Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 31 Jan 2005 09:53:12 -0600
From:      Billy Newsom <smartweb@leadhill.net>
To:        freebsd-stable@freebsd.org
Subject:   Re: 5.3 STABLE kernel compilation oddity
Message-ID:  <41FE5468.3000804@leadhill.net>
In-Reply-To: <a8b8bb510501310731517789dc@mail.gmail.com>
References:  <a8b8bb510501310731517789dc@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
Alex Burke wrote:

> Hi,
> 
> I wasnt quite sure what to call this problem.
> 
> I am running FreeBSD 5.3 STABLE on a quad Pentium Pro IBM Netfinity
> 7000 system, and since the GENERIC kernel does not come SMP enabled I
> compiled a kernel with SMP support and drivers for networks card and
> SCSI RAID adapter built into it (not loaded as modules). This build
> happens fine under the original GENERIC kernel.
> 
> Afterwards, since that machine is quite powerful and its destined to
> be a unix server I wanted to compile a kernel for another box I have.
> I wrote the kernel config file, and started the compilation. At some
> stage the system crashed, on the console it looked as though it had
> tried to reboot itself (3 times!) with various messages about CPUs
> ignoring requests. I dont really understand why under the SMP kernel I
> compiled the I cannot compile another kernel, but under GENERIC it
> works fine.
> 
> I can only think two things, either I have stumbled accross an odd bug
> or my hardware appears to be working fine on the surface and actually
> there is something slightly wrong, although this machine ran the
> GENERIC kernel of 5.2 for a long time without any issues (5.2.1 was
> SMP enabled from the start I believe).
I wish FreeBSD 5.3 was as stable as the STABLE tree implies it was.  My SMP 
problems are quite numerous on 5.3 so far.

I have a similar system to yours, but a dual processor.  One thing I see is 
that as the days have gone by since the original 5.3-Release, the SMP code 
has had some code changes that have alternately caused crashes or 
compile-time errors, at least from my point of view.  I am beginning to 
regret using 5.3, but all of the press releases nudged me to believe that 5.3 
ws ready for live systems.  I have to admit that I had problems with 4.7 and 
4.9, but not like the 5.3-STABLE track.  I hope all of this doesn't give 
FreeBSD a bad reputation, although it might earn it if it continues to be so 
buggy for SMP systems.  Having no experience with 5.2, I almost wish I could 
provide more advice.  But perhaps you can get the generic kernel off the CD 
and boot to it to recompile your kernel.

See this as one documented example:
ftp://ftp.freebsd.org/pub/FreeBSD/ERRATA/notices/FreeBSD-EN-05:03.ipi.asc

My last cvs update was the 30th (yesterday) and it compiled for me using
make -j6 buildworld and took 3:38.  My kernel build took 1:27.  Those are 
with a working SMP kernel, so the generic kernel would take longer.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?41FE5468.3000804>