Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 1 Sep 2000 10:15:37 -0500 (CDT)
From:      Joseph Thomas <jpt@networkcs.com>
To:        mike@ARL.ARMY.MIL (Mike Muuss)
Cc:        freebsd-atm@FreeBSD.ORG, Mike@ARL.ARMY.MIL, CJohnson@ARL.ARMY.MIL
Subject:   Re: FreeBSD 4.1 failure w/Fore PCA-200E
Message-ID:  <200009011515.KAA10949@us.networkcs.com>
In-Reply-To:  <200009010040.aa3011451@CAD.ARL.ARMY.MIL> from "Mike Muuss" at Sep 01, 2000 12:40:41 AM

next in thread | previous in thread | raw e-mail | index | archive | help
> 
> 
> In this past week I've installed two FreeBSD 4.1 systems, and have been
> unable to get either one to work.
> 
> The first one was a brand new dual-CPU Dell server, with a new PCA-200E.
> I didn't fret too much because I assumed that we'd gotten something
> wrong in the configuration or BIOS settings.
> 
> The second one was my desktop machine.  It had been running FreeBSD 4.0
> with ATM very successfully.  I used the "upgrade" option of the CD-ROM,
> then built a new kernel with the exact same ATM kernel config lines
> as I had used for 4.0, left /etc/rc.conf alone, and rebooted.  This
> system now fails to initialize the ATM in exactly the same way as
> the dual-CPU Dell.  This suggests that either (a) some important detail
> of establishing the configuration has changed, or (b) something broke.
> 
> As best as I can tell, it appears that cells are not being received from
> the PCA-200E, but that cells are successfully being transmitted.  I can
> see cells arriving at the PCA-200E on the blinking lights, and on the
> Switch's statistics, but not inside FreeBSD.
> 
> I'm running the Fore PCA-200E interface via fiber into a Fore ASX-200BX
> switch that has the new Pentium SCP processor in it.
> 
> Has anyone else succeeded in getting a similar configuration to work?
> Any hints on what I might be doing wrong?
> 
> 	Thanks!
> 	 -Mike
> 
> 
> Here is what I see:
> 
> Intf      Vendor    Model     Media           Bus   Serial No
> hfa0      Fore      PCA-200E  OC-3c           PCI   4203797
>     MAC address = 00:20:48:40:25:15
>     Hardware version = 2.0.1
>     Firmware version = 3.0.1
> 
> Interface  Sigmgr   State
> hfa0       UNI 3.1  ADR_WAIT			<*** THIS IS TROUBLE
>     ATM address = -
>     Network interfaces: atm0
> 
> Net Intf  Phy Intf  IP Address
> atm0      hfa0      vm.arl.army.mil (128.63.247.191)
> 
>              Input    Input  Input  Output   Output Output    Cmd
> Interface     PDUs    Bytes   Errs    PDUs    Bytes   Errs   Errs
> hfa0             0        0      0      88     4236      0      0
> 
> Fore PCA-200E OC-3c Statistics
> Section     Path    Line      Line     Path     Corr   Uncorr
> BIP8        BIP8    BIP24     FEBE     FEBE     HCS    HCS
> Errs        Errs    Errs      Errs     Errs     Errs   Errs
>       0        0        0        0        0        0        0
> 
> Fore PCA-200E Device Statistics
> Type 1      Type 1      Type 2      Type 2
> Small Buff  Large Buff  Small Buff  Large Buff  Receive     Receive
> Alloc Fail  Alloc Fail  Alloc Fail  Alloc Fail  Queue Full  Carrier
>        604           0           0           0           0  On
> 
> Fore PCA-200E ATM Layer Statistics
>   Cells In   Cells Out   VPI Range  VPI NoConn   VCI Range  VCI NoConn
>          0           0           0           0           0       45100
> 
> Fore PCA-200E AAL 5 Statistics
>                         CRC/Len                              CRC   Proto  PDU
>   Cells In   Cells Out  Errs   Drops    PDUs In   PDUs Out   Errs  Errs   Drops
>          0         176    604      0          0         88      0      0      0
> 
> Fore PCA-200E Device Driver Statistics
>   No  Xmit   Max   Seg          No    No          No    IQ    No   Cmd    No
>  VCC Queue   Seg   Not   Seg   DMA   VCC    No  Mbuf  Full   DMA Queue   DMA
>  Out  Full  Size Align   Pad   Out    In  Buff    In    In   Sup  Full   Cmd
>    0     0     0     0     0     0     0     0     0     0     0     0     0
> 
> Interface  VPI   VCI  AAL   Type Dir    State    Encaps   Owner
> hfa0         0    16  AAL5  PVC  InOut  ACT_DOWN Null     ILMI
> 
>                         Input    Input  Input  Output   Output Output
> Interface  VPI   VCI     PDUs    Bytes   Errs    PDUs    Bytes   Errs
> hfa0         0    16        0        0      0      88     4236      0
> 
> Net Intf  State     ATM Address
> atm0      PEND_ADR  0x47.0005.80.007300.0000.0300.0100.0020481c095f.00
> 
> % netstat -nI atm0
> Name  Mtu   Network       Address            Ipkts Ierrs    Opkts Oerrs  Coll
> atm0  9180  <Link#9>    00:20:48:40:25:15        0     0       89     0     0
> atm0  9180  128.63.240/20 128.63.247.191         0     0       89     0     0
> 
> 
> % dmesg
> 
> hfa0: <FORE Systems PCA-200E ATM> mem 0xd0a00000-0xd0bfffff irq 7 at device 14.0
>  on pci0
> hfa0: driver is using old-style compatability shims
> 
> 
> To Unsubscribe: send mail to majordomo@FreeBSD.org
> with "unsubscribe freebsd-atm" in the body of the message
> 

	There is a known problem with the HARP FORE driver and FBSD 4.1
which was first reported to us by Jim.Pirzyk@disney.com. It is not a
problem in getting the cells in per se, but rather a problem with HARP
and buffers. HARP uses some data space at the head of each buffer for
passing params around (function pointers, token values etc.). With the
additional words in the pkthdr portion of the mbuf header, the calculation
for this space yeilds a negative amount causing the FORE driver to overwrite
portions of the mbuf structure which leads to buffer allocation problems.
[Look at your output for 'atm sh stat int hfa dev' which shows 604 type 1
small buffer alloc failures, and at 'atm sh stat int hfa0 aal5' which shows
604 crc/len errors on output as well.]

	I want to talk with the developer who wrote that part of the
code since I'm a little confused as to why to calculation is being
done the way it is so I don't have an 'official' fix yet.

	In the meantime, one can edit /usr/src/sys/dev/hfa/fore.h
and change the define for BUF1_SM_DOFF from

#define	BUF1_SM_DOFF	((BUF1_SM_HOFF + SIZEOF_Buf_handle) - BUF1_SM_HDR)

to

#define	BUF1_SM_DOFF	16


Please note that I'm seeing some significant throughput problems on my test
system which are not present in FBSD 4.0. I have not had time to investigate
or resolve these as yet so others may also have this problem as well.


-- 
Joseph Thomas                           E/Mail:  jpt@networkcs.com
Network Computing Services, Inc.    	 	 jpt@magic.net
1200 Washington Ave So.			Tel:	 +1 612 337 3558
Minneapolis, MN     55415-1227          FAX:     +1 612 337 3400

	An elephant is a mouse with an operating system.


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-atm" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200009011515.KAA10949>