Re: kerneld/request-route interaction bug with kernel patch
Chris Faylor (cgf@bbc.com)
Mon, 1 Jul 1996 12:24:26 -0400
>Hello!
>
>> I think that the problem is fixed by the patch below. Possibly there is
>> a better way to do this, but this is working for me. With this patch,
>> for example, the first 'rlogin' to a remote system will succeed. Without
>> it, the rlogin will fail.
>
>Oooouch! It means that kerneld_send sleeps... It is disaster.
>Even if your patch seems to be working, it will eventually destroy your system.
>
>You can two choices:
>- comment call to kerneld_send.
>- patch kerneld.h.
>
>static inline int kerneld_route(const char *ip_route)
>{
>+ if (intr_count) return NULL;
> return kerneld_send(KERNELD_REQUEST_ROUTE,
>- 0 | KERNELD_WAIT,
>+ 0 | KERNELD_NOWAIT,
> strlen(ip_route), ip_route, NULL);
>}
>
>
>Your rlogin will not work from first time, just because it is
>impossible for current TCP.
If this is a problem with my patch isn't it a problem for the current
implementation as well? 2.0.0 still calls kerneld_route.
Or is it a problem with *where* kerneld is being called?
--
http://www.bbc.com/ cgf@bbc.com "Strange how unreal
VMS=>UNIX Solutions Boston Business Computing the real can be."