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  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:	Sat, 1 Nov 2008 16:18:18 -0700 (PDT)
From:	Linus Torvalds <torvalds@...ux-foundation.org>
To:	Jonathan Corbet <corbet@....net>
cc:	Yinghai Lu <yinghai@...nel.org>, Ingo Molnar <mingo@...e.hu>,
	Robert Hancock <hancockr@...w.ca>,
	e1000-devel@...ts.sourceforge.net,
	LKML <linux-kernel@...r.kernel.org>,
	Steven Rostedt <rostedt@...dmis.org>
Subject: Re: 2.6.28-rc2 hates my e1000e



On Sat, 1 Nov 2008, Jonathan Corbet wrote:
>
> Looks to me like Linus's patch is the way to go, at least for now...

I'll make an -rc3 tomorrow. However, I suspect that if we have lots more 
regressions, we'll just have to revert the resource handling back to the
2.6.27 state.

The problem with resource handling is that even when we can write code 
that "makes sense", in the end firmware can always do odd things. For 
example, in your case it really does make sense to keep the already 
allocated PCI resources in the reserved region, because the firmware 
obviously did both the reserved region _and_ the PCI BAR allocation.

At the same time, I'm worried that what Windows does is something totally 
different, probably odd, and possibly even dependent on some HAL layer 
motherboard driver or other.  And it's really the case that every single 
time we change resource allocation - never mind how subtly, or how much 
sense it makes - it will break some odd setup somewhere.

And I would not be surprised if we end up finding some machine that really 
had totally _broken_ PCI BAR setup, where it set up some PCI decode to 
overlap with a reserved region and then depended on the OS re-allocating 
the resource. 

As usual, the right answer doesn't necessarily end up being the one that 
makes most sense, but probably the one that matches what Windows ends up 
doing most closely - just because that's the one that was tested against. 
And windows behaviour can in turn easily depend on some internal Windows 
implementation detail, rather than any "thought out" solution.

The good news here is that the particular behavior wrt e820 reserved 
resources and various PCI BAR's should be totally irrelevant for 99.9% of 
all hardware, and we _only_ have to worry about the really odd cases. But 
even just a couple of odd BIOS versions are enough to cause a lot of pain.

So let's see how it turns out in -rc3. It works for _you_, and it looks 
sane to me, but ...

			Linus
--
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/

Powered by blists - more mailing lists