linux-kernel - Re: [PATCH] ARM: mtd-xip: work around clang/llvm bug

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: <CAKwvOdnbVFQZNFaZs7Yh4C=OnR8k3CyrRc=NQEQqFvPL=Qo9Vg@mail.gmail.com>
Date:   Tue, 9 Jul 2019 11:11:37 -0700
From:   Nick Desaulniers <ndesaulniers@...gle.com>
To:     Russell King - ARM Linux admin <linux@...linux.org.uk>
Cc:     Linus Walleij <linus.walleij@...aro.org>,
        Arnd Bergmann <arnd@...db.de>,
        Linux ARM <linux-arm-kernel@...ts.infradead.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        clang-built-linux <clang-built-linux@...glegroups.com>
Subject: Re: [PATCH] ARM: mtd-xip: work around clang/llvm bug

On Tue, Jul 9, 2019 at 5:26 AM Russell King - ARM Linux admin
<linux@...linux.org.uk> wrote:
>
> On Tue, Jul 09, 2019 at 02:17:58PM +0200, Linus Walleij wrote:
> > On Mon, Jul 8, 2019 at 10:31 PM Arnd Bergmann <arnd@...db.de> wrote:
> >
> > > -#define xip_iprefetch()        do { asm volatile (".rep 8; nop; .endr"); } while (0)
> > > +#define xip_iprefetch()        do {                                            \
> > > +        asm volatile ("nop; nop; nop; nop; nop; nop; nop; nop;");      \
> > > +} while (0)                                                            \
> >
> > This is certainly an OK fix since we use a row of inline nop at
> > other places.
> >
> > However after Russell explained the other nops I didn't understand I located
> > these in boot/compressed/head.S as this in __start:
> >
> >                 .rept   7
> >                 __nop
> >                 .endr
> > #ifndef CONFIG_THUMB2_KERNEL
> >                 mov     r0, r0
> > #else
> >
> > And certainly this gets compiled, right?
> >
> > So does .rept/.endr work better than .rep/.endr, is it simply mis-spelled?
> >
> > I.e. s/.rep/.rept/g
> > ?
> >
> > In that case we should explain in the commit that .rep doesn't work
> > but .rept does.
>
> According to the info pages for gas:
>
> 7.96 `.rept COUNT'
> ==================
>
> Repeat the sequence of lines between the `.rept' directive and the next
> `.endr' directive COUNT times.
>
> So yes, ".rep" is mis-spelled, and it brings up the obvious question:
> why isn't gas issuing an error for ".rep"?  There is no mention of
> ".rep" in the manual.

I swear I had looked this up somewhere and found that GNU as and
clang's integrated assembler supported alternative spellings for
assembly directives.  Just checked the manual
https://ftp.gnu.org/old-gnu/Manuals/gas-2.9.1/html_chapter/as_7.html#SEC116
and indeed no mention of the alternatives...must have been looking at
the source...
-- 
Thanks,
~Nick Desaulniers