[llvm] f9649d1 - [JITLink][MachO] Split C-string literal sections on null-terminators.
Lang Hames via llvm-commits
llvm-commits at lists.llvm.org
Tue Jun 8 17:21:28 PDT 2021
Author: Lang Hames
Date: 2021-06-09T10:19:27+10:00
New Revision: f9649d123db64e38bce7bcb4cc5791239820f3e2
URL: https://github.com/llvm/llvm-project/commit/f9649d123db64e38bce7bcb4cc5791239820f3e2
DIFF: https://github.com/llvm/llvm-project/commit/f9649d123db64e38bce7bcb4cc5791239820f3e2.diff
LOG: [JITLink][MachO] Split C-string literal sections on null-terminators.
MachO C-string literal sections should be split on null-terminator boundaries,
rather than the usual symbol boundaries. This patch updates
MachOLinkGraphBuilder to do that.
Added:
Modified:
llvm/lib/ExecutionEngine/JITLink/MachOLinkGraphBuilder.cpp
llvm/lib/ExecutionEngine/JITLink/MachOLinkGraphBuilder.h
Removed:
################################################################################
diff --git a/llvm/lib/ExecutionEngine/JITLink/MachOLinkGraphBuilder.cpp b/llvm/lib/ExecutionEngine/JITLink/MachOLinkGraphBuilder.cpp
index 59db2ad83300d..904f38aac0856 100644
--- a/llvm/lib/ExecutionEngine/JITLink/MachOLinkGraphBuilder.cpp
+++ b/llvm/lib/ExecutionEngine/JITLink/MachOLinkGraphBuilder.cpp
@@ -411,10 +411,16 @@ Error MachOLinkGraphBuilder::graphifyRegularSymbols() {
<< " as it has a custom parser.\n";
});
continue;
+ } else if ((NSec.Flags & MachO::SECTION_TYPE) ==
+ MachO::S_CSTRING_LITERALS) {
+ if (auto Err = graphifyCStringSection(
+ NSec, std::move(SecIndexToSymbols[SecIndex])))
+ return Err;
+ continue;
} else
LLVM_DEBUG({
- dbgs() << " Processing section " << NSec.GraphSection->getName()
- << "...\n";
+ dbgs() << " Graphifying regular section "
+ << NSec.GraphSection->getName() << "...\n";
});
bool SectionIsNoDeadStrip = NSec.Flags & MachO::S_ATTR_NO_DEAD_STRIP;
@@ -525,34 +531,14 @@ Error MachOLinkGraphBuilder::graphifyRegularSymbols() {
bool SymLive =
(NSym.Desc & MachO::N_NO_DEAD_STRIP) || SectionIsNoDeadStrip;
- LLVM_DEBUG({
- dbgs() << " " << formatv("{0:x16}", NSym.Value) << " -- "
- << formatv("{0:x16}", SymEnd) << ": ";
- if (!NSym.Name)
- dbgs() << "<anonymous symbol>";
- else
- dbgs() << NSym.Name;
- if (SymLive)
- dbgs() << " [no-dead-strip]";
- if (LastCanonicalAddr == NSym.Value)
- dbgs() << " [non-canonical]";
- dbgs() << "\n";
- });
+ auto &Sym = createStandardGraphSymbol(NSym, B, SymEnd - NSym.Value,
+ SectionIsText, SymLive,
+ LastCanonicalAddr != NSym.Value);
- auto &Sym =
- NSym.Name
- ? G->addDefinedSymbol(B, NSym.Value - BlockStart, *NSym.Name,
- SymEnd - NSym.Value, NSym.L, NSym.S,
- SectionIsText, SymLive)
- : G->addAnonymousSymbol(B, NSym.Value - BlockStart,
- SymEnd - NSym.Value, SectionIsText,
- SymLive);
- NSym.GraphSymbol = &Sym;
if (LastCanonicalAddr != Sym.getAddress()) {
if (LastCanonicalAddr)
SymEnd = *LastCanonicalAddr;
LastCanonicalAddr = Sym.getAddress();
- setCanonicalSymbol(Sym);
}
}
}
@@ -561,6 +547,41 @@ Error MachOLinkGraphBuilder::graphifyRegularSymbols() {
return Error::success();
}
+Symbol &MachOLinkGraphBuilder::createStandardGraphSymbol(NormalizedSymbol &NSym,
+ Block &B, size_t Size,
+ bool IsText,
+ bool IsNoDeadStrip,
+ bool IsCanonical) {
+
+ LLVM_DEBUG({
+ dbgs() << " " << formatv("{0:x16}", NSym.Value) << " -- "
+ << formatv("{0:x16}", NSym.Value + Size) << ": ";
+ if (!NSym.Name)
+ dbgs() << "<anonymous symbol>";
+ else
+ dbgs() << NSym.Name;
+ if (IsText)
+ dbgs() << " [text]";
+ if (IsNoDeadStrip)
+ dbgs() << " [no-dead-strip]";
+ if (!IsCanonical)
+ dbgs() << " [non-canonical]";
+ dbgs() << "\n";
+ });
+
+ auto &Sym = NSym.Name ? G->addDefinedSymbol(B, NSym.Value - B.getAddress(),
+ *NSym.Name, Size, NSym.L, NSym.S,
+ IsText, IsNoDeadStrip)
+ : G->addAnonymousSymbol(B, NSym.Value - B.getAddress(),
+ Size, IsText, IsNoDeadStrip);
+ NSym.GraphSymbol = &Sym;
+
+ if (IsCanonical)
+ setCanonicalSymbol(Sym);
+
+ return Sym;
+}
+
Error MachOLinkGraphBuilder::graphifySectionsWithCustomParsers() {
// Graphify special sections.
for (auto &KV : IndexToSection) {
@@ -581,5 +602,95 @@ Error MachOLinkGraphBuilder::graphifySectionsWithCustomParsers() {
return Error::success();
}
+Error MachOLinkGraphBuilder::graphifyCStringSection(
+ NormalizedSection &NSec, std::vector<NormalizedSymbol *> NSyms) {
+
+ assert(NSec.GraphSection && "C string literal section missing graph section");
+ assert(NSec.Data && "C string literal section has no data");
+
+ LLVM_DEBUG({
+ dbgs() << " Graphifying C-string literal section "
+ << NSec.GraphSection->getName() << "\n";
+ });
+
+ if (NSec.Data[NSec.Size - 1] != '\0')
+ return make_error<JITLinkError>("C string literal section " +
+ NSec.GraphSection->getName() +
+ " does not end with null terminator");
+
+ /// Sort into reverse order to use as a stack.
+ llvm::sort(NSyms,
+ [](const NormalizedSymbol *LHS, const NormalizedSymbol *RHS) {
+ if (LHS->Value != RHS->Value)
+ return LHS->Value > RHS->Value;
+ if (LHS->L != RHS->L)
+ return LHS->L > RHS->L;
+ if (LHS->S != RHS->S)
+ return LHS->S > RHS->S;
+ if (RHS->Name) {
+ if (!LHS->Name)
+ return true;
+ return *LHS->Name > *RHS->Name;
+ }
+ return false;
+ });
+
+ bool SectionIsNoDeadStrip = NSec.Flags & MachO::S_ATTR_NO_DEAD_STRIP;
+ bool SectionIsText = NSec.Flags & MachO::S_ATTR_PURE_INSTRUCTIONS;
+ JITTargetAddress BlockStart = 0;
+
+ // Scan section for null characters.
+ for (size_t I = 0; I != NSec.Size; ++I)
+ if (NSec.Data[I] == '\0') {
+ JITTargetAddress BlockEnd = I + 1;
+ size_t BlockSize = BlockEnd - BlockStart;
+ // Create a block for this null terminated string.
+ auto &B = G->createContentBlock(*NSec.GraphSection,
+ {NSec.Data + BlockStart, BlockSize},
+ NSec.Address + BlockStart, 1, 0);
+
+ LLVM_DEBUG({
+ dbgs() << " Created block " << formatv("{0:x}", B.getAddress())
+ << " -- " << formatv("{0:x}", B.getAddress() + B.getSize())
+ << " for \"" << StringRef(B.getContent().data()) << "\"\n";
+ });
+
+ // Process any symbols that point into this block.
+ JITTargetAddress LastCanonicalAddr = BlockEnd;
+ Symbol *BlockStartSymbol = nullptr;
+ while (!NSyms.empty() &&
+ NSyms.back()->Value < (B.getAddress() + BlockSize)) {
+ auto &NSym = *NSyms.back();
+ size_t SymSize = (B.getAddress() + BlockSize) - NSyms.back()->Value;
+ bool SymLive =
+ (NSym.Desc & MachO::N_NO_DEAD_STRIP) || SectionIsNoDeadStrip;
+
+ auto &S =
+ createStandardGraphSymbol(NSym, B, SymSize, SectionIsText, SymLive,
+ NSym.Value != LastCanonicalAddr);
+
+ if (S.getAddress() == B.getAddress() && !BlockStartSymbol)
+ BlockStartSymbol = &S;
+
+ NSyms.pop_back();
+ }
+
+ if (!BlockStartSymbol) {
+ auto &S = G->addAnonymousSymbol(B, 0, BlockSize, false, false);
+ setCanonicalSymbol(S);
+ LLVM_DEBUG({
+ dbgs() << " Adding anonymous symbol for "
+ << formatv("{0:x16} -- {1:x16}", S.getAddress(),
+ S.getAddress() + BlockSize)
+ << "\n";
+ });
+ }
+
+ BlockStart += BlockSize;
+ }
+
+ return Error::success();
+}
+
} // end namespace jitlink
} // end namespace llvm
diff --git a/llvm/lib/ExecutionEngine/JITLink/MachOLinkGraphBuilder.h b/llvm/lib/ExecutionEngine/JITLink/MachOLinkGraphBuilder.h
index 73a1f4054ee4f..a6df4458c1114 100644
--- a/llvm/lib/ExecutionEngine/JITLink/MachOLinkGraphBuilder.h
+++ b/llvm/lib/ExecutionEngine/JITLink/MachOLinkGraphBuilder.h
@@ -200,9 +200,21 @@ class MachOLinkGraphBuilder {
/// all defined symbols in sections without custom parsers.
Error graphifyRegularSymbols();
+ /// Create and return a graph symbol for the given normalized symbol.
+ ///
+ /// NSym's GraphSymbol member will be updated to point at the newly created
+ /// symbol.
+ Symbol &createStandardGraphSymbol(NormalizedSymbol &Sym, Block &B,
+ size_t Size, bool IsText,
+ bool IsNoDeadStrip, bool IsCanonical);
+
/// Create graph blocks and symbols for all sections.
Error graphifySectionsWithCustomParsers();
+ /// Graphify cstring section.
+ Error graphifyCStringSection(NormalizedSection &NSec,
+ std::vector<NormalizedSymbol *> NSyms);
+
// Put the BumpPtrAllocator first so that we don't free any of the underlying
// memory until the Symbol/Addressable destructors have been run.
BumpPtrAllocator Allocator;
More information about the llvm-commits
mailing list