[LoongArch] Fix GOT usage for non-dso_local function calls in large code model
#117099
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Created using spr 1.3.5-bogner
|
@llvm/pr-subscribers-backend-loongarch Author: wanglei (wangleiat) ChangesThis commit fixes an issue in the large code model where non-dso_local Full diff: https://github.com/llvm/llvm-project/pull/117099.diff 6 Files Affected:
diff --git a/llvm/lib/Target/LoongArch/LoongArchExpandPseudoInsts.cpp b/llvm/lib/Target/LoongArch/LoongArchExpandPseudoInsts.cpp
index b7b8987e4084b6..30742c79653b52 100644
--- a/llvm/lib/Target/LoongArch/LoongArchExpandPseudoInsts.cpp
+++ b/llvm/lib/Target/LoongArch/LoongArchExpandPseudoInsts.cpp
@@ -564,7 +564,7 @@ bool LoongArchPreRAExpandPseudo::expandFunctionCALL(
? MF->getRegInfo().createVirtualRegister(&LoongArch::GPRRegClass)
: LoongArch::R1;
- bool UseGOT = Func.isGlobal() && !Func.getGlobal()->isDSOLocal();
+ bool UseGOT = Func.getTargetFlags() == LoongArchII::MO_CALL_PLT;
unsigned MO = UseGOT ? LoongArchII::MO_GOT_PC_HI : LoongArchII::MO_PCREL_LO;
unsigned LAOpcode = UseGOT ? LoongArch::LDX_D : LoongArch::ADD_D;
expandLargeAddressLoad(MBB, MBBI, NextMBBI, LAOpcode, MO, Func, AddrReg,
diff --git a/llvm/test/CodeGen/LoongArch/code-models.ll b/llvm/test/CodeGen/LoongArch/code-models.ll
index 14bd0f4df47109..c0120688623343 100644
--- a/llvm/test/CodeGen/LoongArch/code-models.ll
+++ b/llvm/test/CodeGen/LoongArch/code-models.ll
@@ -82,11 +82,11 @@ define void @call_external_sym(ptr %dst) {
; LARGE-NEXT: .cfi_offset 1, -8
; LARGE-NEXT: ori $a2, $zero, 1000
; LARGE-NEXT: move $a1, $zero
-; LARGE-NEXT: pcalau12i $a3, %pc_hi20(memset)
-; LARGE-NEXT: addi.d $ra, $zero, %pc_lo12(memset)
-; LARGE-NEXT: lu32i.d $ra, %pc64_lo20(memset)
-; LARGE-NEXT: lu52i.d $ra, $ra, %pc64_hi12(memset)
-; LARGE-NEXT: add.d $ra, $ra, $a3
+; LARGE-NEXT: pcalau12i $a3, %got_pc_hi20(memset)
+; LARGE-NEXT: addi.d $ra, $zero, %got_pc_lo12(memset)
+; LARGE-NEXT: lu32i.d $ra, %got64_pc_lo20(memset)
+; LARGE-NEXT: lu52i.d $ra, $ra, %got64_pc_hi12(memset)
+; LARGE-NEXT: ldx.d $ra, $ra, $a3
; LARGE-NEXT: jirl $ra, $ra, 0
; LARGE-NEXT: ld.d $ra, $sp, 8 # 8-byte Folded Reload
; LARGE-NEXT: addi.d $sp, $sp, 16
diff --git a/llvm/test/CodeGen/LoongArch/expand-adjacency.ll b/llvm/test/CodeGen/LoongArch/expand-adjacency.ll
index 154d2121a6321e..b00cf2c519a151 100644
--- a/llvm/test/CodeGen/LoongArch/expand-adjacency.ll
+++ b/llvm/test/CodeGen/LoongArch/expand-adjacency.ll
@@ -14,10 +14,10 @@ declare void @llvm.memset.p0.i64(ptr, i8, i64, i1)
define void @call_external_sym(ptr %dst) {
; LARGE-LABEL: call_external_sym:
-; LARGE: pcalau12i [[REG1:\$[a-z0-9]+]], %pc_hi20(memset)
-; LARGE-NEXT: addi.d [[REG2:\$[a-z0-9]+]], $zero, %pc_lo12(memset)
-; LARGE-NEXT: lu32i.d [[REG2]], %pc64_lo20(memset)
-; LARGE-NEXT: lu52i.d [[REG2]], [[REG2]], %pc64_hi12(memset)
+; LARGE: pcalau12i [[REG1:\$[a-z0-9]+]], %got_pc_hi20(memset)
+; LARGE-NEXT: addi.d [[REG2:\$[a-z0-9]+]], $zero, %got_pc_lo12(memset)
+; LARGE-NEXT: lu32i.d [[REG2]], %got64_pc_lo20(memset)
+; LARGE-NEXT: lu52i.d [[REG2]], [[REG2]], %got64_pc_hi12(memset)
entry:
call void @llvm.memset.p0.i64(ptr %dst, i8 0, i64 1000, i1 false)
ret void
diff --git a/llvm/test/CodeGen/LoongArch/machinelicm-address-pseudos.ll b/llvm/test/CodeGen/LoongArch/machinelicm-address-pseudos.ll
index ba72ef5bd7ba4b..fc0c7ad1686ee8 100644
--- a/llvm/test/CodeGen/LoongArch/machinelicm-address-pseudos.ll
+++ b/llvm/test/CodeGen/LoongArch/machinelicm-address-pseudos.ll
@@ -279,11 +279,11 @@ define void @test_la_tls_ld(i32 signext %n) {
; LA64LARGE-NEXT: .LBB3_1: # %loop
; LA64LARGE-NEXT: # =>This Inner Loop Header: Depth=1
; LA64LARGE-NEXT: move $a0, $s0
-; LA64LARGE-NEXT: pcalau12i $a1, %pc_hi20(__tls_get_addr)
-; LA64LARGE-NEXT: addi.d $ra, $zero, %pc_lo12(__tls_get_addr)
-; LA64LARGE-NEXT: lu32i.d $ra, %pc64_lo20(__tls_get_addr)
-; LA64LARGE-NEXT: lu52i.d $ra, $ra, %pc64_hi12(__tls_get_addr)
-; LA64LARGE-NEXT: add.d $ra, $ra, $a1
+; LA64LARGE-NEXT: pcalau12i $a1, %got_pc_hi20(__tls_get_addr)
+; LA64LARGE-NEXT: addi.d $ra, $zero, %got_pc_lo12(__tls_get_addr)
+; LA64LARGE-NEXT: lu32i.d $ra, %got64_pc_lo20(__tls_get_addr)
+; LA64LARGE-NEXT: lu52i.d $ra, $ra, %got64_pc_hi12(__tls_get_addr)
+; LA64LARGE-NEXT: ldx.d $ra, $ra, $a1
; LA64LARGE-NEXT: jirl $ra, $ra, 0
; LA64LARGE-NEXT: ld.w $zero, $a0, 0
; LA64LARGE-NEXT: addi.w $s1, $s1, 1
@@ -445,11 +445,11 @@ define void @test_la_tls_gd(i32 signext %n) nounwind {
; LA64LARGE-NEXT: .LBB5_1: # %loop
; LA64LARGE-NEXT: # =>This Inner Loop Header: Depth=1
; LA64LARGE-NEXT: move $a0, $s0
-; LA64LARGE-NEXT: pcalau12i $a1, %pc_hi20(__tls_get_addr)
-; LA64LARGE-NEXT: addi.d $ra, $zero, %pc_lo12(__tls_get_addr)
-; LA64LARGE-NEXT: lu32i.d $ra, %pc64_lo20(__tls_get_addr)
-; LA64LARGE-NEXT: lu52i.d $ra, $ra, %pc64_hi12(__tls_get_addr)
-; LA64LARGE-NEXT: add.d $ra, $ra, $a1
+; LA64LARGE-NEXT: pcalau12i $a1, %got_pc_hi20(__tls_get_addr)
+; LA64LARGE-NEXT: addi.d $ra, $zero, %got_pc_lo12(__tls_get_addr)
+; LA64LARGE-NEXT: lu32i.d $ra, %got64_pc_lo20(__tls_get_addr)
+; LA64LARGE-NEXT: lu52i.d $ra, $ra, %got64_pc_hi12(__tls_get_addr)
+; LA64LARGE-NEXT: ldx.d $ra, $ra, $a1
; LA64LARGE-NEXT: jirl $ra, $ra, 0
; LA64LARGE-NEXT: ld.w $zero, $a0, 0
; LA64LARGE-NEXT: addi.w $s1, $s1, 1
diff --git a/llvm/test/CodeGen/LoongArch/psabi-restricted-scheduling.ll b/llvm/test/CodeGen/LoongArch/psabi-restricted-scheduling.ll
index a7873f466bee3f..c7de3dcf2ecfd2 100644
--- a/llvm/test/CodeGen/LoongArch/psabi-restricted-scheduling.ll
+++ b/llvm/test/CodeGen/LoongArch/psabi-restricted-scheduling.ll
@@ -102,11 +102,11 @@ define void @foo() nounwind {
; LARGE_NO_SCH-NEXT: lu32i.d $a1, %got64_pc_lo20(gd)
; LARGE_NO_SCH-NEXT: lu52i.d $a1, $a1, %got64_pc_hi12(gd)
; LARGE_NO_SCH-NEXT: add.d $a0, $a1, $a0
-; LARGE_NO_SCH-NEXT: pcalau12i $a1, %pc_hi20(__tls_get_addr)
-; LARGE_NO_SCH-NEXT: addi.d $ra, $zero, %pc_lo12(__tls_get_addr)
-; LARGE_NO_SCH-NEXT: lu32i.d $ra, %pc64_lo20(__tls_get_addr)
-; LARGE_NO_SCH-NEXT: lu52i.d $ra, $ra, %pc64_hi12(__tls_get_addr)
-; LARGE_NO_SCH-NEXT: add.d $ra, $ra, $a1
+; LARGE_NO_SCH-NEXT: pcalau12i $a1, %got_pc_hi20(__tls_get_addr)
+; LARGE_NO_SCH-NEXT: addi.d $ra, $zero, %got_pc_lo12(__tls_get_addr)
+; LARGE_NO_SCH-NEXT: lu32i.d $ra, %got64_pc_lo20(__tls_get_addr)
+; LARGE_NO_SCH-NEXT: lu52i.d $ra, $ra, %got64_pc_hi12(__tls_get_addr)
+; LARGE_NO_SCH-NEXT: ldx.d $ra, $ra, $a1
; LARGE_NO_SCH-NEXT: jirl $ra, $ra, 0
; LARGE_NO_SCH-NEXT: ld.d $zero, $a0, 0
; LARGE_NO_SCH-NEXT: pcalau12i $a0, %ld_pc_hi20(ld)
@@ -114,11 +114,11 @@ define void @foo() nounwind {
; LARGE_NO_SCH-NEXT: lu32i.d $a1, %got64_pc_lo20(ld)
; LARGE_NO_SCH-NEXT: lu52i.d $a1, $a1, %got64_pc_hi12(ld)
; LARGE_NO_SCH-NEXT: add.d $a0, $a1, $a0
-; LARGE_NO_SCH-NEXT: pcalau12i $a1, %pc_hi20(__tls_get_addr)
-; LARGE_NO_SCH-NEXT: addi.d $ra, $zero, %pc_lo12(__tls_get_addr)
-; LARGE_NO_SCH-NEXT: lu32i.d $ra, %pc64_lo20(__tls_get_addr)
-; LARGE_NO_SCH-NEXT: lu52i.d $ra, $ra, %pc64_hi12(__tls_get_addr)
-; LARGE_NO_SCH-NEXT: add.d $ra, $ra, $a1
+; LARGE_NO_SCH-NEXT: pcalau12i $a1, %got_pc_hi20(__tls_get_addr)
+; LARGE_NO_SCH-NEXT: addi.d $ra, $zero, %got_pc_lo12(__tls_get_addr)
+; LARGE_NO_SCH-NEXT: lu32i.d $ra, %got64_pc_lo20(__tls_get_addr)
+; LARGE_NO_SCH-NEXT: lu52i.d $ra, $ra, %got64_pc_hi12(__tls_get_addr)
+; LARGE_NO_SCH-NEXT: ldx.d $ra, $ra, $a1
; LARGE_NO_SCH-NEXT: jirl $ra, $ra, 0
; LARGE_NO_SCH-NEXT: ld.d $zero, $a0, 0
; LARGE_NO_SCH-NEXT: pcalau12i $a0, %ie_pc_hi20(ie)
@@ -158,11 +158,11 @@ define void @foo() nounwind {
; LARGE_SCH-NEXT: lu32i.d $a1, %got64_pc_lo20(gd)
; LARGE_SCH-NEXT: lu52i.d $a1, $a1, %got64_pc_hi12(gd)
; LARGE_SCH-NEXT: add.d $a0, $a1, $a0
-; LARGE_SCH-NEXT: pcalau12i $a1, %pc_hi20(__tls_get_addr)
-; LARGE_SCH-NEXT: addi.d $ra, $zero, %pc_lo12(__tls_get_addr)
-; LARGE_SCH-NEXT: lu32i.d $ra, %pc64_lo20(__tls_get_addr)
-; LARGE_SCH-NEXT: lu52i.d $ra, $ra, %pc64_hi12(__tls_get_addr)
-; LARGE_SCH-NEXT: add.d $ra, $ra, $a1
+; LARGE_SCH-NEXT: pcalau12i $a1, %got_pc_hi20(__tls_get_addr)
+; LARGE_SCH-NEXT: addi.d $ra, $zero, %got_pc_lo12(__tls_get_addr)
+; LARGE_SCH-NEXT: lu32i.d $ra, %got64_pc_lo20(__tls_get_addr)
+; LARGE_SCH-NEXT: lu52i.d $ra, $ra, %got64_pc_hi12(__tls_get_addr)
+; LARGE_SCH-NEXT: ldx.d $ra, $ra, $a1
; LARGE_SCH-NEXT: jirl $ra, $ra, 0
; LARGE_SCH-NEXT: ld.d $zero, $a0, 0
; LARGE_SCH-NEXT: pcalau12i $a0, %ld_pc_hi20(ld)
@@ -170,11 +170,11 @@ define void @foo() nounwind {
; LARGE_SCH-NEXT: lu32i.d $a1, %got64_pc_lo20(ld)
; LARGE_SCH-NEXT: lu52i.d $a1, $a1, %got64_pc_hi12(ld)
; LARGE_SCH-NEXT: add.d $a0, $a1, $a0
-; LARGE_SCH-NEXT: pcalau12i $a1, %pc_hi20(__tls_get_addr)
-; LARGE_SCH-NEXT: addi.d $ra, $zero, %pc_lo12(__tls_get_addr)
-; LARGE_SCH-NEXT: lu32i.d $ra, %pc64_lo20(__tls_get_addr)
-; LARGE_SCH-NEXT: lu52i.d $ra, $ra, %pc64_hi12(__tls_get_addr)
-; LARGE_SCH-NEXT: add.d $ra, $ra, $a1
+; LARGE_SCH-NEXT: pcalau12i $a1, %got_pc_hi20(__tls_get_addr)
+; LARGE_SCH-NEXT: addi.d $ra, $zero, %got_pc_lo12(__tls_get_addr)
+; LARGE_SCH-NEXT: lu32i.d $ra, %got64_pc_lo20(__tls_get_addr)
+; LARGE_SCH-NEXT: lu52i.d $ra, $ra, %got64_pc_hi12(__tls_get_addr)
+; LARGE_SCH-NEXT: ldx.d $ra, $ra, $a1
; LARGE_SCH-NEXT: jirl $ra, $ra, 0
; LARGE_SCH-NEXT: ld.d $zero, $a0, 0
; LARGE_SCH-NEXT: pcalau12i $a0, %ie_pc_hi20(ie)
diff --git a/llvm/test/CodeGen/LoongArch/tls-models.ll b/llvm/test/CodeGen/LoongArch/tls-models.ll
index 4ac6201fdd9d4c..dbd7bf6a81269c 100644
--- a/llvm/test/CodeGen/LoongArch/tls-models.ll
+++ b/llvm/test/CodeGen/LoongArch/tls-models.ll
@@ -55,11 +55,11 @@ define ptr @f1() nounwind {
; LA64LARGEPIC-NEXT: lu32i.d $a1, %got64_pc_lo20(unspecified)
; LA64LARGEPIC-NEXT: lu52i.d $a1, $a1, %got64_pc_hi12(unspecified)
; LA64LARGEPIC-NEXT: add.d $a0, $a1, $a0
-; LA64LARGEPIC-NEXT: pcalau12i $a1, %pc_hi20(__tls_get_addr)
-; LA64LARGEPIC-NEXT: addi.d $ra, $zero, %pc_lo12(__tls_get_addr)
-; LA64LARGEPIC-NEXT: lu32i.d $ra, %pc64_lo20(__tls_get_addr)
-; LA64LARGEPIC-NEXT: lu52i.d $ra, $ra, %pc64_hi12(__tls_get_addr)
-; LA64LARGEPIC-NEXT: add.d $ra, $ra, $a1
+; LA64LARGEPIC-NEXT: pcalau12i $a1, %got_pc_hi20(__tls_get_addr)
+; LA64LARGEPIC-NEXT: addi.d $ra, $zero, %got_pc_lo12(__tls_get_addr)
+; LA64LARGEPIC-NEXT: lu32i.d $ra, %got64_pc_lo20(__tls_get_addr)
+; LA64LARGEPIC-NEXT: lu52i.d $ra, $ra, %got64_pc_hi12(__tls_get_addr)
+; LA64LARGEPIC-NEXT: ldx.d $ra, $ra, $a1
; LA64LARGEPIC-NEXT: jirl $ra, $ra, 0
; LA64LARGEPIC-NEXT: ld.d $ra, $sp, 8 # 8-byte Folded Reload
; LA64LARGEPIC-NEXT: addi.d $sp, $sp, 16
@@ -168,11 +168,11 @@ define ptr @f2() nounwind {
; LA64LARGEPIC-NEXT: lu32i.d $a1, %got64_pc_lo20(ld)
; LA64LARGEPIC-NEXT: lu52i.d $a1, $a1, %got64_pc_hi12(ld)
; LA64LARGEPIC-NEXT: add.d $a0, $a1, $a0
-; LA64LARGEPIC-NEXT: pcalau12i $a1, %pc_hi20(__tls_get_addr)
-; LA64LARGEPIC-NEXT: addi.d $ra, $zero, %pc_lo12(__tls_get_addr)
-; LA64LARGEPIC-NEXT: lu32i.d $ra, %pc64_lo20(__tls_get_addr)
-; LA64LARGEPIC-NEXT: lu52i.d $ra, $ra, %pc64_hi12(__tls_get_addr)
-; LA64LARGEPIC-NEXT: add.d $ra, $ra, $a1
+; LA64LARGEPIC-NEXT: pcalau12i $a1, %got_pc_hi20(__tls_get_addr)
+; LA64LARGEPIC-NEXT: addi.d $ra, $zero, %got_pc_lo12(__tls_get_addr)
+; LA64LARGEPIC-NEXT: lu32i.d $ra, %got64_pc_lo20(__tls_get_addr)
+; LA64LARGEPIC-NEXT: lu52i.d $ra, $ra, %got64_pc_hi12(__tls_get_addr)
+; LA64LARGEPIC-NEXT: ldx.d $ra, $ra, $a1
; LA64LARGEPIC-NEXT: jirl $ra, $ra, 0
; LA64LARGEPIC-NEXT: ld.d $ra, $sp, 8 # 8-byte Folded Reload
; LA64LARGEPIC-NEXT: addi.d $sp, $sp, 16
|
SixWeining
approved these changes
Nov 21, 2024
wangleiat
deleted the
users/wangleiat/spr/loongarch-fix-got-usage-for-non-dso_local-function-calls-in-large-code-model
branch
wangleiat
added a commit
to wangleiat/llvm-project
that referenced
this pull request
Nov 21, 2024
… code model This commit fixes an issue in the large code model where non-dso_local function calls did not use the GOT as expected in PIC mode. Instead, direct PC-relative access was incorrectly applied, leading to linker errors when building shared libraries. For `ExternalSymbol`, it is not possible to determine whether it is dso_local during pseudo-instruction expansion. We use target flags to differentiate whether GOT should be used. Cherry-picked from llvm#117099, used for fix linker errors when bulding shared libraries with large code model.
tru
pushed a commit
to wangleiat/llvm-project
that referenced
this pull request
Nov 25, 2024
… code model This commit fixes an issue in the large code model where non-dso_local function calls did not use the GOT as expected in PIC mode. Instead, direct PC-relative access was incorrectly applied, leading to linker errors when building shared libraries. For `ExternalSymbol`, it is not possible to determine whether it is dso_local during pseudo-instruction expansion. We use target flags to differentiate whether GOT should be used. Cherry-picked from llvm#117099, used for fix linker errors when bulding shared libraries with large code model.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This commit fixes an issue in the large code model where non-dso_local
function calls did not use the GOT as expected in PIC mode. Instead,
direct PC-relative access was incorrectly applied, leading to linker
errors when building shared libraries.
For
ExternalSymbol, it is not possible to determine whether it isdso_local during pseudo-instruction expansion. We use target flags to
differentiate whether GOT should be used.