:1.0 Feb 3 23:28:34 snoopy kernel: usb usb4: hcd_unlink_urb c807f940 fail -22 Feb 3 23:28:34 snoopy kernel: drivers/usb/core/usb.c: usb_hotplug Feb 3 23:28:34 snoopy kernel: usb usb4: unregistering device Feb 3 23:28:34 snoopy kernel: drivers/usb/core/usb.c: usb_hotplug Feb 3 23:28:34 snoopy kernel: ehci_hcd 0000:00:10.3: stop Feb 3 23:28:34 snoopy kernel: ehci_hcd 0000:00:10.3: reset command 01000b (park)=0 ithresh=1 period=256 Reset RUN Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=928,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=936,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=944,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=952,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=960,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=968,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=976,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=984,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=992,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=1000,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=1008,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=1016,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=1024,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=1032,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=1040,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=1048,running=0 ... Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=824,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=832,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=840,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=848,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=856,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=864,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=872,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=880,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=888,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=896,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=904,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=912,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=920,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=921,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=922,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=923,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=924,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=925,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=926,running=0 Feb 3 23:28:34 snoopy kernel: clock=927,now_uframe=927,running=0 Feb 3 23:28:34 snoopy kernel: ehci_hcd 0000:00:10.3: irq normal 8936 err 0 reclaim 7 (lost 0) Feb 3 23:28:34 snoopy kernel: ehci_hcd 0000:00:10.3: complete 8932 unlink 4 Feb 3 23:28:34 snoopy kernel: ehci_hcd 0000:00:10.3: ehci_stop completed status 1000 Halt Feb 3 23:28:34 snoopy kernel: ehci_hcd 0000:00:10.3: USB bus 4 deregistered

scan_periodic (struct ehci_hcd *ehci, struct pt_regs *regs)
{
        unsigned        frame, clock, now_uframe, mod;
        unsigned        modified;

mod = ehci->periodic_size << 3;

        /*
         * When running, scan from last scan point up to "now"
         * else clean up by scanning everything that's left.
         * Touches as few pages as possible:  cache-friendly.
         */
        now_uframe = ehci->next_uframe;
        if (HCD_IS_RUNNING (ehci->hcd.state))
                clock = readl (&ehci->regs->frame_index);
        else
                clock = now_uframe + mod - 1;
        clock %= mod;

        for (;;) {
                union ehci_shadow       q, *q_p;
                u32                     type, *hw_p;
                unsigned                uframes;

                printk("clock=%d,now_uframe=%d,running=%d\n",
                       clock,now_uframe,
                       HCD_IS_RUNNING (ehci->hcd.state));

                /* don't scan past the live uframe */
                frame = now_uframe >> 3;
                if ((frame == (clock >> 3)))
//                              && HCD_IS_RUNNING (ehci->hcd.state))
                        uframes = now_uframe & 0x07;
                else {
                        /* safe to scan the whole frame at once */
                        now_uframe |= 0x07;
                        uframes = 8;
                }

restart:
                /* scan each element in frame's queue for completions */
                q_p = &ehci->pshadow [frame];
                hw_p = &ehci->periodic [frame];
                q.ptr = q_p->ptr;
                type = Q_NEXT_TYPE (*hw_p);

/Bernd

- Dave


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Re: [linux-usb-devel] Re: kernel crash during shutdown, patch

Reply via email to