Message ID | 20191018161044.6983-7-david@redhat.com |
---|---|
State | New |
Headers | show |
Series | s390x/tcg: Vector instruction fixes | expand |
On 10/18/19 9:10 AM, David Hildenbrand wrote: > + /* Isolate the carry to the next doubleword */ > tcg_gen_andi_i64(dl, th, 1); You can remove this now, since the only possible results are 0/1; it was only our subtract implementation that produced -1/0. r~
On 18.10.19 20:55, Richard Henderson wrote: > On 10/18/19 9:10 AM, David Hildenbrand wrote: >> + /* Isolate the carry to the next doubleword */ >> tcg_gen_andi_i64(dl, th, 1); > > You can remove this now, since the only possible results are 0/1; it was only > our subtract implementation that produced -1/0. > > > r~ > Right, we can simply reuse the VACCC implementation now: diff --git a/target/s390x/translate_vx.inc.c b/target/s390x/translate_vx.inc.c index 87b5790db4..49f9916c37 100644 --- a/target/s390x/translate_vx.inc.c +++ b/target/s390x/translate_vx.inc.c @@ -2240,20 +2240,13 @@ static void gen_sbcbi2_i64(TCGv_i64 dl, TCGv_i64 dh, TCGv_i64 al, TCGv_i64 ah, { TCGv_i64 th = tcg_temp_new_i64(); TCGv_i64 tl = tcg_temp_new_i64(); - TCGv_i64 zero = tcg_const_i64(0); - tcg_gen_andi_i64(tl, cl, 1); - tcg_gen_sub2_i64(tl, th, al, zero, tl, zero); - tcg_gen_sub2_i64(tl, th, tl, th, bl, zero); - tcg_gen_andi_i64(th, th, 1); - tcg_gen_sub2_i64(tl, th, ah, zero, th, zero); - tcg_gen_sub2_i64(tl, th, tl, th, bh, zero); - tcg_gen_andi_i64(dl, th, 1); - tcg_gen_mov_i64(dh, zero); + tcg_gen_not_i64(tl, bl); + tcg_gen_not_i64(th, bh); + gen_accc2_i64(dl, dh, al, ah, tl, th, cl, ch); tcg_temp_free_i64(tl); tcg_temp_free_i64(th); - tcg_temp_free_i64(zero); } This works as we only have to compute the bitwise complement.
diff --git a/target/s390x/translate_vx.inc.c b/target/s390x/translate_vx.inc.c index 87b5790db4..2015af9012 100644 --- a/target/s390x/translate_vx.inc.c +++ b/target/s390x/translate_vx.inc.c @@ -2240,17 +2240,30 @@ static void gen_sbcbi2_i64(TCGv_i64 dl, TCGv_i64 dh, TCGv_i64 al, TCGv_i64 ah, { TCGv_i64 th = tcg_temp_new_i64(); TCGv_i64 tl = tcg_temp_new_i64(); + TCGv_i64 sh = tcg_temp_new_i64(); + TCGv_i64 sl = tcg_temp_new_i64(); TCGv_i64 zero = tcg_const_i64(0); tcg_gen_andi_i64(tl, cl, 1); - tcg_gen_sub2_i64(tl, th, al, zero, tl, zero); - tcg_gen_sub2_i64(tl, th, tl, th, bl, zero); + tcg_gen_not_i64(sl, bl); + tcg_gen_not_i64(sh, bh); + + /* Add the borrow to the low doubleword of a */ + tcg_gen_add2_i64(tl, th, al, zero, tl, zero); + /* Add the bit-wise complement of b to the low doubleword */ + tcg_gen_add2_i64(tl, th, tl, th, sl, zero); + /* Isolate the carry to the high doubleword */ tcg_gen_andi_i64(th, th, 1); - tcg_gen_sub2_i64(tl, th, ah, zero, th, zero); - tcg_gen_sub2_i64(tl, th, tl, th, bh, zero); + /* Add the carry to the high doubleword of a */ + tcg_gen_add2_i64(tl, th, ah, zero, th, zero); + /* Add the bit-wise complement of b to the high doubleword */ + tcg_gen_add2_i64(tl, th, tl, th, sh, zero); + /* Isolate the carry to the next doubleword */ tcg_gen_andi_i64(dl, th, 1); tcg_gen_mov_i64(dh, zero); + tcg_temp_free_i64(sl); + tcg_temp_free_i64(sh); tcg_temp_free_i64(tl); tcg_temp_free_i64(th); tcg_temp_free_i64(zero);
The numbers are unsigned, the computation is wrong. "Each operand is treated as an unsigned binary integer". Let's implement as given in the PoP: "A subtraction is performed by adding the contents of the second operand with the bitwise complement of the third operand along with a borrow indication from the rightmost bit of the fourth operand." Fixes: bc725e65152c ("s390x/tcg: Implement VECTOR SUBTRACT WITH BORROW COMPUTE BORROW INDICATION") Signed-off-by: David Hildenbrand <david@redhat.com> --- target/s390x/translate_vx.inc.c | 21 +++++++++++++++++---- 1 file changed, 17 insertions(+), 4 deletions(-)