NAT MCH with TIC6678 EVM in uTCA: IPMI Messaging

Brandy Jabkiewicz

Hello,

In my previous posts on uTCA related topics, I have stated that I was using an all Vadatech chassis, MCH and Power module. It was then noted by TI that the TI EVMs were not really tested with the Vadatech MCH but that a newer release of the MMC code might allow them to work with Vadatech MCH. Well, I ran into another problem with the Vadatech components and have decided to switch to NAT. According to historical posts, everything should work fine becuase TI did all their testing with NAT MCH. However, I am working with NAT and they discovered that TI EVM is failing to initialize properly with their MCH.

Setup:

Vadatech Chassis VT862

NAT MCH Basic Version 3.4 LC

NAT Power Module PM-DC-780

TI EVM C6678 LE

Here are some historical posts for reference:

http://e2e.ti.com/support/dsp/c6000_multi-core_dsps/f/639/t/115240.aspx?pi24928=4

http://e2e.ti.com/support/dsp/c6000_multi-core_dsps/f/639/p/164695/601272.aspx#601272

With NATs help, I was able to log the messages coming from the TI EVM. It is showing that the checksum byte is a) incorrect b) duplicated.

IpmiLib_Verify(0x80->0x20): ERR - found msg CRC 0x9b exp 0x4d GET_DEVICE_SDR_RSP

IpmiLib_PrintMsgRaw: WARN -1 - CRC2 error

GET_DEVICE_SDR_RSP(0x80->0x20):

20 14 cc 80 7c 21 00 01 00 80 01 0c 28 00 00 00

c1 68 00 c9 54 49 2d 4c 43 5f 45 56 9b 9b

ipmiMsgReaderI: invalid msg of len 30 from ipmiDev 8

IpmiLib_Verify(0x84->0x20): ERR - found msg CRC 0x7d exp 0x4d GET_DEVICE_SDR_RSP

IpmiLib_PrintMsgRaw: WARN -1 - CRC2 error

GET_DEVICE_SDR_RSP(0x84->0x20):

20 14 cc 84 90 21 00 01 00 84 01 0c 28 00 00 00

c1 6a 00 c9 54 49 2d 4c 43 5f 45 56 7d 7d

ipmiMsgReaderI: invalid msg of len 30 from ipmiDev 10

R(12,8,1)IpmiLib_Verify(0x80->0x20): ERR - found msg CRC 0x9b exp 0x4d GET_DEVICE_SDR_RSP

IpmiLib_PrintMsgRaw: WARN -1 - CRC2 error

GET_DEVICE_SDR_RSP(0x80->0x20):

20 14 cc 80 7c 21 00 01 00 80 01 0c 28 00 00 00

c1 68 00 c9 54 49 2d 4c 43 5f 45 56 9b 9b

ipmiMsgReaderI: invalid msg of len 30 from ipmiDev 8

R(14,10,1)IpmiLib_Verify(0x84->0x20): ERR - found msg CRC 0x7d exp 0x4d GET_DEVICE_SDR_RSP

IpmiLib_PrintMsgRaw: WARN -1 - CRC2 error

GET_DEVICE_SDR_RSP(0x84->0x20):

20 14 cc 84 90 21 00 01 00 84 01 0c 28 00 00 00

c1 6a 00 c9 54 49 2d 4c 43 5f 45 56 7d 7d

ipmiMsgReaderI: invalid msg of len 30 from ipmiDev 10

Can someone familar with the MSP430 MMC code comment on this and perhaps determine if there is a bug. I think this could be related to why my Vadatech MCH does not consistently power on the EVM either.

Thanks,
Brandy

over 13 years ago

0 Dmitry Cooper over 13 years ago

Prodigy 110 points

Hi, BrandyJ.

I'm living the same problem with MCH NAT. I solved it.

0 Brandy Jabkiewicz over 13 years ago in reply to Dmitry Cooper

Mastermind 6325 points

Hi Dmitry,

Can you point me to the solution?

Thanks,
Brandy

0 Rakesh Modi over 13 years ago

Expert 1680 points

Hi Brandy,

I have attached here .txt file, it is tested with NAT MCH. Please download it and rename it to ipm430.out. Load the the same file on MSP430 and then let me know if still the same issue exist.

Thanks & Regards,

Rakesh Modi5504.ipm430.txt

0 Brandy Jabkiewicz over 13 years ago in reply to Rakesh Modi

Mastermind 6325 points

Rakesh,

Thank you so much for the file, is this newer than the 1.32 version? I have already updated the msp430 with that code.

0 Rakesh Modi over 13 years ago in reply to Brandy Jabkiewicz

Expert 1680 points

Hi,

There is no update after 1.32 version, I have just recompiled and given .out file to you.

0 Brandy Jabkiewicz over 13 years ago in reply to Rakesh Modi

Mastermind 6325 points

Hello,

Is there any help on this from TI?

Thanks,
Brandy

0 Brandy Jabkiewicz over 13 years ago in reply to Brandy Jabkiewicz

Mastermind 6325 points

Hello,

Upon some advice, I verified once again that I have programmed the mmc with the latest version 1.32. I have done this and still encounter the problem.

If I program with the code from this link:

https://gforge.ti.com/gf/project/msp430_mmc_code/

and read the version with my Vadatech MCH I get 1.20. As described in a previous post (http://e2e.ti.com/support/dsp/c6000_multi-core_dsps/f/639/p/164695/601272.aspx#601272) this is because Vadatech does not convert the response from hex to dec. 0x20 = 32.

But just to be sure, I change the version in the code (to 0x19) and reprogrammed the mmc again. When I read the version this time it said 1.19. Therefore I am convinced that I have successfully programmed the mmcs to the latest version.

Please advise on how to best correct this situation to work with NAT MCH, the Vadatech MCH was unreliable with these EVMs and no longer suits my purpose. Since TI has gotten this to work already, some hints would be nice.

Thanks so much!!!!
Brandy

0 Rakesh Modi over 13 years ago in reply to Brandy Jabkiewicz

Expert 1680 points

Hi Brandy,

Please let me know that the .out file I have provided to you is working properly or not with NAT MCH? Because I have tested here the same and its working fine. NAT MCH power up the EVM and reads FRU data correctly.

Thanks & Regards,

Rakesh Modi

0 Brandy Jabkiewicz over 13 years ago in reply to Rakesh Modi

Mastermind 6325 points

Hello Rakesh,

I cannot just use your binary, I am sorry. I need the code to build myself for legal reasons. That is why I provided you with the link to the code set I am using. If you cannot give me your code exactly, perhaps you can compare it to the link I provided and tell me if it is different? Then we will know that I do not have the newest code or there is some change that I do not have.

Is that an option?

thanks,

Brandy

0 Dmitry Cooper over 13 years ago in reply to Brandy Jabkiewicz

Prodigy 110 points

Hello, BrandyJ

In mmc.c v.1.2 need to add a lines:

LinkDescriptorPack(&fru_data.p2p_rec.link_descr[6][0], 0x3F, 1, 0, 2, AMC_LINK_PCI_EXPRESS, 0, 0, 1, 1, 2);
LinkDescriptorPack(&fru_data.p2p_rec.link_descr[7][0], 0x3F, 1, 0, 0, AMC_LINK_PCI_EXPRESS, 0, 0, 1, 1, 2);

Dmitry.

0 Tom Johnson 16214 over 13 years ago in reply to Brandy Jabkiewicz

TI__Mastermind 46460 points

Brandy,

Rakesh works for our EVM development partner eInfochips. They integrated the MMC code for us. We have been communicating internally to get this resolved. He has found a header file entry that is causing the problem you are seeing. He has provided a new build but did not change the software revision. This will cause confusion. I have asked him to increase the revision number and to then resubmit a build. We should get another release today or tomorrow.

Tom

0 Brandy Jabkiewicz over 13 years ago in reply to Tom Johnson 16214

Mastermind 6325 points

Ahh, Tom, thank you for the info.

@Rakesh, I apologize, I did not understand your relationship with TI and again I apologize for not being able to just test your binary. I'll look for the next release of source code tomorrow and close this issue out. Please post the link (and expected version number) when it is available.

Thank you for everyone's help and interest!

Brandy

0 Tom Johnson 16214 over 13 years ago in reply to Brandy Jabkiewicz

TI__Mastermind 46460 points

Brandy,

A new release with a unique version string is being put on Gforge which should resolve your problem.

Tom

0 Brandy Jabkiewicz over 13 years ago in reply to Tom Johnson 16214

Mastermind 6325 points

Thank you Tom and Rakesh!

It will take me a few days to verify as I am having some setup issues and having a hard time programming the part.

I will verify this post as soon as I can.

Thanks!

Brandy

0 tscheck over 13 years ago in reply to Brandy Jabkiewicz

TI__Mastermind 23525 points

Any luck Brandy? I posted the new MMC code to Gforge https://gforge.ti.com/gf/project/msp430_mmc_code/frs/

Hopefully that did the trick for you.

Regards,

Travis

0 Brandy Jabkiewicz over 13 years ago in reply to tscheck

Mastermind 6325 points

I hope so. I have ordered a new Jtag to SBW setup hoping this is my problem with programming. It will be in soon and then I can let you know.

Thanks,
Brandy

0 Brandy Jabkiewicz over 13 years ago in reply to Brandy Jabkiewicz

Mastermind 6325 points

Hello,

I programmed the MSP430 with the new code and yes, it fixes the problems with the NAT MCH. However, now it won't work with the Vadatech MCH at all.

While I appreciate all the effort and the quickness of the fix, I thought that uTCA was supposed to be a standard. Shouldn't I be able to work with both NAT and Vadatech?

Thanks,
Brandy

0 tscheck over 12 years ago in reply to Brandy Jabkiewicz

TI__Mastermind 23525 points

Brandy,

Glad to hear that fixed your issue. Interesting about the Vadatech MCH, I wouldn't have expected that either. I confirmed that the only change made was to the IPMB_RESP_MAX_DATA_LEN value in the IPMI.h file. So if you change it back and recompile, does it still work with the Vadatech MCH? I realize it is not preferred, but sounds like there is some vendor specific limit here.

Regards,

Travis

Processors

Processors forum

NAT MCH with TIC6678 EVM in uTCA: IPMI Messaging