linux-kernel - Re: [GIT PULL] bug fix for devicetree memory parsing

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:	Sun, 6 Jul 2014 12:24:42 -0700
From:	Linus Torvalds <torvalds@...ux-foundation.org>
To:	Grant Likely <grant.likely@...aro.org>
Cc:	Rob Herring <rob.herring@...aro.org>,
	Laura Abbott <lauraa@...eaurora.org>,
	"devicetree@...r.kernel.org" <devicetree@...r.kernel.org>,
	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [GIT PULL] bug fix for devicetree memory parsing

On Sun, Jul 6, 2014 at 2:37 AM, Grant Likely <grant.likely@...aro.org> wrote:
>
> Can you pull this bug fix into your tree please?

I took it, but I think both your explanation and the patch itself is
actually crap. It may fix the issue, but it's seriously confused.

Your explanation says that it's a 32-bit platform issue. No it's not.
Most 32-bit configurations still have a 64-bit phys_addr_t (ie
PAE/LPAE etc).

And the code is crap, because it uses ULONG_MAX etc in ways that
simply make no f*cking sense. And why does it care about sizeof?

Why does the code not just do something like

  #define MAX_PHYS_ADDR ((phys_addr_t) ~0)

and then do

  if (base > MAX_PHYS_ADDR || base + size > MAX_PHYS_ADDR)
    ...

and be done with it? All those sizeof tests are completely pointless.
If it turns out that phys_addr_t is the same size as u64, then the
tests will never be true, and the compiler will happily optimize them
away.

So I think this fixes a problem, but it's all ugly as hell. I ended up
pulling it because I'm lazy and don't have a machine to test a proper
fix on anyway, but I hope this can get cleaned up. And more
importantly, I hope maintainers will spend a bit more time thinking
about things like this. It's not just that the code is unnecessarily
complex, it's WRONG. Comparing things to "ULONG_MAX" makes absolutely
zero sense, since "ULONG_MAX" has nothing to do with anything. It's
just stupid and misleading, and it just so happens to work by random
luck because it so *happens* that phys_addr_t is smaller than "u64"
only when ULONG_MAX happens to be the same size. But even that is not
guaranteed (ie some stupid broken architecture might have a 32-bit
physical address space despite having a 64-bit "long")

                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/