Re: [PATCH 3/5] lkdtm: add function for testing .rodata section

From: Kees Cook <keescook@chromium.org>
To: Laura Abbott <labbott@redhat.com>
Cc: "kernel-hardening@lists.openwall.com" 
	<kernel-hardening@lists.openwall.com>,
	Arnd Bergmann <arnd@arndb.de>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	"David S. Miller" <davem@davemloft.net>,
	Mauro Carvalho Chehab <mchehab@osg.samsung.com>,
	Jiri Slaby <jslaby@suse.cz>, Guenter Roeck <linux@roeck-us.net>,
	LKML <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 3/5] lkdtm: add function for testing .rodata section
Date: Wed, 8 Jun 2016 08:46:02 -0700	[thread overview]
Message-ID: <CAGXu5jKRW1T6uR1ax7Oah_-nHCayALJ-DWj=MMoXwvdZvAfV-w@mail.gmail.com> (raw)
In-Reply-To: <941dc7f5-03a2-8a98-8888-7e92985699e3@redhat.com>

On Tue, Jun 7, 2016 at 6:02 PM, Laura Abbott <labbott@redhat.com> wrote:
> On 06/07/2016 02:57 PM, Kees Cook wrote:
>>
>> This adds a function that lives in the .rodata section. The section
>> flags are corrected using objcopy since there is no way with gcc to
>> declare section flags in an architecture-agnostic way.
>>
>
> Permit me to be the bearer of bad architecture news once again. With
> arm64 cross compiler (both Fedora 6.1.1 and Linaro 5.1)
>
>   CC      drivers/misc/lkdtm_rodata.o
>   OBJCOPY drivers/misc/lkdtm_rodata_objcopy.o
>   LD      drivers/misc/lkdtm.o
> drivers/misc/lkdtm_rodata_objcopy.o: file not recognized: File format not
> recognized
> scripts/Makefile.build:423: recipe for target 'drivers/misc/lkdtm.o' failed
> make[2]: *** [drivers/misc/lkdtm.o] Error 1
> scripts/Makefile.build:440: recipe for target 'drivers/misc' failed
> make[1]: *** [drivers/misc] Error 2
> Makefile:985: recipe for target 'drivers' failed
> make: *** [drivers] Error 2
>
>
> As far as I can tell this is because arm64 defines OBJCOPYFLAGS and they get
> propagated to objcopy
>
> aarch64-linux-gnu-objcopy -O binary -R .note -R .note.gnu.build-id -R
> .comment
>         -S --set-section-flags .text=alloc,readonly
>         --rename-section .text=.rodata drivers/misc/lkdtm_rodata.o
> drivers/misc/lkdtm_rodata_objcopy.o
>
> vs x86
>
> objcopy  --set-section-flags .text=alloc,readonly --rename-section
> .text=.rodata
>         drivers/misc/lkdtm_rodata.o drivers/misc/lkdtm_rodata_objcopy.o
>
>
> specifically it's the -O binary that seems to break things, the same failure
> happens on x86 as well with the the same commands. It works if I clear out
> the OBJCOPYFLAGS variable first but I don't think that's the correct way to
> fix this.
>
> Thanks,
> Laura
>
>
>> Signed-off-by: Kees Cook <keescook@chromium.org>
>> ---
>>  drivers/misc/Makefile       |  7 +++++++
>>  drivers/misc/lkdtm.h        |  6 ++++++
>>  drivers/misc/lkdtm_core.c   | 24 +++++++++++++++++-------
>>  drivers/misc/lkdtm_rodata.c | 10 ++++++++++
>>  4 files changed, 40 insertions(+), 7 deletions(-)
>>  create mode 100644 drivers/misc/lkdtm.h
>>  create mode 100644 drivers/misc/lkdtm_rodata.c
>>
>> diff --git a/drivers/misc/Makefile b/drivers/misc/Makefile
>> index c3cb6ad8cc37..b2d3d68dfa22 100644
>> --- a/drivers/misc/Makefile
>> +++ b/drivers/misc/Makefile
>> @@ -59,3 +59,10 @@ obj-$(CONFIG_CXL_BASE)               += cxl/
>>  obj-$(CONFIG_PANEL)             += panel.o
>>
>>  lkdtm-$(CONFIG_LKDTM)          += lkdtm_core.o
>> +lkdtm-$(CONFIG_LKDTM)          += lkdtm_rodata_objcopy.o
>> +
>> +OBJCOPYFLAGS_lkdtm_rodata_objcopy.o := \
>> +                       --set-section-flags .text=alloc,readonly \
>> +                       --rename-section .text=.rodata
>> +$(obj)/lkdtm_rodata_objcopy.o: $(obj)/lkdtm_rodata.o
>> +       $(call if_changed,objcopy)

Uhhhh... How is arm64 injecting those extra flags? OBJCOPYFLAGS is
being set with := here?

In related news I need to figure out how to get my cross-compiler
builds more well scripted...

-Kees

-- 
Kees Cook
Chrome OS & Brillo Security