Dario Borchers | 3 Oct 15:03 2003
Picon

pppd connection breaks down

Hi, 

i've got the following problem: i'm using pppd 2.4.0 to dial to my
isp(t-online, dsl), wich works quite fine, even with ipnat and ipfilter. The
connection works fine for 5 minutes, but then it just "breaks down": i cannot
route any packets across this link, but the link itself still persists; after
2 minutes pppd says "couldn't set device flags" . after next 2 minutes pppd
claims the link to be down and reconnects, which works fine. But i cannot
route any packets across the new link, too. After killing and restarting the
pppd it works again for roughly 5-10 minutes, then i have the same scenario.

I'm using NetBSD 1.6.1 on an Ultra1 (140 mhz, 196 mb ram) with 2 nics (le0 and
hme0), le0 is conneted tho the modem. This box should replace my old
debian-box (pentium 60, 32 mb ram), which is way too slow to handle
transparent proxying and bandwith management. But: this old debian box is
running pppd 2.4.1 and is just doing fine - no problems at all.

I tried recompiling the pppd from pkgsrc - doesen't compile, would be 2.3.9.
And i downloaded the latest sources from the samba cvs
(http://www.samba.org/ppp/)  - it says "no support for netbsd in this
distibution"

Can anybody point me in the right direction, how to find out, whats really
going on, when the link "fails" (and perhaps how to fix it? ;-)) or how to get
a new "working" pppd compiled? I've found this one :
http://groups.google.de/groups?q=netbsd+pppd+compile&hl=de&lr=&ie=UTF-8&selm=200101080614.f086E7T04160%40charm.icompute.com&rnum=4
but dunno, if its related in any way, since it's from Jan 2001.

Thanks in advance,
Dario Borchers
(Continue reading)

Martin Husemann | 3 Oct 15:20 2003
Picon

Re: pppd connection breaks down

On Fri, Oct 03, 2003 at 01:03:18PM +0000, Dario Borchers wrote:
> I'm using NetBSD 1.6.1 on an Ultra1 (140 mhz, 196 mb ram) with 2 nics (le0 and
> hme0), le0 is conneted tho the modem.

Uhmm, you should definitively *not* use a userland pppd in this setup.

look at "man pppoe" and read http://www.netbsd.org/Documentation/network/pppoe,
this will give you a dramatic performance boost.

Martin

Mike Parson | 8 Oct 22:23 2003

Promise ATA in PCI Sparc


I've got a Tatung Ultra10, UltraSparc-IIi 360Mhz, OpenBoot 3.10.7
SME, but no onboard IDE.

I'm trying to get a Promise Ultra33 card working in the system, but when
I try to run the nvramrc from the FAQ, I get an error:

ok nvedit
   0: dev /packages/SUNW,builtin-drivers
   1: : class018000   
   2: f 9 my-space + " config-b!" $call-parent class010100
   3: ;
   4: device-end
   5: 
ok nvrun
Evaluating: f 9 my-space + " config-b!" $call-parent class010100
class010100 ?

(this is all after set-defaults and a clean reset)

Does anyone have any pointers to how I might be able to work this out?

--

-- 
Michael Parson
mparson <at> bl.org

Manuel Bouyer | 8 Oct 22:53 2003

pciide/atabus changes

Hi,
with today's pciide changes the kernel doesn't find its root any more (sorry,
I should have checked before ... I did all my tests with a netbooted kernel).

The easy way would be to add all the *ide drivers names to bus_class_tab[]
in sparc64/sparc64/autoconf.c, but as we now have atabus between the
controller driver and wd, maybe we could use that instead to avoid
growing bus_class_tab[] too much.

Any idea or comment on this ?

--

-- 
Manuel Bouyer <bouyer <at> antioche.eu.org>
     NetBSD: 24 ans d'experience feront toujours la difference
--

Martin Husemann | 9 Oct 00:28 2003
Picon

Re: pciide/atabus changes

On Wed, Oct 08, 2003 at 10:53:33PM +0200, Manuel Bouyer wrote:
> The easy way would be to add all the *ide drivers names to bus_class_tab[]
> in sparc64/sparc64/autoconf.c

Please do that for now, at least for cmdide.

> but as we now have atabus between the
> controller driver and wd, maybe we could use that instead to avoid
> growing bus_class_tab[] too much.

From a very quick glance it's unclear to me why the dev_compatible checks
on PCI subclasses don't fire:

                case PCI_CLASS_MASS_STORAGE:
                        if (PCI_SUBCLASS(pa->pa_class) ==
                                PCI_SUBCLASS_MASS_STORAGE_IDE) {

If this could be made to catch this case, we wouldn't need the table for it;
or are vendors just lacking to fill in proper info here?

Martin

Rafal Boni | 9 Oct 00:52 2003
Picon

Re: Promise ATA in PCI Sparc

In message <200310082023.h98KNgYe018129 <at> ultra.bl.org>, you write: 

-> I've got a Tatung Ultra10, UltraSparc-IIi 360Mhz, OpenBoot 3.10.7
-> SME, but no onboard IDE.
-> 
-> I'm trying to get a Promise Ultra33 card working in the system, but when
-> I try to run the nvramrc from the FAQ, I get an error:

Avoid this card, even if you can figure our the PROM magic to get it to
boot (IIRC, whoever posted that later said that it was causing him all
sorts of other problems).

Barring the PROM issues, both Matt Green and I have had very poor luck
with these cards.  In my case panics under loads where I was using both
channels -- I don't know if it was because of that or simply just a
function of the load... the panic were due to mysterious accesses to
areas of PCI space the driver shoul not have been touching... I never
got much further with it than that and instead bought a Sil680-based
card on Matt's suggestion.  That's been running a RAID1 with a disk
on each channel just peachy.

See the thread here (and tech-kern) a little while ago.

--rafal

----
Rafal Boni                                                     rafal <at> pobox.com
  We are all worms.  But I do believe I am a glowworm.  -- Winston Churchill

(Continue reading)

matthew green | 9 Oct 01:00 2003
Picon

re: Promise ATA in PCI Sparc


i never got booting from the pciide card working but there appear to
be bugs in the promise pciide driver on sparc64 that do not occur on
other pciide cards.  see the port-sparc64 archives of the past couple
of months for details.

.mrg.

Mike Parson | 9 Oct 01:42 2003

Re: Promise ATA in PCI Sparc

On Wed, Oct 08, 2003 at 06:52:30PM -0400, Rafal Boni wrote:
> In message <200310082023.h98KNgYe018129 <at> ultra.bl.org>, you write: 
> 
> -> I've got a Tatung Ultra10, UltraSparc-IIi 360Mhz, OpenBoot 3.10.7
> -> SME, but no onboard IDE.
> -> 
> -> I'm trying to get a Promise Ultra33 card working in the system, but when
> -> I try to run the nvramrc from the FAQ, I get an error:
> 
> Avoid this card, even if you can figure our the PROM magic to get it to
> boot (IIRC, whoever posted that later said that it was causing him all
> sorts of other problems).
> 
> Barring the PROM issues, both Matt Green and I have had very poor luck
> with these cards.  In my case panics under loads where I was using both
> channels -- I don't know if it was because of that or simply just a
> function of the load... the panic were due to mysterious accesses to
> areas of PCI space the driver shoul not have been touching... I never
> got much further with it than that and instead bought a Sil680-based
> card on Matt's suggestion.  That's been running a RAID1 with a disk
> on each channel just peachy.

I'm less concerned with RAID, I'm just wanting to use IDE disks instead
of SCSI.  I also plan on dual-booting, NetBSD and Solaris 9, so,
something that works under both OSes would be ideal.

> See the thread here (and tech-kern) a little while ago.

googling now

(Continue reading)

Takeshi Nakayama | 9 Oct 14:41 2003
Picon

Re: pciide/atabus changes

>>> Martin Husemann <martin <at> duskware.de> wrote

> On Wed, Oct 08, 2003 at 10:53:33PM +0200, Manuel Bouyer wrote:
> > The easy way would be to add all the *ide drivers names to bus_class_tab[]
> > in sparc64/sparc64/autoconf.c
> 
> Please do that for now, at least for cmdide.

aceride, too. Recent low-end models use ALI M5229.

-- Takeshi Nakayama

Manuel Bouyer | 9 Oct 19:55 2003

Re: pciide/atabus changes

On Thu, Oct 09, 2003 at 09:41:41PM +0900, Takeshi Nakayama wrote:
> >>> Martin Husemann <martin <at> duskware.de> wrote
> 
> > On Wed, Oct 08, 2003 at 10:53:33PM +0200, Manuel Bouyer wrote:
> > > The easy way would be to add all the *ide drivers names to bus_class_tab[]
> > > in sparc64/sparc64/autoconf.c
> > 
> > Please do that for now, at least for cmdide.
> 
> aceride, too. Recent low-end models use ALI M5229.

Done, for both

--

-- 
Manuel Bouyer <bouyer <at> antioche.eu.org>
     NetBSD: 24 ans d'experience feront toujours la difference
--


Gmane