[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: `linux debug' causing hang when Lilo is to be run.



>>>>> "Erik" == Erik Andersen <andersen@xmission.com> writes:

    Erik> On Thu Mar 23, 2000 at 08:08:17PM -0500, Adam Di Carlo wrote:
    >> > 
    >> >  I also ran into the "it won't reboot" problem, and the `dbootstrap'
    >> >  isn't respawned bug.  I have a feeling that in both cases, if
    >> >  `cardmgr' is killed, `init' might continue and execute those
    >> >  actions... ?  It might be blocking trying to write to the logs.
    >> 
    >> I never tried to kill cardmgr but it looks like this is the bug.  I
    >> agree it seems to be syslog somehow.  What cardmgr has to do with
    >> syslog I have no idea -- maybe understanding that would shed light.
    >> 

    Erik> The reason I changed dbootstrap and busybox init to do the 
    Erik>     openlog("foo", 0, LOG_USER);
    Erik>     syslog(LOG_USER|LOG_INFO, theMessage);
    Erik>     closelog();
    Erik> thing was that if process A has the log open, and then
    Erik> process B tries to log stuff, process B loses...

    Erik> I wonder if somebody has the log locked open.

 `cardmgr' does.  From what I gather, you are supposed to be able to
 do `openlog()' at the top of your program, and not have to
 `closelog()' until the end of it.  There's something wrong with the
 way the `syslogd' is working, I think.  After the reboot, `cardmgr'
 is the identical same, and gets along fine with other programs using
 the `syslogd'.  So there must be an important difference of some kind
 in how `busybox' `syslogd' is handling the log socket or something.
 We need to RTFS standard `syslogd', (again) I _guess_.  I have not
 looked at it yet.

 I need to work on fixing the `installkernel' script from
 `debianutils', and on `liloconfig' from `lilo', then I need to fix
 the "lilo.conf" template in `dbootstrap' and the `install.sh' script
 in `boot-floppies'...  They want me to make it use symlinks from "/"
 into "/boot" for the "image=.." in the "lilo.conf" because that's
 where they've "always" been.  I have them in "/boot", which is where
 `installkernel' (which is run by the kernel Makefile when you do
 `make install' from there.) puts them currently.

 As soon as I finish that up, I will start RTFS standard `syslogd' to
 see if I can find what might be wrong.  Maybe you'll finish this up
 before I can get to it?

    Erik> BTW, sorry I didn't get to look into this today at work.

 That's ok.  Hope you have some time this evening for it.  See you on
 IRC?


Reply to: