[PATCH] Tokyo Electron SDIO controller (Ellen) support

[PATCH] Tokyo Electron SDIO controller (Ellen) support

[Vitaly Luban]

[-- Attachment #1: Type: text/plain, Size: 275 bytes --]

Kernel "pci_ids.h" file has data for that card missing.

Also, Ellen needs some control bits flipped before it functions properly 
as SDIO controller by the spec.
Should apply clenly to Linus and Drzeus trees. Please apply.

Signed-off-by:  Vitaly Luban <vitaly@luban.org>



[-- Attachment #2: ellen.patch --]
[-- Type: text/plain, Size: 3543 bytes --]

--- drzeus/drivers/mmc/host/sdhci.c	2007-11-27 19:51:35.000000000 -0800
+++ linux-2.6.23/drivers/mmc/host/sdhci.c	2007-11-30 18:43:02.000000000 -0800
@@ -97,6 +99,13 @@
 				  SDHCI_QUIRK_RESET_CMD_DATA_ON_IOS,
 	},
 
+        {
+                .vendor         = PCI_VENDOR_ID_TOKYO_ELECTRON,
+                .device         = PCI_DEVICE_ID_ELLEN,
+                .subvendor      = PCI_ANY_ID,
+                .subdevice      = PCI_ANY_ID,
+        },
+
 	{	/* Generic SD host controller */
 		PCI_DEVICE_CLASS((PCI_CLASS_SYSTEM_SDHCI << 8), 0xFFFF00)
 	},
@@ -1205,10 +1214,43 @@
  *                                                                           *
 \*****************************************************************************/
 
+static int __devinit sdhci_ellen_init(struct pci_dev *pdev)
+{
+	int                 ret;
+	unsigned int        config;
+	void __iomem *      ioaddr;
+	 
+	if (pci_resource_flags(pdev, 0) & PCI_BASE_ADDRESS_SPACE)
+		return -ENOMEM;
+	
+	ret = pci_request_region(pdev, 0, "EllenPCI");
+	if (ret)
+		return -ENOMEM;
+	
+	ioaddr = ioremap_nocache(pci_resource_start(pdev, 0), 
+	                         pci_resource_len(pdev, 0));
+	if (!ioaddr) {
+		ret = -ENOMEM;
+		goto release;
+	}
+	
+	/* Do some magic passes to enable Ellen PCI interrupts */
+	config = readw(ioaddr + 0x4c);
+	config |= (1) | (1<<3) | (1<<6);
+	writew(config, ioaddr + 0x4c);
+	/* Retire last write by read */
+	config = readw(ioaddr + 0x4c);
+	
+	iounmap(ioaddr);
+release:
+	pci_release_region(pdev, 0);
+	return ret;
+}
+
 static int __devinit sdhci_probe_slot(struct pci_dev *pdev, int slot)
 {
 	int ret;
-	unsigned int version;
+	unsigned int version, temp, iomemsize;
 	struct sdhci_chip *chip;
 	struct mmc_host *mmc;
 	struct sdhci_host *host;
@@ -1225,6 +1267,15 @@
 
 	first_bar &= PCI_SLOT_INFO_FIRST_BAR_MASK;
 
+	if ((pdev->vendor == PCI_VENDOR_ID_TOKYO_ELECTRON) &&
+	    (pdev->device == PCI_DEVICE_ID_ELLEN)) {
+		first_bar = 2;
+		iomemsize = 0x400;
+	}
+	else {
+		iomemsize = 0x100;
+	}
+
 	if (first_bar > 5) {
 		printk(KERN_ERR DRIVER_NAME ": Invalid first BAR. Aborting.\n");
 		return -ENODEV;
@@ -1235,9 +1286,9 @@
 		return -ENODEV;
 	}
 
-	if (pci_resource_len(pdev, first_bar + slot) != 0x100) {
-		printk(KERN_ERR DRIVER_NAME ": Invalid iomem size. "
-			"You may experience problems.\n");
+	if ((temp = pci_resource_len(pdev, first_bar + slot)) != iomemsize) {
+		printk(KERN_ERR DRIVER_NAME ": Invalid iomem size %x. "
+			"You may experience problems.\n", temp);
 	}
 
 	if ((pdev->class & 0x0000FF) == PCI_SDHCI_IFVENDOR) {
@@ -1526,6 +1577,24 @@
 	chip->pdev = pdev;
 	chip->quirks = ent->driver_data;
 
+	if ((pdev->vendor == PCI_VENDOR_ID_TOKYO_ELECTRON) &&
+	    (pdev->device == PCI_DEVICE_ID_ELLEN)) {
+		ret = pci_read_config_byte(pdev, PCI_CLASS_PROG, &rev);
+		if (ret)
+			goto free;
+		
+		if( rev & 0x1 ) {
+			/* Ellen indicates DMA support by this bit */
+			chip->quirks |= SDHCI_QUIRK_FORCE_DMA;
+		}
+		
+		ret = sdhci_ellen_init(pdev);
+		if (ret) {
+			ret = -ENODEV;
+			goto free;
+		}
+	}
+
 	if (debug_quirks)
 		chip->quirks = debug_quirks;
 
--- drzeus/include/linux/pci_ids.h	2007-11-30 19:15:24.000000000 -0800
+++ linux-2.6.23/include/linux/pci_ids.h	2007-11-30 19:15:01.000000000 -0800
@@ -2057,6 +2057,9 @@
 #define PCI_DEVICE_ID_BCM1250_PCI	0x0001
 #define PCI_DEVICE_ID_BCM1250_HT	0x0002
 
+#define PCI_VENDOR_ID_TOKYO_ELECTRON	0x1679
+#define PCI_DEVICE_ID_ELLEN		0x3000
+
 #define PCI_VENDOR_ID_ATHEROS		0x168c
 
 #define PCI_VENDOR_ID_NETCELL		0x169c

Re: [PATCH] Tokyo Electron SDIO controller (Ellen) support

[Pierre Ossman]

[-- Attachment #1: Type: text/plain, Size: 845 bytes --]

On Sat, 01 Dec 2007 00:27:06 -0800
Vitaly Luban <vitaly@luban.org> wrote:

> Kernel "pci_ids.h" file has data for that card missing.
> 
> Also, Ellen needs some control bits flipped before it functions properly 
> as SDIO controller by the spec.
> Should apply clenly to Linus and Drzeus trees. Please apply.
> 
> Signed-off-by:  Vitaly Luban <vitaly@luban.org>
> 
> 

The illiteracy rates for hardware engineers must be through the roof. Nobody seems capable of reading specifications anymore...

As for the patch, it's a big NAK at this point. Vendors who can't be bothered to follow the standard will have to wait for Ben's separation patches before they can be supported. Then they can have some voodoo front-end to handle their mistakes instead of turning sdhci.c into a collection of hacks and workarounds.

Rgds
Pierre

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH] Tokyo Electron SDIO controller (Ellen) support

[Matt Porter]

On Sat, Dec 01, 2007 at 09:50:10PM +0100, Pierre Ossman wrote:
> As for the patch, it's a big NAK at this point. Vendors who can't be bothered to follow
> the standard will have to wait for Ben's separation patches before they can be supported.
> Then they can have some voodoo front-end to handle their mistakes instead of turning
> sdhci.c into a collection of hacks and workarounds.

What's the status of Ben's separation patches? I haven't seen a posting of those
versus a recent kernel. I've got some SDHCI driver glue for the non-pci Arasan core
running in an older kernel tree with those patches. It's just waiting for the separation
patches to hit the mmc or mainline tree.

-Matt

Re: [PATCH] Tokyo Electron SDIO controller (Ellen) support

[Ben Dooks]

On Mon, Dec 03, 2007 at 08:22:07AM -0600, Matt Porter wrote:
> On Sat, Dec 01, 2007 at 09:50:10PM +0100, Pierre Ossman wrote:
> > As for the patch, it's a big NAK at this point. Vendors who can't be bothered to follow
> > the standard will have to wait for Ben's separation patches before they can be supported.
> > Then they can have some voodoo front-end to handle their mistakes instead of turning
> > sdhci.c into a collection of hacks and workarounds.
> 
> What's the status of Ben's separation patches? I haven't seen a posting of those
> versus a recent kernel. I've got some SDHCI driver glue for the non-pci Arasan core
> running in an older kernel tree with those patches. It's just waiting for the separation
> patches to hit the mmc or mainline tree.

I need to go back and try and sort out the last of Pierre's last
comments, and update to the latest kernel version. I was waiting
for 2.6.24-rc4 to re-start the effort to try and ensure there are
fewer changes due to fixes.

-- 
Ben (ben@fluff.org, http://www.fluff.org/)

  'a smiley only costs 4 bytes'

Re: [PATCH] Tokyo Electron SDIO controller (Ellen) support

[Pierre Ossman]

[-- Attachment #1: Type: text/plain, Size: 756 bytes --]

On Mon, 3 Dec 2007 14:39:26 +0000
Ben Dooks <ben@fluff.org> wrote:

> On Mon, Dec 03, 2007 at 08:22:07AM -0600, Matt Porter wrote:
> > 
> > What's the status of Ben's separation patches? I haven't seen a posting of those
> > versus a recent kernel. I've got some SDHCI driver glue for the non-pci Arasan core
> > running in an older kernel tree with those patches. It's just waiting for the separation
> > patches to hit the mmc or mainline tree.
> 
> I need to go back and try and sort out the last of Pierre's last
> comments, and update to the latest kernel version. I was waiting
> for 2.6.24-rc4 to re-start the effort to try and ensure there are
> fewer changes due to fixes.
> 

Any progress on this little project?

Rgds
Pierre

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: [PATCH] Tokyo Electron SDIO controller (Ellen) support

[Pierre Ossman]

[-- Attachment #1: Type: text/plain, Size: 700 bytes --]

On Sat, 01 Dec 2007 00:27:06 -0800
Vitaly Luban <vitaly@luban.org> wrote:

> Kernel "pci_ids.h" file has data for that card missing.
> 
> Also, Ellen needs some control bits flipped before it functions properly 
> as SDIO controller by the spec.
> Should apply clenly to Linus and Drzeus trees. Please apply.
> 
> Signed-off-by:  Vitaly Luban <vitaly@luban.org>
> 
> 

The separation has now finally been performed in case you want to
revisit this patch.

Rgds
-- 
     -- Pierre Ossman

  WARNING: This correspondence is being monitored by the
  Swedish government. Make sure your server uses encryption
  for SMTP traffic and consider using PGP for end-to-end
  encryption.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 197 bytes --]