udev problems

udev problems

[Michal Čihař]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi

I finally decided to give udev a try, but it was not as good as I hoped. I 
have Apacer Mega Steno card reader (as well as other usb storage device) and 
I wanted to have permanent name for it. Okay, lets create some rule in 
udev.rules and script that will generate it. Sometimes the name was generated 
correctly and sometimes not. Afer some experiments I got three rules, that 
seem to work (with line numbers):

21: BUS="scsi", SYSFS{vendor}="Apacer  ", SYSFS{model}="Mega Steno      ", 
PROGRAM="/etc/udev/apacer.sh %k %b %n", NAME="%c{1}", SYMLINK="%k %c{2} 
%c{3}"
22: BUS="scsi", SYSFS{vendor}="Apacer  ", PROGRAM="/etc/udev/apacer.sh %k %b 
%n", NAME="%c{1}", SYMLINK="%k %c{2} %c{3}"
23: BUS="scsi", SYSFS{vendor}="Apacer", PROGRAM="/etc/udev/apacer.sh %k %b 
%n", NAME="%c{1}", SYMLINK="%k %c{2} %c{3}"

related udev log:

configured rule in '/etc/udev/udev.rules' at line 21 applied, added symlink 
'%k %c{2} %c{3}'
configured rule in '/etc/udev/udev.rules' at line 21 applied, 'sda' becomes 
'%c{1}'
creating device node '/dev/scsi/host28/bus0/target0/lun0/disc'
configured rule in '/etc/udev/udev.rules' at line 21 applied, added symlink 
'%k %c{2} %c{3}'
configured rule in '/etc/udev/udev.rules' at line 21 applied, 'sdc' becomes 
'%c{1}'
creating device node '/dev/scsi/host28/bus0/target0/lun2/disc'
configured rule in '/etc/udev/udev.rules' at line 22 applied, added symlink 
'%k %c{2} %c{3}'
configured rule in '/etc/udev/udev.rules' at line 22 applied, 'sdb' becomes 
'%c{1}'
creating device node '/dev/scsi/host28/bus0/target0/lun1/disc'
configured rule in '/etc/udev/udev.rules' at line 23 applied, added symlink 
'%k %c{2} %c{3}'
configured rule in '/etc/udev/udev.rules' at line 23 applied, 'sda1' becomes 
'%c{1}'
creating device node '/dev/scsi/host28/bus0/target0/lun0/part1'

Why does every time match another rule?

Using udev 021 (from Debian unstable) and kernel 2.6.3.

Another topic is that device does not automatically report partitions, so man 
has to read appropriate device to get partitions.

- -- 
Regards
	Michal ÄŒihaÅ™
	http://cihar.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFASxJG3DVS6DbnVgQRAjb1AKDXvj6ChkdblRraH0vkidN3uQfERACgkHu3
BoXQrnBG57RSh81OJ9skwdw=Y4Pt
-----END PGP SIGNATURE-----



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Kay Sievers]

On Sun, 2004-03-07 at 13:15, Michal Èihaø wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hi
> 
> I finally decided to give udev a try, but it was not as good as I hoped. I 
> have Apacer Mega Steno card reader (as well as other usb storage device) and 
> I wanted to have permanent name for it. Okay, lets create some rule in 
> udev.rules and script that will generate it. Sometimes the name was generated 
> correctly and sometimes not. Afer some experiments I got three rules, that 
> seem to work (with line numbers):
> 
> 21: BUS="scsi", SYSFS{vendor}="Apacer  ", SYSFS{model}="Mega Steno      ", 
> PROGRAM="/etc/udev/apacer.sh %k %b %n", NAME="%c{1}", SYMLINK="%k %c{2} 
> %c{3}"

What is this "apacer.sh" for? What's in there?


> Another topic is that device does not automatically report partitions, so man 
> has to read appropriate device to get partitions.

Most removable media devices are unable to report media changes, you
have to poll the device, but thats not a udev task.
So your choice is to dis-/reconnect for partition update or create all
partitions if the main device is connected with NAME{all_partitions}=.

thanks,
Kay




-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Michal Čihař]

[-- Attachment #1: clearsigned data --]
[-- Type: Text/Plain, Size: 929 bytes --]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

No need to cc me, I'm subsribed.

On Sunday 07 of March 2004 13:29, Kay Sievers wrote:
> What is this "apacer.sh" for? What's in there?

See attachment, it just generates device name.

> Most removable media devices are unable to report media changes, you
> have to poll the device, but thats not a udev task.

Okay, this problem is solvable.

> So your choice is to dis-/reconnect for partition update 

This doesn't work, it doesn't report partiotions even if card inserted while 
connecting.

> or create all partitions if the main device is connected with
> NAME{all_partitions}=.

Probably best solution.

- -- 
Regards
	Michal Čihař
	http://cihar.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFASxno3DVS6DbnVgQRAiVOAJ9v4U30sNDYEHpS34h5dab/vQgRzgCggEmF
+xtXrJ6qwjJe8h2L+LjR3kI=
=3+K8
-----END PGP SIGNATURE-----

[-- Attachment #2: apacer.sh --]
[-- Type: application/x-shellscript, Size: 249 bytes --]

Re: udev problems

[Michal Čihař]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sunday 07 of March 2004 13:29, Kay Sievers wrote:
> So your choice is to dis-/reconnect for partition update or create all
> partitions if the main device is connected with NAME{all_partitions}=.

Is there way to make this work with symlinks?

- -- 
Regards
	Michal Èihaø
	http://cihar.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFASx9N3DVS6DbnVgQRAtPGAJ9mn1/aP+YTvLtCgSi1qlhcGNeCVQCeJ2ra
8Jf3NEOThCHjcTDKn8tIc8M=kazo
-----END PGP SIGNATURE-----



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Kay Sievers]

On Sun, 2004-03-07 at 13:47, Michal Èihaø wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> No need to cc me, I'm subsribed.

Yes, but it's quite common.
I just reply to all and do not edit the mail header.
The list needs several hours to deliver some days,
so this isn't a option anyway.
And you receive your own mail too, right? :)

> On Sunday 07 of March 2004 13:29, Kay Sievers wrote:
> > What is this "apacer.sh" for? What's in there?
> 
> See attachment, it just generates device name.

I'll take a look.

> > Most removable media devices are unable to report media changes, you
> > have to poll the device, but thats not a udev task.
> 
> Okay, this problem is solvable.
> 
> > So your choice is to dis-/reconnect for partition update 
> 
> This doesn't work, it doesn't report partiotions even if card inserted while 
> connecting.

Are the partitons in /sys ?
Something like /sys/block/sda/sda1/ ?

> > or create all partitions if the main device is connected with
> > NAME{all_partitions}=.
> 
> Probably best solution.

Yes, but no symlinks.

Kay



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Michal Čihař]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sunday 07 of March 2004 14:44, Kay Sievers wrote:
> I just reply to all and do not edit the mail header.

I just reply to list :-)

> > > So your choice is to dis-/reconnect for partition update
> >
> > This doesn't work, it doesn't report partiotions even if card inserted
> > while connecting.
>
> Are the partitons in /sys ?
> Something like /sys/block/sda/sda1/ ?

No they're not there until I read from /dev/sda*

- -- 
Regards
	Michal Èihaø
	http://cihar.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFASyiA3DVS6DbnVgQRAs05AJ0fshLjTy2Ou529/Vs7ndTNbTMOWwCfYckj
SvbEHJgXNG2UCY5HMAEmqwI=N8d4
-----END PGP SIGNATURE-----



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Kay Sievers]

On Sun, 2004-03-07 at 14:49, Michal Èihaø wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Sunday 07 of March 2004 14:44, Kay Sievers wrote:
> > I just reply to all and do not edit the mail header.
> 
> I just reply to list :-)
> 
> > > > So your choice is to dis-/reconnect for partition update
> > >
> > > This doesn't work, it doesn't report partiotions even if card inserted
> > > while connecting.
> >
> > Are the partitons in /sys ?
> > Something like /sys/block/sda/sda1/ ?
> 
> No they're not there until I read from /dev/sda*

No idea. My partitions are found.
Whats in dmesg? I have the following for my cheap 4slot-reader:

USB Mass Storage device found at 3
usb 1-3: USB disconnect, address 3
usb 1-3: new high speed USB device using address 4
scsi2 : SCSI emulation for USB Mass Storage devices
  Vendor: Zynet     Model: USB Storage-SMC   Rev: I03A
  Type:   Direct-Access                      ANSI SCSI revision: 02
Attached scsi removable disk sda at scsi2, channel 0, id 0, lun 0
Attached scsi generic sg0 at scsi2, channel 0, id 0, lun 0,  type 0
  Vendor: Zynet     Model: USB Storage-CFC   Rev: I03A
  Type:   Direct-Access                      ANSI SCSI revision: 02
sdb: Spinning up disk....ready
SCSI device sdb: 31360 512-byte hdwr sectors (16 MB)
sdb: assuming Write Enabled
sdb: assuming drive cache: write through
 sdb: sdb1 sdb2 sdb3 sdb4
Attached scsi removable disk sdb at scsi2, channel 0, id 0, lun 1
Attached scsi generic sg1 at scsi2, channel 0, id 0, lun 1,  type 0
  Vendor: Zynet     Model: USB Storage-MMC   Rev: I03A
  Type:   Direct-Access                      ANSI SCSI revision: 02
Attached scsi removable disk sdc at scsi2, channel 0, id 0, lun 2
Attached scsi generic sg2 at scsi2, channel 0, id 0, lun 2,  type 0
  Vendor: Zynet     Model: USB Storage-MSC   Rev: I03A
  Type:   Direct-Access                      ANSI SCSI revision: 02
Attached scsi removable disk sdd at scsi2, channel 0, id 0, lun 3
Attached scsi generic sg3 at scsi2, channel 0, id 0, lun 3,  type 0
WARNING: USB Mass Storage data integrity not assured
USB Mass Storage device found at 4




-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Kay Sievers]

On Sun, 2004-03-07 at 13:15, Michal Èihaø wrote:

> 21: BUS="scsi", SYSFS{vendor}="Apacer  ", SYSFS{model}="Mega Steno      ", 
> PROGRAM="/etc/udev/apacer.sh %k %b %n", NAME="%c{1}", SYMLINK="%k %c{2} 
> %c{3}"

> Why does every time match another rule?

Yeah, It's a bug in the trailin whitespace removal.
We mangle the attribut evalue while we compare it and
libsysfs reuses it :(

Just remove your trailing whitespace from "Apacer  " and please tell me
if it works. Anyway I will fix it soon.

thanks for pointing this out,
Kay



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Michal Čihař]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sunday 07 of March 2004 15:05, Kay Sievers wrote:
> Whats in dmesg? I have the following for my cheap 4slot-reader:

[connect reader]

USB Mass Storage device found at 21
usb 1-2: USB disconnect, address 21
usb 1-2: new full speed USB device using address 22
scsi34 : SCSI emulation for USB Mass Storage devices
  Vendor: Apacer    Model: Mega Steno        Rev: 1.03
  Type:   Direct-Access                      ANSI SCSI revision: 02
Attached scsi removable disk sda at scsi34, channel 0, id 0, lun 0
  Vendor: Apacer    Model: Mega Steno        Rev: 1.03
  Type:   Direct-Access                      ANSI SCSI revision: 02
Attached scsi removable disk sdb at scsi34, channel 0, id 0, lun 1
  Vendor: Apacer    Model: Mega Steno        Rev: 1.03
  Type:   Direct-Access                      ANSI SCSI revision: 02
Attached scsi removable disk sdc at scsi34, channel 0, id 0, lun 2
WARNING: USB Mass Storage data integrity not assured

[now I read from /dev/sda]

USB Mass Storage device found at 22
SCSI device sda: 15680 512-byte hdwr sectors (8 MB)
sda: assuming Write Enabled
sda: assuming drive cache: write through
SCSI device sda: 15680 512-byte hdwr sectors (8 MB)
sda: assuming Write Enabled
sda: assuming drive cache: write through
 sda: sda1

> sdb: Spinning up disk....ready

This doesn't happen to me. Related kernel code is in drivers/scsi/sd.c:

                /*
                 * If manual intervention is required, or this is an
                 * absent USB storage device, a spinup is meaningless.
                 */
                if (SRpnt->sr_sense_buffer[2] = NOT_READY &&
                    SRpnt->sr_sense_buffer[12] = 4 /* not ready */ &&
                    SRpnt->sr_sense_buffer[13] = 3) {
                        break;          /* manual intervention required */

                /*
                 * Issue command to spin up drive when not ready
                 */
                } else if (SRpnt->sr_sense_buffer[2] = NOT_READY) {
                        unsigned long time1;
                        if (!spintime) {
                                printk(KERN_NOTICE "%s: Spinning up disk...",
                                       diskname);
                }

- -- 
Regards
	Michal Èihaø
	http://cihar.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFASy/H3DVS6DbnVgQRAnbcAJsHzcFoiAYs1+Vv3zBvhr18ftnlIQCgqZxv
UHJPF3qYs7be1Unn8iqReQE=/Dn5
-----END PGP SIGNATURE-----



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Kay Sievers]

On Sun, 2004-03-07 at 15:20, Michal Èihaø wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Sunday 07 of March 2004 15:05, Kay Sievers wrote:
> > Whats in dmesg? I have the following for my cheap 4slot-reader:
> 
> [connect reader]
> 
> USB Mass Storage device found at 21
> usb 1-2: USB disconnect, address 21
> usb 1-2: new full speed USB device using address 22
> scsi34 : SCSI emulation for USB Mass Storage devices
>   Vendor: Apacer    Model: Mega Steno        Rev: 1.03
>   Type:   Direct-Access                      ANSI SCSI revision: 02
> Attached scsi removable disk sda at scsi34, channel 0, id 0, lun 0
>   Vendor: Apacer    Model: Mega Steno        Rev: 1.03
>   Type:   Direct-Access                      ANSI SCSI revision: 02
> Attached scsi removable disk sdb at scsi34, channel 0, id 0, lun 1
>   Vendor: Apacer    Model: Mega Steno        Rev: 1.03
>   Type:   Direct-Access                      ANSI SCSI revision: 02
> Attached scsi removable disk sdc at scsi34, channel 0, id 0, lun 2
> WARNING: USB Mass Storage data integrity not assured
> 
> [now I read from /dev/sda]
> 
> USB Mass Storage device found at 22
> SCSI device sda: 15680 512-byte hdwr sectors (8 MB)
> sda: assuming Write Enabled
> sda: assuming drive cache: write through
> SCSI device sda: 15680 512-byte hdwr sectors (8 MB)
> sda: assuming Write Enabled
> sda: assuming drive cache: write through
>  sda: sda1
> 
> > sdb: Spinning up disk....ready
> 
> This doesn't happen to me. Related kernel code is in drivers/scsi/sd.c:
> 
>                 /*
>                  * If manual intervention is required, or this is an
>                  * absent USB storage device, a spinup is meaningless.
>                  */
>                 if (SRpnt->sr_sense_buffer[2] = NOT_READY &&
>                     SRpnt->sr_sense_buffer[12] = 4 /* not ready */ &&
>                     SRpnt->sr_sense_buffer[13] = 3) {
>                         break;          /* manual intervention required */
> 
>                 /*
>                  * Issue command to spin up drive when not ready
>                  */
>                 } else if (SRpnt->sr_sense_buffer[2] = NOT_READY) {
>                         unsigned long time1;
>                         if (!spintime) {
>                                 printk(KERN_NOTICE "%s: Spinning up disk...",
>                                        diskname);
>                 }
> 

Ah, nice to know.
You may access you device with your apacer.sh to revalidate. (?)
Just create a temporary node, access and remove it.

Kay



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Michal Čihař]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sunday 07 of March 2004 15:16, Kay Sievers wrote:
> On Sun, 2004-03-07 at 13:15, Michal Èihaø wrote:
> > 21: BUS="scsi", SYSFS{vendor}="Apacer  ", SYSFS{model}="Mega Steno     
> > ", PROGRAM="/etc/udev/apacer.sh %k %b %n", NAME="%c{1}", SYMLINK="%k
> > %c{2} %c{3}"
> >
> > Why does every time match another rule?
>
> Yeah, It's a bug in the trailin whitespace removal.
> We mangle the attribut evalue while we compare it and
> libsysfs reuses it :(
>
> Just remove your trailing whitespace from "Apacer  " and please tell me
> if it works. Anyway I will fix it soon.

When I leave just:

BUS="scsi", SYSFS{vendor}="Apacer", PROGRAM="/etc/udev/apacer.sh %k %b %n", 
NAME="%c{1}", SYMLINK="%k %c{2} %c{3}"

It still behaves very non determinantely, sometimes it catches all three 
slots, sometimes just one :-(. This doesn't depend whether there is another 
rule that will catch this device.

Same happens when I try to match it on serial number (AFAIK no trailing 
space):

BUS="usb", SYSFS_serial="19220DA1018B", PROGRAM="/etc/udev/apacer.sh %k %b 
%n", NAME="%c{1}", SYMLINK="%k %c{2} %c{3}"

- -- 
Regards
	Michal Èihaø
	http://cihar.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFASzHy3DVS6DbnVgQRAjjdAJ9mW+G6v+3+B/8LKAaDE1kKkZG0/wCg8blT
nKjlEfGFPollKnBnA53dswM=jcWO
-----END PGP SIGNATURE-----



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Kay Sievers]

On Sun, 2004-03-07 at 15:30, Michal Èihaø wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Sunday 07 of March 2004 15:16, Kay Sievers wrote:
> > On Sun, 2004-03-07 at 13:15, Michal Èihaø wrote:
> > > 21: BUS="scsi", SYSFS{vendor}="Apacer  ", SYSFS{model}="Mega Steno     
> > > ", PROGRAM="/etc/udev/apacer.sh %k %b %n", NAME="%c{1}", SYMLINK="%k
> > > %c{2} %c{3}"
> > >
> > > Why does every time match another rule?
> >
> > Yeah, It's a bug in the trailin whitespace removal.
> > We mangle the attribut evalue while we compare it and
> > libsysfs reuses it :(
> >
> > Just remove your trailing whitespace from "Apacer  " and please tell me
> > if it works. Anyway I will fix it soon.
> 
> When I leave just:
> 
> BUS="scsi", SYSFS{vendor}="Apacer", PROGRAM="/etc/udev/apacer.sh %k %b %n", 
> NAME="%c{1}", SYMLINK="%k %c{2} %c{3}"
> 
> It still behaves very non determinantely, sometimes it catches all three 
> slots, sometimes just one :-(. This doesn't depend whether there is another 
> rule that will catch this device.
> 
> Same happens when I try to match it on serial number (AFAIK no trailing 
> space):
> 
> BUS="usb", SYSFS_serial="19220DA1018B", PROGRAM="/etc/udev/apacer.sh %k %b 
> %n", NAME="%c{1}", SYMLINK="%k %c{2} %c{3}"

If devices are missing, whats in sysfs '/sys/block/' at this time?
You may use 'tree /sys/block' to look at it.

Hmm, if all devices are there, seems we need a udev version with debug
enabled. Is it possible for you to build udev from the sources and run
'make DEBUG=true'?

Kay



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Michal Čihař]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sunday 07 of March 2004 15:30, Michal Èihaø wrote:
> It still behaves very non determinantely, sometimes it catches all three
> slots, sometimes just one :-(. This doesn't depend whether there is another
> rule that will catch this device.

And the apacer.sh script is always executed and always ends with exit 0.

- -- 
Regards
	Michal Èihaø
	http://cihar.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFASzUu3DVS6DbnVgQRAqUhAJ9CnSaew6bLGICsAnGhil4eboX6vgCg8gpl
2Tsvorst+iMOaT8TXI5D/QU=LrlH
-----END PGP SIGNATURE-----



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Michal Čihař]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sunday 07 of March 2004 15:41, Kay Sievers wrote:
> Hmm, if all devices are there, seems we need a udev version with debug
> enabled. Is it possible for you to build udev from the sources and run
> 'make DEBUG=true'?

Wow, I didn't expect it is so much verbose :-)

The problem seems to be that udev thinks that program ended with nonzero. But 
I'm pretty sure it ends with exit 0...

- -- 
Regards
	Michal Èihaø
	http://cihar.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFASzuW3DVS6DbnVgQRAimQAKDIjDFRzGzQgksJs9eHDBqqXZuDlwCdE2qc
6vq83vdHGI9jhuUTci6YeXQ=1PeY
-----END PGP SIGNATURE-----



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Michal Čihař]

[-- Attachment #1: clearsigned data --]
[-- Type: Text/Plain, Size: 645 bytes --]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sunday 07 of March 2004 16:11, Michal Čihař wrote:
> The problem seems to be that udev thinks that program ended with nonzero.
> But I'm pretty sure it ends with exit 0...

Okay, I found the solution, but not the problem - attached script works. It 
builds name and then echoes complete line, while the previous one echoed it 
by parts.

- -- 
Regards
	Michal Čihař
	http://cihar.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFASz+v3DVS6DbnVgQRAtA3AKCyMzcOS9iJI/cbUZ+d3O3MPf5OxwCfadKY
Gcg0F1+iP+dni2b+99dBaqU=
=kr/L
-----END PGP SIGNATURE-----

[-- Attachment #2: apacer.sh --]
[-- Type: application/x-shellscript, Size: 273 bytes --]

Re: udev problems

[Michal Čihař]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sunday 07 of March 2004 16:28, Michal Èihaø wrote:
> Okay, I found the solution, but not the problem - attached script works. It
> builds name and then echoes complete line, while the previous one echoed it
> by parts.

execute_program in namedev.c doesn't handle correctly situation when it get 
line split into smaller pieces (it should IMHO concatenate it till EOF or 
\n), this leads to setting retval to -1, what is then returned. This doesn't 
happen always, as it depends when pipe buffer is read/written.

Debug log for this case:

match_rule: check PROGRAM
apply_format: format=k, string='/etc/udev/apacer.sh ', tail=' %b %n'
apply_format: substitute kernel name 'sdb'
apply_format: format=b, string='/etc/udev/apacer.sh sdb ', tail=' %n'
apply_format: substitute bus_id '57:0:0:1'
apply_format: format=n, string='/etc/udev/apacer.sh sdb 57:0:0:1 ', tail=''
execute_program: executing '/etc/udev/apacer.sh sdb 57:0:0:1 '
execute_program: execute program '/etc/udev/apacer.sh'
execute_program: result is 'cardreader-'
execute_program: result value already set
execute_program: result value already set
match_rule: PROGRAM returned nozero

- -- 
Regards
	Michal Èihaø
	http://cihar.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFAS0PJ3DVS6DbnVgQRAo0nAJ9WjVuX1G0ou3UC3LrkEGX09HNQZQCfVVPf
HrihustikcY3nE+flge8/4w=BtCH
-----END PGP SIGNATURE-----



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

udev problems

[clemens]

is.

If it is legal, then I have a complaint...

This has happened to me twice now (three times if you count the twice
yesterday while I was trying to dig my way out).

It would seem, that on a reboot (on the way down, on the way up, Im not
sure) that udev proceeds to remove all files that were active at the
time the system went down.  I will assume that it created all of these
files, so it thinks that it can remove them.

Now if it were doing this in a SEPARATE, isolated, file system, say
/udev and not /dev, that would be fine.  However doing it in /dev
is disasterous.  Some of these things existed before udev (re)created
them.

It removed my /dev/hde? files from 0-7, just the ones I have mounted.
It got /dev/console, so even if you get past the previous error,
you get nailed again.
And it removed enough other files to make repair basically impossible.

Hey what ever happened to the MKNOD (?) script that used to be in
UNIX systems?  I dont see it in Fedora2.  That would have helped in
the repair.

I was able to tar over a /dev directory from a working partition
on the same machine, but at this point I can bring up X11 for root,
but not for myself as a user (sigh).

I had this same problem a couple of months ago, and just gave up
and reinstalled the OS.  I hate to do that as it may only take 30-45m
to do the reinstall, but it takes another week to get everything
configured back the way it was...

Needless to say, Im not going to use

        udev_root="/dev/"
        udev_db="/dev/.udev.tdb

again, anthough it sure makes life a lot simpler.

And, Im going to have to go and remove it from some machines at the
lab, before disaster strikes there too.


                                Reg.Clemens
                                reg@dwf.com


[ Side question: There is a .udev.tdb sub directory in an old /udev
directory that hasnt been used since February.  There is an
empty .udev.tdb subdirectory in /dev .  Any chance udev could
have looked at /dev seen the empty subdirectory, and then looked
in /udev ?  GAKKK....




-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Greg KH]

On Mon, Apr 18, 2005 at 12:13:38PM -0600, clemens@dwf.com wrote:
> is.
> 
> If it is legal, then I have a complaint...
> 
> This has happened to me twice now (three times if you count the twice
> yesterday while I was trying to dig my way out).
> 
> It would seem, that on a reboot (on the way down, on the way up, Im not
> sure) that udev proceeds to remove all files that were active at the
> time the system went down.  I will assume that it created all of these
> files, so it thinks that it can remove them.

No, odds are /dev is mounted on a ram or tmpfs filesystem, which means
the whole thing was thrown away when the machine was rebooted.  udev did
not remove anything on it's own.

> Hey what ever happened to the MKNOD (?) script that used to be in
> UNIX systems?  I dont see it in Fedora2.  That would have helped in
> the repair.

Try asking the Fedora people, that's up to them, and how they package
udev.

> I had this same problem a couple of months ago, and just gave up
> and reinstalled the OS.  I hate to do that as it may only take 30-45m
> to do the reinstall, but it takes another week to get everything
> configured back the way it was...

Sounds like you need to learn about backing stuff up :)

> Needless to say, Im not going to use
> 
>         udev_root="/dev/"
>         udev_db="/dev/.udev.tdb
> 
> again, anthough it sure makes life a lot simpler.
> 
> And, Im going to have to go and remove it from some machines at the
> lab, before disaster strikes there too.

Might I suggest a newer distro than Fedora2?  Like one that actually
implementes udev properly?

thanks,

greg k-h


-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[David Zeuthen]

On Mon, 2005-04-18 at 13:49 -0700, Greg KH wrote:
> Might I suggest a newer distro than Fedora2?  Like one that actually
> implementes udev properly?

Actually Fedora didn't start using udev before Fedora Core 3 IIRC.

Cheers,
David




-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Greg KH]

On Mon, Apr 18, 2005 at 05:02:30PM -0400, David Zeuthen wrote:
> On Mon, 2005-04-18 at 13:49 -0700, Greg KH wrote:
> > Might I suggest a newer distro than Fedora2?  Like one that actually
> > implementes udev properly?
> 
> Actually Fedora didn't start using udev before Fedora Core 3 IIRC.

My point exactly :)


-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[clemens]

> On Mon, Apr 18, 2005 at 12:13:38PM -0600, clemens@dwf.com wrote:
> > is.
> > 
> > If it is legal, then I have a complaint...
> > 
> > This has happened to me twice now (three times if you count the twice
> > yesterday while I was trying to dig my way out).
> > 
> > It would seem, that on a reboot (on the way down, on the way up, Im not
> > sure) that udev proceeds to remove all files that were active at the
> > time the system went down.  I will assume that it created all of these
> > files, so it thinks that it can remove them.
> 
> No, odds are /dev is mounted on a ram or tmpfs filesystem, which means
> the whole thing was thrown away when the machine was rebooted.  udev did
> not remove anything on it's own.

No, /dev is just part of the root file system. /dev/shm is on tempf, but
thats irrevelant to the current problem.  I had assumed that the only reason
for udev to retain a database was so that it COULD do this.

As I noted, /dev/hde0-7 (which I mount) are gone, /dev/hde8-9 (which I dont)
are still there.  Very suspicious to me...


> 
> > Hey what ever happened to the MKNOD (?) script that used to be in
> > UNIX systems?  I dont see it in Fedora2.  That would have helped in
> > the repair.
> 
> Try asking the Fedora people, that's up to them, and how they package
> udev.
> 
Yes, I know, I was just asking the question.

> > I had this same problem a couple of months ago, and just gave up
> > and reinstalled the OS.  I hate to do that as it may only take 30-45m
> > to do the reinstall, but it takes another week to get everything
> > configured back the way it was...
> 
> Sounds like you need to learn about backing stuff up :)

Somehow, for the sorts of things that have happened, I wouldnt trust
a backup to fix them.

> 
> > Needless to say, Im not going to use
> > 
> >         udev_root="/dev/"
> >         udev_db="/dev/.udev.tdb
> > 
> > again, anthough it sure makes life a lot simpler.
> > 
> > And, Im going to have to go and remove it from some machines at the
> > lab, before disaster strikes there too.
> 
> Might I suggest a newer distro than Fedora2?  Like one that actually
> implementes udev properly?
> 
OK, sorry, a few lines got truncated off my original message when I
posted it.  The original started with:
OK, Im having problems, and Im sure its UDEV that is doing the deed.

[Fedora 2 and the latest udev ]

First, is it legal to say

        udev_root="/dev/"
        udev_db="/dev/.udev.tdb

If not, there should be a BIG sign on the /udev.conf page saying not to
do this.

If it is legal, ...

So, back to the original question.  IS THIS LEGAL?

And, as I said there, Fedora2 + the most recently released udev.

-- 
                                        Reg.Clemens
                                        reg@dwf.com




-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[Kay Sievers]

On Mon, 2005-04-18 at 15:27 -0600, clemens@dwf.com wrote:
> > On Mon, Apr 18, 2005 at 12:13:38PM -0600, clemens@dwf.com wrote:

> OK, sorry, a few lines got truncated off my original message when I
> posted it.  The original started with:
> OK, Im having problems, and Im sure its UDEV that is doing the deed.
> 
> [Fedora 2 and the latest udev ]
> 
> First, is it legal to say
> 
>         udev_root="/dev/"
>         udev_db="/dev/.udev.tdb

This is valid, sure. But that alone is not what you need to have it
working as you expect.

> If not, there should be a BIG sign on the /udev.conf page saying not to
> do this.
> 
> If it is legal, ...
> 
> So, back to the original question.  IS THIS LEGAL?
> 
> And, as I said there, Fedora2 + the most recently released udev.

You may investigate how Red Hat integrates udev into its distribution
instead of making your own experiments:
  ftp://download.fedora.redhat.com/pub/fedora/linux/core/development/i386/Fedora/RPMS/udev-056-2.i386.rpm

The system needs a lot of tweaking to work reliable with udev. It's not
a simple download and install on an old system.

Good luck,
Kay



-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

Re: udev problems

[clemens]

> 
> The system needs a lot of tweaking to work reliable with udev. It's not
> a simple download and install on an old system.
> 

You would seem to be saying that the instructions in the udev README
are not correct.  Ive never seen that statement made before.

Ive been using udev since it was first released, and never had this
problem till this spring.

-- 
                                        Reg.Clemens
                                        reg@dwf.com




-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728
_______________________________________________
Linux-hotplug-devel mailing list  http://linux-hotplug.sourceforge.net
Linux-hotplug-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel