ppp problems in 2.5.69-bk14

ppp problems in 2.5.69-bk14

[Brett]

Hey,

I got this lovely mess when trying to run ppp under 2.5.69-bk14

using devfs
all ppp is modular

thanks,

	/ Brett

PPP generic driver version 2.4.2
devfs_mk_cdev: could not append to parent for ppp
failed to register PPP device (-17)
Unable to handle kernel paging request at virtual address c38755c0
 printing eip:
c0150743
*pde = 01092067
*pte = 00000000
Oops: 0000 [#1]
CPU:    0
EIP:    0060:[<c0150743>]    Not tainted
EFLAGS: 00010286
EIP is at lookup_chrfops+0x83/0xd0
eax: c38755c0   ebx: c129c600   ecx: c1e4c000   edx: 00000000
esi: 00000000   edi: 00000000   ebp: 0000006c   esp: c1e4def0
ds: 007b   es: 007b   ss: 0068
Process pppd (pid: 402, threadinfo=c1e4c000 task=c145d380)
Stack: c1e4c000 00000000 0000006c 00000000 c01507da 0000006c 00000000 c2048b20 
       c2048b20 ffffffed c21adea0 c0150a42 0000006c 00000000 c2048b20 c21adea0 
       ffffffed c107b220 c01986c9 c21adea0 c2048b20 c2048b20 c21adea0 c10c6084 
Call Trace:
 [<c01507da>] get_chrfops+0x4a/0x70
 [<c0150a42>] chrdev_open+0x22/0xa0
 [<c01986c9>] devfs_open+0xd9/0x110
 [<c0147b8a>] dentry_open+0x1da/0x210
 [<c01479a0>] filp_open+0x50/0x60
 [<c0147e2b>] sys_open+0x3b/0x70
 [<c0108df7>] syscall_call+0x7/0xb

Code: 8b 00 be 01 00 00 00 85 c0 74 24 8b 69 14 45 89 69 14 83 38 
 <6>note: pppd[402] exited with preempt_count 2
bad: scheduling while atomic!
Call Trace:
 [<c0114fc7>] schedule+0x3b7/0x3c0
 [<c0139ddd>] unmap_vmas+0x1ed/0x260
 [<c013de66>] exit_mmap+0x66/0x180
 [<c0116763>] mmput+0x53/0xa0
 [<c011a19a>] do_exit+0x10a/0x420
 [<c0109654>] die+0xc4/0xd0
 [<c0112d51>] do_page_fault+0x111/0x469
 [<c0125247>] queue_work+0x97/0xb0
 [<c012516e>] call_usermodehelper+0xbe/0xd0
 [<c0125060>] __call_usermodehelper+0x0/0x50
 [<c0112c40>] do_page_fault+0x0/0x469
 [<c010905d>] error_code+0x2d/0x40
 [<c0150743>] lookup_chrfops+0x83/0xd0
 [<c01507da>] get_chrfops+0x4a/0x70
 [<c0150a42>] chrdev_open+0x22/0xa0
 [<c01986c9>] devfs_open+0xd9/0x110
 [<c0147b8a>] dentry_open+0x1da/0x210
 [<c01479a0>] filp_open+0x50/0x60
 [<c0147e2b>] sys_open+0x3b/0x70
 [<c0108df7>] syscall_call+0x7/0xb

PPP generic driver version 2.4.2
failed to register PPP device (-16)
ppp_async: Unknown symbol ppp_channel_index
ppp_async: Unknown symbol ppp_register_channel
ppp_async: Unknown symbol ppp_input
ppp_async: Unknown symbol ppp_input_error
ppp_async: Unknown symbol ppp_output_wakeup
ppp_async: Unknown symbol ppp_unregister_channel
ppp_async: Unknown symbol ppp_unit_number

Re: ppp problems in 2.5.69-bk14

[Matthew Harrell]

Same issue under 2.5.70-bk1


> Hey,
> 
> I got this lovely mess when trying to run ppp under 2.5.69-bk14
> 
> using devfs
> all ppp is modular
> 
> thanks,
> 
> 	/ Brett
> 
> PPP generic driver version 2.4.2
> devfs_mk_cdev: could not append to parent for ppp
> failed to register PPP device (-17)
> Unable to handle kernel paging request at virtual address c38755c0
>  printing eip:
> c0150743
> *pde = 01092067
> *pte = 00000000
> Oops: 0000 [#1]
> CPU:    0
> EIP:    0060:[<c0150743>]    Not tainted
> EFLAGS: 00010286
> EIP is at lookup_chrfops+0x83/0xd0
> eax: c38755c0   ebx: c129c600   ecx: c1e4c000   edx: 00000000
> esi: 00000000   edi: 00000000   ebp: 0000006c   esp: c1e4def0
> ds: 007b   es: 007b   ss: 0068
> Process pppd (pid: 402, threadinfo=c1e4c000 task=c145d380)
> Stack: c1e4c000 00000000 0000006c 00000000 c01507da 0000006c 00000000 c2048b20 
>        c2048b20 ffffffed c21adea0 c0150a42 0000006c 00000000 c2048b20 c21adea0 
>        ffffffed c107b220 c01986c9 c21adea0 c2048b20 c2048b20 c21adea0 c10c6084 
> Call Trace:
>  [<c01507da>] get_chrfops+0x4a/0x70
>  [<c0150a42>] chrdev_open+0x22/0xa0
>  [<c01986c9>] devfs_open+0xd9/0x110
>  [<c0147b8a>] dentry_open+0x1da/0x210
>  [<c01479a0>] filp_open+0x50/0x60
>  [<c0147e2b>] sys_open+0x3b/0x70
>  [<c0108df7>] syscall_call+0x7/0xb
> 
> Code: 8b 00 be 01 00 00 00 85 c0 74 24 8b 69 14 45 89 69 14 83 38 
> <6>note: pppd[402] exited with preempt_count 2
> bad: scheduling while atomic!
> Call Trace:
>  [<c0114fc7>] schedule+0x3b7/0x3c0
>  [<c0139ddd>] unmap_vmas+0x1ed/0x260
>  [<c013de66>] exit_mmap+0x66/0x180
>  [<c0116763>] mmput+0x53/0xa0
>  [<c011a19a>] do_exit+0x10a/0x420
>  [<c0109654>] die+0xc4/0xd0
>  [<c0112d51>] do_page_fault+0x111/0x469
>  [<c0125247>] queue_work+0x97/0xb0
>  [<c012516e>] call_usermodehelper+0xbe/0xd0
>  [<c0125060>] __call_usermodehelper+0x0/0x50
>  [<c0112c40>] do_page_fault+0x0/0x469
>  [<c010905d>] error_code+0x2d/0x40
>  [<c0150743>] lookup_chrfops+0x83/0xd0
>  [<c01507da>] get_chrfops+0x4a/0x70
>  [<c0150a42>] chrdev_open+0x22/0xa0
>  [<c01986c9>] devfs_open+0xd9/0x110
>  [<c0147b8a>] dentry_open+0x1da/0x210
>  [<c01479a0>] filp_open+0x50/0x60
>  [<c0147e2b>] sys_open+0x3b/0x70
>  [<c0108df7>] syscall_call+0x7/0xb
> 
> PPP generic driver version 2.4.2
> failed to register PPP device (-16)
> ppp_async: Unknown symbol ppp_channel_index
> ppp_async: Unknown symbol ppp_register_channel
> ppp_async: Unknown symbol ppp_input
> ppp_async: Unknown symbol ppp_input_error
> ppp_async: Unknown symbol ppp_output_wakeup
> ppp_async: Unknown symbol ppp_unregister_channel
> ppp_async: Unknown symbol ppp_unit_number
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

-- 
  Matthew Harrell                          May the bluebird of happiness
  Bit Twiddlers, Inc.                       twiddle your bits.
  mharrell@bittwiddlers.com     

Re: ppp problems in 2.5.69-bk14 - devfs related?

[Matthew Harrell]

My oops output is just marginally different under 2.5.70-bk2.
Unfortunately, I don't seem to have a /proc/ksyms so I don't know what
to point ksymoops to.  I can make this one happen over and over by
just running pppd.  It does not kill the system but it does make it
rather unuseable.


CSLIP: code copyright 1989 Regents of the University of California
PPP generic driver version 2.4.2
devfs_mk_cdev: could not append to parent for ppp
failed to register PPP device (-17)
Unable to handle kernel paging request at virtual address e0927c80
c015a226
*pde = 0162e067
Oops: 0000 [#1]
CPU:    0
EIP:    0060:[<c015a226>]    Not tainted
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010282
eax: dcdb7a00   ebx: e0927c80   ecx: 0000006c   edx: da17df24
esi: 00000000   edi: 00006c00   ebp: dcdb7a00   esp: da17ded8
ds: 007b   es: 007b   ss: 0068
Stack: c015e070 ddc2f940 00000000 c015a12f dcdb7a00 c022ee80 00006c00 dcdb7a00 
       00000000 c015a110 000000ff 00000000 00000000 00000000 defde200 c0159fb2 
       defe3400 00006c00 da17df24 00000000 defde200 ffffffed db6cb500 defe1f00 
Call Trace:
 [<c015e070>] do_lookup+0x30/0xb0
 [<c015a12f>] exact_lock+0xf/0x20
 [<c022ee80>] kobj_lookup+0xe0/0x170
 [<c015a110>] exact_match+0x0/0x10
 [<c0159fb2>] chrdev_open+0xe2/0x150
 [<c019e3d0>] devfs_open+0xb0/0xd0
 [<c0150a20>] dentry_open+0x110/0x1a0
 [<c0150908>] filp_open+0x68/0x70
 [<c0150cdb>] sys_open+0x5b/0x90
 [<c010b14f>] syscall_call+0x7/0xb
Code: 83 3b 02 74 41 ff 83 00 01 00 00 89 04 24 e8 57 c4 06 00 85 


>>EIP; c015a226 <cdev_get+16/60>   <=====

>>eax; dcdb7a00 <__crc_scsi_get_command+e98e4/386137>
>>ebx; e0927c80 <__crc_elv_queue_empty+7279d/135f59>
>>edx; da17df24 <__crc_sock_rfree+2f6ef9/3b231c>
>>ebp; dcdb7a00 <__crc_scsi_get_command+e98e4/386137>
>>esp; da17ded8 <__crc_sock_rfree+2f6ead/3b231c>

Trace; c015e070 <do_lookup+30/b0>
Trace; c015a12f <exact_lock+f/20>
Trace; c022ee80 <kobj_lookup+e0/170>
Trace; c015a110 <exact_match+0/10>
Trace; c0159fb2 <chrdev_open+e2/150>
Trace; c019e3d0 <devfs_open+b0/d0>
Trace; c0150a20 <dentry_open+110/1a0>
Trace; c0150908 <filp_open+68/70>
Trace; c0150cdb <sys_open+5b/90>
Trace; c010b14f <syscall_call+7/b>

Code;  c015a226 <cdev_get+16/60>
00000000 <_EIP>:
Code;  c015a226 <cdev_get+16/60>   <=====
   0:   83 3b 02                  cmpl   $0x2,(%ebx)   <=====
Code;  c015a229 <cdev_get+19/60>
   3:   74 41                     je     46 <_EIP+0x46>
Code;  c015a22b <cdev_get+1b/60>
   5:   ff 83 00 01 00 00         incl   0x100(%ebx)
Code;  c015a231 <cdev_get+21/60>
   b:   89 04 24                  mov    %eax,(%esp,1)
Code;  c015a234 <cdev_get+24/60>
   e:   e8 57 c4 06 00            call   6c46a <_EIP+0x6c46a>
Code;  c015a239 <cdev_get+29/60>
  13:   85 00                     test   %eax,(%eax)


1 warning and 1 error issued.  Results may not be reliable.

-- 
  Matthew Harrell                          Artificial Intelligence is no
  Bit Twiddlers, Inc.                       match for natural stupidity
  mharrell@bittwiddlers.com     

Re: ppp problems in 2.5.69-bk14 - devfs related?

[viro]

On Wed, May 28, 2003 at 05:27:08PM -0400, Matthew Harrell wrote:
> 
> My oops output is just marginally different under 2.5.70-bk2.
> Unfortunately, I don't seem to have a /proc/ksyms so I don't know what
> to point ksymoops to.  I can make this one happen over and over by
> just running pppd.  It does not kill the system but it does make it
> rather unuseable.

Fsck knows why devfs_mk_cdev() fails, but what follows that is obvious -
int __init ppp_init(void)
{
        int err;

        printk(KERN_INFO "PPP generic driver version " PPP_VERSION "\n");
        err = register_chrdev(PPP_MAJOR, "ppp", &ppp_device_fops);
        if (!err) {
                err = devfs_mk_cdev(MKDEV(PPP_MAJOR, 0),
                                S_IFCHR|S_IRUSR|S_IWUSR, "ppp");
        }

        if (err)
                printk(KERN_ERR "failed to register PPP device (%d)\n", err);
        return err;
}
clearly leaves device registered after failed insmod.  open() afterwards
happily finds the device and dies on attempt to do anything with it
(the module is not there, pointers go to hell knows where).

I'd suggest to change that to
	err = devfs_mk_cdev(...)
	if (!err) {
		err = register_chrdev(...)
		if (!err)
			return 0;
		devfs_remove(...)
	}
	printk(...)
	return err;

That will _not_ solve the devfs problem, whatever it is, but it will make sure
that any errors are handled correctly.
	

Re: ppp problems in 2.5.69-bk14 - devfs related?

[Matthew Harrell]

Well, this does stop the oops but it also prevents ppp from working.  I now get
this when I execute pppd

  pppd: This system lacks kernel support for PPP.  This could be because
  the PPP kernel module could not be loaded, or because PPP was not
  included in the kernel configuration.  If PPP was included as a
  module, try `/sbin/modprobe -v ppp'.  If that fails, check that
  ppp.o exists in /lib/modules/`uname -r`/net.
  See README.linux file in the ppp distribution for more details.

and the kernel logs show this

  CSLIP: code copyright 1989 Regents of the University of California
  PPP generic driver version 2.4.2
  devfs_mk_cdev: could not append to parent for ppp


: Fsck knows why devfs_mk_cdev() fails, but what follows that is obvious -
: int __init ppp_init(void)
: {
:         int err;
: 
:         printk(KERN_INFO "PPP generic driver version " PPP_VERSION "\n");
:         err = register_chrdev(PPP_MAJOR, "ppp", &ppp_device_fops);
:         if (!err) {
:                 err = devfs_mk_cdev(MKDEV(PPP_MAJOR, 0),
:                                 S_IFCHR|S_IRUSR|S_IWUSR, "ppp");
:         }
: 
:         if (err)
:                 printk(KERN_ERR "failed to register PPP device (%d)\n", err);
:         return err;
: }
: clearly leaves device registered after failed insmod.  open() afterwards
: happily finds the device and dies on attempt to do anything with it
: (the module is not there, pointers go to hell knows where).

-- 
  Matthew Harrell                          Life is like a diaper -
  Bit Twiddlers, Inc.                       short and loaded.
  mharrell@bittwiddlers.com     

Re: ppp problems in 2.5.69-bk14

[Matthew Harrell]

And just FYI this problem still exists in the latest 2.5.70-bk9.  If someone
would like to give an idea how to fix it and yet still maintain ppp capability
then I would be happy to experiment and try to get something working

-- 
  Matthew Harrell                          I love defenseless animals,
  Bit Twiddlers, Inc.                       especially in a good gravy.
  mharrell@bittwiddlers.com