From nobody@FreeBSD.org  Tue Jan 18 20:07:39 2011
Return-Path: <nobody@FreeBSD.org>
Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34])
	by hub.freebsd.org (Postfix) with ESMTP id 74D4B106566C
	for <freebsd-gnats-submit@FreeBSD.org>; Tue, 18 Jan 2011 20:07:39 +0000 (UTC)
	(envelope-from nobody@FreeBSD.org)
Received: from red.freebsd.org (unknown [IPv6:2001:4f8:fff6::22])
	by mx1.freebsd.org (Postfix) with ESMTP id 64BC58FC1D
	for <freebsd-gnats-submit@FreeBSD.org>; Tue, 18 Jan 2011 20:07:39 +0000 (UTC)
Received: from red.freebsd.org (localhost [127.0.0.1])
	by red.freebsd.org (8.14.4/8.14.4) with ESMTP id p0IK7dsW018400
	for <freebsd-gnats-submit@FreeBSD.org>; Tue, 18 Jan 2011 20:07:39 GMT
	(envelope-from nobody@red.freebsd.org)
Received: (from nobody@localhost)
	by red.freebsd.org (8.14.4/8.14.4/Submit) id p0IK7dsC018399;
	Tue, 18 Jan 2011 20:07:39 GMT
	(envelope-from nobody)
Message-Id: <201101182007.p0IK7dsC018399@red.freebsd.org>
Date: Tue, 18 Jan 2011 20:07:39 GMT
From: Wojciech Puchar <wojtek@tensor.gdynia.pl>
To: freebsd-gnats-submit@FreeBSD.org
Subject: bge driver problems BCM5906 laptop
X-Send-Pr-Version: www-3.1
X-GNATS-Notify:

>Number:         154123
>Category:       kern
>Synopsis:       [bge] bge driver problems BCM5906 laptop
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    yongari
>State:          closed
>Quarter:        
>Keywords:       
>Date-Required:  
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Tue Jan 18 20:10:11 UTC 2011
>Closed-Date:    Mon Feb 14 19:45:25 UTC 2011
>Last-Modified:  Mon Feb 14 19:45:25 UTC 2011
>Originator:     Wojciech Puchar
>Release:        FreeBSD 8.2
>Organization:
TENSOR ap
>Environment:
FreeBSD wojtek.tensor.gdynia.pl 8.2-PRERELEASE FreeBSD 8.2-PRERELEASE #2: Wed Jan 12 23:04:28 CET 2011     root@wojtek.tensor.gdynia.pl:/usr/src/sys/amd64/compile/wojtek  amd64

>Description:
bge face works slow. ping latencies are like 50-100ms when i put traffic on it, 6MB/s was most i can achieve sending to other machine at all, but without any tricks it's at most 2MB/s

The trick is to put CPU speed down by setting dev.cpu.0.freq by sysctl
THE SLOWER CPU the faster network! at 300MHz i was able to do 6MB/s, at full 2300MHz - 2MB/s was max!


>How-To-Repeat:
Buy lenovo G550 :)
and try sending data to other computer
>Fix:
No idea. I don't have any idea how bge driver works and how broadcom chips work, but it seems that something tries to put more data at some moment that it should so it chokes. At lower CPU speed it happens more rarely.

but it's only suggestion, i don't really have idea.

>Release-Note:
>Audit-Trail:
Responsible-Changed-From-To: freebsd-bugs->freebsd-net 
Responsible-Changed-By: linimon 
Responsible-Changed-When: Tue Jan 18 20:36:26 UTC 2011 
Responsible-Changed-Why:  
Over to maintainer(s). 

http://www.freebsd.org/cgi/query-pr.cgi?pr=154123 
State-Changed-From-To: open->feedback 
State-Changed-By: yongari 
State-Changed-When: Tue Jan 18 21:01:07 UTC 2011 
State-Changed-Why:  
Would you show me both dmeg and "pciconf -lcbv" output? 
ifconfig bge0 output also would be helpful. 


Responsible-Changed-From-To: freebsd-net->yongari 
Responsible-Changed-By: yongari 
Responsible-Changed-When: Tue Jan 18 21:01:07 UTC 2011 
Responsible-Changed-Why:  
Grab. 

http://www.freebsd.org/cgi/query-pr.cgi?pr=154123 

From: Pyun YongHyeon <pyunyh@gmail.com>
To: wojtek@tensor.gdynia.pl
Cc: bug-followup@FreeBSD.org
Subject: Re: kern/154123: [bge] bge driver problems BCM5906 laptop
Date: Wed, 19 Jan 2011 12:10:43 -0800

 On Wed, Jan 19, 2011 at 04:56:12PM +0100, wojtek@tensor.gdynia.pl wrote:
 > my mail is wojtek@tensor.gdynia.pl
 > 
 > what's exactly wrong?
 > 
 
 It seems it was corrected now. I don't know what was happened but
 GNAT database showed your address as "wojtek@tensor.gdynia.p" so
 mail I sent was rejected.
 
 > dmesg and pciconv below:
 > 
 
 [...]
 
 > vboxdrv: fAsync=0 offMin=0xe6 offMax=0x3af
 > ipfw2 initialized, divert enabled, nat enabled, rule-based forwarding 
 > enabled, default to accept, logging disabled
 > load_dn_sched dn_sched FIFO loaded
 > load_dn_sched dn_sched PRIO loaded
 > load_dn_sched dn_sched QFQ loaded
 > load_dn_sched dn_sched RR loaded
 > load_dn_sched dn_sched WF2Q+ loaded
 
 It seems you have ipfw/dummynet/vboxdrv in kernel.
 I don't think it could be related with bge(4) poor performance but
 dummynet can add extra latency for the driver.
 Could you disable all these features in kernel and see whether that
 makes any difference?
 
 Honestly driver does not even know CPU clock speed and there is no
 conditional code which may affect performance from clock speed
 changes.
 
 One more thing, if you have activated any kind of CPU clock control
 features, please try disable it

From: User Wojtek <wojtek@tensor.gdynia.pl>
To: Pyun YongHyeon <pyunyh@gmail.com>
Cc: bug-followup@FreeBSD.org
Subject: Re: kern/154123: [bge] bge driver problems BCM5906 laptop
Date: Wed, 19 Jan 2011 21:26:28 +0100 (CET)

 ups. maybe i missed l after p.
 possible still not yet learned this keyboard :)
 
 On Wed, 19 Jan 2011, Pyun YongHyeon wrote:
 
 > On Wed, Jan 19, 2011 at 04:56:12PM +0100, wojtek@tensor.gdynia.pl wrote:
 >> my mail is wojtek@tensor.gdynia.pl
 >>
 >> what's exactly wrong?
 >>
 >
 > It seems it was corrected now. I don't know what was happened but
 > GNAT database showed your address as "wojtek@tensor.gdynia.p" so
 > mail I sent was rejected.
 >
 >> dmesg and pciconv below:
 > [...]
 >
 >> vboxdrv: fAsync=0 offMin=0xe6 offMax=0x3af
 >> ipfw2 initialized, divert enabled, nat enabled, rule-based forwarding
 >> enabled, default to accept, logging disabled
 >> load_dn_sched dn_sched FIFO loaded
 >> load_dn_sched dn_sched PRIO loaded
 >> load_dn_sched dn_sched QFQ loaded
 >> load_dn_sched dn_sched RR loaded
 >> load_dn_sched dn_sched WF2Q+ loaded
 >
 > It seems you have ipfw/dummynet/vboxdrv in kernel.
 
 but it is not turned on. i will delete it from kernel at all and report.
 
 > I don't think it could be related with bge(4) poor performance but
 > dummynet can add extra latency for the driver.
 > Could you disable all these features in kernel and see whether that
 > makes any difference?
 >
 > Honestly driver does not even know CPU clock speed and there is no
 > conditional code which may affect performance from clock speed
 > changes.
 
 i just report what i see - the slower is set the CPU - the faster upload 
 goes.
 
 downloads seem unaffected.
 
 > One more thing, if you have activated any kind of CPU clock control
 > features, please try disable it
 yes i disabled powerd to be able to set up clock speed manually.
 i too tried
 hw.acpi.cpu.cx_lowest=C1
 instead of C2
 
 no effects

From: User Wojtek <wojtek@tensor.gdynia.pl>
To: Pyun YongHyeon <pyunyh@gmail.com>
Cc: bug-followup@FreeBSD.org
Subject: Re: kern/154123: [bge] bge driver problems BCM5906 laptop
Date: Wed, 19 Jan 2011 21:41:26 +0100 (CET)

 removed IPFIREWALL from kernel, changed hw.acpi.cpu.cx_lowest=C1
 stopped powerd
 results (ftp upload to other machine), nothing else going on
 
 2300MHz - average 2MB/s (fluctuates randomly from 1.5 to 2.5MB/s)
 2012MHz - 1.4MB/s
 1400MHz - 1.4MB/s
 1050MHz - same
 600MHz - 1.8MB/s
 450MHz - 3.6-4MB/s (randomly fluctuates)
 300MHz - 3MB/s
 150MHz - 7.7MB/s actually CPU-bound as i have geli encrypted disk, gets 
 full 11MB/s when uploading from ramdisk!
 
 All things like IPFIREWALL doesn't seem to make effect, as well as 
 hw.acpi.cpu.cx_lowest
 
 any ideas? or should i call a shaman? ;)

From: Pyun YongHyeon <pyunyh@gmail.com>
To: User Wojtek <wojtek@tensor.gdynia.pl>
Cc: bug-followup@freebsd.org
Subject: Re: kern/154123: [bge] bge driver problems BCM5906 laptop
Date: Wed, 19 Jan 2011 13:01:53 -0800

 On Wed, Jan 19, 2011 at 09:41:26PM +0100, User Wojtek wrote:
 > removed IPFIREWALL from kernel, changed hw.acpi.cpu.cx_lowest=C1
 > stopped powerd
 > results (ftp upload to other machine), nothing else going on
 > 
 > 2300MHz - average 2MB/s (fluctuates randomly from 1.5 to 2.5MB/s)
 > 2012MHz - 1.4MB/s
 > 1400MHz - 1.4MB/s
 > 1050MHz - same
 > 600MHz - 1.8MB/s
 > 450MHz - 3.6-4MB/s (randomly fluctuates)
 > 300MHz - 3MB/s
 > 150MHz - 7.7MB/s actually CPU-bound as i have geli encrypted disk, gets 
 > full 11MB/s when uploading from ramdisk!
 > 
 
 No, you measured performance incorrectly. There are too many
 factors that may affect performance. In order to get pure network
 driver performance, you should use netperf or iperf benchmark
 program in ports/benchmarks.
 
 I'm under the impression that you're not measuring network driver
 performance but disk/geli performance.
 
 > All things like IPFIREWALL doesn't seem to make effect, as well as 
 > hw.acpi.cpu.cx_lowest
 > 
 > any ideas? or should i call a shaman? ;)
 
 Try check performance again using netperf or iperf.

From: User Wojtek <wojtek@tensor.gdynia.pl>
To: Pyun YongHyeon <pyunyh@gmail.com>
Cc: bug-followup@freebsd.org
Subject: Re: kern/154123: [bge] bge driver problems BCM5906 laptop
Date: Thu, 20 Jan 2011 20:28:53 +0100 (CET)

 >> 1050MHz - same
 >> 600MHz - 1.8MB/s
 >> 450MHz - 3.6-4MB/s (randomly fluctuates)
 >> 300MHz - 3MB/s
 >> 150MHz - 7.7MB/s actually CPU-bound as i have geli encrypted disk, gets
 >> full 11MB/s when uploading from ramdisk!
 >>
 >
 > No, you measured performance incorrectly. There are too many
 > factors that may affect performance. In order to get pure network
 > driver performance, you should use netperf or iperf benchmark
 > program in ports/benchmarks.
 >
 > I'm under the impression that you're not measuring network driver
 > performance but disk/geli performance.
 
 if so - why it works slower when CPU is faster?
 

From: Pyun YongHyeon <pyunyh@gmail.com>
To: User Wojtek <wojtek@tensor.gdynia.pl>
Cc: bug-followup@freebsd.org
Subject: Re: kern/154123: [bge] bge driver problems BCM5906 laptop
Date: Thu, 20 Jan 2011 13:06:25 -0800

 On Thu, Jan 20, 2011 at 08:28:53PM +0100, User Wojtek wrote:
 > >>1050MHz - same
 > >>600MHz - 1.8MB/s
 > >>450MHz - 3.6-4MB/s (randomly fluctuates)
 > >>300MHz - 3MB/s
 > >>150MHz - 7.7MB/s actually CPU-bound as i have geli encrypted disk, gets
 > >>full 11MB/s when uploading from ramdisk!
 > >>
 > >
 > >No, you measured performance incorrectly. There are too many
 > >factors that may affect performance. In order to get pure network
 > >driver performance, you should use netperf or iperf benchmark
 > >program in ports/benchmarks.
 > >
 > >I'm under the impression that you're not measuring network driver
 > >performance but disk/geli performance.
 > 
 > if so - why it works slower when CPU is faster?
 > 
 
 Honestly I have no idea about that. But performance measuring
 method was wrong.

From: User Wojtek <wojtek@tensor.gdynia.pl>
To: Pyun YongHyeon <pyunyh@gmail.com>
Cc: bug-followup@freebsd.org
Subject: Re: kern/154123: [bge] bge driver problems BCM5906 laptop
Date: Fri, 21 Jan 2011 13:00:41 +0100 (CET)

 so i did the RIGHT (as you stated) - and got nearly same results, few% 
 higher on high CPU speed, and 11MB/s at 150MHz
 
 
 On Thu, 20 Jan 2011, Pyun YongHyeon wrote:
 
 > On Thu, Jan 20, 2011 at 08:28:53PM +0100, User Wojtek wrote:
 >>>> 1050MHz - same
 >>>> 600MHz - 1.8MB/s
 >>>> 450MHz - 3.6-4MB/s (randomly fluctuates)
 >>>> 300MHz - 3MB/s
 >>>> 150MHz - 7.7MB/s actually CPU-bound as i have geli encrypted disk, gets
 >>>> full 11MB/s when uploading from ramdisk!
 >>>>
 >>>
 >>> No, you measured performance incorrectly. There are too many
 >>> factors that may affect performance. In order to get pure network
 >>> driver performance, you should use netperf or iperf benchmark
 >>> program in ports/benchmarks.
 >>>
 >>> I'm under the impression that you're not measuring network driver
 >>> performance but disk/geli performance.
 >>
 >> if so - why it works slower when CPU is faster?
 >>
 >
 > Honestly I have no idea about that. But performance measuring
 > method was wrong.
 >
 >

From: User Wojtek <wojtek@tensor.gdynia.pl>
To: Pyun YongHyeon <pyunyh@gmail.com>
Cc: bug-followup@freebsd.org
Subject: Re: kern/154123: [bge] bge driver problems BCM5906 laptop
Date: Fri, 21 Jan 2011 13:07:45 +0100 (CET)

 i will today do some more tests. i probably found a problem, and it is 
 something wrong in autonegotiation with that switch.
 
 I used today my laptop in other network with other switches and other 
 server and all worked great.
 
 
 On Fri, 21 Jan 2011, User Wojtek wrote:
 
 > so i did the RIGHT (as you stated) - and got nearly same results, few% higher 
 > on high CPU speed, and 11MB/s at 150MHz
 >
 >
 > On Thu, 20 Jan 2011, Pyun YongHyeon wrote:
 >
 >> On Thu, Jan 20, 2011 at 08:28:53PM +0100, User Wojtek wrote:
 >>>>> 1050MHz - same
 >>>>> 600MHz - 1.8MB/s
 >>>>> 450MHz - 3.6-4MB/s (randomly fluctuates)
 >>>>> 300MHz - 3MB/s
 >>>>> 150MHz - 7.7MB/s actually CPU-bound as i have geli encrypted disk, gets
 >>>>> full 11MB/s when uploading from ramdisk!
 >>>>> 
 >>>> 
 >>>> No, you measured performance incorrectly. There are too many
 >>>> factors that may affect performance. In order to get pure network
 >>>> driver performance, you should use netperf or iperf benchmark
 >>>> program in ports/benchmarks.
 >>>> 
 >>>> I'm under the impression that you're not measuring network driver
 >>>> performance but disk/geli performance.
 >>> 
 >>> if so - why it works slower when CPU is faster?
 >>> 
 >> 
 >> Honestly I have no idea about that. But performance measuring
 >> method was wrong.
 >> 
 >> 
 >

From: Pyun YongHyeon <pyunyh@gmail.com>
To: User Wojtek <wojtek@tensor.gdynia.pl>
Cc: bug-followup@freebsd.org
Subject: Re: kern/154123: [bge] bge driver problems BCM5906 laptop
Date: Fri, 21 Jan 2011 15:26:57 -0800

 On Fri, Jan 21, 2011 at 01:07:45PM +0100, User Wojtek wrote:
 > i will today do some more tests. i probably found a problem, and it is 
 > something wrong in autonegotiation with that switch.
 > 
 
 That's weird. That does not explain why you get better performance
 with lower CPU clock. Show me the output of "ifconfig bge0" after
 connecting your controller to the guilty switch. By chance, are you
 using forced manual media configuration? Posting the output of
 "sysctl dev.bge.0.stats" would be useful.
  
 > I used today my laptop in other network with other switches and other 
 > server and all worked great.
 > 
 > 
 > On Fri, 21 Jan 2011, User Wojtek wrote:
 > 
 > >so i did the RIGHT (as you stated) - and got nearly same results, few% 
 > >higher on high CPU speed, and 11MB/s at 150MHz
 > >
 > >
 > >On Thu, 20 Jan 2011, Pyun YongHyeon wrote:
 > >
 > >>On Thu, Jan 20, 2011 at 08:28:53PM +0100, User Wojtek wrote:
 > >>>>>1050MHz - same
 > >>>>>600MHz - 1.8MB/s
 > >>>>>450MHz - 3.6-4MB/s (randomly fluctuates)
 > >>>>>300MHz - 3MB/s
 > >>>>>150MHz - 7.7MB/s actually CPU-bound as i have geli encrypted disk, gets
 > >>>>>full 11MB/s when uploading from ramdisk!
 > >>>>>
 > >>>>
 > >>>>No, you measured performance incorrectly. There are too many
 > >>>>factors that may affect performance. In order to get pure network
 > >>>>driver performance, you should use netperf or iperf benchmark
 > >>>>program in ports/benchmarks.
 > >>>>
 > >>>>I'm under the impression that you're not measuring network driver
 > >>>>performance but disk/geli performance.
 > >>>
 > >>>if so - why it works slower when CPU is faster?
 > >>>
 > >>
 > >>Honestly I have no idea about that. But performance measuring
 > >>method was wrong.
 > >>
 > >>
 > >

From: User Wojtek <wojtek@tensor.gdynia.pl>
To: Pyun YongHyeon <pyunyh@gmail.com>
Cc: bug-followup@freebsd.org
Subject: Re: kern/154123: [bge] bge driver problems BCM5906 laptop
Date: Sun, 23 Jan 2011 00:44:00 +0100 (CET)

 tomorrow (actually today it's 0:40 here) i will go there and do this.
 
 i attempted manual mode and it never connected at all. only media auto 
 works at all, "no carrier" otherwise.
 
 
 On Fri, 21 Jan 2011, Pyun YongHyeon wrote:
 
 > On Fri, Jan 21, 2011 at 01:07:45PM +0100, User Wojtek wrote:
 >> i will today do some more tests. i probably found a problem, and it is
 >> something wrong in autonegotiation with that switch.
 >>
 >
 > That's weird. That does not explain why you get better performance
 > with lower CPU clock. Show me the output of "ifconfig bge0" after
 > connecting your controller to the guilty switch. By chance, are you
 > using forced manual media configuration? Posting the output of
 > "sysctl dev.bge.0.stats" would be useful.
 >
 >> I used today my laptop in other network with other switches and other
 >> server and all worked great.
 >>
 >>
 >> On Fri, 21 Jan 2011, User Wojtek wrote:
 >>
 >>> so i did the RIGHT (as you stated) - and got nearly same results, few%
 >>> higher on high CPU speed, and 11MB/s at 150MHz
 >>>
 >>>
 >>> On Thu, 20 Jan 2011, Pyun YongHyeon wrote:
 >>>
 >>>> On Thu, Jan 20, 2011 at 08:28:53PM +0100, User Wojtek wrote:
 >>>>>>> 1050MHz - same
 >>>>>>> 600MHz - 1.8MB/s
 >>>>>>> 450MHz - 3.6-4MB/s (randomly fluctuates)
 >>>>>>> 300MHz - 3MB/s
 >>>>>>> 150MHz - 7.7MB/s actually CPU-bound as i have geli encrypted disk, gets
 >>>>>>> full 11MB/s when uploading from ramdisk!
 >>>>>>>
 >>>>>>
 >>>>>> No, you measured performance incorrectly. There are too many
 >>>>>> factors that may affect performance. In order to get pure network
 >>>>>> driver performance, you should use netperf or iperf benchmark
 >>>>>> program in ports/benchmarks.
 >>>>>>
 >>>>>> I'm under the impression that you're not measuring network driver
 >>>>>> performance but disk/geli performance.
 >>>>>
 >>>>> if so - why it works slower when CPU is faster?
 >>>>>
 >>>>
 >>>> Honestly I have no idea about that. But performance measuring
 >>>> method was wrong.
 >>>>
 >>>>
 >>>
 >
 >

From: Wojciech Puchar <wojtek@tensor.gdynia.pl>
To: Pyun YongHyeon <pyunyh@gmail.com>
Cc: bug-followup@freebsd.org
Subject: Re: kern/154123: [bge] bge driver problems BCM5906 laptop
Date: Thu, 27 Jan 2011 16:38:27 +0100 (CET)

 >
 > That's weird. That does not explain why you get better performance
 > with lower CPU clock. Show me the output of "ifconfig bge0" after
 > connecting your controller to the guilty switch. By chance, are you
 > using forced manual media configuration? Posting the output of
 > "sysctl dev.bge.0.stats" would be useful.
 
 hell that's the difference. i'm not NOW here but i remember getting 
 different that what i get in home.
 in home
 
 bge0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500
 
 options=8009b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,LINKSTATE>
          ether 88:ae:1d:d1:b5:7f
          inet 192.168.1.3 netmask 0xffffff00 broadcast 192.168.1.255
          media: Ethernet autoselect (100baseTX <full-duplex>)
          status: active
 
 
 there i had MORE things after <full-duplex> but i don't remember what!
 
 i will tell it today if i will be there.

From: Pyun YongHyeon <pyunyh@gmail.com>
To: Wojciech Puchar <wojtek@tensor.gdynia.pl>
Cc: bug-followup@freebsd.org
Subject: Re: kern/154123: [bge] bge driver problems BCM5906 laptop
Date: Thu, 27 Jan 2011 09:39:02 -0800

 On Thu, Jan 27, 2011 at 04:38:27PM +0100, Wojciech Puchar wrote:
 > >
 > >That's weird. That does not explain why you get better performance
 > >with lower CPU clock. Show me the output of "ifconfig bge0" after
 > >connecting your controller to the guilty switch. By chance, are you
 > >using forced manual media configuration? Posting the output of
 > >"sysctl dev.bge.0.stats" would be useful.
 > 
 > hell that's the difference. i'm not NOW here but i remember getting 
 > different that what i get in home.
 > in home
 > 
 > bge0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500
 > 
 > options=8009b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,LINKSTATE>
 >         ether 88:ae:1d:d1:b5:7f
 >         inet 192.168.1.3 netmask 0xffffff00 broadcast 192.168.1.255
 >         media: Ethernet autoselect (100baseTX <full-duplex>)
 >         status: active
 > 
 > 
 > there i had MORE things after <full-duplex> but i don't remember what!
 > 
 > i will tell it today if i will be there.
 
 Ok, it's common to see unexpected result when resolved speed/duplex
 is not matched with link partner. When one system use manual
 configuration, speed is correctly resolved but duplex mismatch
 results in poor performance even though it can send/receive frames.
 If one end of link have manually set speed/duplex the other end
 which uses auto-negotiation will suffer from duplex mismatch. So
 please make sure you're using auto-negotiation, from the output
 above I believe you are using it.
 For switches, also check the port is configured to use
 auto-negotiation. If the switch configuration is already
 auto-negotiation, log in to the switch and verify whether the
 resolved speed/duplex of switch is the same thing what you see on
 your box.

From: Wojciech Puchar <wojtek@tensor.gdynia.pl>
To: Pyun YongHyeon <pyunyh@gmail.com>
Cc: bug-followup@freebsd.org
Subject: Re: kern/154123: [bge] bge driver problems BCM5906 laptop
Date: Mon, 14 Feb 2011 14:50:41 +0100 (CET)

 please close this problem.
 
 After checking in many places i found it is not my laptop's and FreeBSD 
 fault.
 
 It's a fault of autonegotiation between switch and server with which i 
 transmitted data. It badly negotiated duplex. The slower data was 
 transmitted - the lower chance of collision - the faster achieved speed.
 
 Changing from this switch to other fixed the problem completely!
State-Changed-From-To: feedback->closed 
State-Changed-By: yongari 
State-Changed-When: Mon Feb 14 19:44:56 UTC 2011 
State-Changed-Why:  
Close by submitter's request. It was caused by bad switch. 

http://www.freebsd.org/cgi/query-pr.cgi?pr=154123 
>Unformatted:
