Forum Discussion

Altera_Forum's avatar
Altera_Forum
Icon for Honored Contributor rankHonored Contributor
19 years ago

uClinux 2.6.14 snapshots

v1.0 committed - http://via.dynalias.org/snapshots/linux-2.6.14-1.0.tar.bz2 (http://via.dynalias.org/snapshots/linux-2.6.14-1.0.tar.bz2)

------------------------------------------------------------------------

r920 | jdhar | 2006-04-13 15:51:43 -0400 (Thu, 13 Apr 2006) | 1 line

recommitting v1.0 - EPCS MTD, I2C Opencores, Open_Eth, MMC 0.1.5, zImage

No CF support yet, I will need a patch for that.

So far, kernel seems ok, although a little more prone to page allocation failures than 2.6.12 - if people try out heavy IO operations with this kernel, it will help.

5 Replies

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    hippo, here is what I am seeing - I am doing a simple test where I copy a large file (35M Linux BZ2 kernel) from one MMC partition to another.

    In 2.6.12, I can telnet in about 3 different sessions before it crashes.

    In 2.6.14, I can maybe get only 2 sessions.

    In 2.6.15, it will bomb as soon as I do anything else. If I have a UART session open, and a telnet session open executing the 'cp' command, and do something like cat /proc/meminfo while the copy is taking place, it gets page allocation failures.

    So I'm not sure if this is a bug, not enough RAM, or the kernels are getting worse at handling small amounts of RAM. I am going to try and define a more conclusive test, but that's teh current status.
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    I tried this on 2.6.11 as well, and it fails with about 2 telnet sessions as well.

    When I check /proc/meminfo, most of it is in 'cached' - I guess because it's doing a large IO transfer. Does this seem normal?
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    --- Quote Start ---

    originally posted by jdhar@Apr 15 2006, 12:30 AM

    i tried this on 2.6.11 as well, and it fails with about 2 telnet sessions as well.

    when i check /proc/meminfo, most of it is in 'cached' - i guess because it's doing a large io transfer. does this seem normal?

    <div align='right'><{post_snapback}> (index.php?act=findpost&pid=14383)

    --- quote end ---

    --- Quote End ---

    It is strange.

    Maybe you can try add a "sync" command to flush write cache before next session.

    Or use CF drive as a cross check.
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    I can verify J&#39;s 2.6.14 snapshot.

    The NIOS serial driver is OK.

    Unfortunately I couldn&#39;t test the LAN91C111 ethernet, since the relevant menuconfig option seems to have disappeared. Any idea what I&#39;m doing wrong?