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] [day] [month] [year] [list]
Date:   Sun, 29 Mar 2020 20:02:40 +0200
From:   Stefan Agner <stefan@...er.ch>
To:     Nick Desaulniers <ndesaulniers@...gle.com>
Cc:     Russell King <linux@...linux.org.uk>,
        Arnd Bergmann <arnd@...db.de>,
        Masahiro Yamada <yamada.masahiro@...ionext.com>,
        Manoj Gupta <manojgupta@...gle.com>,
        Jian Cai <jiancai@...gle.com>,
        Linux ARM <linux-arm-kernel@...ts.infradead.org>,
        LKML <linux-kernel@...r.kernel.org>,
        clang-built-linux <clang-built-linux@...glegroups.com>
Subject: Re: [PATCH] ARM: warn if pre-UAL assembler syntax is used

On 2020-03-17 08:55, Stefan Agner wrote:
> On 2020-03-17 01:00, Nick Desaulniers wrote:
>> Revert "ARM: 8846/1: warn if divided syntax assembler is used"On Thu,
>> Feb 27, 2020 at 4:19 PM Stefan Agner <stefan@...er.ch> wrote:
>>>
>>> Remove the -mno-warn-deprecated assembler flag for GCC versions newer
>>> than 5.1 to make sure the GNU assembler warns in case non-unified
>>> syntax is used.
>>
>> Hi Stefan, sorry for the late reply from me; digging out my backlog.
>> Do you happen to have a godbolt link perhaps that demonstrates this?
>> It sounds like GCC itself is emitting pre-UAL?
> 
> Yes, that is what Russell observed and caused the revert:
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=b752bb405a13
> 
> I do not have a godbolt link at hand, I just built the complete kernel
> using some GCC toolchains I had locally available and noticed that the
> problem persists up to and including GCC 5.0. I did not track down what
> exactly is causing GCC to emit pre-UAL.

Godbolt link:
https://godbolt.org/z/fxYq_s

However, the warning does not appear on godbolt.org, presumably because
the compiler is started such that it does not invoke the assembler (at
least that is what I understand when using -v). However, on my end this
leads to:

$ arm-linux-gnueabihf-gcc -marm -march=armv7-a -O2  -c  test.c -o test
--save-temps
test.s: Assembler messages:
test.s:33: Warning: conditional infixes are deprecated in unified syntax
$ arm-linux-gnueabihf-gcc --version
arm-linux-gnueabihf-gcc (Linaro GCC 4.9-2016.02) 4.9.4 20151028
(prerelease)

And line 33 contains a non-unified assembler mnemonic:
        strneb  r2, [r3]

Note: I do have to explicitly use .syntax unified. It seems that gcc is
not explicitly doing this, but if any inline assembly is changing the
assembler mode (e.g. what including arch/arm/include/asm/unified.h is
doing) then the assembler warnings start to appear.

--
Stefan


> 
>>
>>>
>>> This also prevents a warning when building with Clang and enabling
>>> its integrated assembler:
>>> clang-10: error: unsupported argument '-mno-warn-deprecated' to option 'Wa,'
>>>
>>> This is a second attempt of commit e8c24bbda7d5 ("ARM: 8846/1: warn if
>>> divided syntax assembler is used").
>>
>> Would it be helpful to also make note of
>> commit b752bb405a13 ("Revert "ARM: 8846/1: warn if divided syntax
>> assembler is used"")?
> 
> Sure, I can do that.
> 
>>
>>
>>>
>>> Signed-off-by: Stefan Agner <stefan@...er.ch>
>>> ---
>>>  arch/arm/Makefile | 14 +++++++++-----
>>>  1 file changed, 9 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/arch/arm/Makefile b/arch/arm/Makefile
>>> index db857d07114f..a6c8c9f39185 100644
>>> --- a/arch/arm/Makefile
>>> +++ b/arch/arm/Makefile
>>> @@ -119,21 +119,25 @@ ifeq ($(CONFIG_CC_IS_CLANG),y)
>>>  CFLAGS_ABI     += -meabi gnu
>>>  endif
>>>
>>> -# Accept old syntax despite ".syntax unified"
>>> -AFLAGS_NOWARN  :=$(call as-option,-Wa$(comma)-mno-warn-deprecated,-Wa$(comma)-W)
>>
>> This existing code is quite bad for Clang, which doesn't support
>> `-Wa,-mno-warn-deprecated`, so this falls back to `-Wa,-W`, which
>> disables all warnings from the assembler, which we definitely do not
>> want.  That alone is worth putting in the GCC guard.  But I would like
>> more info about GCC above before signing off.
> 
> FWIW, I submitted this to the patch tracker already, but I don't think
> it got merged already.
> 
> --
> Stefan
> 
>>
>>> -
>>>  ifeq ($(CONFIG_THUMB2_KERNEL),y)
>>> -CFLAGS_ISA     :=-mthumb -Wa,-mimplicit-it=always $(AFLAGS_NOWARN)
>>> +CFLAGS_ISA     :=-mthumb -Wa,-mimplicit-it=always
>>>  AFLAGS_ISA     :=$(CFLAGS_ISA) -Wa$(comma)-mthumb
>>>  # Work around buggy relocation from gas if requested:
>>>  ifeq ($(CONFIG_THUMB2_AVOID_R_ARM_THM_JUMP11),y)
>>>  KBUILD_CFLAGS_MODULE   +=-fno-optimize-sibling-calls
>>>  endif
>>>  else
>>> -CFLAGS_ISA     :=$(call cc-option,-marm,) $(AFLAGS_NOWARN)
>>> +CFLAGS_ISA     :=$(call cc-option,-marm,)
>>>  AFLAGS_ISA     :=$(CFLAGS_ISA)
>>>  endif
>>>
>>> +ifeq ($(CONFIG_CC_IS_GCC),y)
>>> +ifeq ($(call cc-ifversion, -lt, 0501, y), y)
>>> +# GCC <5.1 emits pre-UAL code and causes assembler warnings, suppress them
>>> +CFLAGS_ISA     +=$(call as-option,-Wa$(comma)-mno-warn-deprecated,-Wa$(comma)-W)
>>> +endif
>>> +endif
>>> +
>>>  # Need -Uarm for gcc < 3.x
>>>  KBUILD_CFLAGS  +=$(CFLAGS_ABI) $(CFLAGS_ISA) $(arch-y) $(tune-y) $(call cc-option,-mshort-load-bytes,$(call cc-option,-malignment-traps,)) -msoft-float -Uarm
>>>  KBUILD_AFLAGS  +=$(CFLAGS_ABI) $(AFLAGS_ISA) $(arch-y) $(tune-y) -include asm/unified.h -msoft-float
>>> --

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ