This operation has float16 inputs and thus must use
the FZ16 control not the FZ control.

Cc: qemu-stable@nongnu.org
Reported-by: Daniyal Khan <danikhan632@gmail.com>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2374
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/tcg/translate-sme.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/target/arm/tcg/translate-sme.c b/target/arm/tcg/translate-sme.c
index 46c7fce8b4..185a8a917b 100644
--- a/target/arm/tcg/translate-sme.c
+++ b/target/arm/tcg/translate-sme.c
@@ -304,6 +304,7 @@ static bool do_outprod(DisasContext *s, arg_op *a, MemOp esz,
 }
 
 static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz,
+                            ARMFPStatusFlavour e_fpst,
                             gen_helper_gvec_5_ptr *fn)
 {
     int svl = streaming_vec_reg_size(s);
@@ -319,15 +320,18 @@ static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz,
     zm = vec_full_reg_ptr(s, a->zm);
     pn = pred_full_reg_ptr(s, a->pn);
     pm = pred_full_reg_ptr(s, a->pm);
-    fpst = fpstatus_ptr(FPST_FPCR);
+    fpst = fpstatus_ptr(e_fpst);
 
     fn(za, zn, zm, pn, pm, fpst, tcg_constant_i32(desc));
     return true;
 }
 
-TRANS_FEAT(FMOPA_h, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_h)
-TRANS_FEAT(FMOPA_s, aa64_sme, do_outprod_fpst, a, MO_32, gen_helper_sme_fmopa_s)
-TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64, gen_helper_sme_fmopa_d)
+TRANS_FEAT(FMOPA_h, aa64_sme, do_outprod_fpst, a,
+           MO_32, FPST_FPCR_F16, gen_helper_sme_fmopa_h)
+TRANS_FEAT(FMOPA_s, aa64_sme, do_outprod_fpst, a,
+           MO_32, FPST_FPCR, gen_helper_sme_fmopa_s)
+TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a,
+           MO_64, FPST_FPCR, gen_helper_sme_fmopa_d)
 
 /* TODO: FEAT_EBF16 */
 TRANS_FEAT(BFMOPA, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_bfmopa)
-- 
2.43.0

Richard Henderson <richard.henderson@linaro.org> writes:

> This operation has float16 inputs and thus must use
> the FZ16 control not the FZ control.
>
> Cc: qemu-stable@nongnu.org
> Reported-by: Daniyal Khan <danikhan632@gmail.com>
> Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2374
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

-- 
Alex Bennée
Virtualisation Tech Lead @ Linaro

On 7/15/24 22:58, Richard Henderson wrote:
> This operation has float16 inputs and thus must use
> the FZ16 control not the FZ control.
> 
> Cc: qemu-stable@nongnu.org
> Reported-by: Daniyal Khan <danikhan632@gmail.com>
> Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2374
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
>   target/arm/tcg/translate-sme.c | 12 ++++++++----
>   1 file changed, 8 insertions(+), 4 deletions(-)

Fixes: 3916841ac75 ("target/arm: Implement FMOPA, FMOPS (widening)")


r~