• State TI Thinks Resolved
  • Locked Locked
  • Replies 14 replies
  • Answers 1 answer
  • Subscribers 104 subscribers
  • Views 4557 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

AM437x gp evm - emmc boot - raw or card

Robert Woerle
Intellectual 380 points

Dear all


We are trying to verfiy a new design on the gp-evm board and basicly want to achieve the same bootorder as the beaglebone black.
 There one can switch the bootorder via a button gpio from mmc0/mmc1.

So we are trying to verify on the gp-evm that booting a emmc from mmc1 in file mode is possible.

Sofar we physically made sure the SEL_eMMCORNAND pin is set high to select the emmc.

ive formattted the emmc accordingly ( like the mmc cards i use for booting) and placed a MLO into the first partitions fat filesystem.

I selected sysboot 00100b to get a bootorder for  mmc1 first and then mmc0.

So for my understanding i should at least get to see the first line of u-boots mlo.

Yet i am confused by the TRM saying that the cpu only supports "raw" mode emmc on mmc1 ?

Will the am437x never support emmc on mmc1 booting ?

I read in a different post that the am335x had the same "limitation" yet people report emmc booting in file mode just fine and the TI tech commented this, saying that the emmc might be not detected as emmc but as normal mmc card.

http://e2e.ti.com/support/arm/sitara_arm/f/791/t/336877.aspx


Any hints ?

  • 0
    TI__Guru**** 393215 points

    What was your result? Have you booted from eMMC? I am expecting a clarification about this raw mode statement from the factory team and will update here when this is clarified.

  • 0 in reply to Biser Gatchev-XID
    Intellectual 380 points

    Basicly i didnt get the exspected first line output of u-boot.

    The system is in a way stuck. There are a few things which come to my mind.

    1) The stuck system ( even when mmc0 is also plugged in) tells me that core looked for mmc1, found it , tryed to boot it so far but failed, but got sofar that i doesnt try mmc0 ( which was the supposed bootorder in 00100b sysboot

    2) we doubled checked some pins (RESET_not etc) and also CLK on mmc1 so that also brings me to the point that the core actaully finds mmc1 and gets "quite" far, at least not taking the next device ( mmc0) in considerations.

    tomorrow i will double check u-boot mlo code and see when actually that 1st line is put out. Maybe it dies for a other reason and the exspected 1st line is just too late.

  • 0 in reply to Robert Woerle
    Intellectual 380 points

    Sorry i ve to revert the fact that the system is stuck .. it actually is not stuck and does try the mmc0 after "failing" to boot the emmc on mmc1

    so i am back to the drawing board and the scope again doing some further research.

    The confirmation about emmc only supporting raw mode booting would be nice though.

  • 0 in reply to Robert Woerle
    TI__Genius 15510 points

    I'll have to spend more time on the raw vs. FAT boot. I know you can boot FAT on 2GB and 4GB eMMC.

    As for the GP EVM, I'll have to dig up an e-mail I have that explains a boot ROM issue where eMMC on mmc1 will not boot. There is supposed to be a fix to the boot ROM in the next PG of the part.

    Steve K.

  • 0 in reply to Steve Kipisz
    Intellectual 380 points

    Hi Steve

    Dont bother with the gp evm , as i think i know why it is not working properly

    According to the TRM the rom code muxes mmc1 i to use GPMC_AD8-11 for data.

    Yet the gp evm is connecting GPMC_AD0-7 as 8bit data connection. I guess thats all to the fact that they implemented both NAND and emmc.

    So my tests with the gp evm are done for this part.

    We concluded to use mmc0 for external mmc, ( as using 8bit emmc on mmc0 would destroy 2 ethnert solutions)

    and mmc1 for internal emmc and stick with the TRM pin mux aka layout.

    Also we will have 0000b or 00100b sysboot models ( with a little push button to switch )

    Yet, FAT vs RAW is still a valid point ( of course i love to get FAT access) hence its nice to update stuff from linux just via file copys

    Also it came up that, hopfully a 8bit emmc attached to mmc1 will work in 4bit mode during rom boot ?

    I hope emmc starts out in 4bit mode and must be switched to 8bit later on ( by the kernel etc)

    So getting the FAT vs RAW questions sorted would be nice ( or also why the Beaglebone black made it work and how)

    Do you know any reference design with 8bit  emmc on mmc1 ?

  • 0 in reply to Robert Woerle
    TI__Genius 15510 points

    The boot ROM starts in 1-bit mode, queries the card for what it supports, and if 4-bit is supported, it changes to 4-bit mode. SPL/u-boot will use 4-bit mode. The Linux driver will query the eMMC capabilites and if 8-bit is supported, it switches to 8-bit mode.

    I'm still looking into raw vs. fat.

    Steve K.

  • 0 in reply to Steve Kipisz
    TI__Genius 15510 points

    By the way, I had an EVM modified so that we could test booting eMMC1 and it boots using FAT.

    Steve K.

  • 0 in reply to Steve Kipisz
    Intellectual 380 points

    "an EVM" , which one exactly was that ?

  • 0 in reply to Robert Woerle
    TI__Genius 15510 points

    An AM437x GP EVM.

    Steve K.

  • 0 in reply to Steve Kipisz
    Prodigy 10 points

    Hi Steve

    We are working on AM437x eMMC booting , from your post it seems you are able to boot form eMMC  on AM437x GP EVM using FAT filesystem with some EVM modification, could you please share the details of modifications and was there any updates in SW?

    With Regards

    Srinath

  • 0 in reply to Srinath Rama Iyenngar
    Intellectual 380 points
    Dear Srinath
    During our investigations it tourned out that the gp-evm board is not able to boot from the emmc , as the layout is not made so.
    The Cpu Rom code uses different pins for the particular mmc interface to boot from. See my post above.

    Meanwhile we finalized our own layout and using a emmc connected to the correct pins and can confirm that it does boot it with file mode , so using fat did work for us.
    Yet its unclrear when the rom code decides to "detect" the emmc as sdmmc and therefore beeing able to use file mode instead of raw mode.
    So we are waiting for TI to clear this criteria. Until then we will simply test it with newer emmc chips and hope !
    Regards Robert
  • 0 in reply to Srinath Rama Iyenngar
    Intellectual 460 points
    Any update here? We are also working on same issue
  • 0 in reply to sadasivappa s
    Intellectual 380 points

    As stated above, we did not came across an issue. Our systems use the proper emmc fat loading just fine.

    I gave green lights to my production team and the machines do their job ;-)

    regards rob

  • 0 in reply to Steve Kipisz
    Intellectual 490 points
    Hi Steve,

    Can you please explain the modification