Date: Wed, 22 Aug 2001 00:05:47 -0700 From: Peter Wemm <peter@wemm.org> To: Masahide -mac- NODA <mac@clave.gr.jp> Cc: stable@FreeBSD.ORG Subject: Re: 4.4-RC kernel can't boot on vmware Message-ID: <20010822070547.E996838FD@overcee.netplex.com.au> In-Reply-To: <20010822.153726.126569560.mac@clave.gr.jp>
next in thread | previous in thread | raw e-mail | index | archive | help
Masahide -mac- NODA wrote:
> Now, I cvsup RELENG_4 again(08/22 13:00 JST) and make kernel on vmware
> guest OS.
>
> It works fine! Why? I don't know, but, I'm happy. :)
>
> Probably, recent pci commit has some effect....
>
> Thank you! Peter!
I was half way through replying to the last message when this arrived.
:-)
I am glad it works. I have been talking with the VMWARE people and we came
to the conclusion that the SSE ifdef should not have made any difference.
They think that CPU_ENABLE_SSE should work too.
However, the big doubt is the size of the kernel stack. That ifdef that
I had you add saved about 336 bytes of kernel stack. Also, in the last day,
the PCIBIOS code was removed, and that uses quite a bit of kernel stack...
I think the problem was not with the change itself, but with the fact that
we were right on the wire of stack usage and were probably already
causing some damage. The larger 'union fpusave' plus PCIBIOS was enough
to break it.
In my other email, I was going to ask if you could configure your kernel
after changing sys/i386/include/param.h from:
#define UPAGES 2 /* pages of u-area (8kb) */
to:
#define UPAGES 3 /* pages of u-area (12kb) */
This would increase the kernel stack by 4k. I think this would have also
solved the problem.
This is getting to be a real worry though. I think we are overflowing
the stack more often than we are aware of. I think we should probably
consider increasing the stack size by a small amount before the release.
Cheers,
-Peter
> From: Peter Wemm <peter@wemm.org>
> Subject: Re: 4.4-RC kernel can't boot on vmware
> Date: Mon, 20 Aug 2001 00:31:03 -0700
> Message-ID: <20010820073103.142BD3811@overcee.netplex.com.au>
>
> peter> > peter> The only effect that the commit changed that is not masked by
> peter> > peter> the CPU_ENABLE_SSE that I know about is the change to the pcb
> peter> > peter> structure layout. I would like to know if the following
> peter> > peter> (UNTESTED) patch has any effect:
> peter> >
> peter> > peter> Index: npx.h
> peter> >
> peter> > I test this patch for cvsuped 4.4-RC kernel, compiled without
> peter> > CPU_ENABLE_SSE option. It works fine! (fpu.ko fail to compile)
> peter>
> peter> This is bad news. :-( It looks like vmware "knows" the layout of the
pcb
> peter> structure and is breaking when it doesn't recognize it anymore. :-(
peter>
> peter> I've got a query in with the vmware folks.
>
> --
> mac@clave.gr.jp/mac@jp.freebsd.org
>
>
--
Peter Wemm - peter@FreeBSD.org; peter@yahoo-inc.com; peter@netplex.com.au
"All of this is for nothing if we don't go to the stars" - JMS/B5
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010822070547.E996838FD>