Qlogic (isp) cannot login into fabric after link loss

Matthew Jacob mj at feral.com
Tue Feb 16 15:38:54 UTC 2010


On 2/16/2010 2:02 AM, Oleg Sharoyko wrote:
> Hi!
>
> I'm sorry for cross-posting, but I'm in trouble with isp driver and I
> really need your help.
>
>    

Hmm. You haven't given details of what you're running here (OS rev, 
whether you're loading ispfw from the OS or running it from the BIOS). 
That would help. But this seems to happen with the latest anyway sometimes.
I don't suppose you have a FC trace of this happening?

It's probably step 4 below that there is a logic problem on. I need to 
work on this.


> Here we have IBM blade with Qlogic (seems to be 2462s card) and EMC
> CX3-40 both connected to Brocade DS4900. Everything seems to be ok,
> until FC link between switch and server fails. When it later restores
> isp cannot anymore log into fabric. I have a complete debug log of isp
> output. Here is small excerpt from it which I think is relevant to my
> problem:
>
> isp0: Chan 0 got 3 ports back from name server
> isp0: Chan 0 skip ourselves on @ PortID 0x011500
> isp0: Chan 0 Checking Fabric Port 0x010e00
> isp0: IN mbox 0 = 0x0064
> isp0: IN mbox 1 = 0x0003
> isp0: IN mbox 2 = 0x05e8
> isp0: IN mbox 3 = 0xd000
> isp0: IN mbox 6 = 0x0000
> isp0: IN mbox 7 = 0x0000
> isp0: IN mbox 9 = 0x0000
> isp0: IN mbox 10 = 0x0000
> isp0: RISC2HOST ISR 0x40000010
> isp0: RISC2HOST ISR 0x40068011
> isp0: OUT mbox 0 = 0x4006
> isp0: IOCB LOGX:
> isp0: 0x00000000: 52 01 00 00 ff ff ff ff 00 00 03 00 00 00 00 00
> isp0: 0x00000010: 00 0e 01 00 00 00 00 00 00 00 00 00 00 00 00 00
> isp0: 0x00000020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> isp0: 0x00000030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> isp0: IN mbox 0 = 0x0054
> isp0: IN mbox 1 = 0x0040
> isp0: IN mbox 2 = 0x05e8
> isp0: IN mbox 3 = 0xd000
> isp0: IN mbox 6 = 0x0000
> isp0: IN mbox 7 = 0x0000
> isp0: RISC2HOST ISR 0x40060011
> isp0: RISC2HOST ISR 0x40060011
> isp0: RISC2HOST ISR 0x40060011
> isp0: RISC2HOST ISR 0x40008010
> isp0: OUT mbox 0 = 0x4000
> isp0: IOCB LOGX response:
> isp0: 0x00000000: 52 01 00 00 ff ff ff ff 31 00 03 00 00 00 00 00
> isp0: 0x00000010: 00 0e 01 02 1a 00 00 00 01 00 00 00 00 00 00 00
> isp0: 0x00000020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> isp0: 0x00000030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
> isp0: Chan 0 PLOGX PortID 0x010e00 to N-Port handle 0x3: already logged
> in with N-Port handle 0x1
> isp0: IN mbox 0 = 0x0064
> isp0: IN mbox 1 = 0x0001
> isp0: IN mbox 2 = 0x05e8
> isp0: IN mbox 3 = 0xd000
> isp0: IN mbox 6 = 0x0000
> isp0: IN mbox 7 = 0x0000
> isp0: IN mbox 9 = 0x0000
> isp0: IN mbox 10 = 0x0000
> isp0: RISC2HOST ISR 0x40000010
> isp0: RISC2HOST ISR 0x40008010
> isp0: OUT mbox 0 = 0x4000
> isp0: Chan 0 Port 0x010e00 flags 0x0 curstate 7
> isp0: Chan 0 new device 0x010e00 at 0x1 disappeared
>
> As far as I can tell from this log and isp.c things are happening in the
> following order:
>
> 1. Get the list of devices on fabric
> 2. Try to log into each device (excluding self)
> 3. The result is failure with code "already logged in" and old login
> handle
> 4. Try to login with that handle and expect that to work, but it fails
> with code PDB2400_STATE_PORT_UNAVAIL.
>
> Bear in mind, that I'm not anywhere close to FC expert, so I may be
> totally wrong here.
>
> So who is wrong and would it be possible to do resolve this issue? I
> have an access to all the components (blade, fc switch and storage
> system) and can provide additional information if it's needed.
>
> I'm really stuck and would greatly appreciate any help! Thanks in
> advance.
>
>    



More information about the freebsd-scsi mailing list