Kerry Shu wrote:
Hi Kyle,
Let me make sure I understand your issue first:
Your issue is the BMC connection, i.e. the connection to the server
processor, via one of the two bge ports is broken with SNV, it's not
that bge0/bge1 can't work in SNV after boot, right?
I think that's right. However:
1) While I know networking through bge1 works, I have yet to fully test
network access through bge0 (the one that is shared with the BMC) - I
jumpstart through bge1 to avoud the PXE issue with the bge0/BMC, and I
haven't configured bge0 yet now that it's installed.
2) Also I don't know if the connection to the service processor is
entirely broken, or if it is only the SOL function over that connection
- outside of the SOL console, I'm not sure how to test connectivity to
the SP - any ideas?
If it's former, looks like bge driver loading breaks the chipset
BMC sideband connection here. You can have a try to remove bge driver
and to see if your BMC connection is fine.
# rem_drv bge
# reboot -p
Now that I have the installation done (via the VGA console) I can test
this, though I think it will break my bge1 network connection - I can
live with that for the test though.
I'll post back with the results.
-Kyle
Regards,
Kerry
Kyle McDonald wrote:
Pavan Chandrashekar - Sun Microsystems wrote:
On 11/03/09 17:26, minskey guo wrote:
Even with the same card, it is difficult to reproduce it. Since Kyle
said that s10 works for his hardware, mayb he can try to disable
some features, such as vlan filter, jumbo, even hardware CRC,
(Can those features be disabled in .conf ??) etc, to see if it can
work, in case that it is a bug in BMC or nic firmware, and
driver just triggers the bug.
I remember a long time ago that SOL was not working with IBM blade
centers because the driver was stripping off the CRC, and the fix
was to not strip the CRC for the blade centers. It is one area you
can check, by changing the configurations.
What do I change to test if this is the problem?
-Kyle
Pavan
-minskey
Carson Tan 写道:
Hi Kyle,
Thanks for your information. I really appreciate it.
But I am sorry to say that we don't have this type of NIC on hand,
so I cannot reproduce your problem.
Have you ever tried any other build of SXCE?
Thanks again,
Carson
Kyle McDonald wrote:
Yong Tan wrote:
Hi Kyle,
I am wondering if there is any problem with the bge driver.
Could you send me the output of the attached script?
Hi,
I couldn't run the script exactly, because the problems happen
when booting the machine for JumpStart, so I've never actually
gotten Nevada installed on the machine.
However just now, while attempting to get the machine to boot and
install using the VGA console, I screwed up the JumpStart config,
and I did manage to get a shell prompt. Unfortunately being a VGA
console I coudln't cut and paste your script in, or the results out.
Instead I manually ran the commands in the script, and I've
manually (hopefully with no errors ;) ) typed in the results as
comments in the code of your script below:
#!/bin/sh
echo
hostname
echo
"=============================================================================="
VERSION=`uname -r`
# 5.11
uname -a
# SunOS 5.11 snv_125 i86pc i386 i86pc
prtconf -D | grep "Memory size"
# Memory size: 8192
echo "# psrinfo -v"
psrinfo -v | grep operates
# ksh93: psrinfo: not found
#
# psrinfo must not be included in the miniroot.
# Given the output on another machine I'm guessing this one
would look like:
#
# The i386 processor operates at 3800MHz.
# (repeated for a total of 4 lines)
echo "# ifconfig -a | grep -v lo | grep -v inet6 | grep -v
"127.0.0.1""
ifconfig -a | grep -v lo | grep -v inet6 | grep -v "127.0.0.1"
# bge0: flags=1000842<BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500
index 3
# inet 0.0.0.0 netmask 0
# ether 0:14:5e:3d:5a:f2
# bge1: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,DHCP,IPv4>
mtu 1500 index 1
# inet 172.30.171.123 netmask ffffff00 broadcast 172.30.171.255
# ether 0:14:5e:3d:5a:f3
if [ $VERSION = 5.11 ]; then
echo "# dladm show-ether"
dladm show-ether
# LINK PTYPE STATE AUTO SPEED-DUPLEX PAUSE
# bge0 current up yes 1G-f none
# bge1 current up yes 1G-f none
else
echo "# dladm show-dev"
dladm show-dev
fi
echo "# kstat -m bge | /usr/xpg4/bin/grep -iE 'rev|type|interface'"
kstat -m bge | /usr/xpg4/bin/grep -iE 'rev|type|interface'
# ksh93: kstat: not found
#
# kstat isn't included in the miniroot, but running that command
# on another blade (same HW, but running S10):
#
# kstat -m bge | /usr/xpg4/bin/grep -iE 'rev|type|interface'
# asic_rev 553648128
# bus_type PCI-X
# chip_type 5704
# interface serdes
# revision_id 16
# asic_rev 553648128
# bus_type PCI-X
# chip_type 5704
# interface serdes
# revision_id 16
echo "# prtconf -pv | /usr/xpg4/bin/grep -iE
'14e4,|108e,|17d5,|10de,|10ec,'"
prtconf -pv | /usr/xpg4/bin/grep -iE '14e4'
# compatible: 'pci14e4,16a8.1014.2e8.10' + 'pci14e4,16a8.1014.
# 2e8' + 'pci1014,2e8' + 'pci14e4,16a8.10' + 'pci14e4,16a8' +
'pciclass,020000' +
# 'pciclass,0200'
# compatible: 'pci14e4,16a8.1014.2e8.10' + 'pci14e4,16a8.1014.
# 2e8' + 'pci1014,2e8' + 'pci14e4,16a8.10' + 'pci14e4,16a8' +
'pciclass,020000' +
# 'pciclass,0200'
for i in `echo "::devbindings -q bge" | mdb -k | awk '{print
$1}'`; do
a=`echo $i'$<devinfo' | mdb -k | grep devi_driver_data | awk
'{print $3}'`
echo $a"::print bge_t ifname asf_enabled" | mdb -k
done
# ifname = [ "bge0" ]
# asf_enabled = 1 (B_TRUE)
# ifname = [ "bge1" ]
# asf_enabled = 0 (B_FALSE)
I hope this helps somehow. I'd really like to track this down. :)
-Kyle
_______________________________________________
driver-discuss mailing list
driver-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/driver-discuss
_______________________________________________
driver-discuss mailing list
driver-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/driver-discuss
_______________________________________________
driver-discuss mailing list
driver-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/driver-discuss
_______________________________________________
driver-discuss mailing list
driver-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/driver-discuss
