lkml.org 
[lkml]   [2022]   [Jun]   [20]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: [PATCH v3 0/7] bitops: let optimize out non-atomic bitops on compile-time constants
Date
From: Geert Uytterhoeven <geert@linux-m68k.org>
Date: Mon, 20 Jun 2022 14:07:27 +0200

> Hi Olek,

Hey!

>
> On Fri, Jun 17, 2022 at 6:51 PM Alexander Lobakin
> <alexandr.lobakin@intel.com> wrote:
> > While I was working on converting some structure fields from a fixed
> > type to a bitmap, I started observing code size increase not only in
> > places where the code works with the converted structure fields, but
> > also where the converted vars were on the stack. That said, the
> > following code:
> >
> > DECLARE_BITMAP(foo, BITS_PER_LONG) = { }; // -> unsigned long foo[1];
> > unsigned long bar = BIT(BAR_BIT);
> > unsigned long baz = 0;
> >
> > __set_bit(FOO_BIT, foo);
> > baz |= BIT(BAZ_BIT);
> >
> > BUILD_BUG_ON(!__builtin_constant_p(test_bit(FOO_BIT, foo));
> > BUILD_BUG_ON(!__builtin_constant_p(bar & BAR_BIT));
> > BUILD_BUG_ON(!__builtin_constant_p(baz & BAZ_BIT));
> >
> > triggers the first assertion on x86_64, which means that the
> > compiler is unable to evaluate it to a compile-time initializer
> > when the architecture-specific bitop is used even if it's obvious.
> > I found that this is due to that many architecture-specific
> > non-atomic bitop implementations use inline asm or other hacks which
> > are faster or more robust when working with "real" variables (i.e.
> > fields from the structures etc.), but the compilers have no clue how
> > to optimize them out when called on compile-time constants.
> >
> > So, in order to let the compiler optimize out such cases, expand the
> > test_bit() and __*_bit() definitions with a compile-time condition
> > check, so that they will pick the generic C non-atomic bitop
> > implementations when all of the arguments passed are compile-time
> > constants, which means that the result will be a compile-time
> > constant as well and the compiler will produce more efficient and
> > simple code in 100% cases (no changes when there's at least one
> > non-compile-time-constant argument).
> > The condition itself:
> >
> > if (
> > __builtin_constant_p(nr) && /* <- bit position is constant */
> > __builtin_constant_p(!!addr) && /* <- compiler knows bitmap addr is
> > always either NULL or not */
> > addr && /* <- bitmap addr is not NULL */
> > __builtin_constant_p(*addr) /* <- compiler knows the value of
> > the target bitmap */
> > )
> > /* then pick the generic C variant
> > else
> > /* old code path, arch-specific
> >
> > I also tried __is_constexpr() as suggested by Andy, but it was
> > always returning 0 ('not a constant') for the 2,3 and 4th
> > conditions.
> >
> > The savings are architecture, compiler and compiler flags dependent,
> > for example, on x86_64 -O2:
> >
> > GCC 12: add/remove: 78/29 grow/shrink: 332/525 up/down: 31325/-61560 (-30235)
> > LLVM 13: add/remove: 79/76 grow/shrink: 184/537 up/down: 55076/-141892 (-86816)
> > LLVM 14: add/remove: 10/3 grow/shrink: 93/138 up/down: 3705/-6992 (-3287)
> >
> > and ARM64 (courtesy of Mark[0]):
> >
> > GCC 11: add/remove: 92/29 grow/shrink: 933/2766 up/down: 39340/-82580 (-43240)
> > LLVM 14: add/remove: 21/11 grow/shrink: 620/651 up/down: 12060/-15824 (-3764)
> >
> > And the following:
> >
> > DECLARE_BITMAP(flags, __IP_TUNNEL_FLAG_NUM) = { };
> > __be16 flags;
> >
> > __set_bit(IP_TUNNEL_CSUM_BIT, flags);
> >
> > tun_flags = cpu_to_be16(*flags & U16_MAX);
> >
> > if (test_bit(IP_TUNNEL_VTI_BIT, flags))
> > tun_flags |= VTI_ISVTI;
> >
> > BUILD_BUG_ON(!__builtin_constant_p(tun_flags));
> >
> > doesn't blow up anymore (which is being checked now at build time),
> > so that we can now e.g. use fixed bitmaps in compile-time assertions
> > etc.
> >
> > The series has been in intel-next for a while with no reported issues.
> >
> > From v2[1]:
> > * collect several Reviewed-bys (Andy, Yury);
> > * add a comment to generic_test_bit() that it is atomic-safe and
> > must always stay like that (the first version of this series
> > errorneously tried to change this) (Andy, Marco);
> > * unify the way how architectures define platform-specific bitops,
> > both supporting instrumentation and not: now they define only
> > 'arch_' versions and asm-generic includes take care of the rest;
> > * micro-optimize the diffstat of 0004/0007 (__check_bitop_pr())
> > (Andy);
> > * add compile-time tests to lib/test_bitmap to make sure everything
> > works as expected on any setup (Yury).
>
> Thanks for the update!
>
> Still seeing
> add/remove: 49/13 grow/shrink: 280/137 up/down: 6464/-3328 (3136)

Meh. What about -O2 (OPTIMIZE_FOR_PERFORMANCE)? I have a thought to
make it depend on the config option above, but that would make code
behave differently, so it's not safe.
Are those 3 Kb critical for m68k machines? I'm asking because for
some embedded systems they are :)
Another thing, this could happen due to inlining rebalance. E.g.
the compiler could inline or uninline some functions due to
resolving bit{maps,ops} to compile-time constants. I was seeing
such in the past several times. Also, IIRC you already sent some
bloat-o-meter results here, and that was the case.
On the other hand, if lib/test_bitmap.o builds successfully (I
assume it does), the series works as expected.

>
> on m68k atari_defconfig (i.e.CONFIG_CC_OPTIMIZE_FOR_SIZE=y)
> with gcc version 9.4.0 (Ubuntu 9.4.0-1ubuntu1~20.04).
>
> Gr{oetje,eeting}s,
>
> Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
>
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
> -- Linus Torvalds

Thanks,
Olek

\
 
 \ /
  Last update: 2022-06-20 16:06    [W:0.605 / U:1.036 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site