[parisc-linux] more serial console problems on D310

[parisc-linux] more serial console problems on D310

[djweis]

I grabbed the 20010704 kernel source and net booted my d310 with it
combined with the ramdisk from the 0.9.2 install image. I start with the
serial console on the serial connector labeled "serial console". Output
only comes through there until the kernel starts, then I have to switch
over to the other serial port on the machine to see the rest of the boot
messages. They come quickly until the root filesystem is mounted. It looks
like I lose interrupts after that because everything is written in blocks
of 16 characters with long (15 second?) pauses in between blocks. After
getting this message:
VFS: Mounted root (ext2 filesystem).
serial console detected.  Disabling virtual terminals.

the cursor just tabs along the screen by itself without printing
anything. 

Any ideas? Why is the first serial port not being detected/used?

dave


-- 
Dave Weis             "I believe there are more instances of the abridgement
djweis@sjdjweis.com   of the freedom of the people by gradual and silent
                      encroachments of those in power than by violent 
                      and sudden usurpations."- James Madison

Re: [parisc-linux] more serial console problems on D310

[Grant Grundler]

djweis@sjdjweis.com wrote:
> I grabbed the 20010704 kernel source and net booted my d310 with it
> combined with the ramdisk from the 0.9.2 install image. I start with the
> serial console on the serial connector labeled "serial console". Output
> only comes through there until the kernel starts, then I have to switch
> over to the other serial port on the machine to see the rest of the boot
> messages.

I look at the HW database and it seems the D310 has two LASI (core_adapter)
chips (and WAX, no PCI). I suspect the detection of serial ports misnumbers
them (ie parisc-linux finds them in the wrong order) or we only properly
handle one of the two LASI's.

>  They come quickly until the root filesystem is mounted. It looks
> like I lose interrupts after that because everything is written in blocks
> of 16 characters with long (15 second?) pauses in between blocks.

Probably not getting interrupts in the first place. The initial
console handling is done with PDC calls - ie polled IO.

> After
> getting this message:
> VFS: Mounted root (ext2 filesystem).
> serial console detected.  Disabling virtual terminals.
> 
> the cursor just tabs along the screen by itself without printing
> anything. 

> Any ideas? Why is the first serial port not being detected/used?

Can you post full console output? (or otherwise make it available)
Interested folks can then start guessing what might be the problem.

grant

Grant Grundler
parisc-linux {PCI|IOMMU|SMP} hacker
+1.408.447.7253

Re: [parisc-linux] more serial console problems on D310

[Kirk Bresniker]

| I look at the HW database and it seems the D310 has two LASI (core_adapter)
| chips (and WAX, no PCI). I suspect the detection of serial ports misnumbers
| them (ie parisc-linux finds them in the wrong order) or we only properly
| handle one of the two LASI's.

It may seem like that in the HW database, but there is never more than one
LASI in a d-class system, the LASI is on the main system board, so it's always
there regardless of which processor asssembly gives the d-class its flavor,
and no processor board or add-in card ever contained a second LASI.

The serial port labeled console is attached to the LASI rs-232 unless an
access port card is plugged into the access port capable slot. If an 
access port card is present, then the RS-232 signals from it are routed
to the console connector instead.

Check the system to see if there is an AP card. If there is, then you 
might need to remove it. When I played with some d-classes very early on 
in the port, I had to take the AP card out.  The inside door of your 
d-class system should have a nice diagram showing you how to find the 
AP card. It's an HSC card with two small extra connectors and on the
bulkhead is an LED, a small switch, an serial connector and a tiny
modem connector.

KMB
--
+======================================================================+
|       Kirk Bresniker          (916) 748-2393                         |
|       8000 Foothills Blvd                                            |
|       Roseville, CA 95747-5731                                       |
|       kirkb@rose.hp.com                                              |

Re: [parisc-linux] more serial console problems on D310

[Grant Grundler]

Kirk Bresniker wrote:
> | I look at the HW database and it seems the D310 has two LASI (core_adapter)
> | chips (and WAX, no PCI). I suspect the detection of serial ports misnumbers
> | them (ie parisc-linux finds them in the wrong order) or we only properly
> | handle one of the two LASI's.
> 
> It may seem like that in the HW database, but there is never more than one
> LASI in a d-class system, the LASI is on the main system board, so it's
> always there regardless of which processor asssembly gives the d-class its
> flavor, and no processor board or add-in card ever contained a second LASI.

Kirk,
You are right. I misread the ioscan output.

Here are the HW database contents in a more readable form:

type	driver		HWpath	Description
------	-------------	------	--------------------------
ba	core_adapter	16	Core I/O Adapter
ext_bus	CentIf		16/0	Built-in Parallel Interface
tty	asio0		16/4	Built-in RS-232C
ext_bus	c720		16/5	Built-in SCSI
...
lan	lan2		16/6	Built-in LAN
ps2	ps2		16/7	Built-in Keyboard/Mouse
ba	core_adapter	20	Core I/O Adapter
tty	asio0		20/2	Built-in RS-232C
ba	eisa_adapter	20/5	EISA Bus Adapter
lan	btlan0		20/5/8	EISA card INP0500
...

I had assumed "core_adapter" meant LASI.
The second "Core I/O Adapter" at "20" must be WAX.

> The serial port labeled console is attached to the LASI rs-232 unless an
> access port card is plugged into the access port capable slot.
...

Ok. We still have an ordering issue then between WAX/LASI and how
tty's are numbered/discovered. That's in addition to the problem
of interrupts not getting delivered for the serial port on WAX at "20/2".

grant

Grant Grundler
parisc-linux {PCI|IOMMU|SMP} hacker
+1.408.447.7253

Re: [parisc-linux] more serial console problems on D310

[Dave Weis]

On Thu, 9 Aug 2001, Kirk Bresniker wrote:

> Check the system to see if there is an AP card. If there is, then you 
> might need to remove it. When I played with some d-classes very early on 
> in the port, I had to take the AP card out.  The inside door of your 
> d-class system should have a nice diagram showing you how to find the 
> AP card. It's an HSC card with two small extra connectors and on the
> bulkhead is an LED, a small switch, an serial connector and a tiny
> modem connector.

Yes, it does have one of those cards. I'll yank it tonite and see what
happens. It also has an 8 port serial card, would that make a
difference? I'll yank it also, anyway.

thanks
dave

-- 
Dave Weis             "I believe there are more instances of the abridgement
djweis@sjdjweis.com   of the freedom of the people by gradual and silent
                      encroachments of those in power than by violent 
                      and sudden usurpations."- James Madison

Re: [parisc-linux] more serial console problems on D310

[Dave Weis]

On Thu, 9 Aug 2001, Grant Grundler wrote:

> Can you post full console output? (or otherwise make it available)
> Interested folks can then start guessing what might be the problem.

After pulling the AP card and the 8 port serial card, I have a debian
install going. I did use the serial port labeled console on the back of
the machine. I'll report back if it finishes.

dave

-- 
Dave Weis             "I believe there are more instances of the abridgement
djweis@sjdjweis.com   of the freedom of the people by gradual and silent
                      encroachments of those in power than by violent 
                      and sudden usurpations."- James Madison