[openib-general] [mstflint] firmware upgrade error

[Grant Grundler]

On Thu, Jan 12, 2006 at 12:02:06AM -0500, Benoit Morin wrote:
> Hi,
> 
> I tried upgrading the firmware on a MHX-CE128-T card (firmware 3.2.0) 
> using the firmware fw-23108-3_3_5-MHX-CE128-T.bin found on Mellanox's 
> website. I used mstflint found in revision 4464 of the openib svn.

I just went through this excercise with a PCI-e card.

> I used the -nofs flag following what I read in the "[mstflint] firmware 
> upgrade instructions" thread.
> 
> Now, after a reboot, I get the following :
> 
> ib_mthca: Mellanox InfiniBand HCA driver v0.06 (June 23, 2005)
> ib_mthca: Initializing 0000:02:00.0
> ACPI: PCI Interrupt 0000:02:00.0[A] -> GSI 28 (level, low) -> IRQ 19
> ib_mthca 0000:02:00.0: SYS_EN DDR error: syn=0, sock=0, sladdr=0, SPD 
> source=DIMM
> ib_mthca 0000:02:00.0: SYS_EN returned status 0x07, aborting.
> ACPI: PCI interrupt for device 0000:02:00.0 disabled
> ib_mthca: probe of 0000:02:00.0 failed with error -22
> 
> I no longer have access to anything under /sys/class/infiniband.

That's because the driver didn't sucessfully initialize.

> Is the card toast? Where did I go wrong?

It sounds like you grabbed the wrong firmware.
Maybe one with more or less memory on it.
I'm not sure though.

I've used tvflash regularly to update firmware on PCI-X cards.
Other reports to this mailing list indicate mstflint works fine too.

I'll point out that neither tvflash nor mstflint worked for
updating PCI-e cards on ia64 as I posted a few weeks ago.
Tvflash will fully trash the eeprom. I successfully resurrected
the PCI-e card using "mst" and "flint" (mft-0.5.0 from Mellanox
firmware support web site) on a debian Intel P4 box.

However, your card is likely not toast. Try another firmware.
Worst case you get to learn about the "J5" jumper ("EEPROM NOT PRESENT")
and use "mst" and "flint" tools like I did.

hth,
grant