This code block also needs an adjustment to "something" as we have no need to add these to this target's build in the GCC case.

In D13230#275661, @jhb wrote:

I wonder if this shouldn't be conditional on which compiler is used as cc. That way if you do 'make native-stools WITH_CLANG_AS_CC=yes' you would get clang instead of gcc.

Yep, looks good. Thanks!

What I take away from the conversation in this review is that the solution being proposed does what its supposed to do, and there is even more room for improvement.

Builds. Untested at this time.

Odd, I get a wierd chmod error on FreeBSD 10 when building this port. On 11/12 it builds fine:

@olivier Can you post the sysctl/loader.conf settings you are using from both iterations of the test?

@olivier So a decrease of ~33% with IFLIB?

bump. What's the plan here?

bump. What's up here?

In D12425#266522, @pkanneganti_cavium.com wrote:

Hi Bruno,

Are you expecting me to update the revision with the Makefile change?

Thanks
Prasad

Found while testing the iflib'd version of ixgbe(4).

Ok, my report is due to the test systems disabling AIM in sysctl.conf.

I found pretty bad behavior with LRO enabled on this change.

This looks good to me. Can you verify that this is merged into the ixgbe(4) IFLIB review as well?

Ran iperf3 test between two hosts booted into i386 and all seems to be well.

In D12447#261301, @shurd wrote:

In D12447#259600, @sbruno wrote:

Ok, lets rename prefetch2() to something that indicates why there is a second function (being a bit pedantic here). something like prefetch_align() or whatever.

Do you like prefetch2cachelines() better?

This seems pretty reasonable to me. Any objections? Can I get a transport@ review accept on this?

Does this version work on 32bit FreeBSD? Or are you waiting for us to review/commit the 32bit support review?

Thanks!

The issues I was running into seem to be related to my attempts to use the old way/old documentation. I can tftp a kernel/ramdisk with the default options.

Huh ... I saw this review and noted that its still open. Is this still under evaluation?

I suspect this review needs to have stephen's comments addressed.

Ok, lets rename prefetch2() to something that indicates why there is a second function (being a bit pedantic here). something like prefetch_align() or whatever.

I went back to the man page, just to validate. I used the example from the man page after rebuilding pxeboot with *defaults*. It defintely does not TFTP in this case.

In D12466#258441, @tsoome wrote:

In D12466#258432, @sbruno wrote:

In D12466#258416, @tsoome wrote:

This is busting all the effort of avoiding to define tftp or NFS at build time, isn’t it?

I'm unsure as what I can tell is that the current version of pxeboot won't tftp an MFSroot + Kernel in its current state.

Its possible that I just don't know what the magical incantation is in dhcpd.conf to get it to work right, but tcpdump doesn't show a TFTP req if this code isn't in place.

I can not check the man pxeboot right now, but we have tftp:// prefix etc. I am not sure if the man has received the update, or it is just a,bout the source.

In D12466#258416, @tsoome wrote:

This is busting all the effort of avoiding to define tftp or NFS at build time, isn’t it?

Did microsoft update the referenced link? I don't see the removal notice in it.

Is there a reason that we don't modify prefetch() instead of creating a new prefetch2() function?