[lld] [LLD][COFF] Add Support for auxiliary IAT copy (PR #108610)
Jacek Caban via llvm-commits
llvm-commits at lists.llvm.org
Fri Sep 13 10:26:41 PDT 2024
https://github.com/cjacek created https://github.com/llvm/llvm-project/pull/108610
In addition to the auxiliary IAT, ARM64EC modules also contain a copy of it. At runtime, the auxiliary IAT is filled with the addresses of actual ARM64EC functions when possible. If patching is detected, the OS may use the IAT copy to revert the auxiliary IAT, ensuring that the call checker is used for calls to imported functions.
>From 6f94066dc0a443e2ee380bc836830a4397de0eab Mon Sep 17 00:00:00 2001
From: Jacek Caban <jacek at codeweavers.com>
Date: Mon, 2 Sep 2024 21:58:07 +0200
Subject: [PATCH 1/2] [LLD][COFF] Redirect __imp_ Symbols to __imp_aux_ on
ARM64EC for x64 Sections
On ARM64EC, __imp_ symbols reference the auxiliary IAT, while __imp_aux_ symbols
reference the regular IAT. However, x86_64 code expects both to reference the
regular IAT. This change adjusts the symbols accordingly, matching the behavior
observed in the MSVC linker.
---
lld/COFF/Chunks.cpp | 10 ++++++++++
lld/test/COFF/arm64ec-import.test | 20 ++++++++++++++++++++
2 files changed, 30 insertions(+)
diff --git a/lld/COFF/Chunks.cpp b/lld/COFF/Chunks.cpp
index ee54fa39fc3d6a..81faa1f3de80fc 100644
--- a/lld/COFF/Chunks.cpp
+++ b/lld/COFF/Chunks.cpp
@@ -422,6 +422,16 @@ void SectionChunk::applyRelocation(uint8_t *off,
const coff_relocation &rel) const {
auto *sym = dyn_cast_or_null<Defined>(file->getSymbol(rel.SymbolTableIndex));
+ // On ARM64EC, the __imp_ symbol references the auxiliary IAT, while the
+ // __imp_aux_ symbol references the regular IAT. However, x86_64 code expects
+ // both to reference the regular IAT, so adjust the symbol if necessary.
+ if (sym && getMachine() == AMD64 && isArm64EC(file->ctx.config.machine)) {
+ if (auto impSym = dyn_cast<DefinedImportData>(sym)) {
+ if (impSym->file->impchkThunk && sym == impSym->file->impECSym)
+ sym = impSym->file->impSym;
+ }
+ }
+
// Get the output section of the symbol for this relocation. The output
// section is needed to compute SECREL and SECTION relocations used in debug
// info.
diff --git a/lld/test/COFF/arm64ec-import.test b/lld/test/COFF/arm64ec-import.test
index e403daa41f368c..ac43114f7e7ebb 100644
--- a/lld/test/COFF/arm64ec-import.test
+++ b/lld/test/COFF/arm64ec-import.test
@@ -4,6 +4,7 @@ RUN: split-file %s %t.dir && cd %t.dir
RUN: llvm-mc -filetype=obj -triple=arm64ec-windows test.s -o test.obj
RUN: llvm-mc -filetype=obj -triple=arm64ec-windows icall.s -o icall.obj
RUN: llvm-mc -filetype=obj -triple=arm64ec-windows hybmp.s -o hybmp.obj
+RUN: llvm-mc -filetype=obj -triple=x86_64-windows test.s -o test-x86_64.obj
RUN: llvm-mc -filetype=obj -triple=arm64ec-windows %S/Inputs/loadconfig-arm64ec.s -o loadconfig-arm64ec.obj
RUN: llvm-lib -machine:arm64ec -def:test.def -out:test-arm64ec.lib
RUN: llvm-lib -machine:arm64ec -def:test2.def -out:test2-arm64ec.lib
@@ -17,8 +18,13 @@ Link using x86_64 import library:
RUN: lld-link -machine:arm64ec -dll -noentry -out:out2.dll loadconfig-arm64ec.obj icall.obj hybmp.obj \
RUN: test.obj test-x86_64.lib test2-arm64ec.lib
+Link using x86_64 object file:
+RUN: lld-link -machine:arm64ec -dll -noentry -out:out3.dll loadconfig-arm64ec.obj icall.obj hybmp.obj \
+RUN: test-x86_64.obj test-arm64ec.lib test2-arm64ec.lib
+
RUN: llvm-readobj --coff-imports out.dll | FileCheck --check-prefix=IMPORTS %s
RUN: llvm-readobj --coff-imports out2.dll | FileCheck --check-prefix=IMPORTS %s
+RUN: llvm-readobj --coff-imports out3.dll | FileCheck -check-prefix=IMPORTS %s
IMPORTS: Import {
IMPORTS-NEXT: Name: test.dll
IMPORTS-NEXT: ImportLookupTableRVA:
@@ -36,6 +42,7 @@ IMPORTS-NEXT: }
RUN: llvm-objdump -d out.dll | FileCheck --check-prefix=DISASM %s
RUN: llvm-objdump -d out2.dll | FileCheck --check-prefix=DISASM %s
+RUN: llvm-objdump -d out3.dll | FileCheck -check-prefix=DISASM %s
DISASM: 180001000: 52800000 mov w0, #0x0 // =0
DISASM-NEXT: 180001004: d65f03c0 ret
@@ -74,20 +81,33 @@ TESTSEC-NEXT: 0x180007010 08300000 00500000 10300000 20300000
TESTSEC-NEXT: 0x180007020 14100000 28100000 00200000 08100000
TESTSEC-NEXT: 0x180007030 3c100000
+RUN: llvm-readobj --hex-dump=.test out3.dll | FileCheck -check-prefix=TESTSEC-X64 %s
+TESTSEC-X64: 0x180007000 08300000 00300000 10300000 20300000
+TESTSEC-X64-NEXT: 0x180007010 08300000 00500000 10300000 20300000
+TESTSEC-X64-NEXT: 0x180007020 14100000 28100000 00200000 08100000
+
RUN: llvm-readobj --headers out.dll | FileCheck -check-prefix=HEADERS %s
+RUN: llvm-readobj --headers out2.dll | FileCheck -check-prefix=HEADERS %s
+RUN: llvm-readobj --headers out3.dll | FileCheck -check-prefix=HEADERS %s
HEADERS: LoadConfigTableRVA: 0x4010
HEADERS: IATRVA: 0x3000
HEADERS: IATSize: 0x1000
RUN: llvm-readobj --coff-load-config out.dll | FileCheck -check-prefix=LOADCONFIG %s
+RUN: llvm-readobj --coff-load-config out2.dll | FileCheck -check-prefix=LOADCONFIG %s
+RUN: llvm-readobj --coff-load-config out3.dll | FileCheck -check-prefix=LOADCONFIG %s
LOADCONFIG: AuxiliaryIAT: 0x5000
RUN: llvm-readobj --hex-dump=.rdata out.dll | FileCheck -check-prefix=RDATA %s
+RUN: llvm-readobj --hex-dump=.rdata out2.dll | FileCheck -check-prefix=RDATA %s
+RUN: llvm-readobj --hex-dump=.rdata out3.dll | FileCheck -check-prefix=RDATA %s
RDATA: 0x180005000 00000000 00000000 14100080 01000000
RDATA-NEXT: 0x180005010 28100080 01000000 00000000 00000000
RDATA-NEXT: 0x180005020 48100080 01000000 00000000 00000000
RUN: llvm-readobj --coff-basereloc out.dll | FileCheck -check-prefix=BASERELOC %s
+RUN: llvm-readobj --coff-basereloc out2.dll | FileCheck -check-prefix=BASERELOC %s
+RUN: llvm-readobj --coff-basereloc out3.dll | FileCheck -check-prefix=BASERELOC %s
BASERELOC: BaseReloc [
BASERELOC-NOT: Address: 0x5000
BASERELOC: Address: 0x5008
>From 2185a29718997bf8ae448876e17f9cc9969ee3b2 Mon Sep 17 00:00:00 2001
From: Jacek Caban <jacek at codeweavers.com>
Date: Tue, 3 Sep 2024 22:37:52 +0200
Subject: [PATCH 2/2] [LLD][COFF] Add Support for auxiliary IAT copy
In addition to the auxiliary IAT, ARM64EC modules also contain a copy of it.
At runtime, the auxiliary IAT is filled with the addresses of actual ARM64EC
functions when possible. If patching is detected, the OS may use the IAT copy
to revert the auxiliary IAT, ensuring that the call checker is used for calls
to imported functions.
---
lld/COFF/DLL.cpp | 8 +++++++-
lld/COFF/DLL.h | 1 +
lld/COFF/Driver.cpp | 1 +
lld/COFF/InputFiles.cpp | 6 ++++++
lld/COFF/InputFiles.h | 4 +++-
lld/COFF/Writer.cpp | 7 +++++++
lld/test/COFF/Inputs/loadconfig-arm64ec.s | 2 +-
lld/test/COFF/arm64ec-import.test | 23 +++++++++++++++++++++--
8 files changed, 47 insertions(+), 5 deletions(-)
diff --git a/lld/COFF/DLL.cpp b/lld/COFF/DLL.cpp
index b00a59433319a7..39dcce9fe84837 100644
--- a/lld/COFF/DLL.cpp
+++ b/lld/COFF/DLL.cpp
@@ -730,13 +730,19 @@ void IdataContents::create(COFFLinkerContext &ctx) {
auto chunk = make<AuxImportChunk>(s->file);
auxIat.push_back(chunk);
s->file->impECSym->setLocation(chunk);
+
+ chunk = make<AuxImportChunk>(s->file);
+ auxIatCopy.push_back(chunk);
+ s->file->auxImpCopySym->setLocation(chunk);
}
}
// Terminate with null values.
lookups.push_back(make<NullChunk>(ctx.config.wordsize));
addresses.push_back(make<NullChunk>(ctx.config.wordsize));
- if (ctx.config.machine == ARM64EC)
+ if (ctx.config.machine == ARM64EC) {
auxIat.push_back(make<NullChunk>(ctx.config.wordsize));
+ auxIatCopy.push_back(make<NullChunk>(ctx.config.wordsize));
+ }
for (int i = 0, e = syms.size(); i < e; ++i)
syms[i]->setLocation(addresses[base + i]);
diff --git a/lld/COFF/DLL.h b/lld/COFF/DLL.h
index 48b0f17ca62aeb..afb46f22ec9e14 100644
--- a/lld/COFF/DLL.h
+++ b/lld/COFF/DLL.h
@@ -32,6 +32,7 @@ class IdataContents {
std::vector<Chunk *> hints;
std::vector<Chunk *> dllNames;
std::vector<Chunk *> auxIat;
+ std::vector<Chunk *> auxIatCopy;
};
// Windows-specific.
diff --git a/lld/COFF/Driver.cpp b/lld/COFF/Driver.cpp
index 9994639d5d364a..1b94f10acf80e5 100644
--- a/lld/COFF/Driver.cpp
+++ b/lld/COFF/Driver.cpp
@@ -2448,6 +2448,7 @@ void LinkerDriver::linkerMain(ArrayRef<const char *> argsArr) {
ctx.symtab.addAbsolute("__arm64x_redirection_metadata", 0);
ctx.symtab.addAbsolute("__arm64x_redirection_metadata_count", 0);
ctx.symtab.addAbsolute("__hybrid_auxiliary_iat", 0);
+ ctx.symtab.addAbsolute("__hybrid_auxiliary_iat_copy", 0);
ctx.symtab.addAbsolute("__hybrid_code_map", 0);
ctx.symtab.addAbsolute("__hybrid_code_map_count", 0);
ctx.symtab.addAbsolute("__x64_code_ranges_to_entry_points", 0);
diff --git a/lld/COFF/InputFiles.cpp b/lld/COFF/InputFiles.cpp
index 94ad7f3ceb306a..d9184b04735e83 100644
--- a/lld/COFF/InputFiles.cpp
+++ b/lld/COFF/InputFiles.cpp
@@ -1091,6 +1091,12 @@ void ImportFile::parse() {
}
if (!impECSym)
return;
+
+ StringRef auxImpCopyName = saver().save("__auximpcopy_" + name);
+ auxImpCopySym =
+ ctx.symtab.addImportData(auxImpCopyName, this, auxCopyLocation);
+ if (!auxImpCopySym)
+ return;
}
// If this was a duplicate, we logged an error but may continue;
// in this case, impSym is nullptr.
diff --git a/lld/COFF/InputFiles.h b/lld/COFF/InputFiles.h
index acf221d85ae8f2..5fa93f57ef9e3a 100644
--- a/lld/COFF/InputFiles.h
+++ b/lld/COFF/InputFiles.h
@@ -362,10 +362,12 @@ class ImportFile : public InputFile {
const coff_import_header *hdr;
Chunk *location = nullptr;
- // Auxiliary IAT symbol and chunk on ARM64EC.
+ // Auxiliary IAT symbols and chunks on ARM64EC.
DefinedImportData *impECSym = nullptr;
Chunk *auxLocation = nullptr;
Symbol *auxThunkSym = nullptr;
+ DefinedImportData *auxImpCopySym = nullptr;
+ Chunk *auxCopyLocation = nullptr;
// We want to eliminate dllimported symbols if no one actually refers to them.
// These "Live" bits are used to keep track of which import library members
diff --git a/lld/COFF/Writer.cpp b/lld/COFF/Writer.cpp
index 216db652c10aa8..c2765453aa964e 100644
--- a/lld/COFF/Writer.cpp
+++ b/lld/COFF/Writer.cpp
@@ -916,6 +916,8 @@ void Writer::addSyntheticIdata() {
add(".idata$7", idata.dllNames);
if (!idata.auxIat.empty())
add(".idata$9", idata.auxIat);
+ if (!idata.auxIatCopy.empty())
+ add(".idata$a", idata.auxIatCopy);
}
void Writer::appendECImportTables() {
@@ -2279,6 +2281,11 @@ void Writer::setECSymbols() {
replaceSymbol<DefinedSynthetic>(iatSym, "__hybrid_auxiliary_iat",
idata.auxIat.empty() ? nullptr
: idata.auxIat.front());
+
+ Symbol *iatCopySym = ctx.symtab.findUnderscore("__hybrid_auxiliary_iat_copy");
+ replaceSymbol<DefinedSynthetic>(
+ iatCopySym, "__hybrid_auxiliary_iat_copy",
+ idata.auxIatCopy.empty() ? nullptr : idata.auxIatCopy.front());
}
// Write section contents to a mmap'ed file.
diff --git a/lld/test/COFF/Inputs/loadconfig-arm64ec.s b/lld/test/COFF/Inputs/loadconfig-arm64ec.s
index 8d59d29bbd0212..cb79b5c257e6ec 100644
--- a/lld/test/COFF/Inputs/loadconfig-arm64ec.s
+++ b/lld/test/COFF/Inputs/loadconfig-arm64ec.s
@@ -84,7 +84,7 @@ __chpe_metadata:
.rva __arm64x_extra_rfe_table
.word __arm64x_extra_rfe_table_size
.rva __os_arm64x_dispatch_fptr
- .word 0 // __hybrid_auxiliary_iat_copy
+ .rva __hybrid_auxiliary_iat_copy
.rva __os_arm64x_helper0
.rva __os_arm64x_helper1
.rva __os_arm64x_helper2
diff --git a/lld/test/COFF/arm64ec-import.test b/lld/test/COFF/arm64ec-import.test
index ac43114f7e7ebb..92d7f5517bd426 100644
--- a/lld/test/COFF/arm64ec-import.test
+++ b/lld/test/COFF/arm64ec-import.test
@@ -27,7 +27,7 @@ RUN: llvm-readobj --coff-imports out2.dll | FileCheck --check-prefix=IMPORTS %s
RUN: llvm-readobj --coff-imports out3.dll | FileCheck -check-prefix=IMPORTS %s
IMPORTS: Import {
IMPORTS-NEXT: Name: test.dll
-IMPORTS-NEXT: ImportLookupTableRVA:
+IMPORTS-NEXT: ImportLookupTableRVA: 0x4230
IMPORTS-NEXT: ImportAddressTableRVA: 0x3000
IMPORTS-NEXT: Symbol: data (0)
IMPORTS-NEXT: Symbol: func (0)
@@ -79,12 +79,13 @@ RUN: llvm-readobj --hex-dump=.test out2.dll | FileCheck --check-prefix=TESTSEC %
TESTSEC: 0x180007000 08500000 00300000 10500000 20500000
TESTSEC-NEXT: 0x180007010 08300000 00500000 10300000 20300000
TESTSEC-NEXT: 0x180007020 14100000 28100000 00200000 08100000
-TESTSEC-NEXT: 0x180007030 3c100000
+TESTSEC-NEXT: 0x180007030 3c100000 a0420000
RUN: llvm-readobj --hex-dump=.test out3.dll | FileCheck -check-prefix=TESTSEC-X64 %s
TESTSEC-X64: 0x180007000 08300000 00300000 10300000 20300000
TESTSEC-X64-NEXT: 0x180007010 08300000 00500000 10300000 20300000
TESTSEC-X64-NEXT: 0x180007020 14100000 28100000 00200000 08100000
+TESTSEC-X64-NEXT: 0x180007030 3c100000 a0420000
RUN: llvm-readobj --headers out.dll | FileCheck -check-prefix=HEADERS %s
RUN: llvm-readobj --headers out2.dll | FileCheck -check-prefix=HEADERS %s
@@ -97,10 +98,15 @@ RUN: llvm-readobj --coff-load-config out.dll | FileCheck -check-prefix=LOADCONFI
RUN: llvm-readobj --coff-load-config out2.dll | FileCheck -check-prefix=LOADCONFIG %s
RUN: llvm-readobj --coff-load-config out3.dll | FileCheck -check-prefix=LOADCONFIG %s
LOADCONFIG: AuxiliaryIAT: 0x5000
+LOADCONFIG: AuxiliaryIATCopy: 0x4298
RUN: llvm-readobj --hex-dump=.rdata out.dll | FileCheck -check-prefix=RDATA %s
RUN: llvm-readobj --hex-dump=.rdata out2.dll | FileCheck -check-prefix=RDATA %s
RUN: llvm-readobj --hex-dump=.rdata out3.dll | FileCheck -check-prefix=RDATA %s
+RDATA: 0x180004290 2e646c6c 00000000 00000000 00000000
+RDATA-NEXT: 0x1800042a0 14100080 01000000 28100080 01000000
+RDATA-NEXT: 0x1800042b0 00000000 00000000 48100080 01000000
+RDATA-NEXT: 0x1800042c0 00000000 00000000 00000000 00000000
RDATA: 0x180005000 00000000 00000000 14100080 01000000
RDATA-NEXT: 0x180005010 28100080 01000000 00000000 00000000
RDATA-NEXT: 0x180005020 48100080 01000000 00000000 00000000
@@ -109,6 +115,18 @@ RUN: llvm-readobj --coff-basereloc out.dll | FileCheck -check-prefix=BASERELOC %
RUN: llvm-readobj --coff-basereloc out2.dll | FileCheck -check-prefix=BASERELOC %s
RUN: llvm-readobj --coff-basereloc out3.dll | FileCheck -check-prefix=BASERELOC %s
BASERELOC: BaseReloc [
+Aux IAT copy:
+BASERELOC: Address: 0x42A0
+BASERELOC-NEXT: }
+BASERELOC-NEXT: Entry {
+BASERELOC-NEXT: Type: DIR64
+BASERELOC-NEXT: Address: 0x42A8
+BASERELOC-NEXT: }
+BASERELOC-NEXT: Entry {
+BASERELOC-NEXT: Type: DIR64
+BASERELOC-NEXT: Address: 0x42B8
+BASERELOC-NEXT: }
+Aux IAT:
BASERELOC-NOT: Address: 0x5000
BASERELOC: Address: 0x5008
BASERELOC-NEXT: }
@@ -139,6 +157,7 @@ arm64ec_data_sym:
.rva func
.rva "#func"
.rva "#t2func"
+ .rva __auximpcopy_func
#--- icall.s
.text
More information about the llvm-commits
mailing list