Re: [Linux-usb-users] Re: uhci locks up in 2.4.19-pre6

Johannes Erdfelt (johannes@erdfelt.com)
Sat, 13 Apr 2002 12:15:22 -0400

On Sat, Apr 13, 2002, Stanislav Meduna <stano@meduna.org> wrote:
> trying to load a uhci module from 2.4.19-pre6 at boot locks my computer.
>
> In 2.4.18 I noticed that the USB printing stopped to work
> (see the thread "USB printing via ptal broke between 2.4.17 and .18"
> started on 29 Mar). I the tested 2.4.19-pre5 and loading the uhci
> module locked the machine. In -pre6 the behaviour is the same.
>
> Can I switch some debug mode on to help you seeing where the thing
> locks?
>
> This is on the Abit BP6 motherboard (Celeron SMP box). I see that
> there are quite a few changes in the uhci area - please, don't
> make final 2.4.19 until these issues are resolved.
>
> Using usb-uhci is not an option for me, as it is causing lockups
> when printing to a USB printer (and ever did so). uhci was OK
> in this regard up to 2.4.17.

Apply these patches. They should be in -pre7 (or -rc1, depending on what
Marcelo calls it).

This should fix all outstanding lockups and performance problems with
uhci.c. Let me know how it goes for you.

JE

--- linux-2.4.19-pre6.orig/drivers/usb/uhci.c Sun Apr 7 13:22:10 2002
+++ linux-2.4.19-pre6/drivers/usb/uhci.c Sun Apr 7 13:25:27 2002
@@ -1811,6 +1811,9 @@
} else {
urb->status = -ENOENT;

+ spin_unlock(&urb->lock);
+ spin_unlock_irqrestore(&uhci->urb_list_lock, flags);
+
if (in_interrupt()) { /* wait at least 1 frame */
static int errorcount = 10;

@@ -1820,9 +1823,6 @@
} else
schedule_timeout(1+1*HZ/1000);

- spin_unlock(&urb->lock);
- spin_unlock_irqrestore(&uhci->urb_list_lock, flags);
-
uhci_call_completion(urb);
}
}
@@ -2193,12 +2193,12 @@
OK(0);
case RH_PORT_RESET:
SET_RH_PORTSTAT(USBPORTSC_PR);
- wait_ms(50); /* USB v1.1 7.1.7.3 */
+ mdelay(50); /* USB v1.1 7.1.7.3 */
uhci->rh.c_p_r[wIndex - 1] = 1;
CLR_RH_PORTSTAT(USBPORTSC_PR);
udelay(10);
SET_RH_PORTSTAT(USBPORTSC_PE);
- wait_ms(10);
+ mdelay(10);
SET_RH_PORTSTAT(0xa);
OK(0);
case RH_PORT_POWER:

--- linux-2.4.19-pre6.orig/drivers/usb/uhci.c Sun Apr 7 16:21:04 2002
+++ linux-2.4.19-pre6/drivers/usb/uhci.c Sun Apr 7 16:24:03 2002
@@ -748,7 +748,7 @@

if ((!(urb->transfer_flags & USB_NO_FSBR)) && !urbp->fsbr) {
urbp->fsbr = 1;
- if (!uhci->fsbr++)
+ if (!uhci->fsbr++ && !uhci->fsbrtimeout)
uhci->skel_term_qh->link = uhci->skel_hs_control_qh->dma_handle | UHCI_PTR_QH;
}

@@ -2739,6 +2739,11 @@
/* Reset here so we don't get any interrupts from an old setup */
/* or broken setup */
reset_hc(uhci);
+
+ uhci->fsbr = 0;
+ uhci->fsbrtimeout = 0;
+
+ uhci->is_suspended = 0;

spin_lock_init(&uhci->qh_remove_list_lock);
INIT_LIST_HEAD(&uhci->qh_remove_list);

--- linux-2.4.19-pre6.orig/drivers/usb/uhci.c Sun Apr 7 16:27:23 2002
+++ linux-2.4.19-pre6/drivers/usb/uhci.c Sun Apr 7 16:28:23 2002
@@ -1837,10 +1837,6 @@

uhci_dec_fsbr(uhci, urb);

- /* There is a race with updating IOC in here, but it's not worth */
- /* trying to fix since this is merely an optimization. The only */
- /* time we'd lose is if the status of the packet got updated */
- /* and we'd be turning on FSBR next frame anyway, so it's a wash */
urbp->fsbr_timeout = 1;

head = &urbp->td_list;
@@ -2004,23 +2000,23 @@
tmp = head->next;
while (tmp != head) {
struct urb *u = list_entry(tmp, struct urb, urb_list);
- struct urb_priv *urbp = (struct urb_priv *)u->hcpriv;
+ struct urb_priv *up = (struct urb_priv *)u->hcpriv;

tmp = tmp->next;

- spin_lock(&urb->lock);
+ spin_lock(&u->lock);

/* Check if the FSBR timed out */
- if (urbp->fsbr && !urbp->fsbr_timeout && time_after_eq(jiffies, urbp->fsbrtime + IDLE_TIMEOUT))
+ if (up->fsbr && !up->fsbr_timeout && time_after_eq(jiffies, up->fsbrtime + IDLE_TIMEOUT))
uhci_fsbr_timeout(uhci, u);

/* Check if the URB timed out */
- if (u->timeout && time_after_eq(jiffies, urbp->inserttime + u->timeout)) {
+ if (u->timeout && time_after_eq(jiffies, up->inserttime + u->timeout)) {
list_del(&u->urb_list);
list_add_tail(&u->urb_list, &list);
}

- spin_unlock(&urb->lock);
+ spin_unlock(&u->lock);
}
spin_unlock_irqrestore(&uhci->urb_list_lock, flags);

--- linux-2.4.19-pre6.orig/drivers/usb/uhci.c Sun Apr 7 16:29:57 2002
+++ linux-2.4.19-pre6/drivers/usb/uhci.c Sun Apr 7 16:22:49 2002
@@ -1622,8 +1622,7 @@

if (urb->status != -EINPROGRESS) {
info("uhci_transfer_result: called for URB %p not in flight?", urb);
- spin_unlock_irqrestore(&urb->lock, flags);
- return;
+ goto out;
}

switch (usb_pipetype(urb->pipe)) {
@@ -1643,10 +1642,8 @@

urbp->status = ret;

- if (ret == -EINPROGRESS) {
- spin_unlock_irqrestore(&urb->lock, flags);
- return;
- }
+ if (ret == -EINPROGRESS)
+ goto out;

switch (usb_pipetype(urb->pipe)) {
case PIPE_CONTROL:
@@ -1660,11 +1657,8 @@
break;
case PIPE_INTERRUPT:
/* Interrupts are an exception */
- if (urb->interval) {
- uhci_add_complete(urb);
- spin_unlock_irqrestore(&urb->lock, flags);
- return; /* <-- note return */
- }
+ if (urb->interval)
+ goto out_complete;

/* Release bandwidth for Interrupt or Isoc. transfers */
/* Spinlock needed ? */
@@ -1680,8 +1674,10 @@
/* Remove it from uhci->urb_list */
list_del_init(&urb->urb_list);

+out_complete:
uhci_add_complete(urb);

+out:
spin_unlock_irqrestore(&urb->lock, flags);
}

@@ -2710,6 +2706,7 @@
}

uhci->dev = dev;
+ uhci->irq = dev->irq;
uhci->io_addr = io_addr;
uhci->io_size = io_size;
pci_set_drvdata(dev, uhci);
@@ -2922,8 +2919,6 @@

if (request_irq(dev->irq, uhci_interrupt, SA_SHIRQ, "usb-uhci", uhci))
goto err_request_irq;
-
- uhci->irq = dev->irq;

/* disable legacy emulation */
pci_write_config_word(uhci