[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [microblaze-uclinux] MicroBlaze MMU support test release now available



John Williams wrote:
Hi
I wrote:

I'm pleased to announce that we have a test release available for MicroBlaze MMU support in the PetaLinux 2.6.20 kernel.

Well it's 7 days and > 150 downloads later - any reports so far about the kernel itself? Anyone managed to crash it?

I'm aware of minor tool/build gremlins and they'll be sorted, plus we also need to reimplement the old-style kernel+romfs image.bin (instead of the rootfs.cpio stuff). These will happen soon.

THanks,

John

Hi John,

The only thing I have used to stress this kernel so far is flood pings
from a directly attached PC NIC.  I am running
"ping -f -s 1000 10.25.1.201" as root on the PC to ping the MicroBlaze
with 1000 byte packets with no delay between a reply and the next request.
Here are some observations:

1) Using the same hardware bitstream generated from
    hardware/reference-designs/Xilinx-Spartan3E1600-MMU-edk92
  with updated IP, BlueCat_5.4.2 has a transfer rate 3x higher than
  petalinux-mmu-v0.10.  For the aggregate transmit and receive transfer
  rate, I am seeing:

    BlueCat_5.4.2:       1.143 MBytes/sec
    petalinux-mmu-v0.10: 0.374 MBytes/sec

  BlueCat_5.4.2 also drops fewer packets than petalinux-mmu-v0.10.

2) The interactive response time of the serial console while flood pings
  are active is much better with petalinux-mmu-v0.10 than with
  BlueCat_5.4.2.  The console is nearly unusable with BlueCat_5.4.2.

3) With petalinux-mmu-v0.10 the flood ping will fail with 100% packet
  loss after 7-15 minutes.  Once in this state, regular pings from
  the PC to MicroBlaze also fail but pings from MicroBlaze to PC
  continue to work.

I have not seen a kernel panic with petalinux-mmu-v0.10.  I did see
them on BlueCat_5.4.2 before updating the hardware design as I pointed
out yesterday.  It appears to me that both petalinux-mmu-v0.10 and
BlueCat_5.4.2 use the same Spartan3E1600 hardware design as they both
have outdated MicroBlaze IP in their local pcores directory.

Thanks,
Bob


___________________________
microblaze-uclinux mailing list
microblaze-uclinux@xxxxxxxxxxxxxx
Project Home Page : http://www.itee.uq.edu.au/~jwilliams/mblaze-uclinux
Mailing List Archive : http://www.itee.uq.edu.au/~listarch/microblaze-uclinux/