AF:
NF:0
PS:10
SRH:1
SFN:
DSR:
MID:<20090423183224.550e73b5@ripper.onstor.net>
CFG:
PT:0
S:andy.sharp@onstor.com
RQ:
SSV:mail.onstor.net
NSV:
SSH:
R:<rendell.fong@onstor.com>,<Bill.Fisher@onstor.com>
MAID:1
X-Sylpheed-Privacy-System:
X-Sylpheed-Sign:0
SCF:#mh/Mailbox/sent
RMID:#imap/andys@onstor.net@exch1.onstor.net/INBOX	0	2779531E7C760D4491C96305019FEEB52AC8BC1D4E@exch1.onstor.net
X-Sylpheed-End-Special-Headers: 1
Date: Thu, 23 Apr 2009 18:33:44 -0700
From: Andrew Sharp <andy.sharp@onstor.com>
To: Rendell Fong <rendell.fong@onstor.com>
Cc: Bill Fisher <Bill.Fisher@onstor.com>
Subject: Re: Getting FP loaded/idleing with Linux on TXRX
Message-ID: <20090423183344.6f650b85@ripper.onstor.net>
In-Reply-To: <2779531E7C760D4491C96305019FEEB52AC8BC1D4E@exch1.onstor.net>
References: <20090423174757.599a1f26@ripper.onstor.net>
	<2779531E7C760D4491C96305019FEEB52AC8BC1D4E@exch1.onstor.net>
Organization: Onstor
X-Mailer: Sylpheed-Claws 2.6.0 (GTK+ 2.8.20; x86_64-pc-linux-gnu)
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

On Thu, 23 Apr 2009 18:02:38 -0700 Rendell Fong
<rendell.fong@onstor.com> wrote:

> See comments below.
> 
> 
> > -----Original Message-----
> > From: Andy Sharp
> > Sent: Thursday, April 23, 2009 5:48 PM
> > To: Rendell Fong
> > Cc: Bill Fisher
> > Subject: Re: Getting FP loaded/idleing with Linux on TXRX
> > 
> > All images are loaded by the load_vmlinux program on the ssc or the
> > prom on the SSC.
> > 
> > The individual prom on each node starts that node running.  After that,
> > it's a horse race...
> > 
> > So I don't see any good reason to disable starting the other cores on
> > the FP.  Some other things might need to be disabled...
> 
> If you don't disable the other cores (temporarily), they will crash
> as soon as they are loaded for some reason or another.  The init routine
> is done via smp_entry() which is different than FP0.  So it's a separate
> case that can be addressed later once FP0 is more operational.  I didn't
> want to deal with those additional complications.  Starting with a
> minimally working image worked for me since it allowed me to test the
> code of interest (crashdump code) w/o being affected by other issues.
> But I won't speak for Bill.

Probably were crashing because they wanted to chat with chassid or
something.  Or Mr. WKA wasn't initialized by the TXRX or somesuch.

> > The shared memory area I know nothing about, and WKA is a mystery
> > pseudonym for me.  If the FP tries to write to memory that the TUXRX can
> > see, then it will almost certainly cause TUXRX to have fits.
> > 
> > If we need to set aside some chunk of memory on the tuxrx, then just
> > let me know how big and if there's a preference on where, otherwise
> > I'll just take it off the top, which is the easiest.  Minus "the hole"
> > of course.
> > 
> > 
> > 
> > On Thu, 23 Apr 2009 17:40:02 -0700 Rendell Fong
> > <rendell.fong@onstor.com> wrote:
> > 
> > > I managed to create a hacked up FP image that only starts FP0 and does
> > minimal initialization enough to allow it to get to the FP console prompt.
> > Since the eee communication isn't working and wka hasn't been adjusted
> > appropriately, I've skipped initialization of any code that needs to send
> > msgs and sync with apps running in other cores.
> > >
> > > It was enough to allow me to test the "crashdump panic" on both fp and
> > tuxrx.  I suppose as more components become operational, it can be added
> > back into the startup.  Is this what you are looking for?
> > >
> > > My workaround code changes are in sm-test/test.c, sm-eee/eee-init.c and
> > sm-eee/eee-platform.c. (~rendellf/dev/nfx-tree/code workspace directory)
> > >
> > > Rendell
> > >
> > >
> > > > -----Original Message-----
> > > > From: Bill Fisher
> > > > Sent: Thursday, April 23, 2009 5:19 PM
> > > > To: Andy Sharp
> > > > Cc: Bill Fisher; Rendell Fong; Maxim Kozlovsky
> > > > Subject: Re: Getting FP loaded/idleing with Linux on TXRX
> > > >
> > > > Andy Sharp wrote:
> > > > > Not sure what you're asking.  There's nothing extra to be done on
> > the
> > > > > tuxrx side to get FP image loaded.  In fact that will happen now
> > unless
> > > > > you have it commented out like we all do because the FP code is
> > mucking
> > > > > about in shared memory in a way that bothers Linux and makes it
> > crash.
> > > > > As you know.
> > > > >
> > > > > So what's the question again?
> > > > >
> > > > >
> > > > 	Does it need the Shared memory area allocated AND the WKA set
> > > > 	correctly and what else? There are lots of calls
> > > > 	in the initialization code to things that don't exist
> > > > 	under a linux kernel. The crash dump code, the setup
> > > > 	for halting the processor's, etc.
> > > >
> > > > 	I don't think it as easy as just loading it, after allocating
> > > > 	a shared memory region.
> > >
