linux-kernel - Re: [PATCH 1/1] edac x38: new MC driver module

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 14 Nov 2008 00:15:14 +0900
From:	Hitoshi Mitake <h.mitake@...il.com>
To:	Paul Mundt <lethal@...ux-sh.org>
Cc:	Andrew Morton <akpm@...ux-foundation.org>,
	Hitoshi Mitake <mitake@...stcom.com>,
	Doug Thompson <norsk5@...oo.com>, dougthompson@...ssion.com,
	linux-kernel@...r.kernel.org, ktaka@...stcom.com,
	linux-arch@...r.kernel.org
Subject: Re: [PATCH 1/1] edac x38: new MC driver module

On Tue, 11 Nov 2008 15:11:40 +0900
Paul Mundt <lethal@...ux-sh.org> wrote:

> On Sun, Nov 09, 2008 at 11:26:46AM -0800, Andrew Morton wrote:
> > (cc linux-arch)
> > 
> > > It seems that architectures that provide readq/writeq are
> > > mips, parisc and x86 (and x86_64).
> > > 
> There are more than that, grep arch/*/include also.
> 
> In addition to mips, parisc, and x86, there is ia64, alpha, sh, and
> sparc.

I didn't noticed these functions, thanks.

> 
> > #ifdef readq
> > 
> > Is a suitable way of determining whether the architecture implements
> > readq and writeq.  It isn't pretty, but it will suffice.
> > 
> > A problem with it is that drivers will then do
> > 
> > #ifndef readq
> > <provide a local implementation here>
> > #endif
> > 
> > which rather sucks - we don't want lots of little private readq/writeq
> > implementations all over the tree.
> > 
> > Perhaps it would be better to have a CONFIG_ARCH_HAS_READQ and to then
> > disable these drivers on the architectures which don't provide
> > readq/writeq support.
> > 
> However this is handled, we don't want a rehash of the read/writes{b,w,l} fiasco.
> 
> Allowing drivers to do their own local implementations of these things
> has always been a complete disaster. A Kconfig option will at least take
> care of having these craptastic ifdef lists for architectures in every
> driver that rolls its own implementation.
> 
> Even a sub-optimal asm-generic version would be preferable, if the
> semantics are well enough defined and consistently adhered to.

I found nice source file, lib/iomap.c.
There are architecture independent read/write{b,w,l} (named like ioread8).

I want to implement architecture independent readq/writeq in lib/iomap.c .
Andrew told,

> Perhaps it would be better to have a CONFIG_ARCH_HAS_READQ and to then
> disable these drivers on the architectures which don't provide
> readq/writeq support.

But I want to use x38_edac.c on x86_32 environment,
and I think similar desire may occur in the future.

Is this way good enough? Request for comment.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/