3 Replies Latest reply on May 31, 2017 5:58 AM by Carlos_A

    ATOM E3845 LVDS Display and SSD Problem (FIRSTMEGABYTEPOLICY)

    wollesIntel Community Member

      Hi!

       

      We use a E3845 Prozessor on a so called "computer on a module" based industrial PC. Connected to the display port there is a eDP to LVDS bridge which is
      connected to the TFT display.

       

      Because of our used eal time system based on WinCE 6.0 which is running "below" Windows 10 we need to set the boote code parameter "FIRSTMEGABYTEPOLICY=UseAll".

      So, Windows uses the first 1MB as a "normal" memory.

      If we set FIRSTMEGABYTEPOLICY=UseAll we have the Problem that after rebooting sometimes the display keeps black ! This happens because during boot up, the

      display ports were not detected and the VGA output is used.

      If we set FIRSTMEGABYTEPOLICY=UseNone then directly we have no problems at all.

       

      With the same system we had some problems by using SSD's. We had some crashes were the SSD disapeared and windows start rebooting, but that failed, because the
      SSD is not detected.

       

      Finally we have the situation that two on chip devices were "disappearing" when the boote code parameter FIRSTMEGABYTEPOLICY is set to UseAll, which means

      that the first megabyte is used dynamic by windows.

        • Re: ATOM E3845 LVDS Display and SSD Problem (FIRSTMEGABYTEPOLICY)
          Carlos_A Brown Belt

          Hello, wollesIntel:

           

          Thank you for contacting Intel Embedded Community.

           

          In order to better understand these problems, could you please tell us if the cited design has been developed by you or a third party company? In case that it is a third party design, please give us all the information related to this implementation.

           

          Also, could you please try to replicate these problems using any of the following Operating System (OS)?

           

          - Microsoft Windows 8, Windows Embedded Standard 8 (non-connected standby)

          - Microsoft Windows 7, Windows Embedded Standard 7

          - Linux* Tizen (select in-vehicle infotainment (IVI) customers only)

          - Linux based on Yocto Project

          - Linux based on Fedora distribution

          - Microsoft Embedded Compact 7 and 2013

          - Wind River VxWorks Wind River

          - Android (JB MR2 4.3)

           

          Please let us know this information to have clear ideas of these issues.

           

          Thanks in advance for your cooperation.

           

          Best regards,

          Carlos_A.

            • Re: ATOM E3845 LVDS Display and SSD Problem (FIRSTMEGABYTEPOLICY)
              wollesIntel Community Member

              Hi,

               

              it's a third party module called ETX-BT from the company Adlinktech.

               

              We use Win7 Embedded Standard and Windows 10 IoT  2016 LTSB to run this board.


              On both operating system we use CeWin which is a realtime extention for Windows 7/10. (It's WinCE 6.0)

              The system running Win7 actually we have no problems. Only on the Win10 systems.

               

              What we already found out:

               

              a) the black screen only comes up after restarting windows (warm start). It never appears after a power up.

               

              b) In case of a black screen after reboot, the windows try to run 1024x768 px on the 800x600 TFT display.

               

              c) I have made a "snapshot" image (acronis true image) of a situation where nearly 99% of the restart's the screen is black.
                  But when I restore this image directly, it hapens only 1% or 2% ????
                  So something has been changed by the backup and restore procedure.

               

              d) When I delete the boot parameter FIRSTMEGABYTEPOLICY everything is fine. (except the situation that our real time system is not working :-) )

               

              My suggestion:

               

              Something happend during the shutdown proccess. Maybe the diplay driver didn't do a proper deinit.
              Or the worst case is that the E3845 has a bug that a memory access is mistaken a port I/O access and the dynamic
              allocated memory the sometimes the same address a the display interface.

               

              I want to do a small test but I don't know how to do it :-(

               

              What happens if I can lock a defined area of pysical memory which has the same adress as the display interface.

              When I look at the output of "RamMap" I see that some memory is used by "Driver Locked". How can I do this manualy ?

              I try to write a device driver (I have done this for NT4.0 twenty years ago) but the function MmAllocateContiguousMemory() does not
              specify a exact memory range.

              Also i don't know at which port address the display port is.

               

              Wolfgang