linux-kernel - Re: [PATCH] [v2] coresight: etm4x: avoid build failure with unrolled loops

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: <CAK8P3a3-XoDQ6BfTBUof5ST2H1_6JOL+rK-BQqWXAV0H5jMm9Q@mail.gmail.com>
Date:   Thu, 29 Apr 2021 19:50:46 +0200
From:   Arnd Bergmann <arnd@...nel.org>
To:     Suzuki K Poulose <suzuki.poulose@....com>
Cc:     Mathieu Poirier <mathieu.poirier@...aro.org>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Nathan Chancellor <nathan@...nel.org>,
        Nick Desaulniers <ndesaulniers@...gle.com>,
        Mike Leach <mike.leach@...aro.org>,
        Leo Yan <leo.yan@...aro.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Sai Prakash Ranjan <saiprakash.ranjan@...eaurora.org>,
        Qi Liu <liuqi115@...wei.com>,
        Tingwei Zhang <tingwei@...eaurora.org>,
        coresight@...ts.linaro.org,
        Linux ARM <linux-arm-kernel@...ts.infradead.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        clang-built-linux <clang-built-linux@...glegroups.com>
Subject: Re: [PATCH] [v2] coresight: etm4x: avoid build failure with unrolled loops

On Thu, Apr 29, 2021 at 7:37 PM Suzuki K Poulose <suzuki.poulose@....com> wrote:
> On 29/04/2021 15:57, Arnd Bergmann wrote:
> > From: Arnd Bergmann <arnd@...db.de>
> >
> > clang-12 fails to build the etm4x driver with -fsanitize=array-bounds,
> > where it decides to unroll certain loops in a way that result in a
> > C variable getting put into an inline assembly
> >
> > <instantiation>:1:7: error: expected constant expression in '.inst' directive
> > .inst (0xd5200000|((((2) << 19) | ((1) << 16) | (((((((((((0x160 + (i * 4))))) >> 2))) >> 7) & 0x7)) << 12) | ((((((((((0x160 + (i * 4))))) >> 2))) & 0xf)) << 8) | (((((((((((0x160 + (i * 4))))) >> 2))) >> 4) & 0x7)) << 5)))|(.L__reg_num_x8))
> >        ^
> > drivers/hwtracing/coresight/coresight-etm4x-core.c:702:4: note: while in macro instantiation
> >                          etm4x_relaxed_read32(csa, TRCCNTVRn(i));
> >                          ^
> > drivers/hwtracing/coresight/coresight-etm4x.h:403:4: note: expanded from macro 'etm4x_relaxed_read32'
> >                   read_etm4x_sysreg_offset((offset), false)))
> >                   ^
> > drivers/hwtracing/coresight/coresight-etm4x.h:383:12: note: expanded from macro 'read_etm4x_sysreg_offset'
> >                          __val = read_etm4x_sysreg_const_offset((offset));       \
> >                                  ^
> > drivers/hwtracing/coresight/coresight-etm4x.h:149:2: note: expanded from macro 'read_etm4x_sysreg_const_offset'
> >          READ_ETM4x_REG(ETM4x_OFFSET_TO_REG(offset))
> >          ^
> > drivers/hwtracing/coresight/coresight-etm4x.h:144:2: note: expanded from macro 'READ_ETM4x_REG'
> >          read_sysreg_s(ETM4x_REG_NUM_TO_SYSREG((reg)))
> >          ^
> > arch/arm64/include/asm/sysreg.h:1108:15: note: expanded from macro 'read_sysreg_s'
> >          asm volatile(__mrs_s("%0", r) : "=r" (__val));                  \
> >                       ^
> > arch/arm64/include/asm/sysreg.h:1074:2: note: expanded from macro '__mrs_s'
> > "       mrs_s " v ", " __stringify(r) "\n"                      \
> >   ^
> >
> > This only happened in a few loops in which the array bounds sanitizer
> > added a special case for an array overflow that clang determined to be
> > possible, but any compiler is free to unroll any of the loops in the
> > same way that breaks the sysreg macros.
> >
> > Introduce helper functions that perform a sysreg access with a
> > non-constant register number and use them in each call that passes
> > a loop counter.
>
> You don't need to add this special helper. We have the exact
> infrastructure already. So these could simply be replaced with:
>
> csdev_access_xxx(csa, ...)
>
> see :
>
> include/linux/coresight.h

Ah, nice!

Do you mean replacing only the ones that use a nonconstant
offset, or all of them? I guess changing all would avoid some
really ugly magic macros, but the indirect function call and the
switch() adds a few cycles of overhead every time and the code
looks like it is micro-optimized for fast register access here.

      Arnd