svn commit: r215166 - in head: . sys/conf sys/netinet sys/netinet/cc sys/sys

[Mikolaj Golub]

On Tue, 16 Nov 2010 00:46:45 +1100 Lawrence Stewart wrote:

 LS> Ok thanks. If you are or anyone else is feeling really adventurous, I
 LS> have created a series of patches that add better VIMAGE support to the
 LS> framework.

 LS> If you feel like testing them, they need to be applied in the following
 LS> order:

 LS> 1. Init the CC framework global state only once on system boot by
 LS> switching to a SYSINIT. (You already have this one)
 LS> http://people.freebsd.org/~lstewart/patches/5cc/modcc_vnetpanic_ccinit_v2.patch

 LS> 2. When unloading an algo module, loop through all TCP control blocks in
 LS> all vnets (instead of just the default vnet) to look for connections
 LS> using the algo.
 LS> http://people.freebsd.org/~lstewart/patches/5cc/modcc_vnet_algounloadcleanup.patch

 LS> 3. Move protocol specific implementation detail out of the core CC
 LS> framework (not vnet specific but useful cleanup).
 LS> http://people.freebsd.org/~lstewart/patches/5cc/modcc_tcpalgounload.patch

 LS> 4. Create a per-vnet variable to store a ptr to the vnet's default CC
 LS> algo and hook it up to a vnet aware sysctl handler. vnets can now choose
 LS> their own default from the global list of registered algos independently
 LS> of other vnets.
 LS> http://people.freebsd.org/~lstewart/patches/5cc/modcc_pervnet_defaultalgo.patch

 LS> It would be fine to test all of them in one go. The last one is the most
 LS> intrusive and the one I'd most like to see some testing done with. Only
 LS> if you feel like it of course.

Feeling virtually adventurous, applied them all on VirtualBox :-). This is the
box where the hang is not observed. It works ok for me so far. I will try at
home too (where the hang is observed) and let you know if there are any
issues.

-- 
Mikolaj Golub