On Wed, 23 Feb 2005 16:23:02 -0800, Stephen Hemminger
<[EMAIL PROTECTED]> wrote:
> David S. Miller wrote:
>
> >For the millionth time, you can't just delete the existing sysctls.
> >That is a user visisble interface.
> >
> >Instead, you have to make them at least appear to select a
> >single conge
On Wednesday 23 February 2005 22:57, David S. Miller wrote:
>
> For the millionth time, you can't just delete the existing sysctls.
> That is a user visisble interface.
>
> Instead, you have to make them at least appear to select a
> single congestion control algorithm
[..]
Maybe this could be
On Wed, 23 Feb 2005 16:23:02 -0800
Stephen Hemminger <[EMAIL PROTECTED]> wrote:
> I am heading in a different direction with making the TCP congestion
> protocols
> a real infrastructure. After the initial version gets done, then I will
> go back
> and add compatiablity interface hooks as requi
David S. Miller wrote:
For the millionth time, you can't just delete the existing sysctls.
That is a user visisble interface.
Instead, you have to make them at least appear to select a
single congestion control algorithm as I stated in the
following posting:
http://marc.theaimsgroup.com/?l=linux-ke
For the millionth time, you can't just delete the existing sysctls.
That is a user visisble interface.
Instead, you have to make them at least appear to select a
single congestion control algorithm as I stated in the
following posting:
http://marc.theaimsgroup.com/?l=linux-kernel&m=1109099735303
This patch makes selection of congestion control algorithm simpler by
using a single sysctl for that purpose, rather than a cascade of sysctls.
The patch also does some minor cleanups to avoid cascade actions between
algorithms so that flow control is cleaner.
Possible improvements:
- Use a stri
