USB echi and Intel ICH

  
This updates the Philips tweak so that it also applies
to late-model ICH chips from Intel.  (Or so I'm told ... :)
That's at least three EHCI implementations known to
behave on Linux.  (And one hopes VIA soon too...)

It also cleans up a few comments relating to 64bit DMA;
recent API spec updates make it look like no games are
needed with the PCI DMA mask -- it doesn't change the
segment used by pci_pool allocations, so it just needs
to get turned on if the overall system does 64bit DMA.

I've a query in to see if those Philips/Intel tweaks are
chip quirks, or just something the EHCI 1.0 spec isn't
at all clear about.

drivers/usb/hcd/ehci-hcd.c

@@ -208,6 +208,7 @@ static int ehci_start (struct usb_hcd *hcd)
 	/* controller state:  unknown --> reset */
 
 	/* EHCI spec section 4.1 */
+	// FIXME require STS_HALT before reset...
 	ehci_reset (ehci);
 	writel (INTR_MASK, &ehci->regs->intr_enable);
 	writel (ehci->periodic_dma, &ehci->regs->frame_list);
@@ -215,16 +216,17 @@ static int ehci_start (struct usb_hcd *hcd)
 	/*
 	 * hcc_params controls whether ehci->regs->segment must (!!!)
 	 * be used; it constrains QH/ITD/SITD and QTD locations.
-	 * By default, pci_alloc_consistent() won't hand out addresses
-	 * above 4GB (via pdev->dma_mask) so we know this value.
-	 *
-	 * NOTE:  that pdev->dma_mask setting means that all DMA mappings
-	 * for I/O buffers will have the same restriction, though it's
-	 * neither necessary nor desirable in that case.
+	 * pci_pool consistent memory always uses segment zero.
 	 */
 	if (HCC_64BIT_ADDR (hcc_params)) {
 		writel (0, &ehci->regs->segment);
-		info ("using segment 0 for 64bit DMA addresses ...");
+		/*
+		 * FIXME Enlarge pci_set_dma_mask() when possible.  The DMA
+		 * mapping API spec now says that'll affect only single shot
+		 * mappings, and the pci_pool data will stay safe in seg 0.
+		 * That's what we want:  no extra copies for USB transfers.
+		 */
+		info ("restricting 64bit DMA mappings to segment 0 ...");
 	}
 
 	/* clear interrupt enables, set irq latency */
@@ -234,6 +236,9 @@ static int ehci_start (struct usb_hcd *hcd)
 	temp |= 1 << (16 + log2_irq_thresh);
 	// keeping default periodic framelist size
 	temp &= ~(CMD_IAAD | CMD_ASE | CMD_PSE),
+	// Philips, Intel, and maybe others need CMD_RUN before the
+	// root hub will detect new devices (why?); NEC doesn't
+	temp |= CMD_RUN;
 	writel (temp, &ehci->regs->command);
 	dbg_cmd (ehci, "init", temp);
 
@@ -260,7 +265,7 @@ static int ehci_start (struct usb_hcd *hcd)
 	readl (&ehci->regs->command);	/* unblock posted write */
 
         /* PCI Serial Bus Release Number is at 0x60 offset */
-	pci_read_config_byte(hcd->pdev, 0x60, &tempbyte);
+	pci_read_config_byte (hcd->pdev, 0x60, &tempbyte);
 	temp = readw (&ehci->caps->hci_version);
 	info ("USB %x.%x support enabled, EHCI rev %x.%2x",
 	      ((tempbyte & 0xf0)>>4),
@@ -387,6 +392,8 @@ static int ehci_resume (struct usb_hcd *hcd)
 // restore pci FLADJ value
 	// khubd and drivers will set HC running, if needed;
 	hcd->state = USB_STATE_READY;
+	// FIXME Philips/Intel/... etc don't really have a "READY"
+	// state ... turn on CMD_RUN too
 	for (i = 0; i < ports; i++) {
 		int	temp = readl (&ehci->regs->port_status [i]);
 

drivers/usb/hcd/ehci-hub.c

@@ -309,13 +309,6 @@ static int ehci_hub_control (
 					hcd->bus_name, wIndex + 1);
 				temp |= PORT_OWNER;
 			} else {
-				/* Philips 1562 wants CMD_RUN to reset */
-				if (!HCD_IS_RUNNING(ehci->hcd.state)) {
-					u32 cmd = readl (&ehci->regs->command);
-					cmd |= CMD_RUN;
-					writel (cmd, &ehci->regs->command);	
-					ehci->hcd.state = USB_STATE_RUNNING;
-				}
 				vdbg ("%s port %d reset",
 					hcd->bus_name, wIndex + 1);
 				temp |= PORT_RESET;

drivers/usb/hcd/ehci-q.c

@@ -436,14 +436,6 @@ qh_urb_transaction (
 
 	if (usb_pipecontrol (urb->pipe)) {
 		/* control request data is passed in the "setup" pid */
-
-		/* NOTE:  this isn't smart about 64bit DMA, since it uses the
-		 * default (32bit) mask rather than using the whole address
-		 * space.  we could set pdev->dma_mask to all-ones while
-		 * getting this mapping, locking it and restoring before
-		 * allocating qtd/qh/... or maybe only do that for the main
-		 * data phase (below).
-		 */
 		qtd->buf_dma = pci_map_single (
 					ehci->hcd.pdev,
 					urb->setup_packet,
@@ -472,7 +464,6 @@ qh_urb_transaction (
 	 */
 	len = urb->transfer_buffer_length;
 	if (likely (len > 0)) {
-		/* NOTE:  sub-optimal mapping with 64bit DMA (see above) */
 		buf = map_buf = pci_map_single (ehci->hcd.pdev,
 			urb->transfer_buffer, len,
 			usb_pipein (urb->pipe)