[net/8139cp] still crashes my notebook

[net/8139cp] still crashes my notebook

[Andreas Happe]

hi,

my notebook (hp/compaq nx7000) still crashes when using 8139cp (runs
rock solid with 8139too driver). The computer just locks up, there is no
dmesg output. This has happened since I've got this laptop (around
november '03).

8139too outputs:
| 8139too Fast Ethernet driver 0.9.27
| 8139too: pci dev 0000:02:01.0 (id 10ec:8139 rev 20) is an enhanced 8139C+ chip
| 8139too: Use the "8139cp" driver for improved performance and stability.
| 8139too: 0000:02:01.0: unknown chip version, assuming RTL-8139
| 8139too: 0000:02:01.0: TxConfig = 0x74800000

the "Use the.." line is rather annoying..

lspci lists:
02:01.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+ (rev 20)

	--Andreas

Re: [net/8139cp] still crashes my notebook

[OGAWA Hirofumi]

Andreas Happe <andreashappe@gmx.net> writes:

> hi,
> 
> my notebook (hp/compaq nx7000) still crashes when using 8139cp (runs
> rock solid with 8139too driver). The computer just locks up, there is no
> dmesg output. This has happened since I've got this laptop (around
> november '03).

It seems 8139cp.c has the race condition of rx_poll and interrupt.
Does the following patch fix this problem?

NOTE, since I don't have this device, patch is untested. Sorry.
-- 
OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>

---

 drivers/net/8139cp.c |   12 ++++++++++--
 1 files changed, 10 insertions(+), 2 deletions(-)

diff -puN drivers/net/8139cp.c~8139cp-napi-race-fix drivers/net/8139cp.c
--- linux-2.6.2-rc1/drivers/net/8139cp.c~8139cp-napi-race-fix	2004-01-24 02:22:36.000000000 +0900
+++ linux-2.6.2-rc1-hirofumi/drivers/net/8139cp.c	2004-01-24 02:37:43.000000000 +0900
@@ -615,8 +615,10 @@ rx_next:
 		if (cpr16(IntrStatus) & cp_rx_intr_mask)
 			goto rx_status_loop;
 
-		netif_rx_complete(dev);
+		local_irq_disable();
 		cpw16_f(IntrMask, cp_intr_mask);
+		__netif_rx_complete(dev);
+		local_irq_enable();
 
 		return 0;	/* done */
 	}
@@ -643,6 +645,12 @@ cp_interrupt (int irq, void *dev_instanc
 
 	spin_lock(&cp->lock);
 
+	/* close possible race's with dev_close */
+	if (unlikely(!netif_running(dev))) {
+		cpw16(IntrMask, 0);
+		goto out;
+	}
+
 	if (status & (RxOK | RxErr | RxEmpty | RxFIFOOvr)) {
 		if (netif_rx_schedule_prep(dev)) {
 			cpw16_f(IntrMask, cp_norx_intr_mask);
@@ -664,7 +672,7 @@ cp_interrupt (int irq, void *dev_instanc
 
 		/* TODO: reset hardware */
 	}
-
+out:
 	spin_unlock(&cp->lock);
 	return IRQ_HANDLED;
 }

_

Re: [net/8139cp] still crashes my notebook

[Thomas Svedberg]

fre 2004-01-23 klockan 18.56 skrev OGAWA Hirofumi:
> Andreas Happe <andreashappe@gmx.net> writes:
> 
> > hi,
> > 
> > my notebook (hp/compaq nx7000) still crashes when using 8139cp (runs
> > rock solid with 8139too driver). The computer just locks up, there is no
> > dmesg output. This has happened since I've got this laptop (around
> > november '03).
> 
> It seems 8139cp.c has the race condition of rx_poll and interrupt.
> Does the following patch fix this problem?
> 
> NOTE, since I don't have this device, patch is untested. Sorry.

I can confirm that this patch fixes the same complete lockup I have had
since 2.6.0-test4 or so, (See: "Hard lock with recent 2.6.0-test
kernels").

Tested against 2.6.2-rc1-mm2 which locks har without patch and works
great with it.

-- 
/ Thomas
.......................................................................
 Thomas Svedberg
 Department of Applied Mechanics
 Chalmers University of Technology

 Address: S-412 96 Göteborg, SWEDEN
 E-mail : thsv@bigfoot.com, thsv@am.chalmers.se
 Phone  : +46 31 772 1522
 Fax    : +46 31 772 3827
.......................................................................

Re: [net/8139cp] still crashes my notebook

[Jeff Garzik]

applied to 2.4 and 2.6

Re: [net/8139cp] still crashes my notebook

[OGAWA Hirofumi]

Thomas Svedberg <thsv@am.chalmers.se> writes:

> fre 2004-01-23 klockan 18.56 skrev OGAWA Hirofumi:
> > Andreas Happe <andreashappe@gmx.net> writes:
> > 
> > > hi,
> > > 
> > > my notebook (hp/compaq nx7000) still crashes when using 8139cp (runs
> > > rock solid with 8139too driver). The computer just locks up, there is no
> > > dmesg output. This has happened since I've got this laptop (around
> > > november '03).
> > 
> > It seems 8139cp.c has the race condition of rx_poll and interrupt.
> > Does the following patch fix this problem?
> > 
> > NOTE, since I don't have this device, patch is untested. Sorry.
> 
> I can confirm that this patch fixes the same complete lockup I have had
> since 2.6.0-test4 or so, (See: "Hard lock with recent 2.6.0-test
> kernels").
> 
> Tested against 2.6.2-rc1-mm2 which locks har without patch and works
> great with it.

Thaks for testing.
-- 
OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>

Re: [net/8139cp] still crashes my notebook

[OGAWA Hirofumi]

Jeff Garzik <jgarzik@pobox.com> writes:

> applied to 2.4 and 2.6

Oh, thanks.
-- 
OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>

Re: [net/8139cp] still crashes my notebook

[Andreas Happe]

On 2004-01-24, OGAWA Hirofumi <hirofumi@mail.parknet.co.jp> wrote:
> Thomas Svedberg <thsv@am.chalmers.se> writes:
>> I can confirm that this patch fixes the same complete lockup I have had
>> since 2.6.0-test4 or so, (See: "Hard lock with recent 2.6.0-test
>> kernels").
>> 
>> Tested against 2.6.2-rc1-mm2 which locks har without patch and works
>> great with it.
>
> Thaks for testing.

I just wanted to confirm that this patch has solved my problem (my
computer hasn't crashed with 8139cp since ~48h).

Thanks for the patch,
Andreas