Re: [PATCH 5.15 00/78] 5.15.55-rc1 review

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Oh, btw, how bad would it be to just do

     #define FASTOP_SIZE 16
     static_assert(FASTOP_SIZE >= FASTOP_LENGTH)

and leave it at that?

Afaik both gcc and clang default to -falign-functions=16 *anyway*, and
while on 32-bit x86 we have options to minimize alignment, we don't do
that on x86-64 afaik.

In fact, we have an option to force *bigger* alignment
(DEBUG_FORCE_FUNCTION_ALIGN_64B) but not any way to make it less.

And we use

        .p2align 4

in most of our asm, aling with

    #define __ALIGN          .p2align 4, 0x90

So all the *normal* functions already get 16-byte alignment anyway.

So yeah, it would be less dense, but do we care? Wouldn't the "this is
really simple" be a nice thing? It's not like there are a ton of those
fastop functions anyway. 128 of them? Plus 16 of the "setCC" ones?

               Linus



[Index of Archives]     [Linux Kernel]     [Kernel Development Newbies]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite Hiking]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux