[Click] [PATCH] patches for linux 2.6.24

Joonwoo Park joonwpark81 at gmail.com
Thu Feb 7 23:00:34 EST 2008


2008/2/8, Egi, Norbert <n.egi at lancaster.ac.uk>:
> Hi Joonwoo and Adam,
>
> I tested your patches with the latest e1000 driver, for which I just created the patch to support Click polling, (it seemed easier to patch this latest driver than to fix the patch of the version 7.3.20). So, PollDevice now works, although there is an assertion issue when I install or uninstall a script, but apart from that packet forwarding seems to be stable. I'll trace this assertion problem...
>
> I also tested Click multithreading, which also worked, but somehow all the 8 cores I have in the machine seem to be busy (not all at 100%) even though I only have 6 Click threads. However, when I manually assign the threads each to a separate core, only 6 cores seem to be loaded (which is normal), so this might be somehow related to the Linux scheduler. Do you know whether they have changed anything in the Linux scheduler that makes this behaviour normal or the click patch is more likely to be incomplete? The performances look alright in both cases, at least with that single configuration I've tested so far.
>
> If somebody could test the driver patch in his/her system and let me know how it works, that would be great.
>
> Thanks,
> Norbert
>
> ________________________________
>
> From: click-bounces at pdos.csail.mit.edu on behalf of Joonwoo Park
> Sent: Tue 2/5/2008 08:11
> To: click at pdos.csail.mit.edu
> Subject: [Click] [PATCH] patches for linux 2.6.24
>
>
>
> Hi,
> This is a version of click for the 2.6.24 linux.
> Lots of the work was done by Adam Greenhalgh and I just peppered on it.
> This is a very raw version, I think much more fix might be needed.
>
> I didn't work for e1000 and platforms except the x86-32.
> fyi, I'm attaching my config for linux
>
> things tested (a few):
>  build & install on my x86 laptop (thinkpad t60)
>  simple FromDevice
>  simple ToDevice
>
> Thanks,
> Joonwoo
>
>

Awesome Norbert!

I guess the assertion issue  is releated locking problem, I sent a patch for it.
Since I'm off for this week, I can't test your patch at this moment.
I'll dig these things at next week and I'll try NAPI as a possible
option for e1000 that you patched.

Thanks,
Joonwoo


More information about the click mailing list