From: Unruh on
Martin Gregorie <martin(a)see.sig.for.address.invalid> writes:

>On Thu, 18 Dec 2008 15:46:16 +0000, Geoffrey Clements wrote:

>> That should do it, just was out for any preceding rules which match this
>> device and specify a "last_rule" option.
>>
>> You can use "udevadm test" to see what will happen.

>Thanks for the advice. Permissions are now being set just as I want them.

>The rule ended up as:

>KERNEL=="tty[A-Z]*", GROUP="uucp", MODE="0666", OPTIONS="last_rule"

>in a separate file. I changed the match pattern so the rule only affects
>serial ports and USB-serial adapters. At that point it fixed the USB-
>serial adapter but not the ttyS* ports (and I still can't see why there's
>a difference). After adding the "last_rule" option it fixes the
>permissions on all the serial ports.

>BTW, I think there's a misfeature in the udev pattern matcher:
>"tty[A-Z]*" matches 'tty', 'ttyS0' and 'ttyUSB0' but not 'tty0'.
>IOW its acting as though [A-Z]* means 'match at least one of the letters
>A to Z or the end of the string' rather than the implied (and
>conventional) 'zero or more characters'.

No it says match the letters A-Z and then any remaining letters. It is teh
file matching rules, not the egrep rules that apply ( they are file names).

>Can anybody give me the address of a person or group who should be told
>about it?

About what?
It is working as designed.


>--
>martin@ | Martin Gregorie
>gregorie. | Essex, UK
>org |
From: Martin Gregorie on
On Tue, 16 Dec 2008 07:17:53 +0000, Gordon Henderson wrote:

> Stick one of these:
>
> http://linitx.com/viewproduct.php?prodid=11850
>
> in the PCI slot. If it's longer than a normal PCI slot, it's a PCI-X
> slot, but shorter cards should work just as well. See:
>
The card turned up today and has been duly stuck in the spare PCI slot as
a 4 port setup - no spare holes in the back of the case for the 5th and
6th connectors.

The card's S1 and S2 ports were recognised and assigned as ttyS2 and
ttyS3 but anything else isn't being handled - dmesg sees the next pair of
ports being spotted as a 16550A, then an error is thrown and thats it:

0000:01:08.0: ttyS2 at I/O 0x74c8 (irq = 16) is a 16550A
0000:01:08.0: ttyS3 at I/O 0x74d0 (irq = 16) is a 16550A
Couldn't register serial port 0000:01:08.0: -28

So, now I have 4 serial ports working (two on the mobo and the first two
on this card, but I'd like to get the next two working as well. Is there
a HOWTO for this?

The card came with a mini-CD containing relatively old drivers as C
source but no documentation at all apart from short READMEs for the
drivers. There's no indication on the box as to what model the card might
be either, so without pulling the card out again and reading what's
stencilled on it I'm unsure which driver might suit it.

Is there a generic multi-port driver that comes with kernel 2.6.26 that I
can use or should I compile one off the mini-CD?


--
martin@ | Martin Gregorie
gregorie. | Essex, UK
org |
From: Gordon Henderson on
In article <gigvf7$uj2$3(a)localhost.localdomain>,
Martin Gregorie <martin(a)see.sig.for.address.invalid> wrote:
>On Tue, 16 Dec 2008 07:17:53 +0000, Gordon Henderson wrote:
>
>> Stick one of these:
>>
>> http://linitx.com/viewproduct.php?prodid=11850
>>
>> in the PCI slot. If it's longer than a normal PCI slot, it's a PCI-X
>> slot, but shorter cards should work just as well. See:
>>
>The card turned up today and has been duly stuck in the spare PCI slot as
>a 4 port setup - no spare holes in the back of the case for the 5th and
>6th connectors.
>
>The card's S1 and S2 ports were recognised and assigned as ttyS2 and
>ttyS3 but anything else isn't being handled - dmesg sees the next pair of
>ports being spotted as a 16550A, then an error is thrown and thats it:
>
>0000:01:08.0: ttyS2 at I/O 0x74c8 (irq = 16) is a 16550A
>0000:01:08.0: ttyS3 at I/O 0x74d0 (irq = 16) is a 16550A
>Couldn't register serial port 0000:01:08.0: -28
>
>So, now I have 4 serial ports working (two on the mobo and the first two
>on this card, but I'd like to get the next two working as well. Is there
>a HOWTO for this?
>
>The card came with a mini-CD containing relatively old drivers as C
>source but no documentation at all apart from short READMEs for the
>drivers. There's no indication on the box as to what model the card might
>be either, so without pulling the card out again and reading what's
>stencilled on it I'm unsure which driver might suit it.
>
>Is there a generic multi-port driver that comes with kernel 2.6.26 that I
>can use or should I compile one off the mini-CD?

There is one in the kernel, and it defaults to 4 serial ports.

So, you may need to re-compile the kernel to fix it, or if using modules,
there may be a module load option to increase the number.

The module is called "8250"

Gordon
From: Will Kemp on
Martin Gregorie wrote:
> On Tue, 16 Dec 2008 07:17:53 +0000, Gordon Henderson wrote:
>
>> Stick one of these:
>>
>> http://linitx.com/viewproduct.php?prodid=11850
>>
>> in the PCI slot. If it's longer than a normal PCI slot, it's a PCI-X
>> slot, but shorter cards should work just as well. See:
>>
> The card turned up today and has been duly stuck in the spare PCI slot as
> a 4 port setup - no spare holes in the back of the case for the 5th and
> 6th connectors.
>
> The card's S1 and S2 ports were recognised and assigned as ttyS2 and
> ttyS3 but anything else isn't being handled - dmesg sees the next pair of
> ports being spotted as a 16550A, then an error is thrown and thats it:
>
> 0000:01:08.0: ttyS2 at I/O 0x74c8 (irq = 16) is a 16550A
> 0000:01:08.0: ttyS3 at I/O 0x74d0 (irq = 16) is a 16550A
> Couldn't register serial port 0000:01:08.0: -28
>
> So, now I have 4 serial ports working (two on the mobo and the first two
> on this card, but I'd like to get the next two working as well. Is there
> a HOWTO for this?
>
> The card came with a mini-CD containing relatively old drivers as C
> source but no documentation at all apart from short READMEs for the
> drivers. There's no indication on the box as to what model the card might
> be either, so without pulling the card out again and reading what's
> stencilled on it I'm unsure which driver might suit it.
>
> Is there a generic multi-port driver that comes with kernel 2.6.26 that I
> can use or should I compile one off the mini-CD?

You don't need to worry about drivers for 16550 UARTs - that hardware's
about as standard as you can get! Obsolete, maybe, but standard! ;-)

As Gordon said, by default the kernel is configured to only register 4
serial ports, and this can be changed by rebuilding the kernel. However,
it seems you can change it with a boot time parameter too. From the
kernel config documentation:

Number of 8250/16550 serial ports to register at runtime
(SERIAL_8250_RUNTIME_UARTS)

Set this to the maximum number of serial ports you want
the kernel to register at boot time. This can be overridden
with the module parameter "nr_uarts", or boot-time parameter
8250.nr_uarts


From kernel docs - kernel-parameters.txt:
nr_uarts= [SERIAL] maximum number of UARTs to be registered.


So it looks like you can just add "nr_uarts=6" to the kernel parameters
in grub.conf - e.g.,

kernel /boot/vmlinuz-2.6.27.7-134.fc10.x86_64 ro root=/dev/sda1 rhgb
quiet nr_uarts=6



--
http://SnapAndScribble.com
From: Unruh on
Will Kemp <will(a)xxxx.swaggie.net> writes:

>Martin Gregorie wrote:
>> On Tue, 16 Dec 2008 07:17:53 +0000, Gordon Henderson wrote:
>>
>>> Stick one of these:
>>>
>>> http://linitx.com/viewproduct.php?prodid=11850
>>>
>>> in the PCI slot. If it's longer than a normal PCI slot, it's a PCI-X
>>> slot, but shorter cards should work just as well. See:
>>>
>> The card turned up today and has been duly stuck in the spare PCI slot as
>> a 4 port setup - no spare holes in the back of the case for the 5th and
>> 6th connectors.
>>
>> The card's S1 and S2 ports were recognised and assigned as ttyS2 and
>> ttyS3 but anything else isn't being handled - dmesg sees the next pair of
>> ports being spotted as a 16550A, then an error is thrown and thats it:
>>
>> 0000:01:08.0: ttyS2 at I/O 0x74c8 (irq = 16) is a 16550A
>> 0000:01:08.0: ttyS3 at I/O 0x74d0 (irq = 16) is a 16550A
>> Couldn't register serial port 0000:01:08.0: -28
>>
>> So, now I have 4 serial ports working (two on the mobo and the first two
>> on this card, but I'd like to get the next two working as well. Is there
>> a HOWTO for this?
>>
>> The card came with a mini-CD containing relatively old drivers as C
>> source but no documentation at all apart from short READMEs for the
>> drivers. There's no indication on the box as to what model the card might
>> be either, so without pulling the card out again and reading what's
>> stencilled on it I'm unsure which driver might suit it.
>>
>> Is there a generic multi-port driver that comes with kernel 2.6.26 that I
>> can use or should I compile one off the mini-CD?

>You don't need to worry about drivers for 16550 UARTs - that hardware's
>about as standard as you can get! Obsolete, maybe, but standard! ;-)

>As Gordon said, by default the kernel is configured to only register 4
>serial ports, and this can be changed by rebuilding the kernel. However,
>it seems you can change it with a boot time parameter too. From the
>kernel config documentation:

> Number of 8250/16550 serial ports to register at runtime
> (SERIAL_8250_RUNTIME_UARTS)

> Set this to the maximum number of serial ports you want
> the kernel to register at boot time. This can be overridden
> with the module parameter "nr_uarts", or boot-time parameter
> 8250.nr_uarts


> From kernel docs - kernel-parameters.txt:
> nr_uarts= [SERIAL] maximum number of UARTs to be registered.


>So it looks like you can just add "nr_uarts=6" to the kernel parameters
>in grub.conf - e.g.,

From reading the above, it would seem you want to use
8250.nr_uarts=6 as a kernel parameter, or if you are loading the module as
a module, (Ie not compiled into the kernel) then you would use the nr_uarts
parameter to the module, not the kernel.



>kernel /boot/vmlinuz-2.6.27.7-134.fc10.x86_64 ro root=/dev/sda1 rhgb
>quiet nr_uarts=6



>--
>http://SnapAndScribble.com
First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4 5 6 7 8
Prev: World domination foiled again...
Next: Missing?