Skip to contentSkip to conversation

8139too not working in 2.6

DISCUSSION Mirko Caserta 8 participants 15 messages Apr 26, 2004 Network drivers

lore
Share & tools
lore
#0 lore raw

Yes, I know, it's a damn cheap eth card and I should get it replaced :)

Besides that, this card works just fine with 2.4.25 while it refuses to work on a recent 2.6 kernel. I tried 2.6.5 and even 2.6.5-rc2-mm2-broken-out with no luck.

The card is correctly recognized but the kernel refuses to transmit any packet:

8139too Fast Ethernet driver 0.9.27

eth0: RealTek RTL8139 at 0xe0821000, MAC_ADDR_REMOVED, IRQ 5
eth0:  Identified 8139 chip type 'RTL-8139C'
eth0: link up, 100Mbps, full-duplex, lpa 0x45E1
eth0: Transmit timeout, status 0c 0005 c07f media 10.
eth0: Tx queue start entry 4  dirty entry 0.
eth0:  Tx descriptor 0 is 0008a03c. (queue head)
eth0:  Tx descriptor 1 is 0008a03c.
eth0:  Tx descriptor 2 is 0008a03c.
eth0:  Tx descriptor 3 is 0008a03c.

Someone told me to play around with the driver options but the machine is in production and I cannot play much with reboots :/

Any help would be very appreciated.

Mirko

#1 lore raw
Quoted 6 lines · Mirko Caserta

Yes, I know, it's a damn cheap eth card and I should get it replaced :)

Besides that, this card works just fine with 2.4.25 while it refuses to work on a recent 2.6 kernel. I tried 2.6.5 and even 2.6.5-rc2-mm2-broken-out with no luck.

Just to say that I'm having the exact same problem.

The card works fine with 2.6.3 but not 2.6.5. I haven't tried 2.6.4.

Regards,
David.

Signature · 2 lines
David Johnson
http://www.david-web.co.uk/
#2 lore raw
Quoted 8 lines · David Johnson
Quoted 6 lines · Mirko Caserta

Yes, I know, it's a damn cheap eth card and I should get it replaced :)

Besides that, this card works just fine with 2.4.25 while it refuses to work on a recent 2.6 kernel. I tried 2.6.5 and even 2.6.5-rc2-mm2-broken-out with no luck.

Quoted 7 lines

Just to say that I'm having the exact same problem.

The card works fine with 2.6.3 but not 2.6.5. I haven't tried 2.6.4.

Regards,
David.

I have that card, its worked fine in all 2.6.x kernels.

Signature · 2 lines
Windows - "Its been almost an hour since I crashed. Rebooting just to
make sure everything is OK. All your data will be lost."
#3 lore raw
In reply to David Johnson
Quoted 4 lines · David Johnson
Quoted 3 lines · Mirko Caserta

Yes, I know, it's a damn cheap eth card and I should get it replaced :)

Quoted 8 lines
Quoted 3 lines

Besides that, this card works just fine with 2.4.25 while it refuses to work on a recent 2.6 kernel. I tried 2.6.5 and even 2.6.5-rc2-mm2-broken-out with no luck.

Just to say that I'm having the exact same problem.

The card works fine with 2.6.3 but not 2.6.5. I haven't tried 2.6.4.

Please see my other reply to Mirko.

Quoted 3 lines

Regards,
David.

Signature · 8 lines
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
99.22% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com attornies please note, additions to this message
by Gene Heskett are:
Copyright 2004 by Maurice Eugene Heskett, all rights reserved.
#4 lore raw
In reply to David Johnson

Here's the full information for my machine.

Attached is my dmesg, lspci and the output of dump_pirq.pl. This was when running 2.6.6-rc1.

Regards,
David.

Signature · 2 lines
David Johnson
http://www.david-web.co.uk/
#5 lore raw
Quoted 2 lines · David Johnson

Attached is my dmesg, lspci and the output of dump_pirq.pl. This was when running 2.6.6-rc1.

Looks like 8139too still isn't loaded. Could you apply the attached patch, and send the output of dmesg after the problem was happened?

I'd like to see the debugging message of rtl8139_tx_timeout().

Thanks.

Signature · 1 line
OGAWA Hirofumi <[email protected]>

[PATCH] 8139too: more useful debug info for tx_timeout

	/* disable Tx ASAP, if not already */
	tmp8 = RTL_R8 (ChipCmd);
	if (tmp8 & CmdTxEnb)
		RTL_W8 (ChipCmd, CmdRxEnb);

The above will clear the Tx Descs. So, this prints the debugging info before rtl8139_tx_timeout() does it. And IntrStatus etc. also prints anytime for the debug.

---

 drivers/net/8139too.c |   26 +++++++++++---------------
 1 files changed, 11 insertions(+), 15 deletions(-)
drivers/net/8139too.c renamed from drivers/net/8139too.c~8139too-useful-txtimeout +11 -15
diff -puN drivers/net/8139too.c~8139too-useful-txtimeout drivers/net/8139too.c
@@ -1677,11 +1677,17 @@ static void rtl8139_tx_timeout (struct n
 	u8 tmp8;
 	unsigned long flags;
 
-	DPRINTK ("%s: Transmit timeout, status %2.2x %4.4x "
-		 "media %2.2x.\n", dev->name,
-		 RTL_R8 (ChipCmd),
-		 RTL_R16 (IntrStatus),
-		 RTL_R8 (MediaStatus));
+	printk (KERN_DEBUG "%s: Transmit timeout, status %2.2x %4.4x %4.4x "
+		"media %2.2x.\n", dev->name, RTL_R8 (ChipCmd),
+		RTL_R16(IntrStatus), RTL_R16(IntrMask), RTL_R8(MediaStatus));
+	/* Emit info to figure out what went wrong. */
+	printk (KERN_DEBUG "%s: Tx queue start entry %ld  dirty entry %ld.\n",
+		dev->name, tp->cur_tx, tp->dirty_tx);
+	for (i = 0; i < NUM_TX_DESC; i++)
+		printk (KERN_DEBUG "%s:  Tx descriptor %d is %8.8lx.%s\n",
+			dev->name, i, RTL_R32 (TxStatus0 + (i * 4)),
+			i == tp->dirty_tx % NUM_TX_DESC ?
+				" (queue head)" : "");
 
 	tp->xstats.tx_timeouts++;
 
@@ -1694,15 +1700,6 @@ static void rtl8139_tx_timeout (struct n
 	/* Disable interrupts by clearing the interrupt mask. */
 	RTL_W16 (IntrMask, 0x0000);
 
-	/* Emit info to figure out what went wrong. */
-	printk (KERN_DEBUG "%s: Tx queue start entry %ld  dirty entry %ld.\n",
-		dev->name, tp->cur_tx, tp->dirty_tx);
-	for (i = 0; i < NUM_TX_DESC; i++)
-		printk (KERN_DEBUG "%s:  Tx descriptor %d is %8.8lx.%s\n",
-			dev->name, i, RTL_R32 (TxStatus0 + (i * 4)),
-			i == tp->dirty_tx % NUM_TX_DESC ?
-				" (queue head)" : "");
-
 	/* Stop a shared interrupt from scavenging while we are. */
 	spin_lock_irqsave (&tp->lock, flags);
 	rtl8139_tx_clear (tp);
@@ -1714,7 +1711,6 @@ static void rtl8139_tx_timeout (struct n
 		netif_wake_queue (dev);
 	}
 	spin_unlock(&tp->rx_lock);
-	
 }
 
 

_

#6 lore raw
In reply to Mirko Caserta
Quoted 6 lines · Mirko Caserta

Yes, I know, it's a damn cheap eth card and I should get it replaced :)

Besides that, this card works just fine with 2.4.25 while it refuses to work on a recent 2.6 kernel. I tried 2.6.5 and even 2.6.5-rc2-mm2-broken-out with no luck.

I have tried with 2.6.5 and now with 2.6.6-rc2-mm1 and works perfectly. This message is going through that card ;)

Mine is:
00:0b.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+ (rev 10)

#7 lore raw
Quoted 3 lines · Jorge Bernal (Koke)

I have tried with 2.6.5 and now with 2.6.6-rc2-mm1 and works perfectly. This message is going through that card ;)

Do you mean that both 2.6.5 and 2.6.6-rc2-mm1 worked or that only 2.6.6-rc2-mm1 worked and 2.6.5 didn't?

Quoted 5 lines

Mine is:
00:0b.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+ (rev 10)

I should have said what my card is shouldn't I...

Quoted 1 line

From dmesg:

eth0: RealTek RTL8139 at 0xe0914000, 00:90:f5:25:91:22, IRQ 11
eth0: Identified 8139 chip type 'RTL-8100B/8139D'

Quoted 1 line

From lspci:

00:0a.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+ (rev 10)

Signature · 2 lines
David Johnson
http://www.david-web.co.uk/
#8 lore raw
Quoted 8 lines · David Johnson
Quoted 3 lines · Jorge Bernal (Koke)

I have tried with 2.6.5 and now with 2.6.6-rc2-mm1 and works perfectly. This message is going through that card ;)

Do you mean that both 2.6.5 and 2.6.6-rc2-mm1 worked or that only 2.6.6-rc2-mm1 worked and 2.6.5 didn't?

Both worked and I can't remember exactly but think that 2.6.3 (or .4) also worked.

Signature · 5 lines
"Sólo el éxito diferencia al genio del loco"

Blog: http://www.amedias.org/koke
Web Personal: http://sindominio.net/~koke/
JID: [email protected]
#9 lore raw
In reply to Mirko Caserta
Quoted 5 lines · Mirko Caserta

Yes, I know, it's a damn cheap eth card and I should get it replaced :)

Besides that, this card works just fine with 2.4.25 while it refuses to work on a recent 2.6 kernel. I tried 2.6.5 and even 2.6.5-rc2-mm2-broken-out with no luck.

Mine works fine in 2.6.5:

eth0: RealTek RTL8139 at 0xca844000, xx:xx:xx:xx:xx:xx, IRQ 5
eth0:  Identified 8139 chip type 'RTL-8139C'
eth1: RealTek RTL8139 at 0xca895000, xx:xx:xx:xx:xx:xx, IRQ 3
eth1:  Identified 8139 chip type 'RTL-8139C'
eth0: link up, 10Mbps, half-duplex, lpa 0x0000
eth1: link up, 100Mbps, full-duplex, lpa 0x45E1

lspci:

00:03.0 Ethernet controller: D-Link System Inc RTL8139 Ethernet (rev 10) 00:04.0 Ethernet controller: D-Link System Inc RTL8139 Ethernet (rev 10)

Derek

Signature · 9 lines
+---------------------------------------------------------------+
| Derek Chen-Becker                                             |
| [email protected]                                         |
| http://chen-becker.org                                        |
|                                                               |
| PGP key available on request or from public key servers       |
| ID: 21A7FB53                                                  |
| Fngrprnt: 209A 77CA A4F9 E716 E20C  6348 B657 77EC 21A7 FB53  |
+---------------------------------------------------------------+
#10 lore raw
In reply to Mirko Caserta
Quoted 1 line · Mirko Caserta

eth0: Transmit timeout, status 0c 0005 c07f media 10.

This problem looks like miss configuration of level/edge-triggerd, or IRQ-routing problem.

The attached script may clarify the problem. Also "lspci -vvvxxx" and output of dmesg would be useful.

Signature · 1 line
OGAWA Hirofumi <[email protected]>

#!/usr/bin/perl
#

# dump_pirq 1.21 2001/09/05 02:59:31
#
# A utility to parse the BIOS PCI IRQ Routing Table
#
# Copyright (C) 2000 David A. Hinds -- [email protected]
#

#-----------------------------------------------------------------------

sub dev {
    my($devfn) = @_;
    sprintf "%02x.%d", ($devfn>>3), ($devfn&7);
}

sub print_mask

{
    my($mask) = @_;
    printf "0x%04x [", $mask;
    for (my $i = 0; $i < 16; $i++) {
	next if (!($mask & (1<<$i)));
	$mask &= ~(1<<$i);
	print (($mask) ? "$i," : "$i");
    }
    print "]\n";
}
sub row {
    my($tag, $link, $mask) = @_;
    if ($link != 0) {
	printf "  INT$tag: link 0x%02x, irq mask ", $link;
	print_mask($mask);
    }
}

sub class_of

{
    my($slot) = @_;
    open(IN, "/sbin/lspci -s $slot |");
    $_ = <IN>;
    close(IN);
    return (/^....... ([^:]+):/) ? $1 : "";
}

sub parse_pirq

{
    my($buf) = @_;

    my($p) = index($buf, "\$PIR");
    my($minor,$major,$size,$rbus,$rdev,$mask,$cvd,$mini) =
	unpack "CCSCCSLL", substr($buf, $p+4, 16);

    printf "Interrupt routing table found at address 0xf%04x:\n", $p;
    printf "  Version $major.$minor, size 0x%04x\n", $size;
    printf "  Interrupt router is device %02x:%s\n", $rbus, dev($rdev);
    print "  PCI exclusive interrupt mask: ";
    print_mask($mask);
    if ($cvd) {
	printf("  Compatible router: vendor 0x%04x device 0x%04x\n",
	       ($cvd & 0xffff), ($cvd >> 16));
    }

    $ofs = 32;
    while ($ofs < $size) {
	# Parse a table entry describing a single PCI device
	($bus, $devfn, $la, $ma, $lb, $mb, $lc, $mc, $ld, $md, $slot) =
	    unpack "CCCSCSCSCSC", substr($buf, $p+$ofs, 15);
	$s = sprintf "%02x:%s", $bus, dev($devfn);
	printf "\nDevice $s (slot $slot): %s\n", class_of($s);
	row("A", $la, $ma); row("B", $lb, $mb);
	row("C", $lc, $mc); row("D", $ld, $md);
	push(@{$dev{$la}}, $s . "1");
	push(@{$dev{$lb}}, $s . "2");
	push(@{$dev{$lc}}, $s . "3");
	push(@{$dev{$ld}}, $s . "4");
	$ofs += 16;
    }
    return ($rbus, $rdev, $cvd);
}

#-----------------------------------------------------------------------

# The link values in the interrupt routing table are implementation

# dependent.  Here, we'll try to interpret the link values for some
# known PCI bridge types.

%pIIx = (0x122e8086, "82371FB PIIX",

	 0x70008086, "82371SB PIIX3",
	 0x71108086, "82371AB PIIX4/PIIX4E",
	 0x71988086, "82443MX",
	 0x24108086, "82801AA ICH",
	 0x24208086, "82801AB ICH0",
	 0x24408086, "82801BA ICH2",
	 0x244c8086, "82801BAM ICH2-M");

%via = (0x05861106, "82C586",

	0x05961106, "82C596",
	0x06861106, "82C686");

%opti = (0xc7001045, "82C700");

%pico = (0x00021066, "PT86C523");

%ali = (0x153310b9, "Aladdin M1533");

%sis = (0x04961039, "85C496/497",

	0x00081039, "85C503");

%cyrix = (0x01001078, "5530");

%all_routers = (%pIIx, %via, %opti, %pico, %ali, %sis, %cyrix);

sub outb

{
    my($data,$port) = @_;
    open(IO, ">/dev/port") || die;
    sysseek(IO, $port, 0) || die;
    my $x = pack "C", $data;
    syswrite(IO, $x, 1);
    close(IO);
}

sub inb

{
    my($port) = @_;
    my($data);
    open(IO, "/dev/port") || die;
    sysseek(IO, $port, 0) || die;
    sysread(IO, $data, 1);
    close(IO);
    return unpack "C", $data;
}

sub dump_router

{
    my($rbus, $rdev, $cvd) = @_;
    my($buf, @p, $i, $irq);

    printf "\nInterrupt router at %02x:%s: ", $rbus, dev($rdev);
    $rf = sprintf "/proc/bus/pci/%02x/%s", $rbus, dev($rdev);
    open(IN, $rf);
    if (sysread(IN, $buf, 0x100) != 0x100) {
	print "\nCould not read router info from $rf.\n";
	exit;
    }
    close(IN);
    my $vd = unpack "L", substr($buf, 0, 4);

    if ((defined $pIIx{$vd}) || (defined $pIIx{$cvd})) {

	$name = (defined $pIIx{$vd}) ? $pIIx{$vd} : $pIIx{$cvd};
	printf "Intel $name PCI-to-ISA bridge\n";
	@p = unpack "CCCCC", substr($buf, 0x60, 5);
	for ($i = 0; $i < 4; $i++) {
	    printf "  PIRQ%d (link 0x%02x): ", $i+1, 0x60+$i;
	    print (($p[$i] < 16) ? "irq $p[$i]\n" : "unrouted\n");
	}
	print "  Serial IRQ:";
	print (($p[4] & 0x80) ? " [enabled]" : " [disabled]");
	print (($p[4] & 0x40) ? " [continuous]" : " [quiet]");
	print " [frame=", (($p[4] >> 2) & 15) + 17, "]";
	print " [pulse=", (($p[4] & 3) * 4 + 4), "]\n";

    } elsif ((defined $via{$vd}) || (defined $via{$cvd})) {

	$name = (defined $via{$vd}) ? $via{$vd} : $via{$cvd};
	printf "VIA $name PCI-to-ISA bridge\n";
	$p = unpack "L", substr($buf, 0x55, 4);
	%tag = (1, "A", 2, "B", 3, "C", 5, "D");
	foreach $link (1,2,3,5) {
	    $irq = ($p >> ($link * 4)) & 15;
	    print "  PIRQ$tag{$link} (link 0x0$link): ";
	    print ($irq ? "irq $irq\n" : "unrouted\n");
	}
	$p = unpack "C", substr($buf, 0x54, 1);
	%tag = (0 => "A", 1 => "B", 2 => "C", 3 => "D");
	foreach $n (0, 1, 2, 3) {
	    printf "  PIRQ$tag{$n} : %s\n", ($p & (1 << $n)) ? "edge" : "level";
	}

    } elsif ((defined $opti{$vd}) || (defined $opti{$cvd})) {

	$name = (defined $opti{$vd}) ? $opti{$vd} : $opti{$cvd};
	printf "OPTi $name PCI-to-ISA bridge\n";
	$p = unpack "S", substr($buf, 0xb8, 2);
	for ($i = 0; $i < 4; $i++) {
	    $irq = ($p >> ($i * 4)) & 15;
	    printf "  PCIRQ$i (link 0x%02x): ", ($i<<4)+0x02;
	    print ($irq ? "irq $irq\n" : "unrouted\n");
	}

    } elsif ((defined $pico{$vd} || defined $pico{$cvd})) {

	$name = (defined $pico{$vd}) ? $pico{$vd} : $pico{$cvd};
	printf "PicoPower $name PCI-to-ISA bridge\n";
	outb(0x10, 0x24); $p = inb(0x26);
	outb(0x11, 0x24); $p += inb(0x26)<<8;
	@tag = ("A", "B", "C", "D");
	for ($i = 0; $i < 4; $i++) {
	    $irq = ($p >> ($i * 4)) & 15;
	    print "  INT$tag[$i] (link 0x0", $i+1, "): ";
	    print ($irq ? "irq $irq\n" : "unrouted\n");
	}

    } elsif ((defined $ali{$vd} || defined $ali{$cvd})) {

	$name = (defined $ali{$vd}) ? $ali{$vd} : $ali{$cvd};
	printf "AcerLabs $name PCI-to-ISA bridge\n";
	$p = unpack "L", substr($buf, 0x48, 4);
	$t = unpack "L", substr($buf, 0x4C, 4);
	# This mapping is insane!
	@map = (0, 9, 3, 10, 4, 5, 7, 6, 1, 11, 0, 12, 0, 14, 0, 15);
	for ($i = 0; $i < 8; $i++) {
	    $irq = ($p >> ($i*4)) & 15;
	    print "  INT", $i+1, " (link ", $i+1, "): ";
	    print ($map[$irq] ? "irq $map[$irq]" : "unrouted");
	    printf " (%s)\n", ($t & (1 << $i)) ? "edge" : "level";
	}
	$s = unpack "C", substr($buf, 0x70, 1);
	print "  Serial IRQ:";
	print (($s & 0x80) ? " [enabled]" : " [disabled]");
	print (($s & 0x40) ? " [continuous]" : " [quiet]");
	print " [frame=", (($s >> 2) & 15) + 17, "]";
	print " [pulse=", (($s & 3) * 4 + 4), "]\n";

	%table = (0x44 => IDE1, 0x74 => USB, 0x75 => IDE2, 0x76 => PMU,
		  0x77 => SMB);
	foreach $n (keys %table) {
	    $p = unpack "C", substr($buf, $n, 1);
	    printf "  %s level to edge: %s\n", $table{$n},
		($p & 4) ? "enable" : "disable";
	}

    } elsif ((defined $sis{$vd}) || (defined $sis{$cvd})) {

	$name = (defined $sis{$vd}) ? $sis{$vd} : $sis{$cvd};
	printf "SiS $name PCI-to-ISA bridge\n";
	$base = ($name eq "85C496/497") ? 0xc0 : 0x41;
	@p = unpack "CCCC", substr($buf, $base, 4);
	@tag = ("A", "B", "C", "D");
	for ($i = 0; $i < 4; $i++) {
	    $irq = ($p[$i] & 0x80) ? 0 : ($p[$i] & 0x0f);
	    printf "  INT$tag[$i] (link 0x%02x): ", $i+$base;
	    print ($irq ? "irq $irq\n" : "unrouted\n");
	}

    } elsif ((defined $cyrix{$vd}) || (defined $cyrix{$cvd})) {

	$name = (defined $cyrix{$vd}) ? $cyrix{$vd} : $cyrix{$cvd};
	printf "CYRIX $name PCI-to-ISA bridge\n";
	$p = unpack "S", substr($buf, 0x5c, 2);
	%tag = ("A", "B", "C", "D");
	for ($i = 0; $i < 4; $i++) {
	    $irq = ($p >> ($i * 4)) & 15;
	    printf "  PIRQ$tag{$i} (link 0x%02x): ", $i+1;
	    print ($irq ? "irq $irq\n" : "unrouted\n");
	}

    } else {

	printf("unknown vendor 0x%04x device 0x%04x\n",
	       ($vd & 0xffff), ($vd >> 16));
	foreach $k (sort keys %dev) {
	    next if ($k == 0);
	    printf "  PIRQ? (link 0x%02x): ", $k;
	    $irq = 0;
	    foreach $d (@{$dev{$k}}) {
		$d =~ /(..):(..)\..(.)/;
		($bus,$dev,$pin) = ($1,$2,$3);
		for ($fn = 0; $fn < 8; $fn++) {
		    open(IN, "/proc/bus/pci/$bus/$dev.$fn") || last;
		    sysread(IN, $buf, 0x100);
		    close(IN);
		    ($i,$p) = unpack "CC", substr($buf, 0x3c, 2);
		    $irq = $i if (($p == $pin) && $i && ($i != 255));
		}
	    }
	    print ($irq ? "irq $irq\n" : "unrouted?\n");
	}
    }
    print "  PIC Level mask: "; print_mask((inb(0x4d1)<<8) + inb(0x4d0));
}

#-----------------------------------------------------------------------

# Grab the BIOS from 0xf0000-0xfffff
open(IN, "/dev/mem") || die "cannot open /dev/mem\n"; sysseek(IN, 0xf0000, 0) || die;
die if (sysread(IN, $buf, 0x10000) != 0x10000); close(IN);

if (index($buf, "\$PIR") >= 0) {

    # Dump the PIRQ table, and the router information
    ($rbus, $rdev, $cvd) = parse_pirq($buf);
    dump_router($rbus, $rdev, $cvd);

} else {

    # Scan for any interrupt router device we recognize
    print "No PCI interrupt routing table was found.\n";
    open(DEV, "/proc/bus/pci/devices");
    while (<DEV>) {
	($rbus,$rdev,$vd) = /^(..)(..)\s+(........)/;
	$rbus = hex($rbus); $rdev = hex($rdev); $vd = hex($vd);
	$vd = (($vd & 0xffff0000) >> 16) | (($vd & 0xffff) << 16);
	if (defined $all_routers{$vd}) {
	    dump_router($rbus, $rdev, $vd);
	    $nr++;
	}
    }
    print "\nNo known PCI interrupt routers were found.\n" if (!$nr);

}

#11 lore raw
Quoted 9 lines · OGAWA Hirofumi
Quoted 1 line · Mirko Caserta

eth0: Transmit timeout, status 0c 0005 c07f media 10.

This problem looks like miss configuration of level/edge-triggerd, or IRQ-routing problem.

The attached script may clarify the problem. Also "lspci -vvvxxx" and output of dmesg would be useful.

I have attached a tarball with output from dmesg, lspci -vvvxxx, cat /proc/interrupts and your dump_pirq.pl utility.

Please note that before booting the 2.6.6-rc2-mm2 kernel I got into the BIOS and disabled "Assign IRQ for USB" which is why lspci and dmesg show irq 0 assigned to the USB chips.

Besides that change in the BIOS, everything is just the same between the reboots, apart from the card not properly working in 2.6.6-rc2-mm2 :/

Anyway, it doesn't look like an irq problem to me. It looks more like a wrong detection of the TX triggering level in the driver.

Thanks again, Mirko.

#12 lore raw
Quoted 2 lines · Mirko Caserta

Anyway, it doesn't look like an irq problem to me. It looks more like a wrong detection of the TX triggering level in the driver.

In interrupts-2.6.6-rc2-mm2-broken-out,

CPU0 CPU1

  0:     103394         48    IO-APIC-edge  timer
  1:        157          0    IO-APIC-edge  i8042
  5:          2          1    IO-APIC-edge  eth0
                              ^^^^^^^^^^^^-- wrong
  8:          2          0    IO-APIC-edge  rtc
  9:          0          0   IO-APIC-level  acpi
 11:          3          1    IO-APIC-edge  i91u
 12:         87          0    IO-APIC-edge  i8042
 14:       1068          2    IO-APIC-edge  ide0
 15:        953          1    IO-APIC-edge  ide1

The above must be IO-APIC-level.
And the following is interesting one.

ACPI: ACPI tables contain no PCI IRQ routing entries
PCI: Invalid ACPI-PCI IRQ routing table
PCI: Probing PCI hardware
PCI: Using IRQ router default [1106/3091] at 0000:00:00.0 PCI BIOS passed nonexistent PCI bus 0!
PCI BIOS passed nonexistent PCI bus 0!
PCI BIOS passed nonexistent PCI bus 0!
PCI BIOS passed nonexistent PCI bus 0!
PCI BIOS passed nonexistent PCI bus 0!
PCI BIOS passed nonexistent PCI bus 1!
PCI BIOS passed nonexistent PCI bus 0!

Um.. can you try "pci=noacpi" or "acpi=off"?

Signature · 1 line
OGAWA Hirofumi <[email protected]>
#13 lore raw

Re: 8139too not working in 2.6 (works now)

I tried "acpi=off" and it worked like a charm. He's the new /proc/interrupts:

carbon:~# cat /proc/interrupts
            CPU0       CPU1
   0:     857627         12    IO-APIC-edge  timer
   1:       1073          0    IO-APIC-edge  i8042
   2:          0          0          XT-PIC  cascade
   8:          1          1    IO-APIC-edge  rtc
  12:         87          0    IO-APIC-edge  i8042
  14:       4103          2    IO-APIC-edge  ide0
  15:       4900          2    IO-APIC-edge  ide1
  21:       4505          1   IO-APIC-level  eth0
  27:       1484          0   IO-APIC-level  i91u, uhci_hcd, uhci_hcd,  
uhci_hcd
NMI:          0          0
LOC:     857482     857455
ERR:          0
MIS:          0

Hope this helps someone else too.

Thanks a lot, Mirko.

Quoted 35 lines · OGAWA Hirofumi
Quoted 2 lines · Mirko Caserta

Anyway, it doesn't look like an irq problem to me. It looks more like a wrong detection of the TX triggering level in the driver.

In interrupts-2.6.6-rc2-mm2-broken-out,

CPU0 CPU1

  0:     103394         48    IO-APIC-edge  timer
  1:        157          0    IO-APIC-edge  i8042
  5:          2          1    IO-APIC-edge  eth0
                              ^^^^^^^^^^^^-- wrong
  8:          2          0    IO-APIC-edge  rtc
  9:          0          0   IO-APIC-level  acpi
 11:          3          1    IO-APIC-edge  i91u
 12:         87          0    IO-APIC-edge  i8042
 14:       1068          2    IO-APIC-edge  ide0
 15:        953          1    IO-APIC-edge  ide1

The above must be IO-APIC-level.
And the following is interesting one.

ACPI: ACPI tables contain no PCI IRQ routing entries
PCI: Invalid ACPI-PCI IRQ routing table
PCI: Probing PCI hardware
PCI: Using IRQ router default [1106/3091] at 0000:00:00.0 PCI BIOS passed nonexistent PCI bus 0!
PCI BIOS passed nonexistent PCI bus 0!
PCI BIOS passed nonexistent PCI bus 0!
PCI BIOS passed nonexistent PCI bus 0!
PCI BIOS passed nonexistent PCI bus 0!
PCI BIOS passed nonexistent PCI bus 1!
PCI BIOS passed nonexistent PCI bus 0!

Um.. can you try "pci=noacpi" or "acpi=off"?

#14 lore raw

Re: 8139too not working in 2.6 (works now)

Quoted 2 lines · Mirko Caserta

I tried "acpi=off" and it worked like a charm. He's the new /proc/interrupts:

Sorry, I don't know whether this is ACPI subsystem's bug or not. Probably ACPI peoples can look it...

Thanks.

Signature · 1 line
OGAWA Hirofumi <[email protected]>