Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 27 May 2001 22:39:05 -0700 (PDT)
From:      Matthew Jacob <mjacob@feral.com>
To:        "Daniel C. Sobral" <dcs@newsguy.com>
Cc:        alpha@FreeBSD.ORG
Subject:   Re: Latest on ' HEADS UP: loader broken'
Message-ID:  <Pine.BSF.4.21.0105272236240.14704-100000@beppo.feral.com>
In-Reply-To: <3B11B706.E8032F96@newsguy.com>

next in thread | previous in thread | raw e-mail | index | archive | help


On Sun, 27 May 2001, Daniel C. Sobral wrote:

> Matthew Jacob wrote:
> > 
> > Your surmise about loader.4th is correct, I believe.
> > 
> > The alpha's boot1/loader from after your commits seems to boot my PC164 just
> > fine with the older loader.4th, but chokes on the new one.
> > 
> > The relevant diffs in loader.4th are:
> > --------
> > 49a50,52
> > >
> > > 256 dictthreshold !  \ 256 cells minimum free space
> > > 2048 dictincrease !  \ 2048 additional cells each time
> > --------
> 
> Oh... This is *NOT* what I expected!
> 
> > Can you say what additional memory load for this, and in whay area, this
> > should be? If it's stack, then we're probably screwed at the moment. If it's
> > heap storage, that's probably fairly easy to fix (although I'm kind of
> > *appalled* that we'd be using > 512KB of heap for the loader).
> 
> It's allocated from heap. It's a malloc call. I'm not sure what the hell
> is taking so much space on loader.4th stuff. The initial dictionary size
> is about 10000*sizeof(void*) (and I note there is a comment to be
> fixed...). That above increases it by 2048*sizeof(void*) whenever the
> available space is below 256*sizeof(void*) at the beginning of a new
> definition. This space does not include the memory used to store the
> settings read from the loader.conf files.
> 

hmm. well, when I get a chance tomorrow, I'll make a larger heap and we'll
see. I'm still not convinced it's actually heap. The 12KB stack is way too
small- but that may be harder to fix.


btw- I share David's frustration, but let's get past all of this and fix the
problem. This is, by my recollection, the 2nd time this has happened in
relation to a Forth checkin, and I sure didn't see a call for testing to the
alias.

Still- let's move on. We will desparately need you for the Sparc64 port. So if
we grumble, forgive us.

-matt



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-alpha" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0105272236240.14704-100000>