[PATCH 1/5] examples/l3fwd: fix port group mask generation

[PATCH 1/5] examples/l3fwd: fix port group mask generation

[pbhagavatula]

From: Pavan Nikhilesh <pbhagavatula@marvell.com>

Fix port group mask generation in altivec, vec_any_eq returns
0 or 1 while port_groupx4 expects comparison mask result.

Fixes: 2193b7467f7a ("examples/l3fwd: optimize packet processing on powerpc")
Cc: stable@dpdk.org

Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
---
 examples/common/altivec/port_group.h | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/examples/common/altivec/port_group.h b/examples/common/altivec/port_group.h
index 5e209b02fa..7a6ef390ff 100644
--- a/examples/common/altivec/port_group.h
+++ b/examples/common/altivec/port_group.h
@@ -26,12 +26,19 @@ port_groupx4(uint16_t pn[FWDSTEP + 1], uint16_t *lp,
 		uint16_t u16[FWDSTEP + 1];
 		uint64_t u64;
 	} *pnum = (void *)pn;
+	union u_vec {
+		__vector unsigned short v_us;
+		unsigned short s[8];
+	};
 
+	union u_vec res;
 	int32_t v;
 
-	v = vec_any_eq(dp1, dp2);
-
+	dp1 = vec_cmpeq(dp1, dp2);
+	res.v_us = dp1;
 
+	v = (res.s[0] & 0x1) | (res.s[1] & 0x2) | (res.s[2] & 0x4) |
+	    (res.s[3] & 0x8);
 	/* update last port counter. */
 	lp[0] += gptbl[v].lpv;
 
-- 
2.25.1

[PATCH v2 1/5] examples/l3fwd: fix port group mask generation

[pbhagavatula]

From: Pavan Nikhilesh <pbhagavatula@marvell.com>

Fix port group mask generation in altivec, vec_any_eq returns
0 or 1 while port_groupx4 expects comparison mask result.

Fixes: 2193b7467f7a ("examples/l3fwd: optimize packet processing on powerpc")
Cc: stable@dpdk.org

Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
---
 v2 Changes:
 - Fix PPC, RISC-V, aarch32 compilation.

 examples/common/altivec/port_group.h | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/examples/common/altivec/port_group.h b/examples/common/altivec/port_group.h
index 5e209b02fa..592ef80b7f 100644
--- a/examples/common/altivec/port_group.h
+++ b/examples/common/altivec/port_group.h
@@ -26,12 +26,19 @@ port_groupx4(uint16_t pn[FWDSTEP + 1], uint16_t *lp,
 		uint16_t u16[FWDSTEP + 1];
 		uint64_t u64;
 	} *pnum = (void *)pn;
+	union u_vec {
+		__vector unsigned short v_us;
+		unsigned short s[8];
+	};

+	union u_vec res;
 	int32_t v;

-	v = vec_any_eq(dp1, dp2);
-
+	dp1 = (__vector unsigned short)vec_cmpeq(dp1, dp2);
+	res.v_us = dp1;

+	v = (res.s[0] & 0x1) | (res.s[1] & 0x2) | (res.s[2] & 0x4) |
+	    (res.s[3] & 0x8);
 	/* update last port counter. */
 	lp[0] += gptbl[v].lpv;

--
2.25.1

Re: [PATCH v2 1/5] examples/l3fwd: fix port group mask generation

[David Christensen]

On 9/2/22 2:18 AM, pbhagavatula@marvell.com wrote:
> From: Pavan Nikhilesh <pbhagavatula@marvell.com>
> 
> Fix port group mask generation in altivec, vec_any_eq returns
> 0 or 1 while port_groupx4 expects comparison mask result.
> 
> Fixes: 2193b7467f7a ("examples/l3fwd: optimize packet processing on powerpc")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
> ---
>   v2 Changes:
>   - Fix PPC, RISC-V, aarch32 compilation.
> 
>   examples/common/altivec/port_group.h | 11 +++++++++--
>   1 file changed, 9 insertions(+), 2 deletions(-)
> 
> diff --git a/examples/common/altivec/port_group.h b/examples/common/altivec/port_group.h
> index 5e209b02fa..592ef80b7f 100644
> --- a/examples/common/altivec/port_group.h
> +++ b/examples/common/altivec/port_group.h
> @@ -26,12 +26,19 @@ port_groupx4(uint16_t pn[FWDSTEP + 1], uint16_t *lp,
>   		uint16_t u16[FWDSTEP + 1];
>   		uint64_t u64;
>   	} *pnum = (void *)pn;
> +	union u_vec {
> +		__vector unsigned short v_us;
> +		unsigned short s[8];
> +	};
> 
> +	union u_vec res;
>   	int32_t v;
> 
> -	v = vec_any_eq(dp1, dp2);
> -
> +	dp1 = (__vector unsigned short)vec_cmpeq(dp1, dp2);

Altivec vec_cmpeq() is similar to Intel _mm_cmpeq_*(), so this looks 
right to me.

> +	res.v_us = dp1;
> 
> +	v = (res.s[0] & 0x1) | (res.s[1] & 0x2) | (res.s[2] & 0x4) |
> +	    (res.s[3] & 0x8);

This can be vectorized too.  The Intel _mm_unpacklo_epi16() intrinsic 
can be replaced with the following Altivec code:

extern __inline __m128i __attribute__((__gnu_inline__, 
__always_inline__, __artificial__))
_mm_unpacklo_epi16 (__m128i __A, __m128i __B)
{
   return (__m128i) vec_mergeh ((__v8hi)__A, (__v8hi)__B);
}

The Intel _mm_movemask_ps() intrinsic can be replaced with the following 
Altivec implementation:

/* Creates a 4-bit mask from the most significant bits of the SPFP 
values.  */
extern __inline int __attribute__((__gnu_inline__, __always_inline__, 
__artificial__))
_mm_movemask_ps (__m128  __A)
{
   __vector unsigned long long result;
   static const __vector unsigned int perm_mask =
     {
#ifdef __LITTLE_ENDIAN__
         0x00204060, 0x80808080, 0x80808080, 0x80808080
#else
       0x80808080, 0x80808080, 0x80808080, 0x00204060
#endif
     };

   result = ((__vector unsigned long long)
             vec_vbpermq ((__vector unsigned char) __A,
                          (__vector unsigned char) perm_mask));

#ifdef __LITTLE_ENDIAN__
   return result[1];
#else
   return result[0];
#endif
}

Dave

RE: [EXT] Re: [PATCH v2 1/5] examples/l3fwd: fix port group mask generation

[Pavan Nikhilesh Bhagavatula]

> On 9/2/22 2:18 AM, pbhagavatula@marvell.com wrote:
> > From: Pavan Nikhilesh <pbhagavatula@marvell.com>
> >
> > Fix port group mask generation in altivec, vec_any_eq returns
> > 0 or 1 while port_groupx4 expects comparison mask result.
> >
> > Fixes: 2193b7467f7a ("examples/l3fwd: optimize packet processing on
> powerpc")
> > Cc: stable@dpdk.org
> >
> > Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
> > ---
> >   v2 Changes:
> >   - Fix PPC, RISC-V, aarch32 compilation.
> >
> >   examples/common/altivec/port_group.h | 11 +++++++++--
> >   1 file changed, 9 insertions(+), 2 deletions(-)
> >
> > diff --git a/examples/common/altivec/port_group.h
> b/examples/common/altivec/port_group.h
> > index 5e209b02fa..592ef80b7f 100644
> > --- a/examples/common/altivec/port_group.h
> > +++ b/examples/common/altivec/port_group.h
> > @@ -26,12 +26,19 @@ port_groupx4(uint16_t pn[FWDSTEP + 1], uint16_t
> *lp,
> >   		uint16_t u16[FWDSTEP + 1];
> >   		uint64_t u64;
> >   	} *pnum = (void *)pn;
> > +	union u_vec {
> > +		__vector unsigned short v_us;
> > +		unsigned short s[8];
> > +	};
> >
> > +	union u_vec res;
> >   	int32_t v;
> >
> > -	v = vec_any_eq(dp1, dp2);
> > -
> > +	dp1 = (__vector unsigned short)vec_cmpeq(dp1, dp2);
> 
> Altivec vec_cmpeq() is similar to Intel _mm_cmpeq_*(), so this looks
> right to me.
> 
> > +	res.v_us = dp1;
> >
> > +	v = (res.s[0] & 0x1) | (res.s[1] & 0x2) | (res.s[2] & 0x4) |
> > +	    (res.s[3] & 0x8);
> 
> This can be vectorized too.  The Intel _mm_unpacklo_epi16() intrinsic
> can be replaced with the following Altivec code:
> 
> extern __inline __m128i __attribute__((__gnu_inline__,
> __always_inline__, __artificial__))
> _mm_unpacklo_epi16 (__m128i __A, __m128i __B)
> {
>    return (__m128i) vec_mergeh ((__v8hi)__A, (__v8hi)__B);
> }
> 
> The Intel _mm_movemask_ps() intrinsic can be replaced with the following
> Altivec implementation:
> 
> /* Creates a 4-bit mask from the most significant bits of the SPFP
> values.  */
> extern __inline int __attribute__((__gnu_inline__, __always_inline__,
> __artificial__))
> _mm_movemask_ps (__m128  __A)
> {
>    __vector unsigned long long result;
>    static const __vector unsigned int perm_mask =
>      {
> #ifdef __LITTLE_ENDIAN__
>          0x00204060, 0x80808080, 0x80808080, 0x80808080
> #else
>        0x80808080, 0x80808080, 0x80808080, 0x00204060
> #endif
>      };
> 
>    result = ((__vector unsigned long long)
>              vec_vbpermq ((__vector unsigned char) __A,
>                           (__vector unsigned char) perm_mask));
> 
> #ifdef __LITTLE_ENDIAN__
>    return result[1];
> #else
>    return result[0];
> #endif
> }
> 

Sure I will add this to the next version.

> Dave

Thanks, 
Pavan.

[PATCH v3 1/5] examples/l3fwd: fix port group mask generation

[pbhagavatula]

From: Pavan Nikhilesh <pbhagavatula@marvell.com>

Fix port group mask generation in altivec, vec_any_eq returns
0 or 1 while port_groupx4 expects comparison mask result.

Fixes: 2193b7467f7a ("examples/l3fwd: optimize packet processing on powerpc")
Cc: stable@dpdk.org

Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
---
 v3 Changes:
 - PPC optimize port mask generation.
 - Fix aarch32 compilation.

 v2 Changes:
 - Fix PPC, RISC-V, aarch32 compilation.

 examples/common/altivec/port_group.h | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/examples/common/altivec/port_group.h b/examples/common/altivec/port_group.h
index 5e209b02fa..1c05bc025a 100644
--- a/examples/common/altivec/port_group.h
+++ b/examples/common/altivec/port_group.h
@@ -26,12 +26,17 @@ port_groupx4(uint16_t pn[FWDSTEP + 1], uint16_t *lp,
 		uint16_t u16[FWDSTEP + 1];
 		uint64_t u64;
 	} *pnum = (void *)pn;
-
+	__vector unsigned long long result;
+	const __vector unsigned int perm_mask = {0x00204060, 0x80808080,
+						 0x80808080, 0x80808080};
 	int32_t v;

-	v = vec_any_eq(dp1, dp2);
-
+	dp1 = (__vector unsigned short)vec_cmpeq(dp1, dp2);
+	dp1 = vec_mergeh(dp1, dp1);
+	result = (__vector unsigned long long)vec_vbpermq(
+		(__vector unsigned char)dp1, (__vector unsigned char)perm_mask);

+	v = result[1];
 	/* update last port counter. */
 	lp[0] += gptbl[v].lpv;

--
2.25.1

[PATCH v4 1/5] examples/l3fwd: fix port group mask generation

[pbhagavatula]

From: Pavan Nikhilesh <pbhagavatula@marvell.com>

Fix port group mask generation in altivec, vec_any_eq returns
0 or 1 while port_groupx4 expects comparison mask result.

Fixes: 2193b7467f7a ("examples/l3fwd: optimize packet processing on powerpc")
Cc: stable@dpdk.org

Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
---
 v4 Changes:
 - Fix missing `rte_free`.

 v3 Changes:
 - PPC optimize port mask generation.
 - Fix aarch32 compilation.

 v2 Changes:
 - Fix PPC, RISC-V, aarch32 compilation.

 examples/common/altivec/port_group.h | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/examples/common/altivec/port_group.h b/examples/common/altivec/port_group.h
index 5e209b02fa..1c05bc025a 100644
--- a/examples/common/altivec/port_group.h
+++ b/examples/common/altivec/port_group.h
@@ -26,12 +26,17 @@ port_groupx4(uint16_t pn[FWDSTEP + 1], uint16_t *lp,
 		uint16_t u16[FWDSTEP + 1];
 		uint64_t u64;
 	} *pnum = (void *)pn;
-
+	__vector unsigned long long result;
+	const __vector unsigned int perm_mask = {0x00204060, 0x80808080,
+						 0x80808080, 0x80808080};
 	int32_t v;

-	v = vec_any_eq(dp1, dp2);
-
+	dp1 = (__vector unsigned short)vec_cmpeq(dp1, dp2);
+	dp1 = vec_mergeh(dp1, dp1);
+	result = (__vector unsigned long long)vec_vbpermq(
+		(__vector unsigned char)dp1, (__vector unsigned char)perm_mask);

+	v = result[1];
 	/* update last port counter. */
 	lp[0] += gptbl[v].lpv;

--
2.25.1

[PATCH v5 1/5] examples/l3fwd: fix port group mask generation

[pbhagavatula]

From: Pavan Nikhilesh <pbhagavatula@marvell.com>

Fix port group mask generation in altivec, vec_any_eq returns
0 or 1 while port_groupx4 expects comparison mask result.

Fixes: 2193b7467f7a ("examples/l3fwd: optimize packet processing on powerpc")
Cc: stable@dpdk.org

Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
---
 v5 Changes:
 - Fix compilation errors.

 v4 Changes:
 - Fix missing `rte_free`.

 v3 Changes:
 - PPC optimize port mask generation.
 - Fix aarch32 compilation.

 v2 Changes:
 - Fix PPC, RISC-V, aarch32 compilation.

 examples/common/altivec/port_group.h | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/examples/common/altivec/port_group.h b/examples/common/altivec/port_group.h
index 5e209b02fa..1c05bc025a 100644
--- a/examples/common/altivec/port_group.h
+++ b/examples/common/altivec/port_group.h
@@ -26,12 +26,17 @@ port_groupx4(uint16_t pn[FWDSTEP + 1], uint16_t *lp,
 		uint16_t u16[FWDSTEP + 1];
 		uint64_t u64;
 	} *pnum = (void *)pn;
-
+	__vector unsigned long long result;
+	const __vector unsigned int perm_mask = {0x00204060, 0x80808080,
+						 0x80808080, 0x80808080};
 	int32_t v;

-	v = vec_any_eq(dp1, dp2);
-
+	dp1 = (__vector unsigned short)vec_cmpeq(dp1, dp2);
+	dp1 = vec_mergeh(dp1, dp1);
+	result = (__vector unsigned long long)vec_vbpermq(
+		(__vector unsigned char)dp1, (__vector unsigned char)perm_mask);

+	v = result[1];
 	/* update last port counter. */
 	lp[0] += gptbl[v].lpv;

--
2.25.1

RE: [EXT] [PATCH v5 1/5] examples/l3fwd: fix port group mask generation

[Shijith Thotton]

[-- Attachment #1: Type: text/plain, Size: 1601 bytes --]

>
>Fix port group mask generation in altivec, vec_any_eq returns
>0 or 1 while port_groupx4 expects comparison mask result.
>
>Fixes: 2193b7467f7a ("examples/l3fwd: optimize packet processing on powerpc")
>Cc: stable@dpdk.org
>
>Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>

Acked-by: Shijith Thotton <sthotton@marvell.com>

>---
> v5 Changes:
> - Fix compilation errors.
>
> v4 Changes:
> - Fix missing `rte_free`.
>
> v3 Changes:
> - PPC optimize port mask generation.
> - Fix aarch32 compilation.
>
> v2 Changes:
> - Fix PPC, RISC-V, aarch32 compilation.
>
> examples/common/altivec/port_group.h | 11 ++++++++---
> 1 file changed, 8 insertions(+), 3 deletions(-)
>
>diff --git a/examples/common/altivec/port_group.h
>b/examples/common/altivec/port_group.h
>index 5e209b02fa..1c05bc025a 100644
>--- a/examples/common/altivec/port_group.h
>+++ b/examples/common/altivec/port_group.h
>@@ -26,12 +26,17 @@ port_groupx4(uint16_t pn[FWDSTEP + 1], uint16_t *lp,
> 		uint16_t u16[FWDSTEP + 1];
> 		uint64_t u64;
> 	} *pnum = (void *)pn;
>-
>+	__vector unsigned long long result;
>+	const __vector unsigned int perm_mask = {0x00204060, 0x80808080,
>+						 0x80808080, 0x80808080};
> 	int32_t v;
>
>-	v = vec_any_eq(dp1, dp2);
>-
>+	dp1 = (__vector unsigned short)vec_cmpeq(dp1, dp2);
>+	dp1 = vec_mergeh(dp1, dp1);
>+	result = (__vector unsigned long long)vec_vbpermq(
>+		(__vector unsigned char)dp1, (__vector unsigned
>char)perm_mask);
>
>+	v = result[1];
> 	/* update last port counter. */
> 	lp[0] += gptbl[v].lpv;
>
>--
>2.25.1


[-- Attachment #2: winmail.dat --]
[-- Type: application/ms-tnef, Size: 14616 bytes --]

RE: [EXT] [PATCH v5 1/5] examples/l3fwd: fix port group mask generation

[Pavan Nikhilesh Bhagavatula]

> -----Original Message-----
> From: Shijith Thotton <sthotton@marvell.com>
> Sent: Monday, October 17, 2022 5:36 PM
> To: Pavan Nikhilesh Bhagavatula <pbhagavatula@marvell.com>; Jerin Jacob
> Kollanukkaran <jerinj@marvell.com>; David Christensen
> <drc@linux.vnet.ibm.com>
> Cc: dev@dpdk.org; Pavan Nikhilesh Bhagavatula
> <pbhagavatula@marvell.com>; stable@dpdk.org
> Subject: RE: [EXT] [PATCH v5 1/5] examples/l3fwd: fix port group mask
> generation
> 
> >
> >Fix port group mask generation in altivec, vec_any_eq returns
> >0 or 1 while port_groupx4 expects comparison mask result.
> >
> >Fixes: 2193b7467f7a ("examples/l3fwd: optimize packet processing on
> powerpc")
> >Cc: stable@dpdk.org
> >
> >Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
> 
> Acked-by: Shijith Thotton <sthotton@marvell.com>
> 

Thomas, 

Will this series make it into 22.11 release?

> >---
> > v5 Changes:
> > - Fix compilation errors.
> >
> > v4 Changes:
> > - Fix missing `rte_free`.
> >
> > v3 Changes:
> > - PPC optimize port mask generation.
> > - Fix aarch32 compilation.
> >
> > v2 Changes:
> > - Fix PPC, RISC-V, aarch32 compilation.
> >
> > examples/common/altivec/port_group.h | 11 ++++++++---
> > 1 file changed, 8 insertions(+), 3 deletions(-)
> >
> >diff --git a/examples/common/altivec/port_group.h
> >b/examples/common/altivec/port_group.h
> >index 5e209b02fa..1c05bc025a 100644
> >--- a/examples/common/altivec/port_group.h
> >+++ b/examples/common/altivec/port_group.h
> >@@ -26,12 +26,17 @@ port_groupx4(uint16_t pn[FWDSTEP + 1], uint16_t
> *lp,
> > 		uint16_t u16[FWDSTEP + 1];
> > 		uint64_t u64;
> > 	} *pnum = (void *)pn;
> >-
> >+	__vector unsigned long long result;
> >+	const __vector unsigned int perm_mask = {0x00204060, 0x80808080,
> >+						 0x80808080, 0x80808080};
> > 	int32_t v;
> >
> >-	v = vec_any_eq(dp1, dp2);
> >-
> >+	dp1 = (__vector unsigned short)vec_cmpeq(dp1, dp2);
> >+	dp1 = vec_mergeh(dp1, dp1);
> >+	result = (__vector unsigned long long)vec_vbpermq(
> >+		(__vector unsigned char)dp1, (__vector unsigned
> >char)perm_mask);
> >
> >+	v = result[1];
> > 	/* update last port counter. */
> > 	lp[0] += gptbl[v].lpv;
> >
> >--
> >2.25.1

[PATCH v6 1/5] examples/l3fwd: fix port group mask generation

[pbhagavatula]

From: Pavan Nikhilesh <pbhagavatula@marvell.com>

Fix port group mask generation in altivec, vec_any_eq returns
0 or 1 while port_groupx4 expects comparison mask result.

Fixes: 2193b7467f7a ("examples/l3fwd: optimize packet processing on powerpc")
Cc: stable@dpdk.org

Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Acked-by: Shijith Thotton <sthotton@marvell.com>
---
 v6 Changes:
 - Minor optimiazation to process_dst_port NEON.

 v5 Changes:
 - Fix compilation errors.

 v4 Changes:
 - Fix missing `rte_free`.

 v3 Changes:
 - PPC optimize port mask generation.
 - Fix aarch32 compilation.

 v2 Changes:
 - Fix PPC, RISC-V, aarch32 compilation.

 examples/common/altivec/port_group.h | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/examples/common/altivec/port_group.h b/examples/common/altivec/port_group.h
index 5e209b02fa..1c05bc025a 100644
--- a/examples/common/altivec/port_group.h
+++ b/examples/common/altivec/port_group.h
@@ -26,12 +26,17 @@ port_groupx4(uint16_t pn[FWDSTEP + 1], uint16_t *lp,
 		uint16_t u16[FWDSTEP + 1];
 		uint64_t u64;
 	} *pnum = (void *)pn;
-
+	__vector unsigned long long result;
+	const __vector unsigned int perm_mask = {0x00204060, 0x80808080,
+						 0x80808080, 0x80808080};
 	int32_t v;

-	v = vec_any_eq(dp1, dp2);
-
+	dp1 = (__vector unsigned short)vec_cmpeq(dp1, dp2);
+	dp1 = vec_mergeh(dp1, dp1);
+	result = (__vector unsigned long long)vec_vbpermq(
+		(__vector unsigned char)dp1, (__vector unsigned char)perm_mask);

+	v = result[1];
 	/* update last port counter. */
 	lp[0] += gptbl[v].lpv;

--
2.25.1

Re: [PATCH v6 1/5] examples/l3fwd: fix port group mask generation

[Thomas Monjalon]

25/10/2022 18:05, pbhagavatula@marvell.com:
> From: Pavan Nikhilesh <pbhagavatula@marvell.com>
> 
> Fix port group mask generation in altivec, vec_any_eq returns
> 0 or 1 while port_groupx4 expects comparison mask result.
> 
> Fixes: 2193b7467f7a ("examples/l3fwd: optimize packet processing on powerpc")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
> Acked-by: Shijith Thotton <sthotton@marvell.com>

Series applied, thanks.