<snip>

quoted

On Fri, Apr 30, 2021 at 5:27 PM fengchengwen

[off-list ref] wrote:

quoted

quoted

Hi, ALL
We have a question for your help:
  1. We have two platforms, both of which are ARM64, one of which

supports

quoted

quoted

     both NEON and SVE, the other only support NEON.
  2. We want to run on both platforms with a single binary file, and use the
     highest vector capability of the corresponding platform whenever

possible.

quoted

I see VPP has a similar feature. IMO, it is not present in DPDK.
Basically, In order to do this.
- Compile slow-path code(90% of DPDK) with minimal CPU instruction set
support
- Have fastpath function compile with different CPU instruction set
levels -In slowpath, Attach the fastpath function pointer-based on CPU
instruction-level support.

quoted

  3. So we build the DPDK program with -march=armv8-a+sve+crc (GCC

10.2).

quoted

quoted

     However, it is found that invalid instructions occur when the program
     runs on a machine that does not support SVE (pls see below).
  4. The problem is caused by the introduction of SVE in GCC automatic

vector

quoted

quoted

     optimization.

  So Is there a way to disable GCC automatic vector optimization or use

only

quoted

quoted

  NEON to perform automatic vector optimization?

  BTW: we already test -fno-tree-vectorize (as link below) but found no

effect.

quoted

quoted

https://stackoverflow.com/questions/7778174/how-can-i-disable-vector
ization-while-using-gcc


The GDB output:
     EAL: Detected 128 lcore(s)
     EAL: Detected 4 NUMA nodes
     Option -w, --pci-whitelist is deprecated, use -a, --allow
option instead

     Program received signal SIGILL, Illegal instruction.
     0x0000000000671b88 in eal_adjust_config ()
     (gdb)
     (gdb) where
     #0  0x0000000000671b88 in eal_adjust_config ()
     #1  0x0000000000682840 in rte_eal_init ()
     #2  0x000000000051c870 in main ()
     (gdb)

The disassembly output of eal_adjust_config:
     671b7c:       f8237a81        str     x1, [x20, x3, lsl #3]
     671b80:       f110001f        cmp     x0, #0x400
     671b84:       54ffff21        b.ne    671b68 <eal_adjust_config+0x1f4>  //

b.any

quoted

quoted

     671b88:       043357f5        addvl   x21, x19, #-1
     671b8c:       043457e1        addvl   x1, x20, #-1
     671b90:       910562b5        add     x21, x21, #0x158
     671b94:       04e0e3e0        cntd    x0
     671b98:       914012b5        add     x21, x21, #0x4, lsl #12
     671b9c:       52800218        mov     w24, #0x10                      // #16
     671ba0:       25d8e3e1        ptrue   p1.d
     671ba4:       25f80fe0        whilelo p0.d, wzr, w24
     671ba8:       a5e04020        ld1d    {z0.d}, p0/z, [x1, x0, lsl #3]


Best regards.

Is there a way to use Gcc function multiversioning for this?
https://gcc.gnu.org/onlinedocs/gcc/Function-Multiversioning.html

Not sure if this is only available on all compiler versions that DPDK claims to
support. It looks like it made into GCC 6 and LLVM 7

It looks like it is not fully support for Arm. For ex: 'target_clones' is not supported or automatic dispatcher does not seem to be supported, we need to write our own dispatcher. The following code works and should be sufficient for DPDK. There is no need to pass SVE flag at the command line for the compiler. I do not have a machine with SVE, so the SVE part is not tested.

#include <stdio.h>
#include <sys/auxv.h>

__attribute__((target ("arch=armv8-a+crc")))
int foo_neon ()
{
  printf ("Neon\n");
  return 1;
}

__attribute__((target ("arch=armv8-a+sve")))
int foo_sve ()
{
  printf ("SVE\n");
  return 2;
}

/*
  * The following code can go into IO function selection in DPDK during
  * initialization.
  */
void
foo_selector ()
{
  static int(*foo)(void);

  if (!foo)
     /* The following code can use DPDK wrappers */
     foo = getauxval(AT_HWCAP) & HWCAP_SVE ? foo_sve : foo_neon;

  foo ();
}

int main ()
{
  foo_selector();
  return 0;
}