Francois Romieu wrote:
Andy Furniss <[EMAIL PROTECTED]> :
[...]
They will be the same as connexent haven't touched the drivers for
years, Patrick ported them to 2.6 and a diff doesn't show any difference
involving HZ that I can see.
You are comparing two completely different ->systems<-. Please stop being
focused on HZ, ok ? HZ does not rule the priority of the in-kernel tasks.
OK - It was just that changing HZ sorted it on 2.4.
[...]
I suppose there could be something in the binary that says if HZ <1000
use it if not use 100 but I could test that - remember running 500 on
2.4 "fixed" things.
From your previous description, it seemed that 2.4 was always fine.
Did I miss something ?
It was the same on 2.4 unless I patched and raised HZ to 500 which made
it less noticable 2ms vs 10ms.
I still don't know how the driver would "know" the difference between
local and forwarded traffic?
You have lost me there. What does "local" mean ? If it's LAN traffic, the
adsl driver has no business with it.
The box with the modem is my gateway, on 2.4 if I sshed into it and did
the test from it (=locally generated traffic) I would see a difference
to doing the pings from a LAN PC (=forwarded traffic).
I'll have to revisit that - thinking about it that could be because the
test pings from the box with the modem stayed in sync (and I got +0 to
+10ms steady depending on luck) and the timer on the LAN PC drifted and
caused the pattern like I posted.
On 2.6 I can see the cycling pattern whether pinging from a LAN PC or
the gateway PC.
[...]
That being said, it seems that the 2.6 version of the driver uses
workqueues where 2.4 driver used native kernel threads. You can
experiment with the priorities there.
I'll have to google around that one - but I still secretly hope it's not
the driver anyway.
Issue a ps command, raise/lower the priority of things like ksoftirqd and
see if it makes a difference (on 2.6).
Thanks I'll give that a try.
Andy.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
