MPI msg not completed

Hello,

I am trying to solve a problem that looks suspiciously like
the one mentioned below (from an earlier posting to this list):

>Now for another problem: I believe that several recent problems have
>been a result of the 559 swallowing what it thinks are management
>packets, but that we wanted to e.g. complete an MPI message.  Is there
>any way to turn off the management "feature"?

>Donald Becker				<EMAIL: PROTECTED>
>Scyld Computing Corporation		http://www.scyld.com
>410 Severn Ave. Suite 210		Second Generation Beowulf Clusters
>Annapolis MD 21403			410-990-9993

I am developing an MPI code that runs for a while and then stops in
an MPI_Recv (the matching MPI_Send has already completed). Every 
once and a while, the code will run to completion. This problem
only occurs using lam c2c tcp communication (i.e. it works with the
lamd mode). I have done my best to rule out a send/recv deadlock by
using non-blokcing commuication.

I have tried the

Intel(R) PRO/100 Network Driver - version 2.3.27

and

Intel(R) PRO/100 Fast Ethernet Adapter - Loadable driver, ver 1.8.38

drivers.

While it seems that our chips sets are the i82557, is there (a) any chance
that "message swallowing" could occur under with the i82557 and (b) any
chance that a different driver would work?

The machines are all:
Linux version 2.4.18-3rlx3 (gcc version 2.96 20000731 (Red Hat Linux 7.3 2.96-110))

The diagnostic report for the three cards in the system is
(note that Index #1 is not currently used):

eepro100-diag.c:v2.12 4/15/2003 Donald Becker ([email protected])
 http://www.scyld.com/diag/index.html
Index #1: Found a Intel i82557/8/9 EtherExpressPro100 adapter at 0x1000.
i82557 chip registers at 0x1000:
  00000010 3783c000 00000000 00080002 18217809 000005f0
  No interrupt sources are pending.
   The transmit unit state is 'Idle'.
   The receive unit state is 'Ready'.
  This status is unusual for an activated interface.
EEPROM contents, size 64x16:
    00: 4200 0052 2369 0d00 0000 0201 4701 0000  _BR_i#_______G__
  0x08: aaaa aaaa 4080 2010 1749 0003 0000 0000  _____@_ I_______
      ...
  0x30: 0801 4200 4011 0000 0000 0000 0000 0000  ___B_@__________
  0x38: 0000 0000 0000 0000 0000 0000 0000 a7bb  ________________
 The EEPROM checksum is correct.
Intel EtherExpress Pro 10/100 EEPROM contents:
  Station address 00:42:52:00:69:23.
  Receiver lock-up bug exists. (The driver work-around *is* implemented.)
  Board assembly aaaaaa-170, Physical connectors present: RJ45
  Primary interface chip i82555 PHY #1.
Index #2: Found a Intel i82557/8/9 EtherExpressPro100 adapter at 0x1040.
i82557 chip registers at 0x1040:
  00000050 02d82ce4 00000000 00080002 182541e1 000005f0
  No interrupt sources are pending.
   The transmit unit state is 'Suspended'.
   The receive unit state is 'Ready'.
  This status is normal for an activated but idle interface.
EEPROM contents, size 64x16:
    00: 4200 0052 2269 0d00 0000 0201 4701 0000  _BR_i"_______G__
  0x08: aaaa aaaa 4080 2010 1749 0003 0000 0000  _____@_ I_______
      ...
  0x30: 0801 4200 4011 0000 0000 0000 0000 0000  ___B_@__________
  0x38: 0000 0000 0000 0000 0000 0000 0000 a8bb  ________________
 The EEPROM checksum is correct.
Intel EtherExpress Pro 10/100 EEPROM contents:
  Station address 00:42:52:00:69:22.
  Receiver lock-up bug exists. (The driver work-around *is* implemented.)
  Board assembly aaaaaa-170, Physical connectors present: RJ45
  Primary interface chip i82555 PHY #1.
Index #3: Found a Intel i82557/8/9 EtherExpressPro100 adapter at 0x1080.
i82557 chip registers at 0x1080:
  00000050 02db612c 00000000 00080002 182545e1 000005f0
  No interrupt sources are pending.
   The transmit unit state is 'Suspended'.
   The receive unit state is 'Ready'.
  This status is normal for an activated but idle interface.
EEPROM contents, size 64x16:
    00: 4200 0052 2169 0d00 0000 0201 4701 0000  _BR_i!_______G__
  0x08: aaaa aaaa 40a0 2010 1749 0003 0000 0000  _____@_ I_______
      ...
  0x30: 0801 4200 4011 0000 0000 0000 0000 0000  ___B_@__________
  0x38: 0000 0000 0000 0000 0000 0000 0000 a99b  ________________
 The EEPROM checksum is correct.
Intel EtherExpress Pro 10/100 EEPROM contents:
  Station address 00:42:52:00:69:21.
  Receiver lock-up bug exists. (The driver work-around *is* implemented.)
  Board assembly aaaaaa-170, Physical connectors present: RJ45
  Primary interface chip i82555 PHY #1.


Thanks,

Jason Gans
Bioscience Division (B-1)
Los Alamos National Lab