linux-kernel - Re: Device driver memory 'mmap()' function helper cleanup

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:	Wed, 17 Apr 2013 13:27:02 -0400 (EDT)
From:	David Miller <davem@...emloft.net>
To:	torvalds@...ux-foundation.org
Cc:	tomi.valkeinen@...com, linux-kernel@...r.kernel.org,
	clemens@...isch.de, arnd@...db.de, tiwai@...e.de,
	mchehab@...hat.com, ralf@...ux-mips.org
Subject: Re: Device driver memory 'mmap()' function helper cleanup

From: Linus Torvalds <torvalds@...ux-foundation.org>
Date: Wed, 17 Apr 2013 10:20:43 -0700

> On Wed, Apr 17, 2013 at 10:11 AM, David Miller <davem@...emloft.net> wrote:
>>
>> Yeah, the only thing special we do on sparc is interpret the PFN
>> specially.  We munge it into the real physical address and then
>> pass it all down to remap_pfn_range() to do the real work.
> 
> So the main thing I want to check is that *if* it's given a regular
> RAM physical address, it still works?
> 
> Some drivers basically allocate DMA memory and then pass on the
> resulting physical address to this. Others pass in the PCI BAR
> addresses etc. And some try to use "remap_pfn_range()", and others try
> to use "io_remap_pfn_range()", and quite frankly, from what I can tell
> we can just always use the "io_" version because it ends up being a
> proper superset.
> 
> I'm pretty sure it works fine the way I read it, but I'm just verifying..

All we do is take the top 4 bits and shift them down into the
resulting physical address.

For normal RAM these bits will be clear, so it should all work out.

Passing PCI BARs into these routines is illegal, we have proper
abstractions for mmap()'ing PCI resources via pci_mmap_page_range()
et al.

So, any code doing that needs to be fixed.
--
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/