SOLVED: A tale of bricked LSI 9266-8i controllers, and the secrets of the 2208 SBR!

Kim Bjoern

New Member
Feb 26, 2020
16
1
3
Denmark
Hi all, thanks in advance for reading.

So, zach4485 and I have both managed to brick 9266-8i controllers in attempts to cross flash to HBA. We both lost SBR and SPD backups, and have since been on quests to recover original state (iMR/Megaraid).

We've reached out in mails and pm's, and received copies of 256 and 512 byte copies from IBM M5016 (LSI 9265), and an original LSI 9266-8i card. Many thanks to perdrix and Erwan for their prompt support.

But! no luck!! Both our cards a most likely a Cisco OEM variant (ucs-raid-9266cv), which may be the reason.

We have tried EVERY guide, tool, platform, hack and mod we could find. We modded the sbrtool and patched lsirec from the excellent work of Marcan, to adopt to the different SBR size and byte layout required by the 2208 rev. C1 chip.

Please don't hesitate to suggest any method for recovery. If we haven't already tried it - we will!

We've guessed and tried numerous SBR versions with different vendor and card ID's from the PCI_ID DB. Many, including the files we received, can successfully be written to the controller, survives reboots, and iMR firmware can be flashed with both megarec and Megacli (adpfwflash and/or m0flash). But the firmware/card doesn't survive boot.

What would be our last resort (other than the bin)?

Last, another kind member traderjay, has offered to help with dumps from a 9266-8i in iMR state.

Which tools would you point at to possibly get a full dump of the card?

Any last ideas much appreciated.

Thanks - Kim

PS. lsirec can hostboot the chip with 9205-8e.bin (with the proper SBR), but can't survive a flash with lsiutil.

NB.: lab notes.
Card installed in a Dell PowerEdge R520 (lspci: 09:00.0).
I/OAT DMA Engine and SR-IOV are disabled in BIOS
FreeDOS | UEFI | Linux | bootable from vFlash | Virtual Media | USB | ESXI.
Server managed from Mac client through virtual console, IPMI and SSH.
Megaraid & HBA/SAS tools: megarec, megaflsh, megacli, sas2flsh, lsirec, lsiutil
Firmware: Several versions tried from initial to most recent from LSI, Cisco, SMC, DELL, IBM.

Code:
AVAGO MegaRAID SAS-MFI BIOS
Version 5.50.03.0 (Build July 22, 2015)
Copyright(c) 2015 AVAGO Technologies
Host Adapter - Bus 9 Dev 0:
F/W is in Fault State - MFI Register State 0xF0010002

Adapter at Baseport  is not responding
No MegaRAID Adapter Installed
 
Last edited:

Kim Bjoern

New Member
Feb 26, 2020
16
1
3
Denmark
SOLVED!

So skyjam - another great contributor in the forum, replied to a PM with two sets of SBR/SPD's; from a 9266-8i and a Fujitsu D3116. Funny enough, skyjam had mixed up his files, but I was quickly celebrating when my card successfully ate his 9266-8i SBR file with lsirec, and flashed the most recent LSI firmware package (23.34.0-0019) with DOS:megarec.exe.

The card survived boot and POST; and was recognized as FUJITSU!! - and WORKING!

PCI info: 1000:005b,1734:11d3 (I haven't found it in PCI ID)

So, I assume my card was some Fujitsu/Hitachi/Cisco (or I managed to mess it up completely in previous recovery attempts).

Nevermind. I was now in iMR, and had to see if I could brick it again: So fired up DOS and updated PCI info with MegaSCU.exe:

MegaSCU -AdpFactorySettings -SetPCIData -f pci.ini -a0

With a custom pci.ini txt file:

[GEN2_INI_FILE]
[PCIDATA]
VENDORID=1000
DEVICEID=005B
SUBVENDORID=1000
SUBDEVICEID=9266

Also updated SAS addr. and serial number with megascu.exe, and activated trial period for FastPath.

Many many thanks to all who helped.

/Kim Bjoern

PS.: Should I try to cross-flash it again..?
 

Attachments

VenChron

New Member
Dec 2, 2014
10
3
3
Hey Kim, glad to hear you got this working!

I'm in a similar boat with a 9266-8i pulled from a Cisco UCS C240 M3 that I tried to cross flash to HBA.

My PCI IDs are 1000:0081 which a search brings up as a SAS2208 PCI-Express Fusion-MPT SAS-2.

I'm not as far down the rabbit-hole as you were but any chance you can give more info on the rescue process (and perhaps provide those SBR/SPDs)?
 

Kim Bjoern

New Member
Feb 26, 2020
16
1
3
Denmark
Hey Kim, glad to hear you got this working!

I'm in a similar boat with a 9266-8i pulled from a Cisco UCS C240 M3 that I tried to cross flash to HBA.

My PCI IDs are 1000:0081 which a search brings up as a SAS2208 PCI-Express Fusion-MPT SAS-2.

I'm not as far down the rabbit-hole as you were but any chance you can give more info on the rescue process (and perhaps provide those SBR/SPDs)?
Hi VenChron, I still have occasional nightmares about the time spend in that rabbit-hole then. I'll gladly share a .zip of the SBR's & SPD's collected then, but I've repressed the details of process (-:

Happy recoveries - Kim Bjoern
 

Attachments