question about PCIe cards...

Discussion in 'General Chat' started by BLinux, Jun 5, 2019.

  1. BeTeP

    BeTeP Member

    Joined:
    Mar 23, 2019
    Messages:
    93
    Likes Received:
    15
    I am not angry. I am just being direct and calling things as I see them. If my accurate description of your actions and qualities offended you - than you are the one who should worry about low self-esteem.

    I was genuinely trying to help you and I provided very concise answer to your most glaring problem. Too bad it went over your head. I provided some additional info and recommended to read the protocol specs for even more details. Instead of thanking me you got cocky. And now you got the nerve to claim the moral high ground and criticize my manners. Here is another word for your list - hypocrite.

    By the way, me recognizing your less than stellar English skill was never about feeling superior. It was my way of saying that since a part of your lack of comprehension skill can be attributed to your language incompetence, in a way, it makes you less of a moron than I thought originally. Still an arrogant one though.
     
    #21
    Last edited: Jun 6, 2019
  2. Terry Kennedy

    Terry Kennedy Well-Known Member

    Joined:
    Jun 25, 2015
    Messages:
    992
    Likes Received:
    459
    That would appear to be the fab / date code, per this document. K226K is:
    K: ST Shenzen
    2: Year 2012
    26: Week 26
    K: Process technology K

    To try to add some light to the heat here... If the SBR parts are identical, yet copying the contents from one to the other does not work, the difference has to be somewhere else. It could either be some other component on the board, or the "normal" LSI flash utility does something different when writing the SBR (possibly choosing from multiple alternate versions stored in the flashable firmware file) beyond simply copying from the file to the SBR chip (which it seems is what you are doing). Trying to see what you get when doing "readsbr" with the LSI utility vs. reading it via your methof (for both the "working" and "non-working" SBR images) might be informative.
     
    #22
  3. pricklypunter

    pricklypunter Well-Known Member

    Joined:
    Nov 10, 2015
    Messages:
    1,460
    Likes Received:
    414
    I had been wondering if the reason for the differing lengths and repeated blocks was the way the reader interpreted the e2prom paging, or indeed, if the e2proms were addressed differently.

    I would start by making sure that both cards are configured in an identical way, using the same firmware files, flashing tools and process. I might also try blanking the device with the programmer, to remove any possible random data bytes, and then testing to see if the card can still be flashed, and if so, if this results in identical sbr contents. If it does, great, you can get to work on figuring out what each part of the block controls or does, but if not, it would suggest that the controller chip or some other influence is playing a part in determining how things are done and what gets stored where. Check the revision on the controller chips to see if they are the same for example. Are the board revisions identical? When attempting to determine useful information, by doing comparisons and seeing what is different or has changed, you need equal partners to begin with :)
     
    #23
  4. Patrick

    Patrick Administrator
    Staff Member

    Joined:
    Dec 21, 2010
    Messages:
    11,438
    Likes Received:
    4,378
    Can we please keep things civil here?
     
    #24
    mmo, Patriot and BLinux like this.
  5. BLinux

    BLinux Well-Known Member

    Joined:
    Jul 7, 2016
    Messages:
    2,210
    Likes Received:
    723
    still digging into this...

    both H710 cards have the same firmware versions:

    for the one with 512B SBR:
    Code:
    ==============================================================================
                        Versions
                    ================
    Product Name    : PERC H710 Mini
    Serial No       : 36F022C
    FW Package Build: 21.3.5-0002
                        Mfg. Data
                    ================
    Mfg. Date       : 06/17/13
    Rework Date     : 06/17/13
    Revision No     : A00
    Battery FRU     : N/A
                    Image Versions in Flash:
                    ================
    BIOS Version       : 5.42.00.1_4.12.05.00_0x05290003
    Ctrl-R Version     : 4.04-0003
    Preboot CLI Version: 05.00-03:#%00008
    FW Version         : 3.131.05-8148
    NVDATA Version     : 2.1108.03-0097
    Boot Block Version : 2.03.00.00-0004
    BOOT Version       : 06.253.57.219
    
    for the one with 256B SBR:
    Code:
    ==============================================================================
                        Versions
                    ================
    Product Name    : PERC H710 Mini
    Serial No       : 28N0024
    FW Package Build: 21.3.5-0002
                        Mfg. Data
                    ================
    Mfg. Date       : 08/25/12
    Rework Date     : 08/25/12
    Revision No     : A00
    Battery FRU     : N/A
                    Image Versions in Flash:
                    ================
    BIOS Version       : 5.42.00.1_4.12.05.00_0x05290003
    Ctrl-R Version     : 4.04-0003
    Preboot CLI Version: 05.00-03:#%00008
    FW Version         : 3.131.05-8148
    NVDATA Version     : 2.1108.03-0097
    Boot Block Version : 2.03.00.00-0004
    BOOT Version       : 06.253.57.219
    
    As far as firmware is concerned, all the versions are the same from what I can tell. The only differences above are Mfg Date and serial number.

    BUT!!! I did find one difference that doesn't explain how the SBR is read off the card (the EEPROM chips are the same, and i'm not asking about how to read the the chip via I2C, but rather, how does whatever is reading the chip know to read only 256B vs 512B out of the 8K available on the chip), but at least correlates that the 2 cards really are different:

    The 512B SBR has this:
    Code:
                    PCI Info
                    ================
    Controller Id   : 0000
    Vendor Id       : 1000
    Device Id       : 005b
    SubVendorId     : 1028
    SubDeviceId     : 1f38
    Host Interface  : PCIE
    ChipRevision    : D1
    
    The 256B SBR has this:
    Code:
                    PCI Info
                    ================
    Controller Id   : 0000
    Vendor Id       : 1000
    Device Id       : 005b
    SubVendorId     : 1028
    SubDeviceId     : 1f38
    Host Interface  : PCIE
    ChipRevision    : B0
    
    So, the chip revision is D1 (512B SBR) vs B0 (256B SBR). I've seen discussions about the SAS2308 where rev D1 supported PCIe 3.0 vs rev B0 only had PCIe 2.0. I wonder if that difference also applies to the SAS2208 here? Since they are very similar in design...

    Also, the more I think about this, I don't think the firmware has a lot to do with my question. Even when a card has the flash erased (no firmware), the OS is still able to detect that there's a PCIe card and knows the PCI ids off the EEPROM chip, no matter the 256B or 512B format. So, regardless of firmware, the mechanism to detect the PCI ids off the EEPROM chip of the PCIe card must have a way to distinguish between reading 256B vs 512B; that's what I'd like to know...
     
    #25
    RageBone likes this.
  6. pricklypunter

    pricklypunter Well-Known Member

    Joined:
    Nov 10, 2015
    Messages:
    1,460
    Likes Received:
    414
    I touched on this in my last post. How likely is it that you have an H710 supplied as vendor specific and the other one a retail version? It's possible that the additional space in the e2prom could be used to configure the controller for specific disk parameters. Just a thought. It's much more likely though, that the later version of the controller, has several of the parameters that were stored in a configuration prom, now baked into the hardware itself. It's impossible to know for sure exactly what's going on, without full documentation on the devices and their revisions. This would be a good topic for one of the LSI/ Avago engineers though, if one could find such a bod who is both bored and hanging around forums ;)
     
    #26
    RageBone likes this.
Similar Threads: question PCIe
Forum Title Date
General Chat Home Network and server set up and questions Mar 15, 2019
General Chat Dell racking question Mar 7, 2018
General Chat HP Blade servers/blade Servers questions for a Noob Jan 5, 2018
General Chat I have a question, please help Dec 21, 2017
General Chat Satellite Question Nov 14, 2017

Share This Page