From: Edulix <edulix@tumundoweb.com>(by way of Edulix <edulix@tumundoweb.com>)
To: lartc@vger.kernel.org
Subject: Re: [LARTC] filter ingress policy rates -> slow!!
Date: Sat, 03 Jul 2004 22:25:34 +0000 [thread overview]
Message-ID: <200407040025.34742.edulix@tumundoweb.com> (raw)
In-Reply-To: <200407011810.01786.edulix@tumundoweb.com>
El Sábado, 3 de Julio de 2004 17:02, gypsy escribió:
> Edulix wrote:
> > BTW, Should I warn the Fedora devels about the JIFFIES problem?
>
> No. It is set to JIFFIES because that is the setting required so that
> all the CPUs in the kernel .config list work. If you select 386 or 486
> in 'make config', make bzImage will fail because the CPU does not
> support a clock. 'cat /proc/cpuinfo | grep tsc' - if that would work in
> a .h file - would return TRUE for CPUs that do support a clock.
> Otherwise, the Time Of Day clock (IRQ 8) or "jiffies" (IRQ 0) should be
> used.
Well, a bit more info: I've tried to reproduce the same thing in my sister's
computer. Whereas I have a Athlon XP 2000+, she has a Celeron 2.4 Ghz.
Whereas I use FC1, she uses SuSE 9.0. I saw in the sources that she also has
JIFFIES configured.
Basically, exactly the same problem happens in her computer, her rate instead
of be of 44~ kb/s was of about... 400-700kb/s.
After that, I just remembered that I've got gentoo installed in other
partition of my PC. I rebooted, and went to kernel sources (using vanilla
2.6.1). I aw that it was using JIFFIES; so I changed to CPU. Then I executed
make menuconfig to be sure that I selected Athlon as my cpu type and that
QoS modules was selected (they were). So I didn't need to change anything in
the conifguration: Finally, I executed "make" and "make modules_install",
did a "modprobe -r sch_ingress; modprobe sch_ingress.
I executed : "eshaper start downlink" and assured with a "eshaper status"
that the rules were there (no packets droped still). Then I tried copying
file from my sister's computer via scp and it went at about a 700 Kb/s to
1.2 Mb/s rate! Note that in Fedora, in the pretty same pc, it went at only
40 Kb/s.
"eshaper status" told me that there were many many packets dropped. I tried
to transfer files without any tc rules (executed "eshaper stop" for cleaning
them) and it went at 10 Mb/s - files transfered much faster.
I haven't checked CPU usage in this tests though... it might be a good idea.
Any ideas? What to do now ? I'm still looking for compiling my sister's kernel
with CPU instead of JIFFIES, because maybe it will go at desired speed -
remember that even with JIFFIES the rate was of ~400-700 kb/s!
Thanks for your time,
Edulix.
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
next prev parent reply other threads:[~2004-07-03 22:25 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-07-01 16:10 [LARTC] filter ingress policy rates -> slow!! Edulix
2004-07-01 20:40 ` Stef Coene
2004-07-02 8:36 ` Edulix
2004-07-02 14:13 ` gypsy
2004-07-03 4:14 ` gypsy
2004-07-03 9:44 ` Edulix
2004-07-03 11:09 ` Ed Wildgoose
2004-07-03 11:47 ` Edulix
2004-07-03 13:13 ` Ed Wildgoose
2004-07-03 15:02 ` gypsy
2004-07-03 22:25 ` Edulix [this message]
2004-07-04 19:47 ` gypsy
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=200407040025.34742.edulix@tumundoweb.com \
--to=edulix@tumundoweb.com \
--cc=lartc@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.