[PATCH 0/3] lib/string: optimized mem* functions

[Matteo Croce]

On Fri, Jun 25, 2021 at 7:45 PM Nick Desaulniers
<ndesaulniers at google.com> wrote:
>
> On Thu, Jun 24, 2021 at 6:02 PM Matteo Croce <mcroce at linux.microsoft.com> wrote:
> >
> > From: Matteo Croce <mcroce at microsoft.com>
> >
> > Rewrite the generic mem{cpy,move,set} so that memory is accessed with
> > the widest size possible, but without doing unaligned accesses.
> >
> > This was originally posted as C string functions for RISC-V[1], but as
> > there was no specific RISC-V code, it was proposed for the generic
> > lib/string.c implementation.
> >
> > Tested on RISC-V and on x86_64 by undefining __HAVE_ARCH_MEM{CPY,SET,MOVE}
> > and HAVE_EFFICIENT_UNALIGNED_ACCESS.
> >
> > Further testing on big endian machines will be appreciated, as I don't
> > have such hardware at the moment.
>
> Hi Matteo,
> Neat patches.  Do you have you any benchmark data showing the claimed
> improvements? Is it worthwhile to define these only when
> CC_OPTIMIZE_FOR_PERFORMANCE/CC_OPTIMIZE_FOR_PERFORMANCE_O3 are
> defined, not CC_OPTIMIZE_FOR_SIZE? I'd be curious to know the delta in
> ST_SIZE of these functions otherwise.
>

I compared the current versions with the new one with bloat-o-meter,
the kernel grows by ~400 bytes on x86_64 and RISC-V

x86_64

$ scripts/bloat-o-meter vmlinux.orig vmlinux
add/remove: 0/0 grow/shrink: 4/1 up/down: 427/-6 (421)
Function                                     old     new   delta
memcpy                                        29     351    +322
memset                                        29     117     +88
strlcat                                       68      78     +10
strlcpy                                       50      57      +7
memmove                                       56      50      -6
Total: Before=8556964, After=8557385, chg +0.00%

RISC-V

$ scripts/bloat-o-meter vmlinux.orig vmlinux
add/remove: 0/0 grow/shrink: 4/2 up/down: 432/-36 (396)
Function                                     old     new   delta
memcpy                                        36     324    +288
memset                                        32     148    +116
strlcpy                                      116     132     +16
strscpy_pad                                   84      96     +12
strlcat                                      176     164     -12
memmove                                       76      52     -24
Total: Before=1225371, After=1225767, chg +0.03%

I will post benchmarks made on a RISC-V machine which can't handle
unaligned accesses, and it will be the first user of the new
functions.

> For big endian, you ought to be able to boot test in QEMU.  I think
> you'd find out pretty quickly if any of the above had issues.
> (Enabling KASAN is probably also a good idea for a test, too). Check
> out
> https://github.com/ClangBuiltLinux/boot-utils
> for ready made images and scripts for launching various architectures
> and endiannesses.
>

Will do!

-- 
per aspera ad upstream