[Qemu-devel] [PULL] tilegx queued patch

[Qemu-devel] [PULL] tilegx queued patch

[Richard Henderson]

Just one mergable patch since the last update.
I guess this will be it before hard-freeze.


r~


The following changes since commit b803894e2c4d744ccc113ca6cbe6654ec80c1dc6:

  Merge remote-tracking branch 'remotes/afaerber/tags/qom-cpu-for-peter' into staging (2015-10-22 17:33:54 +0100)

are available in the git repository at:

  git://github.com/rth7680/qemu.git tags/pull-tile-20151030

for you to fetch changes up to 2a080ce26682f35517b0e20f4ad10559e9270b5d:

  target-tilegx: Implement prefetch instructions in pipe y2 (2015-10-22 07:51:49 -1000)

----------------------------------------------------------------
Prefetch in y2 pipe

----------------------------------------------------------------
Chen Gang (1):
      target-tilegx: Implement prefetch instructions in pipe y2

 target-tilegx/translate.c | 22 ++++++++++++++--------
 1 file changed, 14 insertions(+), 8 deletions(-)

[Qemu-devel] [PULL] target-tilegx: Implement prefetch instructions in pipe y2

[Richard Henderson]

From: Chen Gang <gang.chen.5i5j@gmail.com>

Originally, tilegx qemu only implement prefetch instructions in pipe x1,
did not implement them in pipe y2.

Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
Signed-off-by: Richard Henderson <rth@twiddle.net>
---
 target-tilegx/translate.c | 22 ++++++++++++++--------
 1 file changed, 14 insertions(+), 8 deletions(-)

diff --git a/target-tilegx/translate.c b/target-tilegx/translate.c
index 34d45f8..354f25a 100644
--- a/target-tilegx/translate.c
+++ b/target-tilegx/translate.c
@@ -2105,38 +2105,44 @@ static TileExcp decode_y2(DisasContext *dc, tilegx_bundle_bits bundle)
     unsigned srcbdest = get_SrcBDest_Y2(bundle);
     const char *mnemonic;
     TCGMemOp memop;
+    bool prefetch_nofault = false;
 
     switch (OEY2(opc, mode)) {
     case OEY2(LD1S_OPCODE_Y2, MODE_OPCODE_YA2):
         memop = MO_SB;
-        mnemonic = "ld1s";
+        mnemonic = "ld1s"; /* prefetch_l1_fault */
         goto do_load;
     case OEY2(LD1U_OPCODE_Y2, MODE_OPCODE_YA2):
         memop = MO_UB;
-        mnemonic = "ld1u";
+        mnemonic = "ld1u"; /* prefetch, prefetch_l1 */
+        prefetch_nofault = (srcbdest == TILEGX_R_ZERO);
         goto do_load;
     case OEY2(LD2S_OPCODE_Y2, MODE_OPCODE_YA2):
         memop = MO_TESW;
-        mnemonic = "ld2s";
+        mnemonic = "ld2s"; /* prefetch_l2_fault */
         goto do_load;
     case OEY2(LD2U_OPCODE_Y2, MODE_OPCODE_YA2):
         memop = MO_TEUW;
-        mnemonic = "ld2u";
+        mnemonic = "ld2u"; /* prefetch_l2 */
+        prefetch_nofault = (srcbdest == TILEGX_R_ZERO);
         goto do_load;
     case OEY2(LD4S_OPCODE_Y2, MODE_OPCODE_YB2):
         memop = MO_TESL;
-        mnemonic = "ld4s";
+        mnemonic = "ld4s"; /* prefetch_l3_fault */
         goto do_load;
     case OEY2(LD4U_OPCODE_Y2, MODE_OPCODE_YB2):
         memop = MO_TEUL;
-        mnemonic = "ld4u";
+        mnemonic = "ld4u"; /* prefetch_l3 */
+        prefetch_nofault = (srcbdest == TILEGX_R_ZERO);
         goto do_load;
     case OEY2(LD_OPCODE_Y2, MODE_OPCODE_YB2):
         memop = MO_TEQ;
         mnemonic = "ld";
     do_load:
-        tcg_gen_qemu_ld_tl(dest_gr(dc, srcbdest), load_gr(dc, srca),
-                           dc->mmuidx, memop);
+        if (!prefetch_nofault) {
+            tcg_gen_qemu_ld_tl(dest_gr(dc, srcbdest), load_gr(dc, srca),
+                               dc->mmuidx, memop);
+        }
         qemu_log_mask(CPU_LOG_TB_IN_ASM, "%s %s, %s", mnemonic,
                       reg_names[srcbdest], reg_names[srca]);
         return TILEGX_EXCP_NONE;
-- 
2.4.3

Re: [Qemu-devel] [PULL] tilegx queued patch

[Chen Gang]

On 10/31/15 04:44, Richard Henderson wrote:
> Just one mergable patch since the last update.
> I guess this will be it before hard-freeze.
>

Thank you for your work.

I have finished coding for the implementation of tilegx floating point
instructions, at present, I am just testing and fixing the related
issues.

Hope I can finish today and send patches for it (although I am not quite
sure).


Thanks.
--
Chen Gang (陈刚)

Open, share, and attitude like air, water, and life which God blessed
 		 	   		  

Re: [Qemu-devel] [PULL] tilegx queued patch

[Peter Maydell]

On 30 October 2015 at 20:44, Richard Henderson <rth@twiddle.net> wrote:
> Just one mergable patch since the last update.
> I guess this will be it before hard-freeze.
>
>
> r~
>
>
> The following changes since commit b803894e2c4d744ccc113ca6cbe6654ec80c1dc6:
>
>   Merge remote-tracking branch 'remotes/afaerber/tags/qom-cpu-for-peter' into staging (2015-10-22 17:33:54 +0100)
>
> are available in the git repository at:
>
>   git://github.com/rth7680/qemu.git tags/pull-tile-20151030
>
> for you to fetch changes up to 2a080ce26682f35517b0e20f4ad10559e9270b5d:
>
>   target-tilegx: Implement prefetch instructions in pipe y2 (2015-10-22 07:51:49 -1000)
>
> ----------------------------------------------------------------
> Prefetch in y2 pipe
>
> ----------------------------------------------------------------
> Chen Gang (1):
>       target-tilegx: Implement prefetch instructions in pipe y2
>

Applied, thanks.

-- PMM