Re: tbench regression in 2.6.25-rc1

On Fri, 2008-02-15 at 07:05 +0100, Eric Dumazet wrote:
Zhang, Yanmin a écrit :
Comparing with kernel 2.6.24, tbench result has regression with
2.6.25-rc1.

1) On 2 quad-core processor stoakley: 4%.
2) On 4 quad-core processor tigerton: more than 30%.

bisect located below patch.

b4ce92775c2e7ff9cf79cca4e0a19c8c5fd6287b is first bad commit
commit b4ce92775c2e7ff9cf79cca4e0a19c8c5fd6287b
Author: Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx>
Date: Tue Nov 13 21:33:32 2007 -0800

[IPV6]: Move nfheader_len into rt6_info

The dst member nfheader_len is only used by IPv6. It's also currently
creating a rather ugly alignment hole in struct dst. Therefore this patch
moves it from there into struct rt6_info.


As tbench uses ipv4, so the patch's real impact on ipv4 is it deletes
nfheader_len in dst_entry. It might change cache line alignment.

To verify my finding, I just added nfheader_len back to dst_entry in 2.6.25-rc1
and reran tbench on the 2 machines. Performance could be recovered completely.

I started cpu_number*2 tbench processes. On my 16-core tigerton:
#./tbench_srv &
#./tbench 32 127.0.0.1

-yanmin

Yup. struct dst is sensitive to alignements, especially for benches.

In the real world, we need to make sure that next pointer start at a cache
line bondary (or a litle bit after), so that RT cache lookups use one cache
line per entry instead of two. This permits better behavior in DDOS attacks.

(check commit 1e19e02ca0c5e33ea73a25127dbe6c3b8fcaac4b for reference)

Are you using a 64 or a 32 bit kernel ?
64bit x86-64 machine. On another 4-way Madison Itanium machine, tbench has the
similiar regression.

-yanmin


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Relevant Pages

  • Re: tbench regression in 2.6.25-rc1
    ... bisect located below patch. ... b4ce92775c2e7ff9cf79cca4e0a19c8c5fd6287b is first bad commit ... As tbench uses ipv4, so the patch's real impact on ipv4 is it deletes ... It might change cache line alignment. ...
    (Linux-Kernel)
  • Re: tbench regression in 2.6.25-rc1
    ... On 4 quad-core processor tigerton: ... bisect located below patch. ... As tbench uses ipv4, so the patch's real impact on ipv4 is it deletes ... In the real world, we need to make sure that next pointer start at a cache line bondary, so that RT cache lookups use one cache line per entry instead of two. ...
    (Linux-Kernel)
  • Re: tbench regression in 2.6.25-rc1
    ... On 4 quad-core processor tigerton: ... bisect located below patch. ... b4ce92775c2e7ff9cf79cca4e0a19c8c5fd6287b is first bad commit ... In the real world, we need to make sure that next pointer start at a cache line bondary, so that RT cache lookups use one cache line per entry instead of two. ...
    (Linux-Kernel)
  • Re: Performance problem since updating from 6.0-RELEASE to 6.0-STABLE last friday
    ... > cache for retrieval to reduce its influence, and its expiration ... > However, acpi_smbat.c doesn't use cache. ... So, I made a patch. ... nate> The patch looks good to me, you should commit please. ...
    (freebsd-stable)
  • [GIT PULL] Please pull the first batch of NFS client changes (and cachefs merge)...
    ... commit e13a5357ab5961844e64ec4ade6e4e13bfc33355 ... SUNRPC: Ensure IPV6_V6ONLY is set on the socket before binding to a port ... NFS: Start PF_INET6 callback listener only if IPv6 support is available ... Close-to-open cache consistency rules really only require us to flush out ...
    (Linux-Kernel)