[Raw Msg Headers][Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: strange scheduler log (again)





  Your   news->email  gateway does not carry the headers properly over ?

> Newsgroups: local.mailgate.zmailer
> Path: acli
> From: acli@www.mingpaoxpress.com (Ambrose Li)
> Subject: Re: strange scheduler log (again)
> Distribution: local
> Organization: Ming Pao Daily News (Canada)
> Message-ID: <E2B77K.G32@www.mingpaoxpress.com>
> References: <Pine.LNX.3.91.961209004923.3312A-100000@monet.mingpaoxpress.com> <d6wiv6b8qbp.fsf@lupinella.troll.no>
> Date: Thu, 12 Dec 1996 16:30:56 GMT
> 
> In article <d6wiv6b8qbp.fsf@lupinella.troll.no>,
> Arnt Gulbrandsen  <arnt@gulbrandsen.priv.no> wrote:
> >> 	...
> >>         libc.so.4 (DLL Jump 4.6pl27) => /lib/libc.so.4.6.27
> >
> >For some reason you're exec()'ing a program and giving it a null argv.
> >(The above message is shown if **argv == '\0'.)
> 
> Turns out that we need the following patch for the scheduler:
...

	Patch taken in.
...
> I still don't know why this doesn't happen on Linux box at home.
> But that's irrelevant now.

	Different dynamic library ?  That seems to indicate that
	you have an A.OUT system (DLL Jump) instead of the newer
	ELF system.

> Now another question: Why do we exec /bin/false and then _exit(0)?
> Wouldn't _exit(1) do the same thing? Please excuse my ignorance if
> this is obvious.

	It isn't obvious;  It is to silence the memory access
	debuggers when it has just done an accept() on the
	mailq -socket, and processed (with some allocations)
	the thing.  Now doing plain exit() would generate
	heaps of reports on what memory blocks were still in use
	at the time of the free() ...

> -- 
> Ambrose Li. acli@mingpaoxpress.com. Ming Pao Newspapers (Canada) Ltd., EDP

	/Matti Aarnio <mea@nic.funet.fi>