Re: [PATCH] sungem: add support for G5 PowerMac, some PM fixes

[Benjamin Herrenschmidt <benh@kernel.crashing.org>]

On Thu, 2004-01-22 at 17:08, David S. Miller wrote:
> On Thu, 22 Jan 2004 16:50:43 +1100
> Benjamin Herrenschmidt <benh@kernel.crashing.org> wrote:
> 
> > +			if (gp->pdev->device == PCI_DEVICE_ID_APPLE_K2_GMAC)
> > +				break;
> > +
> > +       			for (i = 0; i < 32; i++) {
> 
> Please fix this indentation, and I'll happily apply this ;-)
> 
> I assume you'll have a 2.4.x variant of this patch too?

Ohh, this was just a tab vs. space issue actually, the resulting indent
was fine ;) Anyway, here's the fixed version. The 2.4 backport will
come next week hopefully when I'll walk through all my pending 2.4
stuffs, I want to dbl check a few things in it first (and I'm not yet
100% certain I will merge the G5 support in 2.4 upstream anyway).

Note: the new PCI ID isn't yet in Linus tree, but it's in the patches
I'm currently sending to Andrew.

Ben.

diff -urN linux-2.5-merge/drivers/net/sungem.c linuxppc-2.5-benh/drivers/net/sungem.c
--- linux-2.5-merge/drivers/net/sungem.c	2004-01-23 10:10:00.836189096 +1100
+++ linuxppc-2.5-benh/drivers/net/sungem.c	2004-01-23 10:10:53.087245728 +1100
@@ -103,6 +103,8 @@
 	  PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0UL },
 	{ PCI_VENDOR_ID_APPLE, PCI_DEVICE_ID_APPLE_UNI_N_GMAC2,
 	  PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0UL },
+	{ PCI_VENDOR_ID_APPLE, PCI_DEVICE_ID_APPLE_K2_GMAC,
+	  PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0UL },
 	{0, }
 };
 
@@ -778,6 +780,10 @@
 	struct gem *gp = dev->priv;
 	u32 gem_status = readl(gp->regs + GREG_STAT);
 
+	/* Swallow interrupts when shutting the chip down */
+	if (gp->hw_running == 0)
+		goto out;
+
 	spin_lock(&gp->lock);
 
 	if (gem_status & GREG_STAT_ABNORMAL) {
@@ -1240,6 +1246,12 @@
 		gp->lstate = link_force_ok;
 		return 0;
 	case link_aneg:
+		/* We try forced modes after a failed aneg only on PHYs that don't
+		 * have "magic_aneg" bit set, which means they internally do the
+		 * while forced-mode thingy. On these, we just restart aneg
+		 */
+		if (gp->phy_mii.def->magic_aneg)
+			return 1;
 		if (netif_msg_link(gp))
 			printk(KERN_INFO "%s: switching to forced 100bt\n",
 				gp->dev->name);
@@ -1497,18 +1509,26 @@
 		 * to schedule instead
 		 */
 		pmac_call_feature(PMAC_FTR_GMAC_PHY_RESET, gp->of_node, 0, 0);
-		mdelay(10);
 		for (j = 0; j < 3; j++) {
 			/* Some PHYs used by apple have problem getting back to us,
-			 * we _know_ it's actually at addr 0, that's a hack, but
+			 * we _know_ it's actually at addr 0 or 1, that's a hack, but
 			 * it helps to do that reset now. I suspect some motherboards
 			 * don't wire the PHY reset line properly, thus the PHY doesn't
 			 * come back with the above pmac_call_feature.
 			 */
 			gp->mii_phy_addr = 0;
 			phy_write(gp, MII_BMCR, BMCR_RESET);
+			gp->mii_phy_addr = 1;
+			phy_write(gp, MII_BMCR, BMCR_RESET);
 			/* We should probably break some locks here and schedule... */
 			mdelay(10);
+			
+			/* On K2, we only probe the internal PHY at address 1, other
+			 * addresses tend to return garbage.
+			 */
+			if (gp->pdev->device == PCI_DEVICE_ID_APPLE_K2_GMAC)
+				break;
+
 			for (i = 0; i < 32; i++) {
 				gp->mii_phy_addr = i;
 				if (phy_read(gp, MII_BMCR) != 0xffff)
@@ -1770,6 +1790,8 @@
 /* Must be invoked under gp->lock. */
 static void gem_init_pause_thresholds(struct gem *gp)
 {
+       	u32 cfg;
+
 	/* Calculate pause thresholds.  Setting the OFF threshold to the
 	 * full RX fifo size effectively disables PAUSE generation which
 	 * is what we do for 10/100 only GEMs which have FIFOs too small
@@ -1786,17 +1808,28 @@
 		gp->rx_pause_on = on;
 	}
 
-	{
-		u32 cfg;
 
-		cfg  = 0;
+	/* Configure the chip "burst" DMA mode & enable some
+	 * HW bug fixes on Apple version
+	 */
+       	cfg  = 0;
+       	if (gp->pdev->vendor == PCI_VENDOR_ID_APPLE)
+		cfg |= GREG_CFG_RONPAULBIT | GREG_CFG_ENBUG2FIX;
 #if !defined(CONFIG_SPARC64) && !defined(CONFIG_ALPHA)
-		cfg |= GREG_CFG_IBURST;
+       	cfg |= GREG_CFG_IBURST;
 #endif
-		cfg |= ((31 << 1) & GREG_CFG_TXDMALIM);
-		cfg |= ((31 << 6) & GREG_CFG_RXDMALIM);
+       	cfg |= ((31 << 1) & GREG_CFG_TXDMALIM);
+       	cfg |= ((31 << 6) & GREG_CFG_RXDMALIM);
+       	writel(cfg, gp->regs + GREG_CFG);
+
+	/* If Infinite Burst didn't stick, then use different
+	 * thresholds (and Apple bug fixes don't exist)
+	 */
+	if (readl(gp->regs + GREG_CFG) & GREG_CFG_IBURST) {
+		cfg = ((2 << 1) & GREG_CFG_TXDMALIM);
+		cfg = ((8 << 6) & GREG_CFG_RXDMALIM);
 		writel(cfg, gp->regs + GREG_CFG);
-	}
+	}	
 }
 
 static int gem_check_invariants(struct gem *gp)
@@ -1931,18 +1964,10 @@
 	u16 cmd;
 	u32 mif_cfg;
 
+	mb();
 	pmac_call_feature(PMAC_FTR_GMAC_ENABLE, gp->of_node, 0, 1);
 
-	current->state = TASK_UNINTERRUPTIBLE;
-	schedule_timeout((21 * HZ) / 1000);
-
-	pci_read_config_word(gp->pdev, PCI_COMMAND, &cmd);
-	cmd |= PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER | PCI_COMMAND_INVALIDATE;
-    	pci_write_config_word(gp->pdev, PCI_COMMAND, cmd);
-    	pci_write_config_byte(gp->pdev, PCI_LATENCY_TIMER, 6);
-    	pci_write_config_byte(gp->pdev, PCI_CACHE_LINE_SIZE, 8);
-
-	mdelay(1);
+	udelay(3);
 	
 	mif_cfg = readl(gp->regs + MIF_CFG);
 	mif_cfg &= ~(MIF_CFG_PSELECT|MIF_CFG_POLL|MIF_CFG_BBMODE|MIF_CFG_MDI1);
@@ -1950,8 +1975,6 @@
 	writel(mif_cfg, gp->regs + MIF_CFG);
 	writel(PCS_DMODE_MGM, gp->regs + PCS_DMODE);
 	writel(MAC_XIFCFG_OE, gp->regs + MAC_XIFCFG);
-
-	mdelay(1);
 }
 
 /* Turn off the chip's clock */
@@ -1962,10 +1985,17 @@
 
 #endif /* CONFIG_PPC_PMAC */
 
-/* Must be invoked under gp->lock. */
+/* Must be invoked with no lock held. */
 static void gem_stop_phy(struct gem *gp)
 {
 	u32 mifcfg;
+	unsigned long flags;
+
+	/* Let the chip settle down a bit, it seems that helps
+	 * for sleep mode on some models
+	 */
+	set_current_state(TASK_UNINTERRUPTIBLE);
+	schedule_timeout(HZ/100);
 
 	/* Make sure we aren't polling PHY status change. We
 	 * don't currently use that feature though
@@ -1976,17 +2006,28 @@
 
 	if (gp->wake_on_lan) {
 		/* Setup wake-on-lan */
-	} else
+	} else {
 		writel(0, gp->regs + MAC_RXCFG);
+		(void)readl(gp->regs + MAC_RXCFG);
+		/* Machine sleep will die in strange ways if we
+		 * dont wait a bit here, looks like the chip takes
+		 * some time to really shut down
+		 */
+		set_current_state(TASK_UNINTERRUPTIBLE);
+		schedule_timeout(HZ/100);
+	}
+
 	writel(0, gp->regs + MAC_TXCFG);
 	writel(0, gp->regs + MAC_XIFCFG);
 	writel(0, gp->regs + TXDMA_CFG);
 	writel(0, gp->regs + RXDMA_CFG);
 
 	if (!gp->wake_on_lan) {
+		spin_lock_irqsave(&gp->lock, flags);
 		gem_stop(gp);
 		writel(MAC_TXRST_CMD, gp->regs + MAC_TXRST);
 		writel(MAC_RXRST_CMD, gp->regs + MAC_RXRST);
+		spin_unlock_irqrestore(&gp->lock, flags);
 	}
 
 	if (found_mii_phy(gp) && gp->phy_mii.def->ops->suspend)
@@ -2008,31 +2049,33 @@
 /* Shut down the chip, must be called with pm_sem held.  */
 static void gem_shutdown(struct gem *gp)
 {
-	/* Make us not-running to avoid timers respawning */
+	/* Make us not-running to avoid timers respawning
+	 * and swallow irqs 
+	 */
 	gp->hw_running = 0;
+	wmb();
 
 	/* Stop the link timer */
 	del_timer_sync(&gp->link_timer);
 
 	/* Stop the reset task */
 	while (gp->reset_task_pending)
-		schedule();
+		yield();
 	
 	/* Actually stop the chip */
-	spin_lock_irq(&gp->lock);
 	if (gp->pdev->vendor == PCI_VENDOR_ID_APPLE) {
 		gem_stop_phy(gp);
 
-		spin_unlock_irq(&gp->lock);
-
 #ifdef CONFIG_PPC_PMAC
 		/* Power down the chip */
 		gem_apple_powerdown(gp);
 #endif /* CONFIG_PPC_PMAC */
-	} else {
-		gem_stop(gp);
+	} else{
+		unsigned long flags;
 
-		spin_unlock_irq(&gp->lock);
+		spin_lock_irqsave(&gp->lock, flags);
+		gem_stop(gp);
+		spin_unlock_irqrestore(&gp->lock, flags);	
 	}
 }
 
@@ -2692,6 +2735,7 @@
 	 * not have properly shut down the PHY.
 	 */
 #ifdef CONFIG_PPC_PMAC
+	gp->of_node = pci_device_to_OF_node(pdev);
 	if (pdev->vendor == PCI_VENDOR_ID_APPLE)
 		gem_apple_powerup(gp);
 #endif
@@ -2725,9 +2769,6 @@
 		goto err_out_iounmap;
 	}
 
-#ifdef CONFIG_PPC_PMAC
-	gp->of_node = pci_device_to_OF_node(pdev);
-#endif	
 	if (gem_get_device_address(gp))
 		goto err_out_free_consistent;
 
diff -urN linux-2.5-merge/drivers/net/sungem.h linuxppc-2.5-benh/drivers/net/sungem.h
--- linux-2.5-merge/drivers/net/sungem.h	2004-01-23 10:10:00.853186512 +1100
+++ linuxppc-2.5-benh/drivers/net/sungem.h	2004-01-23 10:10:15.303989656 +1100
@@ -28,6 +28,9 @@
 #define GREG_CFG_IBURST		0x00000001	/* Infinite Burst		*/
 #define GREG_CFG_TXDMALIM	0x0000003e	/* TX DMA grant limit		*/
 #define GREG_CFG_RXDMALIM	0x000007c0	/* RX DMA grant limit		*/
+#define GREG_CFG_RONPAULBIT	0x00000800	/* Use mem read multiple for PCI read
+						 * after infinite burst (Apple) */
+#define GREG_CFG_ENBUG2FIX	0x00001000	/* Fix Rx hang after overflow */
 
 /* Global Interrupt Status Register.
  *
diff -urN linux-2.5-merge/drivers/net/sungem_phy.c linuxppc-2.5-benh/drivers/net/sungem_phy.c
--- linux-2.5-merge/drivers/net/sungem_phy.c	2004-01-23 10:10:00.866184536 +1100
+++ linuxppc-2.5-benh/drivers/net/sungem_phy.c	2004-01-23 10:10:15.323986616 +1100
@@ -72,7 +72,7 @@
 	int limit = 10000;
 	
 	val = __phy_read(phy, phy_id, MII_BMCR);
-	val &= ~BMCR_ISOLATE;
+	val &= ~(BMCR_ISOLATE | BMCR_PDOWN);
 	val |= BMCR_RESET;
 	__phy_write(phy, phy_id, MII_BMCR, val);
 
@@ -157,7 +157,7 @@
 	data |= MII_BCM5400_GB_CONTROL_FULLDUPLEXCAP;
 	phy_write(phy, MII_BCM5400_GB_CONTROL, data);
 	
-	mdelay(10);
+	udelay(100);
 
 	/* Reset and configure cascaded 10/100 PHY */
 	(void)reset_one_mii_phy(phy, 0x1f);
@@ -217,7 +217,7 @@
 	data |= MII_BCM5400_GB_CONTROL_FULLDUPLEXCAP;
 	phy_write(phy, MII_BCM5400_GB_CONTROL, data);
 
-	mdelay(10);
+	udelay(10);
 
 	/* Reset and configure cascaded 10/100 PHY */
 	(void)reset_one_mii_phy(phy, 0x1f);
@@ -258,7 +258,7 @@
 	data |= MII_BCM5400_GB_CONTROL_FULLDUPLEXCAP;
 	phy_write(phy, MII_BCM5400_GB_CONTROL, data);
 
-	mdelay(10);
+	udelay(10);
 
 	/* Reset and configure cascaded 10/100 PHY */
 	(void)reset_one_mii_phy(phy, 0x1f);
@@ -302,6 +302,15 @@
 	return 0;
 }
 
+static int bcm5421k2_init(struct mii_phy* phy)
+{
+	/* Init code borrowed from OF */
+	phy_write(phy, 4, 0x01e1);
+	phy_write(phy, 9, 0x0300);
+
+	return 0;
+}
+
 static int bcm54xx_setup_aneg(struct mii_phy *phy, u32 advertise)
 {
 	u16 ctl, adv;
@@ -647,7 +656,7 @@
 	.phy_id_mask	= 0xfffffff0,
 	.name		= "BCM5201",
 	.features	= MII_BASIC_FEATURES,
-	.magic_aneg	= 0,
+	.magic_aneg	= 1,
 	.ops		= &bcm5201_phy_ops
 };
 
@@ -666,7 +675,7 @@
 	.phy_id_mask	= 0xfffffff0,
 	.name		= "BCM5221",
 	.features	= MII_BASIC_FEATURES,
-	.magic_aneg	= 0,
+	.magic_aneg	= 1,
 	.ops		= &bcm5221_phy_ops
 };
 
@@ -746,6 +755,25 @@
 	.ops		= &bcm5421_phy_ops
 };
 
+/* Broadcom BCM 5421 built-in K2 */
+static struct mii_phy_ops bcm5421k2_phy_ops = {
+	.init		= bcm5421k2_init,
+	.suspend	= bcm5411_suspend,
+	.setup_aneg	= bcm54xx_setup_aneg,
+	.setup_forced	= bcm54xx_setup_forced,
+	.poll_link	= genmii_poll_link,
+	.read_link	= bcm54xx_read_link,
+};
+
+static struct mii_phy_def bcm5421k2_phy_def = {
+	.phy_id		= 0x002062e0,
+	.phy_id_mask	= 0xfffffff0,
+	.name		= "BCM5421-K2",
+	.features	= MII_GBIT_FEATURES,
+	.magic_aneg	= 1,
+	.ops		= &bcm5421k2_phy_ops
+};
+
 /* Marvell 88E1101 (Apple seem to deal with 2 different revs,
  * I masked out the 8 last bits to get both, but some specs
  * would be useful here) --BenH.
@@ -790,6 +818,7 @@
 	&bcm5401_phy_def,
 	&bcm5411_phy_def,
 	&bcm5421_phy_def,
+	&bcm5421k2_phy_def,
 	&marvell_phy_def,
 	&genmii_phy_def,
 	NULL
@@ -813,8 +842,8 @@
 		goto fail;
 
 	/* Read ID and find matching entry */	
-	id = (phy_read(phy, MII_PHYSID1) << 16 | phy_read(phy, MII_PHYSID2))
-			 	& 0xfffffff0;
+	id = (phy_read(phy, MII_PHYSID1) << 16 | phy_read(phy, MII_PHYSID2));
+	printk(KERN_DEBUG "PHY ID: %x, addr: %x\n", id, mii_id);
 	for (i=0; (def = mii_phy_table[i]) != NULL; i++)
 		if ((id & def->phy_id_mask) == def->phy_id)
 			break;