Tulip in 2.4.1-ac14 still poorly

Tulip in 2.4.1-ac14 still poorly

[Stephen Thomas]

[-- Attachment #1: Type: text/plain, Size: 196 bytes --]

On trying 2.4.1-ac13 I hit the tulip driver problems reported elsewhere,
and ac14 does not seem to fix the problem on my machine.  Attached is an
extract from my /var/log/messages.

Stephen Thomas

[-- Attachment #2: messages-2.4.1-ac14 --]
[-- Type: text/plain, Size: 2281 bytes --]

Feb 15 16:07:32 triumph kernel: Linux version 2.4.1-ac14 (root@triumph.isltd.insignia.com) (gcc version 2.95.2 19991024 (release)) #1 Thu Feb 15 15:53:08 GMT 2001 
[ ... ]
Feb 15 16:07:33 triumph kernel: Detected 400.915 MHz processor. 
[ ... ]
Feb 15 16:07:33 triumph kernel: Memory: 126652k/131008k available (1009k kernel code, 3968k reserved, 372k data, 188k init, 0k highmem) 
[ ... ]
Feb 15 16:07:35 triumph kernel: CPU: Intel Pentium II (Deschutes) stepping 01 
[ ... ]
Feb 15 16:07:39 triumph kernel: Linux Tulip driver version 0.9.13b (January 24, 2001) 
Feb 15 16:07:39 triumph kernel: PCI: Found IRQ 10 for device 00:10.0 
Feb 15 16:07:39 triumph kernel: eth0: Lite-On 82c168 PNIC rev 32 at 0xe400, 00:C0:F0:2C:2B:3A, IRQ 10. 
Feb 15 16:07:39 triumph kernel: eth0:  MII transceiver #1 config 3100 status 7829 advertising 01e1. 
[ ... ]
Feb 15 16:07:54 triumph amd[601]: vim:/ mounted fstype host on /.automount/vim/root
Feb 15 16:08:03 triumph ypbind[373]: unknown: RPC: Timed out  
Feb 15 16:08:06 triumph kernel: nfs: server vim not responding, still trying 
Feb 15 16:08:06 triumph last message repeated 2 times
Feb 15 16:08:07 triumph kernel: NETDEV WATCHDOG: eth0: transmit timed out 
Feb 15 16:08:14 triumph PAM-securetty[741]: Couldn't open /etc/securetty
Feb 15 16:08:15 triumph kernel: NETDEV WATCHDOG: eth0: transmit timed out 
Feb 15 16:08:19 triumph amd[601]: file server vim type nfs is down
Feb 15 16:08:23 triumph kernel: NETDEV WATCHDOG: eth0: transmit timed out 
Feb 15 16:08:55 triumph last message repeated 4 times
Feb 15 16:08:57 triumph ypbind[373]: broadcast: RPC: Timed out. 
Feb 15 16:09:03 triumph kernel: NETDEV WATCHDOG: eth0: transmit timed out 
Feb 15 16:09:35 triumph last message repeated 4 times
Feb 15 16:09:51 triumph last message repeated 2 times
Feb 15 16:09:55 triumph kernel: nfs: task 99 can't get a request slot 
Feb 15 16:09:59 triumph kernel: NETDEV WATCHDOG: eth0: transmit timed out 
Feb 15 16:10:01 triumph ypbind[373]: broadcast: RPC: Timed out. 
Feb 15 16:10:07 triumph kernel: NETDEV WATCHDOG: eth0: transmit timed out 
Feb 15 16:10:11 triumph PAM-securetty[772]: Couldn't open /etc/securetty
Feb 15 16:10:15 triumph kernel: NETDEV WATCHDOG: eth0: transmit timed out 
Feb 15 16:10:47 triumph last message repeated 4 times

Re: Tulip in 2.4.1-ac14 still poorly

[Manfred Spraul]

[-- Attachment #1: Type: text/plain, Size: 370 bytes --]

Stephen Thomas wrote:
> 
> On trying 2.4.1-ac13 I hit the tulip driver problems reported elsewhere,
> and ac14 does not seem to fix the problem on my machine.  Attached is an
> extract from my /var/log/messages.
>

Could you try the attached oneliner patches? 

patch-tulip-fix1 is integrated in -ac15, and I send patch-tulip-typo to
Alan a few seconds ago.

--
	Manfred

[-- Attachment #2: patch-tulip-typo --]
[-- Type: text/plain, Size: 409 bytes --]

diff -u 2.4/drivers/net/tulip/pnic.c build-2.4/drivers/net/tulip/pnic.c
--- 2.4/drivers/net/tulip/pnic.c	Fri Feb 16 11:17:03 2001
+++ build-2.4/drivers/net/tulip/pnic.c	Fri Feb 16 11:18:08 2001
@@ -84,7 +84,7 @@
 		tp->full_duplex_lock;
 
 	new_csr6 = tp->csr6;
-	if (negotiated & 0x038)	/* 100mbps. */
+	if (negotiated & 0x0380)	/* 100mbps. */
 		new_csr6 &= ~0x00400000;
 	 else
 		new_csr6 |= 0x00400000;


[-- Attachment #3: patch-tulip-fix1 --]
[-- Type: text/plain, Size: 814 bytes --]

diff -u 2.4/drivers/net/tulip/pnic.c build-2.4/drivers/net/tulip/pnic.c
--- 2.4/drivers/net/tulip/pnic.c	Thu Feb 15 00:51:38 2001
+++ build-2.4/drivers/net/tulip/pnic.c	Thu Feb 15 01:05:59 2001
@@ -93,8 +93,6 @@
 	 else	
 		new_csr6 &= ~0x0200;
 	if (new_csr6 != tp->csr6) {
-		/* stop the transceiver*/
-		tulip_stop_rxtx(tp, tp->csr6);
 		tp->full_duplex = duplex;
 		tp->csr6 = new_csr6;
 		if (tulip_debug > 0)
@@ -102,10 +100,7 @@
 				   "#%d link partner capability of %4.4x.\n",
 				   dev->name, tp->full_duplex ? "full" : "half",
 				   tp->phys[0], mii_reg5);
-		/* When the transceiver is stopped it triggeres
-		 * a "Transmit stopped interrupt" (misnamed as TxDied).
-		 * The interrupt handler will restart the transceiver
-		 */
+		tulip_restart_rxtx(tp, tp->csr6);
 		return 1;
 	}
 	return 0;

Re: Tulip in 2.4.1-ac14 still poorly

[Stephen Thomas]

Manfred Spraul wrote:
> Could you try the attached oneliner patches?
> 
> patch-tulip-fix1 is integrated in -ac15, and I send patch-tulip-typo to
> Alan a few seconds ago.

Just booted ac15 with the typo patch applied, seems to work well enough
for me to be sending you this and there are no obvious nasty messages in
my syslog.  Thanks!

Stephen