Messages in this thread | | | Date | Thu, 28 Oct 1999 11:00:47 +0200 | From | Ralf Baechle <> | Subject | Re: 2.3.23pre5: drivers/char/bttv.c and drivers/usb/cpia.c do not compile |
| |
On Mon, Oct 25, 1999 at 12:46:24PM +0200, Roman Zippel wrote:
> There are more bugs, like in mm/swapfile.c:unuse_pte(), where a page > struct ptr is used as an argument to MAP_NR. :( > I tried to update the m68k kernel to 2.3.23 during the weekend and gave up > because of the mess. Why the hell has pte_page() a different meaning now, > but still the same name (without a single comment)? > I'm wondering how that works at all... :(
Linus did communicate with the port maintainers to fix related problems before the 2.3.24 release, below his email.
I haven't yet updated MIPS because > 2.3.9 are an unstable mess, I need to fix that first.
Ralf
Date: Wed, 20 Oct 1999 18:23:19 -0700 (PDT) From: Linus Torvalds <torvalds@transmeta.com> To: "David S. Miller" <davem@redhat.com>, rmk@arm.linux.org.uk, Paul Mackerras <paulus@linuxcare.com>, Jes.Sorensen@cern.ch, ralf@gnu.org, gniibe@chroot.org Subject: architecture bootup changes.. Message-ID: <Pine.LNX.4.10.9910201813380.836-100000@penguin.transmeta.com>
Just a heads-up that in order to avoid various nasty wrapping issues etc with more than 1GB of memory on 32-bit machines (virtual address wraparound) and in order to handle more than 4GB at all (physical address wraparound), the bootup sequence memory management got changed the last few days..
In most cases, it really shouldn't matter all that much, but generally the architecture-specific "init_mem()" call needs some changes to inform the new bootup memory allocator about what areas it can use, etc. And some things are more "struct page *" based than based on virtual kernel addresses, as the high memory support got cleaned up and better integrated in the memory management.
In short, it mainly only impacts architecture initialization (although we also added debugging code to catch "suspicious" use of IO pointers for the IO access functions - that causes debugging messages for ISA drivers that depend on old behaviour to be printed out, but they should work the same way they did before).
So at some point it would be nice if you would take a look at your respective architectures - see arch/i386/kernel/setup.c and the "setup_arch()" function changes for memory handling in particular. For most of you I would assume that this allows you to clean things up rather than make things worse, but who knows..
I haven't released a real kernel with this yet, but it's in
pre-patch-2.3.23-5.gz
on ftp.kernel.org in "testing" in case you want to sync up. No real hurry, although it's obviously less embarrassing for me the quicker the architectures catch up.
Linus
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |