Actions
Forum Thread Details
  • State Verified Answer
  • Replies 6 replies
  • Subscribers 335 subscribers
  • Views 1186 views
  • Users 0 members are here
Related

Mini-ITX self-test - PL DDR issue?

Former Member

Hello everyone,

Just wondering whether there's a self-test boot image available for the 7045 mini-ITX board. I'm seeing some odd behaviour (more below) which I'm 99% sure it just a configuration/software error but I'd like to be completely eliminate the hardware as a cause. I've found the Out-Of-Box design, but that's only a very basic test (essentially checks that the PS starts up).

 

The problem I'm having: I can only access 512MB of the PL DDR3 RAM. It's mapped into address range 0x8000_0000 to 0xBFFF_FFFF (ie 1GB address space) but everything from 0xA000_0000 to 0xBFFF_FFFF is the same as 0x8000_0000 to 0x9FFF_FFFF. If I write to 0x8000_0000, the same data appears at 0xA000_0000 and vice versa.

Software in use is Vivado 2016.1, Petalinux 2015.4, and a heavily-modified version of the Zynq 7045 Mini-ITX 2015.2.1 Petalinux reference design (I cut out almost all the original hardware blocks to put my own in, but I haven't touched the MIG).

 

Any ideas?

 

Thanks,

Evan

Parents
  • 0

    Evan,

      If you just want to validatate that all the memory is available in bare metal, the SDK can generate a memory test program.  By default, it only tests a portion of available memory (for a quick and speedy check that things are working), but you can easily modify the source to validate the entire available memory.

    Ron

  • 0 in reply to Former Member

    Ron,

     

    Thanks very much for the reply. I didn't know about that feature in the SDK; it'll definitely be useful. I've done a bit of software-level testing using peek/poke from Petalinux and that is definitely not able to access the upper 512MB of PL RAM.

     

    The problem I have is that I suspect there's actually a problem below the software level, either in the physical hardware (unlikely) or in the bitstream (more likely). What I was hoping to get was a "known-good" self-test bitstream which would distinguish between those scenarios; if a known-good bitstream that works on other mini-ITX boards works fine here then the hardware is fine and somehow Vivado is getting the bitstream generation messed-up; if it fails here then there must be a hardware issue with the board and we should look into replacing it.

    It looks like I can take the Petalinux reference design (which includes a pre-built bitstream) and use that with the SDK, so I'll give that a shot.

     

    Cheers,

    Evan

Reply
  • 0 in reply to Former Member

    Ron,

     

    Thanks very much for the reply. I didn't know about that feature in the SDK; it'll definitely be useful. I've done a bit of software-level testing using peek/poke from Petalinux and that is definitely not able to access the upper 512MB of PL RAM.

     

    The problem I have is that I suspect there's actually a problem below the software level, either in the physical hardware (unlikely) or in the bitstream (more likely). What I was hoping to get was a "known-good" self-test bitstream which would distinguish between those scenarios; if a known-good bitstream that works on other mini-ITX boards works fine here then the hardware is fine and somehow Vivado is getting the bitstream generation messed-up; if it fails here then there must be a hardware issue with the board and we should look into replacing it.

    It looks like I can take the Petalinux reference design (which includes a pre-built bitstream) and use that with the SDK, so I'll give that a shot.

     

    Cheers,

    Evan

Children
No Data