u64-ull

u64-ull

[Matthew Wilcox]

On Mon, Jul 28, 2008 at 03:53:40AM -0700, Andrew Morton wrote:
> On Mon, 28 Jul 2008 20:03:23 +1000 Benjamin Herrenschmidt <benh@kernel.crashing.org> wrote:
> 
> > Andrew, what was your decision vs. changing once for all u64 and s64 on
> > all archs to be long long's ?
> 
> We should do it.  Matthew's initial patch was busted and I don't
> _really_ want to have to take on all the fallout.  It's a project for
> someone: get the core patch working then start fixing up fallout.  At
> least make it all compile, hopefully fix most of the warnings too.

Yeah, I just threw it out as a trial balloon without thinking about it
properly.  We need to test:

include/asm-alpha/types.h:#include <asm-generic/int-l64.h>
include/asm-ia64/types.h:#include <asm-generic/int-l64.h>
include/asm-mips/types.h:# include <asm-generic/int-l64.h>
include/asm-powerpc/types.h:# include <asm-generic/int-l64.h>
include/asm-s390/types.h:# include <asm-generic/int-l64.h>
include/asm-sparc/types.h:#include <asm-generic/int-l64.h>

I can take care of ia64 easily enough.  Ben is, I presume, willing to
take care of powerpc.  Just need to find suckers^W people willing to fix
up alpha, mips, s390 and sparc.

I don't think this needs to go in during the merge window, I'm happy to
wait until -rc2 or something.  Anyway, initial patch at
http://git.kernel.org/?p=linux/kernel/git/willy/misc.git;a=shortlog;h=u64-ull
so if anyone wants to get started early ...

-- 
Intel are signing my paycheques ... these opinions are still mine
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."

Re: warning in scheduler

[Matthew Wilcox]

On Mon, Jul 28, 2008 at 08:40:22PM +1000, Benjamin Herrenschmidt wrote:
> I tend to agree. I might look into doing it for powerpc around rc3 or
> so, when things have settled a bit.

It'd be nice if allmodconfig even compiled:

drivers/char/pcmcia/ipwireless/hardware.c:571: error: invalid use of undefined type 'struct ipw_network'

Not to mention the qla_attr patch:

+#include <linux/delay.h>

There's probably others, I just haven't got far enough in the
compilation process yet.

But I am seeing lots of warnings _without_ the patch applied:

fs/omfs/inode.c: In function 'omfs_fill_super':
fs/omfs/inode.c:495: warning: format '%llx' expects type 'long long
unsigned int', but argument 2 has type 'u64'
fs/proc/base.c: In function 'do_io_accounting':
fs/proc/base.c:2429: warning: format '%llu' expects type 'long long
unsigned int', but argument 3 has type 'u64'
fs/ubifs/dir.c: In function 'ubifs_readdir':
fs/ubifs/dir.c:437: warning: format '%llu' expects type 'long long
unsigned int', but argument 5 has type '__le64'

(other warnings snipped for brevity).

-- 
Intel are signing my paycheques ... these opinions are still mine
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours.  We can't possibly take such
a retrograde step."

Re: warning in scheduler

[Benjamin Herrenschmidt]

On Mon, 2008-07-28 at 12:39 -0600, Matthew Wilcox wrote:
> On Mon, Jul 28, 2008 at 08:40:22PM +1000, Benjamin Herrenschmidt wrote:
> > I tend to agree. I might look into doing it for powerpc around rc3 or
> > so, when things have settled a bit.
> 
> It'd be nice if allmodconfig even compiled:
> 
> drivers/char/pcmcia/ipwireless/hardware.c:571: error: invalid use of undefined type 'struct ipw_network'
> 
> Not to mention the qla_attr patch:
> 
> +#include <linux/delay.h>
> 
> There's probably others, I just haven't got far enough in the
> compilation process yet.

Yup, it's been a pretty bad merge window (at least it's my impression as
my first one as maintainer), mostly fixing compile bugs all the time.

> But I am seeing lots of warnings _without_ the patch applied:
> 
> fs/omfs/inode.c: In function 'omfs_fill_super':
> fs/omfs/inode.c:495: warning: format '%llx' expects type 'long long
> unsigned int', but argument 2 has type 'u64'
> fs/proc/base.c: In function 'do_io_accounting':
> fs/proc/base.c:2429: warning: format '%llu' expects type 'long long
> unsigned int', but argument 3 has type 'u64'
> fs/ubifs/dir.c: In function 'ubifs_readdir':
> fs/ubifs/dir.c:437: warning: format '%llu' expects type 'long long
> unsigned int', but argument 5 has type '__le64'
> 
> (other warnings snipped for brevity).

I have a handful on powerpc too.

BTW. I'm looking at adding a %pR to print a struct resource (or maybe
a %pr for a resource_size_t and %pR for the whole resource). That would
help with the whole business of the thing being either 32 or 64 bits.

Ben.