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:   Wed, 15 Apr 2020 22:06:27 +0200
From:   Stefan Agner <stefan@...er.ch>
To:     Nick Desaulniers <ndesaulniers@...gle.com>
Cc:     Jian Cai <caij2003@...il.com>, Manoj Gupta <manojgupta@...gle.com>,
        Russell King <linux@...linux.org.uk>,
        Enrico Weigelt <info@...ux.net>,
        Kate Stewart <kstewart@...uxfoundation.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        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: replace the sole use of a symbol with its definition

On 2020-04-13 20:29, Nick Desaulniers wrote:
> On Tue, Apr 7, 2020 at 12:09 PM Jian Cai <caij2003@...il.com> wrote:
>>
>> ALT_UP_B macro sets symbol up_b_offset via .equ to an expression
>> involving another symbol. The macro gets expanded twice when
>> arch/arm/kernel/sleep.S is assembled, creating a scenario where
>> up_b_offset is set to another expression involving symbols while its
>> current value is based on symbols. LLVM integrated assembler does not
>> allow such cases, and based on the documentation of binutils, "Values
>> that are based on expressions involving other symbols are allowed, but
>> some targets may restrict this to only being done once per assembly", so
>> it may be better to avoid such cases as it is not clearly stated which
>> targets should support or disallow them. The fix in this case is simple,
>> as up_b_offset has only one use, so we can replace the use with the
>> definition and get rid of up_b_offset.
>>
>> Signed-off-by: Jian Cai <caij2003@...il.com>

Thanks for tackling this!

> 
> Probably didn't need the extra parens, but it's fine (unless another
> reviewer would like a v2).  Maybe Stefan has some thoughts?

Since this is a processor macro I actually prefer to have parentheses
here. All use sites of ALT_UP_B pass just a label, but still, just to be
on the safe side.

I was wondering why equ has been used in first place. I don't see an
advantage other than having a symbol which can be checked. But given
that this code is stable and don't really need debugging at this point,
I am fine replacing this to make it work for clang.


> Reviewed-by: Nick Desaulniers <ndesaulniers@...gle.com>
> 
> Please add Link tags if these correspond to issues in our link
> tracker, they help us track when and where patches land.
> Link: https://github.com/ClangBuiltLinux/linux/issues/920

Agreed, please add the link. You can add this when submitting.

With that:
Reviewed-by: Stefan Agner <stefan@...er.ch>

--
Stefan

> 
>> ---
>>  arch/arm/include/asm/assembler.h | 3 +--
>>  1 file changed, 1 insertion(+), 2 deletions(-)
>>
>> diff --git a/arch/arm/include/asm/assembler.h b/arch/arm/include/asm/assembler.h
>> index 99929122dad7..adee13126c62 100644
>> --- a/arch/arm/include/asm/assembler.h
>> +++ b/arch/arm/include/asm/assembler.h
>> @@ -269,10 +269,9 @@
>>         .endif                                                  ;\
>>         .popsection
>>  #define ALT_UP_B(label)                                        \
>> -       .equ    up_b_offset, label - 9998b                      ;\
>>         .pushsection ".alt.smp.init", "a"                       ;\
>>         .long   9998b                                           ;\
>> -       W(b)    . + up_b_offset                                 ;\
>> +       W(b)    . + (label - 9998b)                                     ;\
>>         .popsection
>>  #else
>>  #define ALT_SMP(instr...)
>> --
>> 2.26.0.292.g33ef6b2f38-goog
>>

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ