Re: Problem in ppp & patch to solve

Mailing lists (lists@cnti.uanet.kharkov.ua)
Tue, 29 Sep 1998 12:26:34 +0300


In message <199809280956.LAA08015@oboe.it.uc3m.es> Peter T. Breuer
writes:

>"A month of sundays ago Mailing lists wrote:"
>>
>> In message <199809251039.NAA05055@sanders.piton.net> elf@piton.net
>> writes:
>>
>> >>I use small machine (P60) under linux-2.1 as a small dial-in modem
>> >>pool. Under 2.1.88 ppp works fine, but under 2.1.115-2.1.119 the
>> >>first users conection after reboot is OK, but after day of work
>> >>pppd don't sent LCP requests. I run pppd under strace and see
>> >>that syscall write() with LCP packet works OK, but no one byte
>> >>sent to modem. After reboot all again works correct for a some time.
>>
>> > I have the same problem with kernel 2.1.122 ;((
>> >>
>> >> Stanislav Voronyi.
>>
>> > Did you solve that problem? Have you any idea about cause?
>>
>> This is a small patch to fix this problem:
>> ------------------------------------------------------------------
>>
>> --- ppp.c.orig Mon Sep 28 11:02:28 1998
>> +++ ppp.c Fri Sep 25 20:52:07 1998
>> @@ -1289,6 +1289,9 @@
>>
>> CHECK_PPP_MAGIC(ppp);
>>
>> + /* ppp_dev_close may be called with tbusy==1 so we must set it to 0 */
>> + dev->tbusy=0;
>> +
>> MOD_DEC_USE_COUNT;
>>
>> return 0;

>Are you saying that all kernels from way back when have not been able to run
>ppp more than once per reboot?? That is quite an extraordinary thing to say
> ... and it makes me wonder. I have difficulties in 2.0.* running ppp more
>than once per session, but I put it down to my modem (usr sportster 14.4k)
>since resetting only the modem often cures it. Often, but not always. Maybe
>the remaining percentage come from an exit path from ppp that doesn't
>reset tbusy. I see several possible exits.

No, I am not say that ALL kernels have this bug. At least 2.1.88-
haven't (may be 2.1.9x too, I don't test it). And ppp not run only once.
But if I run havy loaded ppp dial-in machine some times hangup occur while
tbusy flag set to 1 & no one of function called during hangup don't set it
back to 0. So next call wich get same pppX device don't send LCP requsts
to the line (why so see ppp.c). As you can see from quoting this problems
occur not only for me. Withot this patch my macine can't work properly
more than 1 day. With this patch look - 12:22pm up 3 days, 15:19 -
and no one problem with sending LCP config requsts.

SY, Stanislav Voronyi.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/