[tulip] mii taints kernel, big EEPROM patch, missing tulip.txt

[Pavel Roskin]

Hi, Jeff!

In Linux-2.4.18-pre7, the module mii has no license.  Also having a
description would be nice.

==========================
--- linux.orig/drivers/net/mii.c
+++ linux/drivers/net/mii.c
@@ -168,6 +168,10 @@ int mii_nway_restart (struct mii_if_info
 	return r;
 }
 
+MODULE_AUTHOR ("Jeff Garzik <jgarzik@mandrakesoft.com>");
+MODULE_DESCRIPTION ("Kernel part of Ethtool");
+MODULE_LICENSE("GPL");
+
 EXPORT_SYMBOL(mii_link_ok);
 EXPORT_SYMBOL(mii_nway_restart);
 EXPORT_SYMBOL(mii_ethtool_gset);
==========================

By the way, I have finally made a working driver for Conexant, but I
didn't have time to send it from the office today and forgot to copy it to
an external site.  I'll send it on Monday.

Basically it's the patch that I already sent, but with EEPROM_SIZE changed
to 512.  That's because RS7112 keeps the MAC address near the end of
EEPROM.  I feel uneasy that ee_data is now quite big for the stack (512
bytes) in tulip_init_one(), but it works.

Why doesn't tulip_init_one() use tp->eeprom immediately?  tp->eeprom is
allocated in the kernel memory.  Ok, I just made a patch that makes
exactly that and will make it easier to increase EEPROM_SIZE.  I tested it
on my Linksys card, which is determined by the driver as
eth0: ADMtek Comet rev 17 at 0xd800, 00:04:5A:43:9C:F0, IRQ 11.
and listed by lspci as
00:08.0 Ethernet controller: Bridgecom, Inc: Unknown device 0985 (rev 11)

This message will be sent through the patched driver :-)

===========================
--- linux.orig/drivers/net/tulip/tulip_core.c
+++ linux/drivers/net/tulip/tulip_core.c
@@ -1354,7 +1354,6 @@ static int __devinit tulip_init_one (str
 	u8 chip_rev;
 	int i, irq;
 	unsigned short sum;
-	u8 ee_data[EEPROM_SIZE];
 	struct net_device *dev;
 	long ioaddr;
 	static int board_idx = -1;
@@ -1573,17 +1572,17 @@ static int __devinit tulip_init_one (str
 		int sa_offset = 0;
 		int ee_addr_size = tulip_read_eeprom(ioaddr, 0xff, 8) & 0x40000 ? 8 : 6;
 
-		for (i = 0; i < sizeof(ee_data)/2; i++)
-			((u16 *)ee_data)[i] =
+		for (i = 0; i < sizeof(tp->eeprom)/2; i++)
+			((u16 *)tp->eeprom)[i] =
 				le16_to_cpu(tulip_read_eeprom(ioaddr, i, ee_addr_size));
 
 		/* DEC now has a specification (see Notes) but early board makers
 		   just put the address in the first EEPROM locations. */
 		/* This does  memcmp(eedata, eedata+16, 8) */
 		for (i = 0; i < 8; i ++)
-			if (ee_data[i] != ee_data[16+i])
+			if (tp->eeprom[i] != tp->eeprom[16+i])
 				sa_offset = 20;
-		if (ee_data[0] == 0xff  &&  ee_data[1] == 0xff &&  ee_data[2] == 0) {
+		if (tp->eeprom[0] == 0xff && tp->eeprom[1] == 0xff && tp->eeprom[2] == 0) {
 			sa_offset = 2;		/* Grrr, damn Matrox boards. */
 			multiport_cnt = 4;
 		}
@@ -1604,8 +1603,8 @@ static int __devinit tulip_init_one (str
                }
 #endif
 		for (i = 0; i < 6; i ++) {
-			dev->dev_addr[i] = ee_data[i + sa_offset];
-			sum += ee_data[i + sa_offset];
+			dev->dev_addr[i] = tp->eeprom[i + sa_offset];
+			sum += tp->eeprom[i + sa_offset];
 		}
 	}
 	/* Lite-On boards have the address byte-swapped. */
@@ -1676,8 +1675,6 @@ static int __devinit tulip_init_one (str
 	}
 
 	if (tp->flags & HAS_MEDIA_TABLE) {
-		memcpy(tp->eeprom, ee_data, sizeof(tp->eeprom));
-
 		sprintf(dev->name, "tulip%d", board_idx);	/* hack */
 		tulip_parse_eeprom(dev);
 		strcpy(dev->name, "eth%d");			/* un-hack */
===========================

The last thing.  Documentation/networking/00-INDEX in Linux-2.4.18-pre7 
mentions tulip.txt, but there is no such file in the kernel sources.  
Please add it if you have it.

-- 
Regards,
Pavel Roskin