On 5/28/07, Arnaldo Carvalho de Melo <acme@xxxxxxxxxxxxxxxxxx> wrote:
On 5/11/07, Ian McDonald <ian.mcdonald@xxxxxxxxxxx> wrote:
> This shifts ccid3_li_hist.
>
> We temporarily export the symbol but remove this in later patch.
>
> Signed-off-by: Ian McDonald <ian.mcdonald@xxxxxxxxxxx>
Ian,
Finally looking at this, sorry for taking so long. But I have
some issues with this loss intervals code reshuflling.
The idea of having dccp_tfrc_lib.ko is to have things that are
not specific to any CCID implementation, so that if some new CCID
uses the TFRC algorithms it would share code with CCID3, etc.
So it is not really correct for the code in net/dccp/ccids/lib/
to access the half connection internal state, as it really don't know
what is the layout of such internal state.
In your patches you move ccid3_hc_rx_update_li from
net/dccp/ccids/ccid3.c to net/dccp/ccids/lib/loss_interval.c and then
renamed it to dccp_li_update_li, but this function still uses
ccid3_hc_rx_sk, not good.
So I'm looking at the intent of the patches and will post here
how I think it should have been done and the sequence of patches
preparing for such change, etc.
Please take a look my tree at:
http://git.kernel.org/?p=linux/kernel/git/acme/net-2.6.git
I rewrote your patches to achieve the same result while keeping
loss_interval.c CCID agnostic Please also take a look at how the
patches go on paving the way for the next ones, without having to
make dccp_li_hist temporarily exported, etc.
This tree has just some copyright and compiler warning fixes besides
these csets, so if you agree on it just provide me the signed-off or
acked by lines and I'll add those and republish the tree.
- Arnaldo
-
To unsubscribe from this list: send the line "unsubscribe dccp" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
