lkml.org 
[lkml]   [2019]   [Sep]   [30]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
SubjectRe: [PATCH] compiler: enable CONFIG_OPTIMIZE_INLINING forcibly
On Mon, Sep 30, 2019 at 5:18 AM Will Deacon <will@kernel.org> wrote:
>
> On Mon, Sep 30, 2019 at 09:05:11PM +0900, Masahiro Yamada wrote:
> > On Mon, Sep 30, 2019 at 8:26 PM Will Deacon <will@kernel.org> wrote:
> > > On Fri, Sep 27, 2019 at 03:38:44PM -0700, Linus Torvalds wrote:
> > > > Soem of that code is pretty subtle. They have fixed register usage
> > > > (but the asm macros actually check them). And the inline asms clobber
> > > > the link register, but they do seem to clearly _state_ that they
> > > > clobber it, so who knows.
> > > >
> > > > Just based on the EFAULT, I'd _guess_ that it's some interaction with
> > > > the domain access control register (so that get/set_domain() thing).
> > > > But I'm not even sure that code is enabled for the Rpi2, so who
> > > > knows..
> > >
> > > FWIW, we've run into issues with CONFIG_OPTIMIZE_INLINING and local
> > > variables marked as 'register' where GCC would do crazy things and end
> > > up corrupting data, so I suspect the use of fixed registers in the arm
> > > uaccess functions is hitting something similar:
> > >
> > > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91111
> >
> > No. Not similar at all.
>
> They're similar in that enabling CONFIG_OPTIMIZE_INLINING causes register
> variables to go wrong. I agree that the ARM code looks dodgy with
> that call to uaccess_save_and_enable(), but there are __asmeq macros
> in there to try to catch that, so it's still very fishy.
>
> > I fixed it already. See
> > https://lore.kernel.org/patchwork/patch/1132459/
>
> You fixed the specific case above for 32-bit ARM, but the arm64 case
> is due to a compiler bug. As it happens, we've reworked our atomics
> in 5.4 so that particular issue no longer triggers, but the fact remains
> that GCC has been shown to screw up explicit register allocation for
> perfectly legitimate code when giving the flexibility to move code out
> of line.

So __attribute__((always_inline)) doesn't guarantee that code will be
inlined. For instance in LLVM's inliner, it asks/answers "should I
inline" and "can I inline." "Should" has to do with a cost model, and
is very heuristic-y. "Can" has more to do with the transforms, and
whether they're all implemented and safe. If you if you say
__attribute__((always_inline)), the answer to "can I inline this" can
still be *no*. The only way to guarantee inlining is via the C
preprocessor. The only way to prevent inlining is via
__attribute__((no_inline)). inline and __attribute__((always_inline))
are a heuristic laden mess and should not be relied upon. I would
also look closely at code that *requires* inlining or the lack there
of to be correct. That the kernel no longer compiles at -O0 is not a
good thing IMO, and hurts developers that want a short
compile/execute/debug cycle.

In this case, if there's a known codegen bug in a particular compiler
or certain versions of it, I recommend the use of either the C
preprocessor or __attribute__((no_inline)) to get the desired behavior
localized to the function in question, and for us to proceed with
Masahiro's cleanup.

The comment above the use of CONFIG_OPTIMIZE_INLINING in
include/linux/compiler_types.h says:
* Force always-inline if the user requests it so via the .config.
Which makes me grimace (__attribute__((always_inline)) doesn't *force*
anything as per above), and the idea that forcing things marked inline
to also be __attribute__((always_inline)) is an "optimization" (re:
the name of the config; CONFIG_OPTIMIZE_INLINING) is also highly
suspect. Aggressive inlining leads to image size bloat, instruction
cache and register pressure; it is not exclusively an optimization.

>
> > The problems are fixable by writing correct code.
>
> Right, in the compiler ;)
>
> > I think we discussed this already.
>
> We did?
>
> > - There is nothing arch-specific in CONFIG_OPTIMIZE_INLINING
>
> Apart from the bugs... and even then, that's just based on reports.
>
> > - 'inline' is just a hint. It does not guarantee function inlining.
> > This is standard.
> >
> > - The kernel macrofies 'inline' to add __attribute__((__always_inline__))
> > This terrible hack must end.
>
> I'm all for getting rid of hacks, but not at the cost of correctness.
>
> > - __attribute__((__always_inline__)) takes aways compiler's freedom,
> > and prevents it from optimizing the code for -O2, -Os, or whatever.
>
> s/whatever/miscompiling the code/
>
> If it helps, here is more information about the arm64 failure which
> triggered the GCC bugzilla:
>
> https://www.spinics.net/lists/arm-kernel/msg730329.html
>
> Will



--
Thanks,
~Nick Desaulniers

\
 
 \ /
  Last update: 2019-09-30 23:51    [W:0.176 / U:0.372 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site