[TG3]: Speed up SRAM access

Speed up SRAM read and write functions if possible by using MMIO
instead of config. cycles. With this change, the post reset signature
done at the end of D3 power change must now be moved before the D3
power change.

IBM reported a problem on powerpc blades during ethtool self test
that was caused by the memory test taking excessively long. Config.
cycles are very slow on powerpc and the memory test can take more
than 10 seconds to complete using config. cycles. As a result, NETDEV
WATCHDOG can be triggered during self test and the chip can end up in
a funny state.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index 3c5c9fa..c504ff2 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -497,33 +497,40 @@
 	unsigned long flags;
 
 	spin_lock_irqsave(&tp->indirect_lock, flags);
-	pci_write_config_dword(tp->pdev, TG3PCI_MEM_WIN_BASE_ADDR, off);
-	pci_write_config_dword(tp->pdev, TG3PCI_MEM_WIN_DATA, val);
+	if (tp->write32 != tg3_write_indirect_reg32) {
+		tw32_f(TG3PCI_MEM_WIN_BASE_ADDR, off);
+		tw32_f(TG3PCI_MEM_WIN_DATA, val);
 
-	/* Always leave this as zero. */
-	pci_write_config_dword(tp->pdev, TG3PCI_MEM_WIN_BASE_ADDR, 0);
+		/* Always leave this as zero. */
+		tw32_f(TG3PCI_MEM_WIN_BASE_ADDR, 0);
+	} else {
+		pci_write_config_dword(tp->pdev, TG3PCI_MEM_WIN_BASE_ADDR, off);
+		pci_write_config_dword(tp->pdev, TG3PCI_MEM_WIN_DATA, val);
+
+		/* Always leave this as zero. */
+		pci_write_config_dword(tp->pdev, TG3PCI_MEM_WIN_BASE_ADDR, 0);
+	}
 	spin_unlock_irqrestore(&tp->indirect_lock, flags);
 }
 
-static void tg3_write_mem_fast(struct tg3 *tp, u32 off, u32 val)
-{
-	/* If no workaround is needed, write to mem space directly */
-	if (tp->write32 != tg3_write_indirect_reg32)
-		tw32(NIC_SRAM_WIN_BASE + off, val);
-	else
-		tg3_write_mem(tp, off, val);
-}
-
 static void tg3_read_mem(struct tg3 *tp, u32 off, u32 *val)
 {
 	unsigned long flags;
 
 	spin_lock_irqsave(&tp->indirect_lock, flags);
-	pci_write_config_dword(tp->pdev, TG3PCI_MEM_WIN_BASE_ADDR, off);
-	pci_read_config_dword(tp->pdev, TG3PCI_MEM_WIN_DATA, val);
+	if (tp->write32 != tg3_write_indirect_reg32) {
+		tw32_f(TG3PCI_MEM_WIN_BASE_ADDR, off);
+		*val = tr32(TG3PCI_MEM_WIN_DATA);
 
-	/* Always leave this as zero. */
-	pci_write_config_dword(tp->pdev, TG3PCI_MEM_WIN_BASE_ADDR, 0);
+		/* Always leave this as zero. */
+		tw32_f(TG3PCI_MEM_WIN_BASE_ADDR, 0);
+	} else {
+		pci_write_config_dword(tp->pdev, TG3PCI_MEM_WIN_BASE_ADDR, off);
+		pci_read_config_dword(tp->pdev, TG3PCI_MEM_WIN_DATA, val);
+
+		/* Always leave this as zero. */
+		pci_write_config_dword(tp->pdev, TG3PCI_MEM_WIN_BASE_ADDR, 0);
+	}
 	spin_unlock_irqrestore(&tp->indirect_lock, flags);
 }
 
@@ -1367,12 +1374,12 @@
 		}
 	}
 
+	tg3_write_sig_post_reset(tp, RESET_KIND_SHUTDOWN);
+
 	/* Finally, set the new power state. */
 	pci_write_config_word(tp->pdev, pm + PCI_PM_CTRL, power_control);
 	udelay(100);	/* Delay after power state change */
 
-	tg3_write_sig_post_reset(tp, RESET_KIND_SHUTDOWN);
-
 	return 0;
 }
 
@@ -6537,11 +6544,11 @@
 		if (tp->tg3_flags & TG3_FLAG_ENABLE_ASF) {
 			u32 val;
 
-			tg3_write_mem_fast(tp, NIC_SRAM_FW_CMD_MBOX,
-					   FWCMD_NICDRV_ALIVE2);
-			tg3_write_mem_fast(tp, NIC_SRAM_FW_CMD_LEN_MBOX, 4);
+			tg3_write_mem(tp, NIC_SRAM_FW_CMD_MBOX,
+				      FWCMD_NICDRV_ALIVE2);
+			tg3_write_mem(tp, NIC_SRAM_FW_CMD_LEN_MBOX, 4);
 			/* 5 seconds timeout */
-			tg3_write_mem_fast(tp, NIC_SRAM_FW_CMD_DATA_MBOX, 5);
+			tg3_write_mem(tp, NIC_SRAM_FW_CMD_DATA_MBOX, 5);
 			val = tr32(GRC_RX_CPU_EVENT);
 			val |= (1 << 14);
 			tw32(GRC_RX_CPU_EVENT, val);