Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 17 Apr 2000 00:31:50 -0600
From:      Warner Losh <imp@village.org>
To:        Mitsuru IWASAKI <iwasaki@jp.FreeBSD.org>
Cc:        FreeBSD-mobile@FreeBSD.org
Subject:   Re: [NEWCARD] patch against panics (Re: cvs commit: src/sys/dev/pcic i82365.c i82365_isa.c i82365var.h src/sys/dev/pccard card_if.m pccard.c pccard_cis.c pccardvar.h) 
Message-ID:  <200004170631.AAA25878@billy-club.village.org>
In-Reply-To: Your message of "Mon, 17 Apr 2000 06:31:26 %2B0900." <200004162131.GAA82367@tasogare.imasy.or.jp> 
References:  <200004162131.GAA82367@tasogare.imasy.or.jp>  <200004130642.XAA39181@freefall.freebsd.org> 

next in thread | previous in thread | raw e-mail | index | archive | help
In message <200004162131.GAA82367@tasogare.imasy.or.jp> Mitsuru IWASAKI writes:
: I've just tried the latest NEWCARD, and noticed that someting wrong with
: bus_space stuff.  Here is the patch to avoid panics, not for the solution :-)

OK.  Looks like there was a minor problem with tuple.ptr not getting
initialized to 0.  That helps, but there are other issues with the
code once that has happened.  That's what caused the panic.  Fixing
this shows another panic.

There's a disconnect between the bus space stuff and the memory
allocation.  The bus_alloc_memory returns one area obtained from the
system (say 0xd4000), while the do_chip_mem_alloc assumes that
subregions are allocated from the memory that the pcic allocated at
probe time.  Since these are different, bad things hapen when you
start to access the CIS. We need to pick one way or the other and
stick to that way exclusively throughout.  I think that I'll kill
subregions and go with asking the parent bus for the resoruce.  If we
centralize it in the pcic layer, we can make it possible to change our
minds later if we find it desirable to have the pcic bridge do this
memory management.

I've implemented part of this change, but must get some sleep now or
I'll be useless tomorrow.  I'll implement the rest of the change when
i can sit down and think a little more about the code.

Once I have CIS working on NEWCARD, I plan to backport the ability to
map the CIS or attribute memory to OLDCARD, which should make the xe
and ray driver people happy :-).

Warner


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




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200004170631.AAA25878>