Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 20 Sep 2000 22:19:49 -0400 (EDT)
From:      Antoine Beaupre <beaupran@IRO.UMontreal.CA>
To:        Mark Ovens <marko@freebsd.org>
Cc:        freebsd-questions@freebsd.org
Subject:   Re: PnP modem not recognized at boot, but by pnpinfo(8)
Message-ID:  <Pine.LNX.4.21.0009202150110.12634-100000@phobos.IRO.UMontreal.CA>
In-Reply-To: <20000921013138.G1612@parish>

next in thread | previous in thread | raw e-mail | index | archive | help
Here we go again.

On Thu, 21 Sep 2000, Mark Ovens wrote:

> On Wed, Sep 20, 2000 at 07:50:59PM -0400, Antoine Beaupre wrote:
> > Hi. These are the latest news about my modem / sio comm problems.
> > Included here is a sample dmesg and pnpinfo output, and my kernel
> > configuration. I hope this is not overkill.
> 
> No, the more info the better.

Cool. I always felt strangely shy of sending 20K or 40K of config files
and debug messages to mailing lists, but then again, people send
HTML. Just joking... :)
 
> [snip]
> 
> > These are my current [possible] BIOS settings (only the relevant are
> > included here):
> > 
> > -- PnP/PCI Configuration --
> > PnP OS Installed:			NO [YES]
> > Resources controled by:		Automatic [Manual*]
> > Reset configuration data:	Disabled [Enabled**]
> > 
> > [* The following appears only when "resources are controlled by "Manual":
> > 
> > IRQ X: 		 PCI/PnP [Legacy/ISA]
> > 
> > for X in 3 to 15]
> > 
> > [** This setting is a toggle which when "enabled" is automatically
> > toggled back to "Disabled" with any power cycle. I don't know its use is]
> 
> That looks familiar! That is exactly what I see. Is this by any chance
> an Award 4.5x BIOS? The "Reset configuration data" entry is to do with
> forcing the BIOS to update it's "resources table". I'm not an expert
> but the way I believe it works is that if your modem for example is
> allocated IRQ10 0x2f8 the first time it is found then the BIOS
> remembers this and allocates the same resources on subsequent boots
> (even if you add more hardware). If you set it to "Enabled" you should
> see something like "Updating ECSD......Done" at the end of the BIOS
> messages (just before the OS starts). I see the same message if
> Windows finds new hardware or I change the resource settings (this is
> what I was meaning about Windows f*****g with the BIOS settings).

Yep. Award 4.51PG BIOS. 

The Updating ECSD message appears when booting after I hit the
Reset.. toggle. Strangely , it also appears the second time I reboot, but
not the third. :) Yay.

All this, of course absolutely not affectingthe probe of my
modem. :) Always on good old IRQ3 (argh!).

> > -- Integrated peripherals --
> > [Both COM1 and COM2 have the same set of possible settings]
> > Onboard serial port 1:	3F8/IRQ4 [only way to have the serial mouse working
> > on FreeBSD]
> > Onboard serial port 2: [one of:
> > 	Disabled | Auto
> > 	3F8/IRQ4 | 3E8/IRQ4 | 2E8/IRQ4 |2F8/IRQ3
> > 	See below]
> > [this only appears if serial port 2 is activated]
> > Ur Mode: standard | IrDA 1.0 | Ask IR | MIR 0.57M | MIR 1.15M | FIR
> > Duplex mode: Full | Half
> > 
> > I haven't tried fiddling with the Ur Mode setting as I really don't know
> > what it is.
> 
> I don't have that. It appears to be connected with Infra-red devices
> (IrDA).

That's what I tought. There is probably no IR devices on my board but the
software is still on chip.
 
> > Here are the experiments I made with various combinations of COM2 settings
> > and kernel configs:
> > 
> > COM2 to Disbled or Auto
> > =======================
> > 
> > When I set COM2 to Disbled or Auto, I get the following boot messages:
> > 
> > sio0 at port 0x3f8-0x3ff irq 4 flags 0x10 on isa0
> > sio0: type 16550A
> > sio1: configured irq 3 not in bitmap of probed irqs 0
> > sio2: configured irq 3 not in bitmap of probed irqs 0
> > sio2: <CIRRUS LOGIC PnP V34 MODEM> at port 0x2f8-0x2ff irq 3 on isa0
> > sio2: type 16550A
> 
> [kernel config entries moved to here for clarity]
> 
> # Serial (COM) ports
> device          sio0    at isa? port IO_COM1 flags 0x10 irq 4
> device          sio1    at isa? port IO_COM2 irq 3
> #device         sio2    at isa? port IO_COM3 irq 5
> #device         sio3    at isa? disable port IO_COM4 irq 9
> 
> OK, the "configured irq 3 not in bitmap of probed irqs 0" appear to be
> caused by the conflict that you have because your config file
> specifies IRQ3 for sio1 but the BIOS has set up the modem on IRQ3. Try
> changing sio1 to IRQ5 in the kernel config file (and building yet
> another kernel).

#33 now. :) I always tought about writing a GUI for that, maybe if the
config file syntax stabalize, I'll start hacking into that. :))

Anyways, I have some odd results regarding this. I tried to recompile with
sio1 on irq 9, obviouslyy not a good idea since my SCSI adapter is on irq
9... So I tried to move it back to irq5, but surprise! after a whole new
"config SHALL && cd ../../compile/SHALL && make depend && make && make
install" cycle with the sio1's irq tweaked to 5, at boot, it still gets
listed ('ls' in userconfig (-c)) as irq 9!!!

> [snip]
> 
> > 
> > The problem is the following now. When connected to the serial device
> > (sio2) associated with the modem with ppp(4) or cu(1), the caracters
> > do not echo properly to the terminal. A character must be sent in
> > order to have one sent back. So, for example, if you type "AT" only
> > "A" appears. You must type another key to have the "T" sent back from
> > the modem.
> 
> Aha! a classic case of IRQ conflicts. 

Or so I've been told. But with what? The sio1? It seems so. But then, the
kernel should switch the modem to another irq! I can't move it away from
irq3, whatever I try! (get ready for the classical "Windoze can do
it!?") On windos, it's on irq 10 and port 0x3e8 - 0x3fe. Shouldn't the
fbsd kernel be able to do the same acrobatics?

And if I "resolve" the IRQ conflict (either disable the sio1 or the BIOS
onboard port setting), nothing is solved?

I feel that the modem should be on another IRQ. When I was using FBSD
2.x/3.x, it was on IRQ5 and my sio1 on IRQ3. And the sky was blue. :)

And I wasn't using that stupid hyperterminal.
 
> > Same goes when connected to a remote machine, except the data get sent by
> > bigger packets (i.e. more characters get sent back when a key is pressed).
> > 
> > <Who again?> suggested that this might be an IRQ conflict problem. I
> > tought first that this might be related to the message:
> > 
> > sio1: configured irq 3 not in bitmap of probed irqs 0
> > 
> > (i.e. the SIO1 port is "configured" on irq 3, this means there's a
> > COM2 port on irq 3, right?)
> > 
> > So I figured that this might be my COM2 bugging off the modem. I tried to
> > disable COM2 in the bios and got the same error messages and modem
> > behavior. I tried disabling the sio1 device, but it only removes the sio1
> > error message. I also tried to remove the sio1 driver from the kernel,
> > nothing would solve the problem. So I don't think this is the problem.
> > 
> > COM2 Set to 2F8/IRQ3
> > ====================
> > 
> > Setting the "Onboard serial port 2" setting to 2F8/IRQ3 gets the sio1
> > probe to run smoothly.
> > 
> > sio1 at port 0x2f8-0x2ff irq 3 on isa0
> > sio1: type 16550A
> > 
> > But then the modem is simply not detected (unknown: <CIRRUS...>). However,
> > if I disable the sio1 device upon boot, the modem is normally detected,
> > without any error message (standard 2-liner). However, impossible to get
> > communication with modem.
> > 
> 
> Ah, but did it find the modem sio1 as when you disabled sio1? If so
> then you need to change the ``set device /dev/cuaa2'' line in
> /etc/ppp/ppp.conf to ``set device /dev/cuaa1'' (and make sure
> /dev/cuaa1 exists). If the modem is found as sio1 in this situation and
> changing ppp.conf gets it working then I'd leave it like that (if you
> can live without the second serial port).

Ah-ah! Yes, it is _still_ on sio2!! Strange eh? I already made the check,
I should have made this clearer. The only thing I always managed to change
was to remove the sio1 message altogether (or its warning), but never
affecting the darn sio2 probe apart from making it fail altogether
(unknown: <CIRRUS...>).
 
> BTW, try to avoid references to COM1 etc. unless you are referring to
> Windows. These are DOS device names and it can get rather confusing
> (partly because they are numbered from 1 not 0).

Yes. It was destinated as an abbreviation for IRQ3/... IRQ4/... But I
agree. The kernel config file doesn't, however:

device		sio1	at isa? port IO_COM2 irq 3
                                     ^^^^^^^
eh-eh.. ;)

> As you have seen,
> IRQs and addresses are not cast in tablets of stone anymore (as they
> were in the DOS days) so saying "COM1 is IRQ4 0x3f8" is not wholly
> accurate.

Yah... I guess it was a kind of nostalgia... (this is getting weird. :)
  
> > ===
> > 
> > Also note the odd message at boot:
> > 
> > isa0: too many dependant configs (8)
> > isa0: too many dependant configs (8)
> 
> Don't worry about this. I get it too, it's a known and harmless
> message.

But does it mean that it's not searching further in the configs for more
IRQ/port settings? Just a shot in the dark here...

> > isa0: unexpected small tag 14
> 
> Not sure about this one though.

Hey, who knows! Doesn't mean much in itself. I would like to be a kernel
developper or at least have some documentation on these messages.

I think they're quite important to know about. Probably that browsing the
source could do it... Yeah. Viva the www.freebsd.org search engine. In
/isa/pnpparse.c line 233... It seems that a resource doesn't have a known
"tag" ID. Doesn't seem very harmful. If it's just a tag that skipped, as I
could understand from the source. But then again, if the whole modem
config is skipped then maybe that's why it can't switch to another irq
than 3???

I'm so tired. But the problem is that I don't have much of a choice
here. I don't feel the 3.x line is going to be here that long, from
what I can tell from the move from 2.2 to 3.x to 4.x... (i.e. 2.2 is dead,
and 3.x is only little patching here and there).

I have to get this working.

And I have to sleep. :)

A.

Si l'image donne l'illusion de savoir
C'est que l'adage pretend que pour croire,
L'important ne serait que de voir

Lofofora



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.LNX.4.21.0009202150110.12634-100000>