Skip to contentSkip to conversation

New: pppd "IPCP: timeout sending Config-Requests"

Bug 3592 Russell King 4 participants 13 messages Oct 19, 2004 – Nov 20, 2004

lore
Share & tools
lore
#0 lore raw

Can someone else look at this - I presently have no time for any "community" work from the end of this week onwards, and shall be bouncing all bugs to lkml. I'm expecting this situation to continue until maybe Christmas time.

I'm going to try to flush as much stuff from my patch queues as possible for the remainder of this week, and then I think it'll be up to others.

mbligh - please assign the default owner for:
- PCMCIA bugs to [email protected]
- serial bugs to, I guess, [email protected]

I'm not sure at present what's going to happen with "community" ARM kernel work, but I think it will certainly not be progressing very quickly. I'll do what I can but I fear it won't be very much.

----- Forwarded message from [email protected] -----

Date: Tue, 19 Oct 2004 04:53:50 -0700
From: [email protected]
To: [email protected]
Subject: [Bug 3592] New: pppd "IPCP: timeout sending Config-Requests"

http://bugme.osdl.org/show_bug.cgi?id=3592

Summary: pppd "IPCP: timeout sending Config-Requests"
Kernel Version: 2.6.9-rc4
Status: NEW
Severity: normal
Owner: [email protected]
Submitter: [email protected]

Distribution: Gentoo
Hardware Environment:
Software Environment: ppp-2.4.1 ( tested with ppp-2.4.2 - the same error ) Problem Description: With 2.6.9-rc4 I'm getting the error when dialing to my provider. With older kernel versions everything works fine. After remote modem cosed the connection (rcvd [CHAP Success id=0x29 ""]) pppd keeps sending Config-Requests
sent [IPCP ConfReq id=0x1 <addr 192.168.1.2> <compress VJ 0f 01>]

sent [CCP ConfReq id=0x1 <deflate 15> <deflate(old#) 15> <bsd v1 15>]

Steps to reproduce:
1) Install kernel 2.6.9-rc4
2) use the following script to connect ( chap1 )
REPORT 'CONNECT'

ABORT 'BUSY'

ABORT 'ERROR'

ABORT 'NO ANSWER'

ABORT 'NO CARRIER'

ABORT 'NO DIALTONE'

ABORT 'Invalid Login'

ABORT 'Login incorrect'

TIMEOUT 45

'' 'ATZ'

'OK' 'ATS32=98'

'OK' 'ATS0=1'

'OK' 'ATDT0*******'

'CONNECT' ''

'~--' ''
3) pppd command:
/usr/sbin/pppd debug nodetach lock modem crtscts asyncmap 00000000 mru 768 mtu 768 user username remotename NT4RAS /dev/ttyS4 115200 noauth linkname ppp0 connect /usr/sbin/chat -v -f /etc/ppp/chat1

------- You are receiving this mail because: ------- You are the assignee for the bug, or are watching the assignee.

----- End forwarded message -----

Signature · 4 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
                 2.6 Serial core
#1 lore raw
Quoted 5 lines · Russell King

Software Environment: ppp-2.4.1 ( tested with ppp-2.4.2 - the same error ) Problem Description: With 2.6.9-rc4 I'm getting the error when dialing to my provider. With older kernel versions everything works fine. After remote modem cosed the connection (rcvd [CHAP Success id=0x29 ""]) pppd keeps sending Config-Requests

This looks like the tty locking changes from Alan Cox.

The tty_io.c do_tty_hangup() no longer switches the line discipline back to N_TTY, so ppp_async.c is not aware of the hangup (ldisc->close not called).

The following is a snippet from tty_io.c:

	/* Defer ldisc switch */
	/* tty_deferred_ldisc_switch(N_TTY);
	
	  This should get done automatically when the port closes and
	  tty_release is called */

I'll setup a test connection and verify this.

Signature · 2 lines
Paul Fulghum
[email protected]
#2 lore raw
Quoted 8 lines · Paul Fulghum

This looks like the tty locking changes from Alan Cox.

The tty_io.c do_tty_hangup() no longer switches the line discipline back to N_TTY, so ppp_async.c is not aware of the hangup (ldisc->close not called).

The following is a snippet from tty_io.c:

Quoted 7 lines
	/* Defer ldisc switch */
	/* tty_deferred_ldisc_switch(N_TTY);
	
	  This should get done automatically when the port closes and
	  tty_release is called */

I'll setup a test connection and verify this.

I have verified the problem as described above.

PPP line disciplines rely on the previous behavior of calling ldisc->close on hangup as a method for indicating hangup to the line discipline.
This is explicitly called out in the PPP ldisc comments. Other line disciplines may also rely on this behavior.

Alan's changes also added the ldisc->hangup() method to indicate hangup, but all the line disciplines must be modified to implement this method.

Signature · 2 lines
Paul Fulghum
[email protected]
#3 lore raw
Quoted 4 lines · Paul Fulghum

PPP line disciplines rely on the previous behavior of calling ldisc->close on hangup as a method for indicating hangup to the line discipline.
This is explicitly called out in the PPP ldisc comments.

I had no choice about that really with the current locking. It's on the list to do further work although I'd not realised some odder pppd configurations relied upon it until the bug report.

Once I've put out -ac1 to fix the other bugs I consider urgent (not tty) I'll see what I can do. Really it would nice if the ppp maintainer would look at this and also fix all the horrible things the code does wrongly if for example the first byte of a received buffer is an error marker - in general serial error processing is not robust in the ppp code it appears.

Alan

#4 lore raw
Quoted 9 lines · Alan Cox
Quoted 4 lines · Paul Fulghum

PPP line disciplines rely on the previous behavior of calling ldisc->close on hangup as a method for indicating hangup to the line discipline.
This is explicitly called out in the PPP ldisc comments.

I had no choice about that really with the current locking. It's on the list to do further work although I'd not realised some odder pppd configurations relied upon it until the bug report.

OK

I'm not sure I would characterize using DCD
for a serial connection indicator as odd.

Quoted 6 lines

Once I've put out -ac1 to fix the other bugs I consider urgent (not tty) I'll see what I can do. Really it would nice if the ppp maintainer would look at this and also fix all the horrible things the code does wrongly if for example the first byte of a received buffer is an error marker - in general serial error processing is not robust in the ppp code it appears.

I'll look at what is needed to implement the new ldisc->hangup() method for the PPP line disciplines.

Signature · 2 lines
Paul Fulghum
[email protected]
#5 lore raw
Quoted 2 lines · Paul Fulghum

I'm not sure I would characterize using DCD
for a serial connection indicator as odd.

The default setup trusts LCP closes sent between the two ends

Quoted 5 lines
Quoted 2 lines

in general serial error processing is not robust in the ppp code it appears.

I'll look at what is needed to implement the new ldisc->hangup() method for the PPP line disciplines.

Thanks

#6 lore raw
Quoted 5 lines · Alan Cox
Quoted 2 lines · Paul Fulghum

I'm not sure I would characterize using DCD
for a serial connection indicator as odd.

The default setup trusts LCP closes sent between the two ends

Yes.

In the case of this bug report, it looks like Window NT4 is dropping the connection without sending the LCP close. That is crappy behavior.

DCD makes a good check on the physical layer in case your POTS line is on the crackly side.
May I never suffer dialup again :-P

Signature · 2 lines
Paul Fulghum
[email protected]
#7 lore raw
In reply to Russell King
Quoted 7 lines · Russell King

----- Forwarded message from [email protected] -----

Date: Tue, 19 Oct 2004 04:53:50 -0700
From: [email protected]
To: [email protected]
Subject: [Bug 3592] New: pppd "IPCP: timeout sending Config-Requests"

Quoted 8 lines

http://bugme.osdl.org/show_bug.cgi?id=3592

Summary: pppd "IPCP: timeout sending Config-Requests"
Kernel Version: 2.6.9-rc4
Status: NEW
Severity: normal
Owner: [email protected]
Submitter: [email protected]

So, what can I do with this bug? Just close or reject it, or what? Maybe Alan or Paul would like to assign this bug to themselves?

Thanks.

Signature · 4 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
                 2.6 Serial core
#8 lore raw
Quoted 3 lines · Russell King

So, what can I do with this bug? Just close or reject it, or what? Maybe Alan or Paul would like to assign this bug to themselves?

I thought that was fixed in 10rc and 2.6.9-ac

#9 lore raw
Quoted 6 lines · Alan Cox
Quoted 3 lines · Russell King

So, what can I do with this bug? Just close or reject it, or what? Maybe Alan or Paul would like to assign this bug to themselves?

I thought that was fixed in 10rc and 2.6.9-ac

Maybe, I have no idea - no one updated the bug, and I'd just like to get rid of it one way or the other.

I'm tempted to reject it with a comment "status unknown, please resubmit if it remains a problem" - which kind of makes a mockery of bug tracking systems.

Signature · 4 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
                 2.6 Serial core
#10 lore raw
Quoted 8 lines · Russell King
Quoted 6 lines · Alan Cox
Quoted 3 lines · Russell King

So, what can I do with this bug? Just close or reject it, or what? Maybe Alan or Paul would like to assign this bug to themselves?

Quoted 5 lines
Quoted 1 line

I thought that was fixed in 10rc and 2.6.9-ac

Maybe, I have no idea - no one updated the bug, and I'd just like to get rid of it one way or the other.

This was fixed (post 2.6.9) with the addition of the ldisc->hangup method in the ppp_async and ppp_synctty line disciplines.

The bug reporter has a Windows NT server that terminates the connection without sending the proper LCP packets. The Linux box had previously relied on DCD negation to indicate loss of connection via hangup. Alan's locking changes removed the ldisc->close from do_hangup() so the line discipline was no longer aware of the hangup. Adding the ldisc->hangup method restored detection of connection loss.

I reproduced this in lab, and tested the fix successfully. I don't remember which version the patch was accepted.

--
Paul Fulghum
[email protected]

#11 lore raw
Quoted 9 lines · Paul Fulghum
Quoted 8 lines · Russell King
Quoted 6 lines · Alan Cox
Quoted 3 lines · Russell King

So, what can I do with this bug? Just close or reject it, or what? Maybe Alan or Paul would like to assign this bug to themselves?

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

I thought that was fixed in 10rc and 2.6.9-ac

Maybe, I have no idea - no one updated the bug, and I'd just like to get rid of it one way or the other.

This was fixed (post 2.6.9) with the addition of the ldisc->hangup method in the ppp_async and ppp_synctty line disciplines.

Right, I've closed the bug then. It's a shame someone else couldn't have done that when the bug was known to have been fixed.

Signature · 4 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
                 2.6 Serial core
#12 lore raw
Quoted 2 lines · Russell King

Right, I've closed the bug then. It's a shame someone else couldn't have done that when the bug was known to have been fixed.

My bad.

I had not previously participated in this particular bugzilla database. In the future, if I supply a fix I know to be associated with a bug tracker, and I am allowed access to change that bug, I will do so.

--
Paul Fulghum
[email protected]