Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
Sun QFE and Netra T1
View unanswered posts
View posts from last 24 hours

 
Reply to topic    Gentoo Forums Forum Index Gentoo on Sparc
View previous topic :: View next topic  
Author Message
linucs
n00b
n00b


Joined: 04 Sep 2003
Posts: 4

PostPosted: Thu Sep 04, 2003 8:27 am    Post subject: Sun QFE and Netra T1 Reply with quote

HI all,
I'm currently experiencing a problem trying to use a Sun Quad FastEthernet (PCI) on a Netra T1 105.
I'm using a 2.4.21 kernel, that correctly recognizes the onboard hme and qfe cards; ifconfig works ok, but nothing gets actually sent or reveived over the qfe interfaces (while the two onboard hme work fine).
I tried to recompile the drivers as a module (sunhme.o), no luck.
I saw a message about an experimental liveCD solving some problems on Netra X1 network drivers; is it a patched kernel?

Lino
Back to top
View user's profile Send private message
Weeve
Retired Dev
Retired Dev


Joined: 30 Oct 2002
Posts: 641

PostPosted: Thu Sep 04, 2003 7:28 pm    Post subject: Reply with quote

Hi Linucs,

At the time I put that together, yes that was a patched kernel. However you can get all the patches I used (minus the hme patch for ultra 1 enterprises) in the 2.4.23_pre series. The driver that was added was for the tg3 series adapters.

Do the link lights on the card come up when cables are plugged in? Does linux seem to auto-negotiate a speed for the interfaces when they are in the UP state?
Back to top
View user's profile Send private message
linucs
n00b
n00b


Joined: 04 Sep 2003
Posts: 4

PostPosted: Fri Sep 05, 2003 6:19 am    Post subject: Reply with quote

Jason,
everything really seems to be "normal" (the lights come up, linux negotiates the speed etc); some time ago, I tried with a Debian based on 2.4.21, resulting in a number of hangs (the system got freezed, just being able to poweroff to lom>) when I just tried to ping some host!
I'm trying with a vanilla 2.4.22 kernel...
Back to top
View user's profile Send private message
linucs
n00b
n00b


Joined: 04 Sep 2003
Posts: 4

PostPosted: Fri Sep 05, 2003 10:15 am    Post subject: Reply with quote

Jason,
I'm still having the same problem with 2.4.22...
I recompiled the sunhme driver with debugging enabled; on my eth0 (QFE/PCI):

---
happy_meal_open: to happy_meal_init
happy_meal_init: happy_flags[00001b86] to happy_meal_poll_stop
not polling, return
happy_meal_init: to happy_meal_stop
happy_meal_stop: reset, done
happy_meal_init: to happy_meal_init_rings
happy_meal_init_rings: counters to zero, clean, init rxring, init txring, done
happy_meal_init: Disable all MIF irqs (old[0000ffff]), use frame old[00000100], to happy_meal_transceiver_check
happy_meal_init: internal, using MII, tx/rx reset, happy_meal_tx_reset: reset, done
happy_meal_rx_reset: reset, done
jsize/ipg1/ipg2, rseed/macaddr, htable, ring ptrs rxr[c3ff0000] txr[c3ff0800]
happy_meal_init: old[00000000] bursts<64>, enable global interrupts, tx rsize=32 oreg[0000000f], tx dma enable old[000003fe], erx regs bug old[00]
happy_meal_init: enable hash rx_cfg_old[00000101], BIGMAC init, XIF config old[00000000], tx old[00000100] and rx [00000a00] ON!
firewall root # eth0: Link is up using internal transceiver at 10Mb/s, Half Duplex.
---

But nothing... no interrupts???
If I do the same on eth4 (Happy Meal, onboard):

---
happy_meal_open: to happy_meal_init
happy_meal_init: happy_flags[00000b86] to happy_meal_poll_stop
not polling, return
happy_meal_init: to happy_meal_stop
happy_meal_stop: reset, done
happy_meal_init: to happy_meal_init_rings
happy_meal_init_rings: counters to zero, clean, init rxring, init txring, done
happy_meal_init: Disable all MIF irqs (old[0000ffff]), use frame old[00000300], to happy_meal_transceiver_check
happy_meal_init: external, disable MII, tx/rx reset, happy_meal_tx_reset: reset, done
happy_meal_rx_reset: reset, done
jsize/ipg1/ipg2, rseed/macaddr, htable, ring ptrs rxr[c3fe8000] txr[c3fe8800]
happy_meal_init: old[00000000] bursts<64>, enable global interrupts, tx rsize=32 oreg[0000000f], tx dma enable old[000003fe], erx regs bug old[00]
happy_meal_init: enable hash rx_cfg_old[00000101], BIGMAC init, XIF config old[00000008], tx old[00000100] and rx [00000a00] ON!
firewall root # eth4: Link is up using external transceiver at 10Mb/s, Half Duplex.
happy_meal_interrupt: status=00010000 RXTOHOST RX<[0 len=60 csum=4511]>done
happy_meal_interrupt: status=00010000 RXTOHOST RX<[1 len=60 csum=4512]>done
happy_meal_interrupt: status=00010000 RXTOHOST RX<[2 len=92 csum=872a]>done
happy_meal_interrupt: status=00010000 RXTOHOST RX<[3 len=60 csum=b945]>done
happy_meal_interrupt: status=00010000 RXTOHOST RX<[4 len=275 csum=85c8]>done
happy_meal_interrupt: status=00010000 RXTOHOST RX<[5 len=60 csum=8c58]>done
happy_meal_interrupt: status=00010000 RXTOHOST RX<[6 len=60 csum=4515]>done
happy_meal_interrupt: status=00010000 RXTOHOST RX<[7 len=60 csum=4498]>done
happy_meal_interrupt: status=00010000 RXTOHOST RX<[8 len=60 csum=44e5]>done
happy_meal_interrupt: status=00010000 RXTOHOST RX<[9 len=92 csum=8723]>done
---

and so on... is there something on how interrupt are masked?

Lino
Back to top
View user's profile Send private message
linucs
n00b
n00b


Joined: 04 Sep 2003
Posts: 4

PostPosted: Fri Sep 05, 2003 12:09 pm    Post subject: Reply with quote

Jason,
I finally got the solution: the problem seems to be related to interrupt routing; I followed the instruction at http://news.gw.com/openbsd.sparc/2401 and set my eth0 as follows from the OpenBoot PROM:

" /pci@1f,0/pci@1/pci@1/pci@f/SUNW,qfe@0,1" begin-select-dev 1 encode-int " interrupts" property end-select-dev

(/pci@1f,0/pci@1/pci@1/pci@f/SUNW,qfe@0,1 is where my first interface QFE is mapped)
..."magically" everything started to work, even with the 2.4.21-r1 series on the Gentoo liveCD.
Should I report this as an "official" bug? Is it already known?
Thanks

Lino
Back to top
View user's profile Send private message
Weeve
Retired Dev
Retired Dev


Joined: 30 Oct 2002
Posts: 641

PostPosted: Fri Sep 05, 2003 7:24 pm    Post subject: Reply with quote

Glad to hear you found a fix.

I'll add this into the list of things for the sparc FAQ.
Back to top
View user's profile Send private message
H0bb3z
Tux's lil' helper
Tux's lil' helper


Joined: 28 May 2003
Posts: 121
Location: in the transmogrifier

PostPosted: Sun Sep 28, 2003 10:03 pm    Post subject: Reply with quote

This is *exactly* what I was looking for. I have an old SparcStation5 with an SBUS QFE card, and up until now I was stumped on how to get the networking up (this is slated to be a firewall as well).

Unfortunately, it didn't completely make the issue go away...

After setting up the nvramrc on the box and booting into Gentoo, I'm able to ping anything on the local networks (yay! finally!), but as soon as I try to do anything with DNS (like an rsync or a ping www.somewhere.com), the QFE-specific interfaces are somehow affected. I can still ping the actual interfaces from the local box, but nothing on the local networks or anywhere else (I get 100% packet loss).

Here's how I have it setup:
eth0: xxx.xxx.xxx.6/29 -> Public subnet and DSL connection
eth1: 192.168.43.2/24 -> Internal LAN subnet
eth2: 10.100.43.1/28 -> DMZ
eth3: 10.132.231.1/24 - Test DMZ

My /etc/resolv.conf is setup correctly to point to my ISPs primary and secondary DNS servers -- I use the same configuration on my existing firewall with no issues.

On a LAN workstation, I can run a persistent ping to the eth1 interface address successfully. The second I try to do something like "ping www.yahoo.com" from the sparcstation, the persistent ping begins to timeout to the previously available interface. If I then try to ping the IP address of the workstation from the sparcstation, I get 100% packet loss and errors. Prior to pinging yahoo, the sparcstation was able to successfully ping all the IPs I tried...

The only way to get the sparcstation to talk on layer 3 again is to restart it. I can use the built-in HME interface without these issues, but I need to be able to leverage the QFE interfaces to make this firewall anything worth using...

Any ideas on this? Is this a related issue, or did I miss something? It seems like a DNS issue almost, but that shouldn't take down a network interface like it appears to be here...
_________________
-H0bb3z-
-----------
Intel QX9770 | 6Gb DDR3 | ATI 4870 HD 1Gb | 2xSeagate 1Tb SATAs
Running Funtoo Core2 build
Back to top
View user's profile Send private message
H0bb3z
Tux's lil' helper
Tux's lil' helper


Joined: 28 May 2003
Posts: 121
Location: in the transmogrifier

PostPosted: Mon Sep 29, 2003 4:46 pm    Post subject: Reply with quote

An interesting additional bit of information: I disabled DNS lookups by changing the "hosts" line in /etc/nsswitch.conf from
Code:
hosts   files dns
to
Code:
hosts   files

Now the interfaces always stay up on the QFE card. Of course, I can't resolve anything on the firewall, but that may be just fine. Anyone heard of QFE cards having this issue with DNS?
_________________
-H0bb3z-
-----------
Intel QX9770 | 6Gb DDR3 | ATI 4870 HD 1Gb | 2xSeagate 1Tb SATAs
Running Funtoo Core2 build
Back to top
View user's profile Send private message
xming
Guru
Guru


Joined: 02 Jul 2002
Posts: 441

PostPosted: Mon Sep 29, 2003 7:55 pm    Post subject: Reply with quote

I am using 2 QFE in 2 E220R, and they work. But when I try to use Ethernet bonding one of the machine would hang, not immediately but after 10's GB traffic, so there might be some problems

xming
Back to top
View user's profile Send private message
H0bb3z
Tux's lil' helper
Tux's lil' helper


Joined: 28 May 2003
Posts: 121
Location: in the transmogrifier

PostPosted: Thu Oct 02, 2003 6:28 am    Post subject: Reply with quote

E220R's are likely PCI though, so it may be related to SBUS versions of the card. I think I saw somewhere in the documentation where SBUS QFE cards were not fully supported. I may be out of luck...
_________________
-H0bb3z-
-----------
Intel QX9770 | 6Gb DDR3 | ATI 4870 HD 1Gb | 2xSeagate 1Tb SATAs
Running Funtoo Core2 build
Back to top
View user's profile Send private message
biggyL
Tux's lil' helper
Tux's lil' helper


Joined: 31 Jan 2005
Posts: 120
Location: Israel

PostPosted: Wed Apr 26, 2006 9:58 am    Post subject: Quad FastEthernet (PCI) Reply with quote

Hello All,

I'm going to install Gentoo on sun4u Sun Fire 280R (UltraSPARC-III+)

The problem I have is that my Quad FastEthernet (PCI) card is recognized by LiveCD not as one QFE card but as 4 separate network cards (eth1-eth4):
Code:

# ifconfig -a
eth0      Link encap:Ethernet  HWaddr 00:03:BA:18:6E:07
          inet addr:195.84.41.6  Bcast:195.84.41.255  Mask:255.255.255.0
          UP BROADCAST NOTRAILERS RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:1640 errors:0 dropped:0 overruns:0 frame:0
          TX packets:31 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:443074 (432.6 Kb)  TX bytes:6489 (6.3 Kb)
          Interrupt:96 Base address:0x7800

eth1      Link encap:Ethernet  HWaddr 00:03:BA:01:5A:65
          inet addr:132.74.40.137  Bcast:132.74.41.255  Mask:255.255.254.0
          UP BROADCAST NOTRAILERS RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:13773 errors:0 dropped:0 overruns:0 frame:0
          TX packets:122 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:3808205 (3.6 Mb)  TX bytes:18031 (17.6 Kb)
          Interrupt:224 Base address:0xe800

eth2      Link encap:Ethernet  HWaddr 00:03:BA:01:5A:66
          BROADCAST MULTICAST  MTU:1500  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:4 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:0 (0.0 b)  TX bytes:2360 (2.3 Kb)
          Base address:0xf800

eth3      Link encap:Ethernet  HWaddr 00:03:BA:01:5A:67
          BROADCAST MULTICAST  MTU:1500  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:4 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:0 (0.0 b)  TX bytes:2360 (2.3 Kb)
          Interrupt:32 Base address:0xc800

eth4      Link encap:Ethernet  HWaddr 00:03:BA:01:5A:68
          BROADCAST MULTICAST  MTU:1500  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:4 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:0 (0.0 b)  TX bytes:2360 (2.3 Kb)
          Interrupt:192 Base address:0xd800


I've never worked with QFE cards, but as I understand I can use it as some load balancing tool?
QFE should get a separate IP and any others could get their IP from the DHCP Server?

Could someone give me some tips on how to solve this issue?
Back to top
View user's profile Send private message
H0bb3z
Tux's lil' helper
Tux's lil' helper


Joined: 28 May 2003
Posts: 121
Location: in the transmogrifier

PostPosted: Wed Apr 26, 2006 1:46 pm    Post subject: Reply with quote

You should actually see it as 4 separate interfaces -- each RJ45 in the quad card is a separate network interface. You can apply any networking configuration to one or all of them independently as you would if you had four single interface cards.

I believe there is also a way to bridge two or more interfaces into a single logical interface to increase throughput/resiliency, but I haven't had to venture into that realm at this point...
_________________
-H0bb3z-
-----------
Intel QX9770 | 6Gb DDR3 | ATI 4870 HD 1Gb | 2xSeagate 1Tb SATAs
Running Funtoo Core2 build
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Gentoo on Sparc All times are GMT
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum