The Arm MVE VDUP implementation would like to be able to emit code to
duplicate a byte or halfword value into an i32.  We have code to do
this already in tcg-op-gvec.c, so all we need to do is change the
function from static to global.

For consistency, expose both the _i32 and _i64 forms.

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
I picked tcg.h just because that's where dup_const (defined
next to gen_dup_i32() and gen_dup_i64()) has its prototype.
Happy to move the prototypes elsewhere.
---
 include/tcg/tcg.h | 3 +++
 tcg/tcg-op-gvec.c | 4 ++--
 2 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index 064dab383bc..018f8965f81 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -1331,6 +1331,9 @@ uint64_t dup_const(unsigned vece, uint64_t c);
         : (qemu_build_not_reached_always(), 0))                    \
      : dup_const(VECE, C))
 
+/* Replicate a value of size @vece from @in to all the lanes in @out */
+void gen_dup_i32(unsigned vece, TCGv_i32 out, TCGv_i32 in);
+void gen_dup_i64(unsigned vece, TCGv_i64 out, TCGv_i64 in);
 
 /*
  * Memory helpers that will be used by TCG generated code.
diff --git a/tcg/tcg-op-gvec.c b/tcg/tcg-op-gvec.c
index 498a959839f..abe9e07116d 100644
--- a/tcg/tcg-op-gvec.c
+++ b/tcg/tcg-op-gvec.c
@@ -386,7 +386,7 @@ uint64_t (dup_const)(unsigned vece, uint64_t c)
 }
 
 /* Duplicate IN into OUT as per VECE.  */
-static void gen_dup_i32(unsigned vece, TCGv_i32 out, TCGv_i32 in)
+void gen_dup_i32(unsigned vece, TCGv_i32 out, TCGv_i32 in)
 {
     switch (vece) {
     case MO_8:
@@ -404,7 +404,7 @@ static void gen_dup_i32(unsigned vece, TCGv_i32 out, TCGv_i32 in)
     }
 }
 
-static void gen_dup_i64(unsigned vece, TCGv_i64 out, TCGv_i64 in)
+void gen_dup_i64(unsigned vece, TCGv_i64 out, TCGv_i64 in)
 {
     switch (vece) {
     case MO_8:
-- 
2.20.1

On 6/14/21 8:09 AM, Peter Maydell wrote:
> The Arm MVE VDUP implementation would like to be able to emit code to
> duplicate a byte or halfword value into an i32.  We have code to do
> this already in tcg-op-gvec.c, so all we need to do is change the
> function from static to global.
> 
> For consistency, expose both the _i32 and _i64 forms.
> 
> Suggested-by: Richard Henderson <richard.henderson@linaro.org>
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
> I picked tcg.h just because that's where dup_const (defined
> next to gen_dup_i32() and gen_dup_i64()) has its prototype.
> Happy to move the prototypes elsewhere.

I would definitely rename them to tcg_gen_*.

I would put the declarations in either tcg-op.h, with the other scalar tcg_gen_* 
functions, or at the end of tcg-op-gvec.h, with the other vector-related but still scalar 
functions.

Otherwise,
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>


r~