debug fix for upstream #194

Alexehv77 · 2025-12-15T09:29:55Z

dejagnu fix

The Synopsys RMX-100 has a short, three-stage, in-order execution pipeline. The option -mmpy-option was added to control which version of the MPY unit the core has and what the latency of multiply instructions should be similar to ARCv2 cores (see gcc/config/arc/arc.opt:60). Authored-by: Artemiy Volkov <[email protected]> Signed-off-by: Michiel Derhaeg <[email protected]>

The Synopsys RHX-100 has a 10-stage, dual-issue, in-order execution pipeline. It has support for instruction fusion, which will be addressed by subsequent patches.

…eries

…axt fusion

Signed-off-by: Luis Silva <[email protected]>

luismgsilva

Please use the contrib/check_GNU_style.sh script and fix the syntax according to the codebase.

$ git format-patch -1
0001...
$ bash contribu/check_GNU_style.sh 0001...

luismgsilva · 2025-12-15T09:36:28Z

gcc/config/riscv/arcv.cc

+/*return the next possible fusible insn. */
+
+rtx_insn *
+next_p_fusible_insn (rtx_insn *insn)


Please add the "arcv_" prefix in the function name.

luismgsilva · 2025-12-15T09:36:59Z

gcc/config/riscv/arcv.cc

+	 continue;     
+
+      if(GET_CODE(insn) == BARRIER)
+           continue;


Fix indentation.

luismgsilva · 2025-12-15T09:38:01Z

gcc/config/riscv/arcv.cc

-		  || !NONDEBUG_INSN_P (next_insn (ready[*n_readyp - i]))
-		  || !SCHED_GROUP_P (next_insn (ready[*n_readyp - i])))
+          rtx_insn* next_insn = next_p_fusible_insn (ready[*n_readyp - i]);
+	  /*try to fuse the last_scheduled_insn with */	


If you want to leave the comments these must follow the correct structure.

/* This is a comment. */

luismgsilva · 2025-12-15T09:38:38Z

gcc/config/riscv/arcv.cc

+	  /*try to fuse the last_scheduled_insn with */	
+	  if (NONDEBUG_INSN_P (ready[*n_readyp - i]) /*fuse only with nondebug insn*/
+	      && !SCHED_GROUP_P (ready[*n_readyp - i]) /*which have not been already fused*/
+	      && (!next_insn 


Suggested change

&& (!next_insn

&& (!next_insn || !SCHED_GROUP_P (next_insn))

luismgsilva · 2025-12-15T09:38:57Z

gcc/config/riscv/arcv.cc

-	      && arcv_macro_fusion_pair_p (sched_state.last_scheduled_insn,
-					   ready[*n_readyp - i]))
+	      && active_insn_p (ready[*n_readyp - i])         		      
+              && (!next_insn  


Suggested change

&& (!next_insn

&& (!next_insn || !SCHED_GROUP_P (next_insn))

luismgsilva · 2025-12-15T10:00:04Z

gcc/config/riscv/arcv.cc

+	      && active_insn_p (ready[*n_readyp - i])         		      
+              && (!next_insn  
+                  || !SCHED_GROUP_P (next_insn))
+              && arcv_macro_fusion_pair_p (sched_state.last_scheduled_insn, ready[*n_readyp - i]))


No need to change this line.

luismgsilva · 2025-12-15T10:02:37Z

gcc/config/riscv/arcv.cc

  sched_state.last_scheduled_insn = 0;
 }

+/*return the next possible fusible insn. */


Fix the comment.

luismgsilva · 2025-12-15T10:03:18Z

gcc/config/riscv/arcv.cc

+              && arcv_macro_fusion_pair_p (sched_state.last_scheduled_insn, ready[*n_readyp - i]))
 	    {
-	      if (get_attr_type (ready[*n_readyp - i]) == TYPE_LOAD
+	      if (GET_CODE(PATTERN(ready[*n_readyp - i])) == USE)


luismgsilva · 2025-12-15T10:04:08Z

gcc/config/riscv/arcv.cc

-	    && recog_memoized (next_insn (ready[*n_readyp - i])) >= 0
-	    && get_attr_type (next_insn (ready[*n_readyp - i])) != TYPE_LOAD
-	    && get_attr_type (next_insn (ready[*n_readyp - i])) != TYPE_STORE)))
+             && (!next_insn || !SCHED_GROUP_P (next_insn)))


Align indentation with the above conditions.

luismgsilva · 2025-12-15T10:05:01Z

gcc/config/riscv/arcv.cc

+      if(GET_CODE(insn) == BARRIER)
+           continue;
+
+      if(GET_CODE(PATTERN(insn)) == USE) 


Fix syntax.

MichielDerhaeg · 2025-12-15T09:36:11Z

gcc/config/riscv/arcv.cc


+/*return the next possible fusible insn. */
+
+rtx_insn *


Suggested change

rtx_insn *

static rtx_insn *

MichielDerhaeg · 2025-12-15T09:37:16Z

gcc/config/riscv/arcv.cc

-		  || !SCHED_GROUP_P (next_insn (ready[*n_readyp - i])))
-	      && arcv_macro_fusion_pair_p (sched_state.last_scheduled_insn,
-					   ready[*n_readyp - i]))
+	      && active_insn_p (ready[*n_readyp - i])         		      


Why do we need to check active_insn_p here?

Doesn't that imply NONDEBUG_INSN_P?

MichielDerhaeg · 2025-12-15T09:42:24Z

gcc/config/riscv/arcv.cc

+/*return the next possible fusible insn. */
+
+rtx_insn *
+next_p_fusible_insn (rtx_insn *insn)


Suggested change

next_p_fusible_insn (rtx_insn *insn)

next_fusible_insn (rtx_insn *insn)

MichielDerhaeg · 2025-12-15T10:11:45Z

gcc/config/riscv/arcv.cc

+	      if (GET_CODE(PATTERN(ready[*n_readyp - i])) == USE)
+		  continue;    


Why do we need to check for USE?

luismgsilva · 2025-12-15T17:07:56Z

FYI: You need to rebase your patch on top of the branch again because the patches on michiel/fusion-trunk-3 have changed.

artemiy-volkov and others added 9 commits November 28, 2025 21:03

RISC-V: Add Synopsys RHX-100 series pipeline description

4c5886d

The Synopsys RHX-100 has a 10-stage, dual-issue, in-order execution pipeline. It has support for instruction fusion, which will be addressed by subsequent patches.

RISC-V: Implement riscv_macro_fusion_pair_p for Synopsys RHX-100 series

4364c6a

RISC-V: Implement TARGET_SCHED_FUSION_PRIORITY for Synopsys RHX-100 s…

4d1d83a

…eries

RISC-V: Implement scheduling for Synopsys RHX-100 series

47ccdf4

RISC-V: Add instruction patterns for 32-bit multiply-add and bit-extr…

86cc974

…axt fusion

fixup! RISC-V: Implement scheduling for Synopsys RHX-100 series

73219a4

Signed-off-by: Luis Silva <[email protected]>

tmp: arcv: protos.

d479345

Signed-off-by: Luis Silva <[email protected]>

debug fix for upstream

a975850

Alexehv77 requested review from MichielDerhaeg and luismgsilva December 15, 2025 09:29

Alexehv77 self-assigned this Dec 15, 2025

luismgsilva requested changes Dec 15, 2025

View reviewed changes

MichielDerhaeg reviewed Dec 15, 2025

View reviewed changes

luismgsilva force-pushed the michiel/fusion-trunk-3 branch 4 times, most recently from 39e1a5e to b64f328 Compare December 15, 2025 16:58

luismgsilva force-pushed the michiel/fusion-trunk-3 branch 5 times, most recently from a33ab39 to 1f24597 Compare December 16, 2025 15:31

	next_p_fusible_insn (rtx_insn *insn)
	next_fusible_insn (rtx_insn *insn)

		if (GET_CODE(PATTERN(ready[*n_readyp - i])) == USE)
		continue;

debug fix for upstream #194

Are you sure you want to change the base?

debug fix for upstream #194

Uh oh!

Conversation

Alexehv77 commented Dec 15, 2025

Uh oh!

luismgsilva left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

luismgsilva commented Dec 15, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants