[RISC-V] Always report call target in range by tomeksowi · Pull Request #119304 · dotnet/runtime

tomeksowi · 2025-09-03T10:22:50Z

auipc + jalr instruction combo has 32-bit range. Further calls are kept absolute by the same condition as x64:

Lines 6866 to 6876 in a6740f4

    
           case IAT_VALUE: 
        
               // IsCallTargetInRange always return true on x64. It wants to use rip-based addressing 
        
               // for this call. Unfortunately, in case of already resolved pinvokes to external libs, 
        
               // which are identified via accessType: IAT_VALUE, the relative offset is unlikely to 
        
               // fit into int32 and we will have to turn fAllowRel32 off globally. To prevent that 
        
               // we'll create a wrapper node and force LSRA to allocate a register so RIP relative 
        
               // isn't used and we don't need to pessimize other callsites. 
        
               if (!comp->IsAot() || !IsCallTargetInRange(addr)) 
        
               { 
        
                   result = AddrGen(addr); 
        
               }

Part of #84834, cc @dotnet/samsung

tomeksowi · 2025-09-03T10:24:26Z

No regressions, more cruft removed.

Diffs are based on 104,367 contexts (2,270 MinOpts, 102,097 FullOpts).

MISSED contexts: 2 (0.00%)

Overall (-3,163,368 bytes)

Collection	Base size (bytes)	Diff size (bytes)	PerfScore in Diffs
linux.riscv64.Checked.3.mch	71,889,656	-3,163,368	-3.25%

MinOpts (-502,760 bytes)

Collection	Base size (bytes)	Diff size (bytes)	PerfScore in Diffs
linux.riscv64.Checked.3.mch	16,685,964	-502,760	-2.34%

FullOpts (-2,660,608 bytes)

Collection	Base size (bytes)	Diff size (bytes)	PerfScore in Diffs
linux.riscv64.Checked.3.mch	55,203,692	-2,660,608	-3.30%

Example diffs

linux.riscv64.Checked.3.mch

-32 (-32.00%) : 695.dasm - TestLibrary.HostPolicyMock:Mock_corehost_resolve_component_dependencies(int,System.String,System.String,System.String):TestLibrary.HostPolicyMock+MockValues_corehost_resolve_component_dependencies (FullOpts)

@@ -26,33 +26,25 @@ G_M6441_IG01:        ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref,
 						;; size=16 bbWeight=1 PerfScore 9.00
 G_M6441_IG02:        ; bbWeight=1, gcrefRegs=3800 {a1 a2 a3}, byrefRegs=0000 {}, byref
             ; gcrRegs +[a1-a3]
-            lui            a4, 0xD1FFAB1E
-            addiw          a4, a4, 0xD1FFAB1E
-            slli           a4, a4, 12
-            addi           a4, a4, 0xD1FFAB1E
-            slli           a4, a4, 3
-            jalr           ra, 0xD1FFAB1E(a4)		// <unknown method>
+            auipc          ra, 0xD1FFAB1E
+            jalr           ra		// <unknown method>
             ; gcrRegs -[a1-a3]
             lui            a0, 0xD1FFAB1E
             addiw          a0, a0, 0xD1FFAB1E
             slli           a0, a0, 12
             addi           a0, a0, 0xD1FFAB1E
             slli           a0, a0, 3
-            lui            a1, 0xD1FFAB1E
-            addiw          a1, a1, 0xD1FFAB1E
-            slli           a1, a1, 12
-            addi           a1, a1, 0xD1FFAB1E
-            slli           a1, a1, 3
-            jalr           ra, 0xD1FFAB1E(a1)		// CORINFO_HELP_NEWSFAST
+            auipc          ra, 0xD1FFAB1E
+            jalr           ra		// CORINFO_HELP_NEWSFAST
             ; gcrRegs +[a0]
-						;; size=68 bbWeight=1 PerfScore 21.00
+						;; size=36 bbWeight=1 PerfScore 13.00
 G_M6441_IG03:        ; bbWeight=1, epilog, nogc, extend
             ld             ra, 8(sp)
             ld             fp, 0(sp)
             addi           sp, sp, 16
             ret						;; size=16 bbWeight=1 PerfScore 7.50
 
-; Total bytes of code 100, prolog size 16, PerfScore 37.50, instruction count 13, allocated bytes for code 100 (MethodHash=7421e6d6) for method TestLibrary.HostPolicyMock:Mock_corehost_resolve_component_dependencies(int,System.String,System.String,System.String):TestLibrary.HostPolicyMock+MockValues_corehost_resolve_component_dependencies (FullOpts)
+; Total bytes of code 68, prolog size 16, PerfScore 29.50, instruction count 11, allocated bytes for code 68 (MethodHash=7421e6d6) for method TestLibrary.HostPolicyMock:Mock_corehost_resolve_component_dependencies(int,System.String,System.String,System.String):TestLibrary.HostPolicyMock+MockValues_corehost_resolve_component_dependencies (FullOpts)
 ; ============================================================
 
 Unwind Info:
@@ -63,7 +55,7 @@ Unwind Info:
   E bit             : 0
   X bit             : 0
   Vers              : 0
-  Function Length   : 50 (0x00032) Actual length = 100 (0x000064)
+  Function Length   : 34 (0x00022) Actual length = 68 (0x000044)
   ---- Epilog scopes ----
   ---- Scope 0
   Epilog Start Offset        : 3523193630 (0xd1ffab1e) Actual offset = 3523193630 (0xd1ffab1e) Offset from main function begin = 3523193630 (0xd1ffab1e)

-16 (-30.77%) : 52577.dasm - JIT.HardwareIntrinsics.Arm.Helpers:RoundToPositiveInfinity(double):double (FullOpts)

@@ -21,21 +21,15 @@ G_M38667_IG01:        ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref,
             sd             ra, 8(sp)
             mv             fp, sp
 						;; size=16 bbWeight=1 PerfScore 9.00
-G_M38667_IG02:        ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
-            lui            a0, 0xD1FFAB1E
-            addiw          a0, a0, 0xD1FFAB1E
-            slli           a0, a0, 12
-            addi           a0, a0, 0xD1FFAB1E
-            slli           a0, a0, 3
-						;; size=20 bbWeight=1 PerfScore 5.00
-G_M38667_IG03:        ; bbWeight=1, epilog, nogc, extend
+G_M38667_IG02:        ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, epilog, nogc
             ld             ra, 8(sp)
             ld             fp, 0(sp)
             addi           sp, sp, 16
-            jr             a0		// <unknown method>
-						;; size=16 bbWeight=1 PerfScore 7.50
+            auipc          t2, 0xD1FFAB1E
+            jr             t2		// <unknown method>
+						;; size=20 bbWeight=1 PerfScore 8.50
 
-; Total bytes of code 52, prolog size 16, PerfScore 21.50, instruction count 9, allocated bytes for code 52 (MethodHash=70aa68f4) for method JIT.HardwareIntrinsics.Arm.Helpers:RoundToPositiveInfinity(double):double (FullOpts)
+; Total bytes of code 36, prolog size 16, PerfScore 17.50, instruction count 8, allocated bytes for code 36 (MethodHash=70aa68f4) for method JIT.HardwareIntrinsics.Arm.Helpers:RoundToPositiveInfinity(double):double (FullOpts)
 ; ============================================================
 
 Unwind Info:
@@ -46,7 +40,7 @@ Unwind Info:
   E bit             : 0
   X bit             : 0
   Vers              : 0
-  Function Length   : 26 (0x0001a) Actual length = 52 (0x000034)
+  Function Length   : 18 (0x00012) Actual length = 36 (0x000024)
   ---- Epilog scopes ----
   ---- Scope 0
   Epilog Start Offset        : 3523193630 (0xd1ffab1e) Actual offset = 3523193630 (0xd1ffab1e) Offset from main function begin = 3523193630 (0xd1ffab1e)

-16 (-30.77%) : 102261.dasm - ComWrappersTests.GlobalInstance.Program+<>c:b__11_0():System.Object:this (FullOpts)

@@ -19,21 +19,15 @@ G_M30715_IG01:        ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref,
             sd             ra, 8(sp)
             mv             fp, sp
 						;; size=16 bbWeight=1 PerfScore 9.00
-G_M30715_IG02:        ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref
-            lui            a0, 0xD1FFAB1E
-            addiw          a0, a0, 0xD1FFAB1E
-            slli           a0, a0, 12
-            addi           a0, a0, 0xD1FFAB1E
-            slli           a0, a0, 3
-						;; size=20 bbWeight=1 PerfScore 5.00
-G_M30715_IG03:        ; bbWeight=1, epilog, nogc, extend
+G_M30715_IG02:        ; bbWeight=1, gcrefRegs=0000 {}, byrefRegs=0000 {}, byref, epilog, nogc
             ld             ra, 8(sp)
             ld             fp, 0(sp)
             addi           sp, sp, 16
-            jr             a0		// <unknown method>
-						;; size=16 bbWeight=1 PerfScore 7.50
+            auipc          t2, 0xD1FFAB1E
+            jr             t2		// <unknown method>
+						;; size=20 bbWeight=1 PerfScore 8.50
 
-; Total bytes of code 52, prolog size 16, PerfScore 21.50, instruction count 9, allocated bytes for code 52 (MethodHash=1a2f8804) for method ComWrappersTests.GlobalInstance.Program+<>c:<ValidatePInvokes>b__11_0():System.Object:this (FullOpts)
+; Total bytes of code 36, prolog size 16, PerfScore 17.50, instruction count 8, allocated bytes for code 36 (MethodHash=1a2f8804) for method ComWrappersTests.GlobalInstance.Program+<>c:<ValidatePInvokes>b__11_0():System.Object:this (FullOpts)
 ; ============================================================
 
 Unwind Info:
@@ -44,7 +38,7 @@ Unwind Info:
   E bit             : 0
   X bit             : 0
   Vers              : 0
-  Function Length   : 26 (0x0001a) Actual length = 52 (0x000034)
+  Function Length   : 18 (0x00012) Actual length = 36 (0x000024)
   ---- Epilog scopes ----
   ---- Scope 0
   Epilog Start Offset        : 3523193630 (0xd1ffab1e) Actual offset = 3523193630 (0xd1ffab1e) Offset from main function begin = 3523193630 (0xd1ffab1e)

+0 (0.00%) : 103392.dasm - Microsoft.Diagnostics.Symbols.SymbolReader:get_SourceCacheDirectory():System.String:this (FullOpts)

No diffs found?

+0 (0.00%) : 103360.dasm - Microsoft.Diagnostics.Symbols.SymbolPath:DefaultSymbolCache(bool):System.String:this (FullOpts)

No diffs found?

+0 (0.00%) : 102096.dasm - ILGEN_0x496bb0b:Main():int (FullOpts)

No diffs found?

Details

Size improvements/regressions per collection

Collection	Contexts with diffs	Improvements	Regressions	Same size	Improvements (bytes)	Regressions (bytes)
linux.riscv64.Checked.3.mch	48,912	47,459	0	1,453	-3,163,368	+0

PerfScore improvements/regressions per collection

Collection	Contexts with diffs	Improvements	Regressions	Same PerfScore	Improvements (PerfScore)	Regressions (PerfScore)	PerfScore Overall in FullOpts
linux.riscv64.Checked.3.mch	48,912	38,001	0	10,911	-4.17%	0.00%	-1.5206%

Context information

Collection	Diffed contexts	MinOpts	FullOpts	Missed, base	Missed, diff
linux.riscv64.Checked.3.mch	104,367	2,270	102,097	2 (0.00%)	2 (0.00%)

jit-analyze output

tomeksowi · 2025-09-03T11:01:30Z

@risc-vv /run

risc-vv · 2025-09-03T11:16:19Z

0bc7710 is being scheduled for building and testing

GIT: 0bc771099d5df4179d1e165b8355e8a3254f5eae
REPO: dotnet/runtime
BRANCH: main

risc-vv · 2025-09-08T08:44:53Z

RISC-V pull_request-CLR-QEMU: 9109 / 9139 (99.67%)

=======================
      passed: 9109
      failed: 2
     skipped: 586
      killed: 28
------------------------
 TOTAL tests: 9725
VIRTUAL time: 80h 39min 22s 181ms
   REAL time: 1h 19min 4s 620ms
=======================

report.xml, report.md, failures.xml, testclr_details.tar.zst

RISC-V pull_request-CLR-VF2: 9109 / 9139 (99.67%)

=======================
      passed: 9109
      failed: 2
     skipped: 586
      killed: 28
------------------------
 TOTAL tests: 9725
VIRTUAL time: 32h 7min 58s 459ms
   REAL time: 2h 7min 40s 139ms
=======================

report.xml, report.md, failures.xml, testclr_details.tar.zst

RISC-V pull_request-FX-QEMU: 0 / 0 (100.00%)

=======================
      passed: 0
      failed: 0
     skipped: 14
      killed: 0
------------------------
 TOTAL tests: 14
VIRTUAL time: 0ms
   REAL time: 4h 30min 0s 551ms
=======================

report.xml, report.md, failures.xml, testclr_details.tar.zst

RISC-V pull_request-FX-VF2: 0 / 71 (0.00%)

=======================
      passed: 0
      failed: 0
     skipped: 14
      killed: 71
------------------------
 TOTAL tests: 85
VIRTUAL time: 1min 26s 303ms
   REAL time: 12h 0min 12s 299ms
=======================

report.xml, report.md, failures.xml, testclr_details.tar.zst

Build information and commands

GIT: 0bc771099d5df4179d1e165b8355e8a3254f5eae
CI: 0dd2da79b21e6d27378407c3af1305973fd300aa
REPO: tomeksowi/runtime
BRANCH: call-target-in-range
CONFIG: Checked
LIB_CONFIG: Release

JulieLeeMSFT · 2025-10-06T19:56:52Z

We will review .NET 11 PRs when we are done with .NET 10 last minute works and less busy. Please give us a few more weeks.
CC @jakobbotsch.

Always report call target in range

0bc7710

github-actions bot added the needs-area-label An area label is needed to ensure this gets routed to the appropriate area owners label Sep 3, 2025

dotnet-policy-service bot added the community-contribution Indicates that the PR has been added by a community member label Sep 3, 2025

clamp03 assigned tomeksowi Sep 3, 2025

clamp03 added the arch-riscv Related to the RISC-V architecture label Sep 3, 2025

tomeksowi marked this pull request as ready for review September 8, 2025 12:53

steveisok added this to the 11.0.0 milestone Oct 6, 2025

jakobbotsch approved these changes Oct 21, 2025

View reviewed changes

jakobbotsch merged commit f34b97a into dotnet:main Oct 21, 2025
115 checks passed

dotnet-maestro bot mentioned this pull request Oct 21, 2025

[main] Source code updates from dotnet/runtime dotnet/dotnet#3024

Merged

github-actions bot locked and limited conversation to collaborators Nov 21, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[RISC-V] Always report call target in range#119304

[RISC-V] Always report call target in range#119304
jakobbotsch merged 1 commit intodotnet:mainfrom
tomeksowi:call-target-in-range

tomeksowi commented Sep 3, 2025

Uh oh!

tomeksowi commented Sep 3, 2025

Size improvements/regressions per collection

PerfScore improvements/regressions per collection

Context information

jit-analyze output

Uh oh!

tomeksowi commented Sep 3, 2025

Uh oh!

risc-vv commented Sep 3, 2025

Uh oh!

risc-vv commented Sep 8, 2025 •

edited

Loading

Uh oh!

JulieLeeMSFT commented Oct 6, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

	case IAT_VALUE:
	// IsCallTargetInRange always return true on x64. It wants to use rip-based addressing
	// for this call. Unfortunately, in case of already resolved pinvokes to external libs,
	// which are identified via accessType: IAT_VALUE, the relative offset is unlikely to
	// fit into int32 and we will have to turn fAllowRel32 off globally. To prevent that
	// we'll create a wrapper node and force LSRA to allocate a register so RIP relative
	// isn't used and we don't need to pessimize other callsites.
	if (!comp->IsAot() \|\| !IsCallTargetInRange(addr))
	{
	result = AddrGen(addr);
	}

Conversation

tomeksowi commented Sep 3, 2025

Uh oh!

tomeksowi commented Sep 3, 2025

Size improvements/regressions per collection

PerfScore improvements/regressions per collection

Context information

jit-analyze output

Uh oh!

tomeksowi commented Sep 3, 2025

Uh oh!

risc-vv commented Sep 3, 2025

Uh oh!

risc-vv commented Sep 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

JulieLeeMSFT commented Oct 6, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

risc-vv commented Sep 8, 2025 •

edited

Loading