RE: [PATCH v3 7/7][X86] Add Sub-byte element extration and Symmetric-signed saturation narrow support.

> From: Dipesh Sharma <[email protected]>
> Sent: Thursday, August 13, 2026 6:12 PM

Patch LGTM with minor changes.

> diff --git a/gcc/config/i386/i386-builtin.def b/gcc/config/i386/i386-
> builtin.def
> index ad963e9011b..84f23239326 100644
> --- a/gcc/config/i386/i386-builtin.def
> +++ b/gcc/config/i386/i386-builtin.def
> @@ -523,6 +523,12 @@ BDESC (OPTION_MASK_ISA_64BIT,
> OPTION_MASK_ISA2_MOVRS | OPTION_MASK_ISA2_AVX10_2,
>  BDESC (OPTION_MASK_ISA_64BIT, OPTION_MASK_ISA2_MOVRS |
> OPTION_MASK_ISA2_AVX10_2, CODE_FOR_avx10_2_vmovrsqv2di_mask,
> "__builtin_ia32_vmovrsq128_mask", IX86_BUILTIN_VMOVRSQ_128,
> UNKNOWN, (int) V2DI_FTYPE_PCV2DI_V2DI_UQI)
>  BDESC (OPTION_MASK_ISA_64BIT, OPTION_MASK_ISA2_MOVRS |
> OPTION_MASK_ISA2_AVX10_2, CODE_FOR_avx10_2_vmovrswv8hi_mask,
> "__builtin_ia32_vmovrsw128_mask", IX86_BUILTIN_VMOVRSW_128,
> UNKNOWN, (int) V8HI_FTYPE_PCV8HI_V8HI_UQI)
> 
> +/* AVX10V2AUX.  */
> +BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX,
> CODE_FOR_avx10v2aux_sym_truncatev4siv4qi2_mask_store_2,
> "__builtin_ia32_vpmovssdb128mem_mask",
> IX86_BUILTIN_VPMOVSSDB128_MEM, UNKNOWN, (int)
> VOID_FTYPE_PUSI_V4SI_UQI)
> +BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX,
> CODE_FOR_avx10v2aux_sym_truncatev8siv8qi2_mask_store_2,
> "__builtin_ia32_vpmovssdb256mem_mask",
> IX86_BUILTIN_VPMOVSSDB256_MEM, UNKNOWN, (int)
> VOID_FTYPE_PUDI_V8SI_UQI)
> +BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX,
> CODE_FOR_avx10v2aux_sym_truncatev16siv16qi2_mask_store,
> "__builtin_ia32_vpmovssdb512mem_mask",
> IX86_BUILTIN_VPMOVSSDB512_MEM, UNKNOWN, (int)
> VOID_FTYPE_PV16QI_V16SI_UHI)
> +
> +

Please do not add two blank lines here, just one and similar for ...

>  BDESC_END (SPECIAL_ARGS, PURE_ARGS)
> 
>  /* AVX */
> @@ -3427,6 +3433,13 @@ BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX,
> CODE_FOR_vcvtbf62hf8v64qi_mask, "__builti
>  BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX,
> CODE_FOR_vcvthf62hf8v16qi_mask,
> "__builtin_ia32_vcvthf62hf8128_mask",
> IX86_BUILTIN_VCVTHF62HF8128_MASK, UNKNOWN, (int)
> V16QI_FTYPE_V16QI_V16QI_UHI)
>  BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX,
> CODE_FOR_vcvthf62hf8v32qi_mask,
> "__builtin_ia32_vcvthf62hf8256_mask",
> IX86_BUILTIN_VCVTHF62HF8256_MASK, UNKNOWN, (int)
> V32QI_FTYPE_V32QI_V32QI_USI)
>  BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX,
> CODE_FOR_vcvthf62hf8v64qi_mask,
> "__builtin_ia32_vcvthf62hf8512_mask",
> IX86_BUILTIN_VCVTHF62HF8512_MASK, UNKNOWN, (int)
> V64QI_FTYPE_V64QI_V64QI_UDI)
> +BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX,
> CODE_FOR_vunpackbv16qi_mask, "__builtin_ia32_vunpackb128_mask",
> IX86_BUILTIN_VUNPACKB128_MASK, UNKNOWN, (int)
> V16QI_FTYPE_V16QI_INT_V16QI_UHI)
> +BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX,
> CODE_FOR_vunpackbv32qi_mask, "__builtin_ia32_vunpackb256_mask",
> IX86_BUILTIN_VUNPACKB256_MASK, UNKNOWN, (int)
> V32QI_FTYPE_V32QI_INT_V32QI_USI)
> +BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX,
> CODE_FOR_vunpackbv64qi_mask, "__builtin_ia32_vunpackb512_mask",
> IX86_BUILTIN_VUNPACKB512_MASK, UNKNOWN, (int)
> V64QI_FTYPE_V64QI_INT_V64QI_UDI)
> +BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX,
> CODE_FOR_avx10v2aux_sym_truncatev4siv4qi2_mask,
> "__builtin_ia32_vpmovssdb128_mask",
> IX86_BUILTIN_VPMOVSSDB128_MASK, UNKNOWN, (int)
> V16QI_FTYPE_V4SI_V16QI_UQI)
> +BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX,
> CODE_FOR_avx10v2aux_sym_truncatev8siv8qi2_mask,
> "__builtin_ia32_vpmovssdb256_mask",
> IX86_BUILTIN_VPMOVSSDB256_MASK, UNKNOWN, (int)
> V16QI_FTYPE_V8SI_V16QI_UQI)
> +BDESC (0, OPTION_MASK_ISA2_AVX10V2AUX,
> CODE_FOR_avx10v2aux_sym_truncatev16siv16qi2_mask,
> "__builtin_ia32_vpmovssdb512_mask",
> IX86_BUILTIN_VPMOVSSDB512_MASK, UNKNOWN, (int)
> V16QI_FTYPE_V16SI_V16QI_UHI)
> +

... here.

> 
>  /* Builtins with rounding support.  */
>  BDESC_END (ARGS, ROUND_ARGS)
> diff --git a/gcc/config/i386/i386-expand.cc b/gcc/config/i386/i386-expand.cc
> index fb4224af0da..7971ddce080 100644
> --- a/gcc/config/i386/i386-expand.cc
> +++ b/gcc/config/i386/i386-expand.cc
> @@ -13322,6 +13322,9 @@ ix86_expand_args_builtin (const struct
> builtin_description *d,
>      case V4DF_FTYPE_V8DF_INT_V4DF_UQI:
>      case V4SF_FTYPE_V16SF_INT_V4SF_UQI:
>      case V8DI_FTYPE_V8DI_INT_V8DI_UQI:
> +    case V16QI_FTYPE_V16QI_INT_V16QI_UHI:
> +    case V32QI_FTYPE_V32QI_INT_V32QI_USI:
> +    case V64QI_FTYPE_V64QI_INT_V64QI_UDI:
>        nargs = 4;
>        mask_pos = 2;
>        nargs_constant = 1;
> @@ -13453,6 +13456,8 @@ ix86_expand_args_builtin (const struct
> builtin_description *d,
>        else if ((mask_pos && (nargs - i - mask_pos) == nargs_constant) ||
>  	       (!mask_pos && (nargs - i) <= nargs_constant))
>  	{
> +
> +

And also here redundant blank lines.

>  	  if (!match)
>  	    switch (icode)
>  	      {
> diff --git a/gcc/testsuite/g++.dg/other/i386-2.C
> b/gcc/testsuite/g++.dg/other/i386-2.C
> index aae565053dc..31775c203b9 100644
> --- a/gcc/testsuite/g++.dg/other/i386-2.C
> +++ b/gcc/testsuite/g++.dg/other/i386-2.C
> @@ -1,5 +1,5 @@
>  /* { dg-do compile { target i?86-*-* x86_64-*-* } } */
> -/* { dg-options "-O -pedantic-errors -march=k8 -msse4a -m3dnow -mavx -
> mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -
> mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -
> mprfchw -madx -mfxsr -mxsaveopt -msha -mxsavec -mxsaves -mclflushopt -
> mclwb -mmwaitx -mclzero -mpku -msgx -mrdpid -mgfni -mpconfig -
> mwbnoinvd -menqcmd -mavx512vp2intersect -mserialize -mtsxldtrk -mamx-
> tile -mamx-int8 -mamx-bf16 -mkl -mwidekl -mavxvnni -mavxifma -
> mavxvnniint8 -mavxneconvert -mcmpccxadd -mamx-fp16 -mprefetchi -
> mraoint -mamx-complex -mavxvnniint16 -msm3 -msha512 -msm4 -
> mavx10.2 -mamx-avx512 -mamx-fp8 -mmovrs -mamx-movrs" } */
> +/* { dg-options "-O -pedantic-errors -march=k8 -msse4a -m3dnow -mavx -
> mavx2 -mfma4 -mxop -maes -mpclmul -mpopcnt -mabm -mlzcnt -mbmi -
> mbmi2 -mtbm -mlwp -mfsgsbase -mrdrnd -mf16c -mfma -mrtm -mrdseed -
> mprfchw -madx -mfxsr -mxsaveopt -msha -mxsavec -mxsaves -mclflushopt -
> mclwb -mmwaitx -mclzero -mpku -msgx -mrdpid -mgfni -mpconfig -
> mwbnoinvd -menqcmd -mavx512vp2intersect -mserialize -mtsxldtrk -mamx-
> tile -mamx-int8 -mamx-bf16 -mkl -mwidekl -mavxvnni -mavxifma -
> mavxvnniint8 -mavxneconvert -mcmpccxadd -mamx-fp16 -mprefetchi -
> mraoint -mamx-complex -mavxvnniint16 -msm3 -msha512 -msm4 -
> mavx10.2 -mamx-avx512 -mamx-fp8 -mmovrs -mamx-movrs -
> mavx10v2aux" } */
>  /* { dg-skip-if "requires hosted libstdc++ for cstdlib malloc" { ! hostedlib } } */
> 
>  /* Test that {,x,e,p,t,s,w,a,b,i}mmintrin.h, mm3dnow.h, fma4intrin.h,
> @@ -7,7 +7,7 @@
>     popcntintrin.h, fmaintrin.h, pkuintrin.h, avx512vpopcntdqintrin.h,
>     gfniintrin.h, avx512bitalgintrin.h, avx512vp2intersectintrin.h,
>     tsxldtrkintrin.h, amxtileintrin.h, amxint8intrin.h, amxbf16intrin.h,
> -   avx512vp2intersectvlintrin.h and mm_malloc.h.h are usable
> +   avx512vp2intersectvlintrin.h, avx10v2auxintrin.h and mm_malloc.h.h are
> usable

Please don't touch these lines. They are not adding new files anymore
or the list will be extreme huge. Similar for other cases.

Thx,
Haochen