devfs & usb

devfs & usb

Post by Gerald Pollac » Sun, 24 Feb 2002 22:59:33



I'm  attempting to synch my Palm M105 (serial) to my thinkpad X21 (USB),
via a keyspan serial-usb pda adaptor. I've been able to load the
appropriate modules:

keyspan_pda            10448   0  (unused)
usbserial              18288   0  [keyspan_pda visor]
usbcore                50752   1  [keyspan_pda visor usbserial uhci usb-storage]

But the usb port is not being created:

   Unable to bind to port /dev/ttyUSB1
   pi_bind: No such file or directory

/etc/devfsd.conf contains the following:
REGISTER        usb/tts/[13579] EXECUTE /etc/dynamic/scripts/visor.script add $devpath
UNREGISTER      usb/tts/[13579] EXECUTE /etc/dynamic/scripts/visor.script del $devpath

If someone can tell me how to get this working, I'd be vey grateful.
Thanks,
Jerry Pollack
Mandrake 8.1, kernel 2.4.8

 
 
 

devfs & usb

Post by Gerald Pollac » Mon, 25 Feb 2002 04:13:16



> NOTE: This message was sent thru a mail2news gateway. No effort was made
> to verify the identity of the sender.
> --------------------------------------------------------



>> pilot-xfer -p /dev/ttyUSB1 -l

>>    Unable to bind to port /dev/ttyUSB1
>>    pi_bind: No such file or directory

> It tells you that the appropriate entry in the directory /dev cannot be
> found. Perhaps you (or your installation program) forgot to create it?

> [ From /your/path/to/src/linux/Documentation/devices.txt ] 188 char
>   USB serial converters
>                   0 = /dev/ttyUSB0      First USB serial converter 1 =
>                   /dev/ttyUSB1      Second USB serial converter

> You may need to execute (as root):
>    for i in 0 1 ; do
>            mknod /dev/ttyUSB$i c 188 $i ; chmod go+rw /dev/ttyUSB$i
>    done

Thanks, but I thought that devfsd is supposed to create devices on the
fly; they don't exist until they're needed. Or am I mistaken?

 
 
 

devfs & usb

Post by Gerald Pollac » Mon, 25 Feb 2002 04:20:31



> On Sat, 23 Feb 2002 08:59:33 -0500 in comp.os.linux.misc, Gerald Pollack


>> pilot-xfer -p /dev/ttyUSB1 -l

>>    Unable to bind to port /dev/ttyUSB1
>>    pi_bind: No such file or directory

> If you are using devfs, the device names are different from the
> traditional Linux device names.  Just look in /dev and compare with (for
> example) the linux kernel Documentation/devices.txt.  You likely won't
> find /dev/ttyUSB1 anywhere in your /dev directory, unless you disable
> devfs, use a traditional /dev directory and create the device file.

> It should not be too hard to examine the /dev hierarchy to find the
> correct device name, then use that.  If you have difficulties doing
> this, perhaps post again to see if anyone more familiar with devfs can
> help.

Thanks for the reply, but according to the kernel documentation,
/dev/ttyUSB0 and /dev/ttyUSB1 are the correct names. These don't exist on
my system. I was under the impression that they should be created by
devfsd when needed; is that not correct?
 
 
 

devfs & usb

Post by Molchu » Mon, 25 Feb 2002 05:51:57




>>On Sat, 23 Feb 2002 08:59:33 -0500 in comp.os.linux.misc, Gerald Pollack


>>>pilot-xfer -p /dev/ttyUSB1 -l

>>>   Unable to bind to port /dev/ttyUSB1
>>>   pi_bind: No such file or directory

>>If you are using devfs, the device names are different from the
>>traditional Linux device names.  Just look in /dev and compare with (for
>>example) the linux kernel Documentation/devices.txt.  You likely won't
>>find /dev/ttyUSB1 anywhere in your /dev directory, unless you disable
>>devfs, use a traditional /dev directory and create the device file.

>>It should not be too hard to examine the /dev hierarchy to find the
>>correct device name, then use that.  If you have difficulties doing
>>this, perhaps post again to see if anyone more familiar with devfs can
>>help.

> Thanks for the reply, but according to the kernel documentation,
> /dev/ttyUSB0 and /dev/ttyUSB1 are the correct names. These don't exist on
> my system. I was under the impression that they should be created by
> devfsd when needed; is that not correct?

Yes they should be created but the names are different. It's very easy
to check, just examine your /dev entries before loading your module and
then examine them after you load the module, I bet you'll see some extra
device entries there - they're the ones you need. That's the beauty of
devfs it doesn't have hundreds of entries and the names are quite
logical, so just put some effort in it :-) it'll take just a couple of
minutes much less then waiting for reply to your post.
An other thing to remember is that even if you find the right device
(I'm sure you will) your program, the one your trying to use with this
device might not know anything about the new names and still try to
access the device by it's old name, so you will need to tell your
program to use the right  name. I don't know anything about palms but
it's pretty much the same, there are a couple of ways you can tell your
program to use the right name. First, by examining the config files,
where you can find an entry which points the application to the right
name - edit it. Second, you can do it during compile time by editing the
Makefile or some other config.h file (info should be in the README).
I'm sure there're other ways but the ones I told you about are pretty
common. And don't forget some programs just refuse to be configured to
use different device 'cos the name of the device is hard coded and
buried somewhere deep inside  the code jungle :-). I think that's the
main reason why devfs is not widespread yet - compatibility with old
programs.

Good luck.

Ruslan.

 
 
 

devfs & usb

Post by Gerald Pollac » Mon, 25 Feb 2002 22:57:02



> Yes they should be created but the names are different. It's very easy
> to check, just examine your /dev entries before loading your module and
> then examine them after you load the module, I bet you'll see some extra
> device entries there - they're the ones you need. That's the beauty of
> devfs it doesn't have hundreds of entries and the names are quite
> logical, so just put some effort in it :-) it'll take just a couple of
> minutes much less then waiting for reply to your post.

Thanks for the hint.The only difference in /dev that I can detect, either
upon loading the module or on attempting to hotsync (see below) is in
/dev/ptmx; its time-stamp changes. Any idea what that is, or how to use
it?
No new devices or links are added to /dev (I've tested for this by doing
'ls /dev > somefile' before and after loading the module, and hotsyncing,
and then using diff to look for changes). The significance of the hotsync
attempt is that the Palm OS destktop HOWTO says that the links in /dev are
not set up properly until the hotsync is initiated by the palm device. It
also states that there are some problems with earlier kernels, so I've
upgraded to 2.4.17, but the problem persists. According to the HOWTO, when
I attempt to hotsync, devfsd should create a link, in /dev, to
/dev/tts/usb/1, but the latter does not exist on my system. I've tried
creating it manually, with mknod, but it's still not recognized when I
attempt to communicate between palm and pc. (So, you see, I have put some
effort into this). My guess is that something's wrong in the config files
that control the behavior of devfsd (and unfortunately I have neither the
time nor inclination to become an expert on this). I normally sync via the
serial port on my laptop's docking station, and that works fine.I had
hoped to be able to get the usb sync working for an upcoming trip, when I
won't have the docking station with me. But, since usb-syncing works fine
under Windows, I may just have to boot Windows every couple of days
(something I almost never do).

Thanks,
Jerry

 
 
 

devfs & usb

Post by Peter T. Breue » Mon, 25 Feb 2002 23:39:13




>> Yes they should be created but the names are different. It's very easy
>> to check, just examine your /dev entries before loading your module and
>> then examine them after you load the module, I bet you'll see some extra
>> device entries there - they're the ones you need. That's the beauty of
>> devfs it doesn't have hundreds of entries and the names are quite
>> logical, so just put some effort in it :-) it'll take just a couple of
>> minutes much less then waiting for reply to your post.
> Thanks for the hint.The only difference in /dev that I can detect, either
> upon loading the module or on attempting to hotsync (see below) is in
> /dev/ptmx; its time-stamp changes. Any idea what that is, or how to use
> it?

Look elsewhere.

As to /dev/ptmx, I have a perfectly good idea what it is, and so do
you! Read devices.txt in the kernel source directory. It's the
pseudo-tty "multiplexer" (i.e. control device).

Quote:> No new devices or links are added to /dev (I've tested for this by doing
> 'ls /dev > somefile' before and after loading the module, and hotsyncing,

This is not correct as a test. You seem unaware of the possibility that
subdirs will change! Why?

Peter

 
 
 

devfs & usb

Post by Molchu » Tue, 26 Feb 2002 01:52:51




>>Yes they should be created but the names are different. It's very easy
>>to check, just examine your /dev entries before loading your module and
>>then examine them after you load the module, I bet you'll see some extra
>>device entries there - they're the ones you need. That's the beauty of
>>devfs it doesn't have hundreds of entries and the names are quite
>>logical, so just put some effort in it :-) it'll take just a couple of
>>minutes much less then waiting for reply to your post.

> Thanks for the hint.The only difference in /dev that I can detect, either
> upon loading the module or on attempting to hotsync (see below) is in
> /dev/ptmx; its time-stamp changes. Any idea what that is, or how to use
> it?
> No new devices or links are added to /dev (I've tested for this by doing
> 'ls /dev > somefile' before and after loading the module, and hotsyncing,
> and then using diff to look for changes).

Devfs groups everything nicely in subdirs. Use 'ls -l /dev' to find out
which one is a dir and which one is a file (I can't believe I'm telling
you how to use 'ls' :) ) then look deeper in subdirs, just don't waist
your time by looking in some totally unrelated dirs (although it'll be
beneficial to know what in them for future) look in the dirs that
contain character devices. As a last resort get an X server and start
clicking the mouse :).

 The significance of the hotsync

Quote:> attempt is that the Palm OS destktop HOWTO says that the links in /dev are
> not set up properly until the hotsync is initiated by the palm device. It
> also states that there are some problems with earlier kernels, so I've
> upgraded to 2.4.17, but the problem persists. According to the HOWTO, when
> I attempt to hotsync, devfsd should create a link, in /dev, to
> /dev/tts/usb/1, but the latter does not exist on my system.

Maybe you're missing some other modules like 'usbcore' or 'uhci'. How do
you load your modules, what do you use 'modprobe' or 'insmod' the former
will pull all the dependent modules as well the later will not.

 I've tried

Quote:> creating it manually, with mknod, but it's still not recognized when I

Why would you use mknod with devfs? DON'T. Your entries are created by
module itself upon loading (and not by devfsd).

Quote:> attempt to communicate between palm and pc. (So, you see, I have put some
> effort into this). My guess is that something's wrong in the config files
> that control the behavior of devfsd

Devfsd has got nothing to do with it. Entries are created by modules
with or without devfsd. Devfsd just solves the compatibility issue by
providing the old names for the progs that devfs unaware (well, it does
a couple of other things like demand loading too).

(and unfortunately I have neither the

Quote:> time nor inclination to become an expert on this).

I don't want to be rude but I think you should not show things like that
in a Linux NG 'cos the people who help you here put lots of effort and
hard work to get the knowledge they have, they will not help you next
time. People giving you an advice here 'cos they expect you to help them
next time in something they won't know. (That's how Linux community
works :) )

 I normally sync via the

Quote:> serial port on my laptop's docking station, and that works fine.I had
> hoped to be able to get the usb sync working for an upcoming trip, when I
> won't have the docking station with me. But, since usb-syncing works fine
> under Windows, I may just have to boot Windows every couple of days
> (something I almost never do).

If you feel more comfortable with win* use it. Linux will force you to
be an expert with time whether you want it or not if you're not prepared
for it give it up now.

Good luck.

Ruslan.

- Show quoted text -

Quote:

> Thanks,
> Jerry

 
 
 

devfs & usb

Post by Gerald Pollac » Wed, 27 Feb 2002 10:50:11


Quote:> Devfs groups everything nicely in subdirs. Use 'ls -l /dev' to find out
> which one is a dir and which one is a file (I can't believe I'm telling
> you how to use 'ls' :) ) then look deeper in subdirs, just don't waist
> your time by looking in some totally unrelated dirs (although it'll be
> beneficial to know what in them for future) look in the dirs that
> contain character devices. As a last resort get an X server and start
> clicking the mouse :).

No new devices are created in subdirectories either; the only changes are
to timestamps (to /dev/ptmx and /dev/pts/1).
Quote:

> Maybe you're missing some other modules like 'usbcore' or 'uhci'. How do
> you load your modules, what do you use 'modprobe' or 'insmod' the former
> will pull all the dependent modules as well the later will not.

I've loaded the modules with modprobe, and all seem to be installed
(uscore, usbserial)
Quote:

> Devfsd has got nothing to do with it. Entries are created by modules
> with or without devfsd. Devfsd just solves the compatibility issue by
> providing the old names for the progs that devfs unaware (well, it does
> a couple of other things like demand loading too).

Thanks very much for the above; this prompted me to look further into the
module I'm loading, and I think it may be the key. I have been assuming
that the keyspan adaptor (USA-19Qi) is linux-compatible, because that's
what it says clearly on the packaging (compatible with kernels >= 2.4.1).
But keyspan's website says that a linux driver should be available "soon".

Thanks again for your help.Your comments have been productive and
informative (unlike some of the others that have been posted).

 
 
 

devfs & usb

Post by Gerald Pollac » Wed, 27 Feb 2002 20:28:09



> On Sun, 24 Feb 2002 15:39:13 +0100 in comp.os.linux.misc, Peter T.

>>> No new devices or links are added to /dev (I've tested for this by
>>> doing 'ls /dev > somefile' before and after loading the module, and
>>> hotsyncing,

>> This is not correct as a test. You seem unaware of the possibility that
>> subdirs will change! Why?

> I missed this, perhaps he skimmed my instructions.  Specifically, the
> -1R options were suggested:

> ls -1R

I did the test after reading your first post (response to AARG!), which
didn't mention the -R switch. I've since repeated the test, using -R
(thanks  for the clarification), and that confirms that nothing has been
created. I've also since found out that, despite package labeling, the
module I'm' using, keyspan_pda, may in fact not support the device I'm
attempting to use, a keyspan USA-19Qi usb->serial adaptor. I don't know
whether this is the root cause, but I'll look in to it.
Quote:

> This will list all files and subdirectories in the hierarchy and will
> make clear where and what files are created.  The -R option is key.

> Perhaps in the time taken waiting for replies, you could have searched
> the entire devfs hierarchy by hand and had results by now.

Huh? as revealed by ls -lR /dev, my /dev tree has 1441 entries, a bit
much to scan by eye without missing something. This
system has been upgraded at least once, and maybe twice (Mandrake
8.0->8.1, possibly also 7.2->8.1); perhaps there are hangers-on from the
earlier installs?
 
 
 

devfs & usb

Post by Peter T. Breue » Wed, 27 Feb 2002 20:48:52





>>>> No new devices or links are added to /dev (I've tested for this by
>>>> doing 'ls /dev > somefile' before and after loading the module, and
>>>> hotsyncing,

>>> This is not correct as a test. You seem unaware of the possibility that
>>> subdirs will change! Why?

>> I missed this, perhaps he skimmed my instructions.  Specifically, the
>> -1R options were suggested:

>> ls -1R
> I did the test after reading your first post (response to AARG!), which
> didn't mention the -R switch. I've since repeated the test, using -R

You shouldn't need it mentioned, because you want to look at ALL
entries, after all, not just those directly below the root of your search
path.

Quote:> (thanks  for the clarification), and that confirms that nothing has been
> created. I've also since found out that, despite package labeling, the

How does it "confirm it"? You haven't shown me anything! Be specific,
please.

Quote:> module I'm' using, keyspan_pda, may in fact not support the device I'm
> attempting to use, a keyspan USA-19Qi usb->serial adaptor. I don't know

Never heard of either. Does the module support devfs? Grep inside its
code for references to devfs, and we'll know.

Quote:> whether this is the root cause, but I'll look in to it.

Root cause? Your dmesg output would tell you.

Quote:> Huh? as revealed by ls -lR /dev, my /dev tree has 1441 entries, a bit
> much to scan by eye without missing something. This

Why are you scanning "by eye"?

Quote:> system has been upgraded at least once, and maybe twice (Mandrake
> 8.0->8.1, possibly also 7.2->8.1); perhaps there are hangers-on from the
> earlier installs?

You seem not to understand.  You want to find entries that are newly
created.  For that, obviously the thing you want to do is list the
directory entries in reverse time order, and tell us what the last ten
or so are (he hopes fervently that devfs supports file creation
timestamps).

Peter

 
 
 

devfs & usb

Post by G. Pollac » Wed, 27 Feb 2002 23:34:23


Quote:>> (thanks  for the clarification), and that confirms that nothing has
>> been created. I've also since found out that, despite package labeling,
>> the

> How does it "confirm it"? You haven't shown me anything! Be specific,
> please.

As I've stated in other posts in this thread, no new devices or links are
created; the time stamps on two files change: /dev/ptmx and /dev/pts/1.
I'm unable to communicate over the interface with either of these.

Quote:> Never heard of either. Does the module support devfs? Grep inside its
> code for references to devfs, and we'll know.

As I stated earlier, some of us are not computer professionals,
Quote:>> whether this is the root cause, but I'll look in to it.

> Root cause? Your dmesg output would tell you.

I'm not on that system at present so I can't give all the specifics, but I
did check dmesg, and it reported no errors.

Quote:

>> Huh? as revealed by ls -lR /dev, my /dev tree has 1441 entries, a bit
>> much to scan by eye without missing something. This

> Why are you scanning "by eye"?

I'm not; I'm using diff. I'm replying here to other suggestions that it
might have been easier for me to "manually" detect new devices (whatever
that means) than to wait for replies from the newsgroup. My point was that
there are too many entries for this to be pratical.

Quote:>> system has been upgraded at least once, and maybe twice (Mandrake
>> 8.0->8.1, possibly also 7.2->8.1); perhaps there are hangers-on from
>> the earlier installs?

> You seem not to understand.  You want to find entries that are newly
> created.

I believe that I'm doing this by comparing the output of 'ls -lR /dev >
somefile" before and after loading the module. This too appeared in an
earlier post in the thread. My point here was that perhaps the reason my
system has a large number of entries is that there are leftovers from
earlier installs. To quote from one of Mochun's replies: "That's the
beauty of
Quote:>>devfs it doesn't have hundreds of entries .."; well, my system _does_

have hundreds of entries; I was merely making a guess as to why this might
be.

I had intended not to editorialize on the tone of your comments, but  I
can no longer resist. Your email address indicates that you may be in
Spain, and perhaps English is not your native language, but you should be
aware that your tone is condescending in the extreme (on the other hand,
perhaps that's your intention). Your email address also includes the
string 'it';am I correct in deducing that this refers to information
technology? You seem not to be aware that linux is used not only by
computer professionals, but also by others who are less knowledgable about
it's inner workings. You may wish that this were otherwise, but the fact
is that several organizations, e.g. Redhat, Mandrake, Ximian, are
explicitly trying to make linux accessible to non-programmers, etc. The
consequence is that many users will occasionally ask for guidance on
issues beyond their current competence. I am among these. I am neither
stupid nor lazy (some of your comments seem to imply both). I have
expertise in several other areas of specialized knowledge,and I am happy
to share this with others in a non-confrontational manner. I hope for the
same in this newsgroup.

 
 
 

devfs & usb

Post by Peter T. Breue » Thu, 28 Feb 2002 00:18:41



>>> (thanks  for the clarification), and that confirms that nothing has
>>> been created. I've also since found out that, despite package labeling,
>>> the

>> How does it "confirm it"? You haven't shown me anything! Be specific,
>> please.
> As I've stated in other posts in this thread, no new devices or links are
> created; the time stamps on two files change: /dev/ptmx and /dev/pts/1.

How do you know? I think I _asked_!

Quote:> I'm unable to communicate over the interface with either of these.

One is the pseudo-tty multiplexer, and the other is a ptty (one half
of).  The latter entry does indicate that you have devfs up and running,
however!  That's good.

Quote:>> Never heard of either. Does the module support devfs? Grep inside its
>> code for references to devfs, and we'll know.

> As I stated earlier, some of us are not computer professionals,

And? How does that stop you grepping in its source code for mention
of the word "devfs"?

Quote:>> Root cause? Your dmesg output would tell you.
> I'm not on that system at present so I can't give all the specifics, but I
> did check dmesg, and it reported no errors.

The module must report either error or success. It cannot do any third
thing.

Quote:>>> Huh? as revealed by ls -lR /dev, my /dev tree has 1441 entries, a bit
>>> much to scan by eye without missing something. This

>> Why are you scanning "by eye"?
> I'm not; I'm using diff. I'm replying here to other suggestions that it

OK, that's good. But I'd still expect you only to be interested in
recent entries.

Quote:> might have been easier for me to "manually" detect new devices (whatever

It is: ls -lrtR /dev | tail

Quote:> that means) than to wait for replies from the newsgroup. My point was that
> there are too many entries for this to be pratical.

There are not - you have a computer to help you.

Quote:>> You seem not to understand.  You want to find entries that are newly
>> created.
> I believe that I'm doing this by comparing the output of 'ls -lR /dev >
> somefile" before and after loading the module. This too appeared in an

diffing, you mean (cmp is something else).  Yes, that helps, but I
wouldn't call it quite definitive.  The ls -lrtR | tail output would
tell us about devfs, and grepping the driver source would tell you if
the driver used devfs or not. Looking at dmesg would show you what the
driver says about it all.

Quote:> earlier post in the thread. My point here was that perhaps the reason my
> system has a large number of entries is that there are leftovers from

There can't be; devfs starts from zero each boot.

Quote:> earlier installs. To quote from one of Mochun's replies: "That's the
> beauty of
>>>devfs it doesn't have hundreds of entries .."; well, my system _does_
> have hundreds of entries; I was merely making a guess as to why this might
> be.

Doesn't sound as though you have devfs at all .. or if you have, you
seem to be overwriting it with some other info. You might want to clean
out whatever excess devfsd puts in there.

Quote:> I had intended not to editorialize on the tone of your comments, but  I
> can no longer resist. Your email address indicates that you may be in
> Spain, and perhaps English is not your native language, but you should be
> aware that your tone is condescending in the extreme (on the other hand,

It's not my intention, and my tone ain't condescending. I'm replying
fast and without concern for your feelings. That's all. If you feel
aggrieved by that, tough tooties. I'm not responsible for your huffs.
And I'm not going to take time out to sooth your ego.

Quote:> perhaps that's your intention). Your email address also includes the
> string 'it';am I correct in deducing that this refers to information
> technology? You seem not to be aware that linux is used not only by

No. It's "telecommunications engineering", more or less. Doesn't make
any difference, though.

Quote:> computer professionals, but also by others who are less knowledgable about
> it's inner workings. You may wish that this were otherwise, but the fact
> is that several organizations, e.g. Redhat, Mandrake, Ximian, are
> explicitly trying to make linux accessible to non-programmers, etc. The
> consequence is that many users will occasionally ask for guidance on
> issues beyond their current competence. I am among these. I am neither
> stupid nor lazy (some of your comments seem to imply both). I have

You read too much into what you do read. But answering me correctly and
competently will convince me of what you assert, not the assertions
themselves, so don't waste the e-ink! You've done quite well so far,
but you can do better still.

Quote:> expertise in several other areas of specialized knowledge,and I am happy

I have no "specialized knowledge".

Quote:> to share this with others in a non-confrontational manner. I hope for the
> same in this newsgroup.

Peter
 
 
 

devfs & usb

Post by Molchu » Thu, 28 Feb 2002 00:34:25



>>>(thanks  for the clarification), and that confirms that nothing has
>>>been created. I've also since found out that, despite package labeling,
>>>the

>>How does it "confirm it"? You haven't shown me anything! Be specific,
>>please.

> As I've stated in other posts in this thread, no new devices or links are
> created; the time stamps on two files change: /dev/ptmx and /dev/pts/1.
> I'm unable to communicate over the interface with either of these.

>>Never heard of either. Does the module support devfs? Grep inside its
>>code for references to devfs, and we'll know.

> As I stated earlier, some of us are not computer professionals,

>>>whether this is the root cause, but I'll look in to it.

>>Root cause? Your dmesg output would tell you.

> I'm not on that system at present so I can't give all the specifics, but I
> did check dmesg, and it reported no errors.

>>>Huh? as revealed by ls -lR /dev, my /dev tree has 1441 entries, a bit
>>>much to scan by eye without missing something. This

>>Why are you scanning "by eye"?

> I'm not; I'm using diff. I'm replying here to other suggestions that it
> might have been easier for me to "manually" detect new devices (whatever
> that means) than to wait for replies from the newsgroup. My point was that
> there are too many entries for this to be pratical.

>>>system has been upgraded at least once, and maybe twice (Mandrake
>>>8.0->8.1, possibly also 7.2->8.1); perhaps there are hangers-on from
>>>the earlier installs?

>>You seem not to understand.  You want to find entries that are newly
>>created.

> I believe that I'm doing this by comparing the output of 'ls -lR /dev >
> somefile" before and after loading the module. This too appeared in an
> earlier post in the thread. My point here was that perhaps the reason my
> system has a large number of entries is that there are leftovers from
> earlier installs. To quote from one of Mochun's replies: "That's the
> beauty of

>>>devfs it doesn't have hundreds of entries .."; well, my system _does_

> have hundreds of entries; I was merely making a guess as to why this might
> be.

> I had intended not to editorialize on the tone of your comments, but  I
> can no longer resist. Your email address indicates that you may be in
> Spain, and perhaps English is not your native language, but you should be
> aware that your tone is condescending in the extreme (on the other hand,
> perhaps that's your intention). Your email address also includes the
> string 'it';am I correct in deducing that this refers to information
> technology? You seem not to be aware that linux is used not only by
> computer professionals, but also by others who are less knowledgable about
> it's inner workings. You may wish that this were otherwise, but the fact
> is that several organizations, e.g. Redhat, Mandrake, Ximian, are
> explicitly trying to make linux accessible to non-programmers, etc. The
> consequence is that many users will occasionally ask for guidance on
> issues beyond their current competence. I am among these. I am neither
> stupid nor lazy (some of your comments seem to imply both). I have
> expertise in several other areas of specialized knowledge,and I am happy
> to share this with others in a non-confrontational manner. I hope for the
> same in this newsgroup.

Hey, don't worry about it. That's Peter for you. If you spend some time
in this NG you'll get used to his tone :). Nevertheless, he DOES provide
valuable information, so just ignore the tone and try to get the most
out of his posts. To go a bit off the subject, I had a math teacher at
uni who failed me 5 times, I cursed him and was ready to hire an
assassin :) but if it wasn't him I would never know math.
 
 
 

devfs & usb

Post by Molchu » Thu, 28 Feb 2002 01:00:00



>>Devfs groups everything nicely in subdirs. Use 'ls -l /dev' to find out
>>which one is a dir and which one is a file (I can't believe I'm telling
>>you how to use 'ls' :) ) then look deeper in subdirs, just don't waist
>>your time by looking in some totally unrelated dirs (although it'll be
>>beneficial to know what in them for future) look in the dirs that
>>contain character devices. As a last resort get an X server and start
>>clicking the mouse :).

> No new devices are created in subdirectories either; the only changes are
> to timestamps (to /dev/ptmx and /dev/pts/1).

>>Maybe you're missing some other modules like 'usbcore' or 'uhci'. How do
>>you load your modules, what do you use 'modprobe' or 'insmod' the former
>>will pull all the dependent modules as well the later will not.

> I've loaded the modules with modprobe, and all seem to be installed
> (uscore, usbserial)

>>Devfsd has got nothing to do with it. Entries are created by modules
>>with or without devfsd. Devfsd just solves the compatibility issue by
>>providing the old names for the progs that devfs unaware (well, it does
>>a couple of other things like demand loading too).

> Thanks very much for the above; this prompted me to look further into the
> module I'm loading, and I think it may be the key. I have been assuming
> that the keyspan adaptor (USA-19Qi) is linux-compatible, because that's
> what it says clearly on the packaging (compatible with kernels >= 2.4.1).
> But keyspan's website says that a linux driver should be available "soon".

> Thanks again for your help.Your comments have been productive and
> informative (unlike some of the others that have been posted).

No worries. There're lots of other things to consider. Maybe your
physical devise is not within serviceable range of the driver (module)
you're trying to load. Maybe your module doesn't see the device and you
need to provide it with some parameters upon loading or from within
/etc/modules.conf. Always run 'dmesg' and see what your modules "talk"
about, if your module sees the physical device it should report I/O
ports and memory ranges maybe some other stuff. If you connect your
devise and there is a driver (module) that can service it the USB
subsystem will report which module "grabs" the device otherwise it'll
say that no driver claimed "responsibility" for this device. You can see
all that by running 'dmesg'. If module is loaded and physical device is
connected but USB subsystem does not report any "good" news, it's either
not the right module or it needs extra parameters (some modules will not
even load if the device they service does not exist).

Good luck.

Ruslan.

 
 
 

devfs & usb

Post by David A. Desrosier » Thu, 28 Feb 2002 15:09:26



> pilot-xfer -p /dev/ttyUSB1 -l

        What version of pilot-link? 0.10.1 is the only one with
USB at this point, since I just added it since the 0.9.5 release.

Quote:>    Unable to bind to port /dev/ttyUSB1
>    pi_bind: No such file or directory

        Also, you are aware that you _MUST_ hit HotSync on the cradle
_FIRST_, before the device is created in Linux, right? On devfs, that
device is /dev/usb/tts/1, non-devfs, it's /dev/ttyUSB1.

        Not sure why you're using the keyspan adapter, since both
serial and USB work natively (USB was added in 0.10.1, which I will be
releasing very soon)

/d

 
 
 

1. devfs and temrinal creation problems (devfs & pts)

You might have problems creating new xterms/kterms/eterms/whatever if
you're using devfs; the problem happens because of the saving and restoring
of permissions of devices in /dev/pts.  To solve this, add the following 3
lines above the geneirc REGISTER/CHANGE/CREATE rules in the /etc/devfs.conf
file (this sollution is borrowed from another post):

REGISTER        ^pts/.*         IGNORE
CHANGE          ^pts/.*         IGNORE
CREATE          ^pts/.*         IGNORE

The post from which this sollution was taken can be found at
http://groups.google.com/groups?q=linux+pts+devfs&num=100&hl=en&safe=...

2. Too late me thinks

3. Visor && USB && coldsync

4. XWindows with SiS6202??

5. USB sticks and devfs

6. iBCS module errors on boot.

7. devfs (3/7) - cleanup devfs use in ide

8. Is it normal for freezing while...

9. device driver using devfs on system with not mounted devfs

10. 2.4.17,2.4.18 ide-scsi+usb-storage+devfs Oops

11. BUG in 2.4.20: sbp2 and usb-storage SCSI emulation with devfs

12. USB/devfs problem

13. USB Scanner devfs support