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: <510C169C.8050400@wwwdotorg.org>
Date:	Fri, 01 Feb 2013 12:25:16 -0700
From:	Stephen Warren <swarren@...dotorg.org>
To:	Rob Herring <robherring2@...il.com>
CC:	Michal Marek <mmarek@...e.cz>, Stephen Warren <swarren@...dia.com>,
	srinivas.kandagatla@...com, devicetree-discuss@...ts.ozlabs.org,
	Mark Brown <broonie@...nsource.wolfsonmicro.com>,
	linux-kernel@...r.kernel.org, Scott Wood <scottwood@...escale.com>,
	Grant Likely <glikely@...retlab.ca>,
	Sam Ravnborg <sam@...nborg.org>
Subject: Re: [PATCH V7] kbuild: create a rule to run the pre-processor on
 *.dts files

On 02/01/2013 10:27 AM, Stephen Warren wrote:
> On 02/01/2013 09:51 AM, Rob Herring wrote:
>> On 02/01/2013 03:01 AM, Srinivas KANDAGATLA wrote:
>>> Hi Stephen,
>>> Not sure if you have already noticed this but,
>>> I did try this patch on my 3.8, and it looks like the intermediate dts
>>> file replaces all instances of linux with 1 because of predefined macros
>>> in gcc.
>>> As a result
>>>     linux,stdout-path = "/soc/stm-asc2";
>>> is changed to.
>>>     1,stdout-path = "/soc/stm-asc2";
>>>
>>> On my version of compiler(gcc version 4.6.3) I have
>>>
>>> armv7-linux-gcc -E -dM - < /dev/null | grep -v _
>>> #define unix 1
>>> #define linux 1
>>>
>>> Which might be true with most compiler versions aswell.
>>> As we are using linux as prefix for some device tree properties it makes
>>> sense to undef the linux gcc define.
>>> Adding -Ulinux to cmd_dtc_cpp should fix it.
>>>
>>> -cmd_dtc_cpp = $(CPP) $(cpp_flags) -D__DTS__ -x assembler-with-cpp -o
>>> $(dtc-tmp) $< ; \
>>> +cmd_dtc_cpp = $(CPP) $(cpp_flags) -D__DTS__  -Ulinux -x
>>> assembler-with-cpp -o $(dtc-tmp) $< ; \
>>>     $(objtree)/scripts/dtc/dtc -O dtb -o $@ -b 0 $(DTC_FLAGS) $(dtc-tmp)
>>
>> That's a hackish solution that seems fragile as well. Is there no way to
>> turn off all built-in defines?
> 
> I'm pretty sure there is; I'll go find it.

Hmmm. I can't actually find one.

> But we do want to keep some of the built-in defines. for example, -x
> assembler-with-cpp turns on __ASSEMBLY__ or similar, which headers can
> use to determine whether to only set up #defines, or also C-oriented
> stuff like types/prototypes. So, at least that one would need to be
> explicitly re-defined.

I grep'd through the kernel's include/ and there are quite a few hits
for some of the pre-define macros such as __linux__, __GNUC__,
__STRICT_ANSI__, __KERNEL__, __arm__ (and presumably other arch macros),
etc. I'd guess that an explicit blacklisting of -Dlinux and -Dunix might
be the most manageable path. Thoughts?
--
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

Powered by Openwall GNU/*/Linux Powered by OpenVZ