linux-kernel - Re: [PATCH 0/3] Add API for weak DMA masks

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]

Message-Id: <200805011916.07340.mb@bu3sch.de>
Date:	Thu, 1 May 2008 19:16:06 +0200
From:	Michael Buesch <mb@...sch.de>
To:	Jesse Barnes <jbarnes@...tuousgeek.org>
Cc:	Christoph Hellwig <hch@...radead.org>,
	John Linville <linville@...driver.com>,
	Andi Kleen <andi@...stfloor.org>,
	David Miller <davem@...emloft.net>,
	Alan Cox <alan@...rguk.ukuu.org.uk>,
	Ingo Molnar <mingo@...e.hu>, bcm43xx-dev@...ts.berlios.de,
	linux-wireless@...r.kernel.org,
	linux-kernel <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 0/3] Add API for weak DMA masks

On Thursday 01 May 2008 18:30:04 Jesse Barnes wrote:
> On Thursday, May 01, 2008 9:07 am Michael Buesch wrote:
> > On Thursday 01 May 2008 17:58:26 Christoph Hellwig wrote:
> > > On Thu, May 01, 2008 at 05:47:26PM +0200, Michael Buesch wrote:
> > > > We've discussed that and this behaviour is not acceptable, as the
> > > > driver must know about a possible fallback in case it can do 32bit DMA
> > > > more efficiently than 64bit DMA, for example.
> > >
> > > That's what we have dma_get_required_mask() for.  See
> > > Documentation/DMA-API.txt.
> >
> > So well. I'm still unsure about the advantage of having some opencoded
> > probe loop in the driver, instead of implementing it in a common place
> > and doing all of it with a single API call.
> > We can still call dma_get_required_mask() and adjust the mask to that
> > in dma_set_mask_weak(). That can _additionally_ be done there.
> 
> So I think we're agreed that we want some core function to fall back to 
> different mask values, but yeah, making it take dma_get_required_mask into 
> account would also be good.

Ok, will redo the patches with that added and the name changed.

> Most drivers just do the fallback themselves, right?

Right.

> So it makes sense to  
> just update the current code to fallback, and update drivers wanting specific 
> mask values to check afterwards.  I hate to inflict that kind of driver wide 
> update on Michael though... :)

Well, that's a lot of work and I'm not sure it's worth it.
I could live with having dma_set_mask as an API that fails on bad masks
and dma_request_mask as an API above that which retries. I think that's
just fine. Drivers can be migrated over time to the new API (or not. That
can be the driver maintainer's choice).

-- 
Greetings Michael.
--
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/