1//===- DataFlowSanitizer.cpp - dynamic data flow analysis -----------------===//
2//
3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4// See https://llvm.org/LICENSE.txt for license information.
5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6//
7//===----------------------------------------------------------------------===//
8//
9/// \file
10/// This file is a part of DataFlowSanitizer, a generalised dynamic data flow
11/// analysis.
12///
13/// Unlike other Sanitizer tools, this tool is not designed to detect a specific
14/// class of bugs on its own. Instead, it provides a generic dynamic data flow
15/// analysis framework to be used by clients to help detect application-specific
16/// issues within their own code.
17///
18/// The analysis is based on automatic propagation of data flow labels (also
19/// known as taint labels) through a program as it performs computation.
20///
21/// Argument and return value labels are passed through TLS variables
22/// __dfsan_arg_tls and __dfsan_retval_tls.
23///
24/// Each byte of application memory is backed by a shadow memory byte. The
25/// shadow byte can represent up to 8 labels. On Linux/x86_64, memory is then
26/// laid out as follows:
27///
28/// +--------------------+ 0x800000000000 (top of memory)
29/// | application 3 |
30/// +--------------------+ 0x700000000000
31/// | invalid |
32/// +--------------------+ 0x610000000000
33/// | origin 1 |
34/// +--------------------+ 0x600000000000
35/// | application 2 |
36/// +--------------------+ 0x510000000000
37/// | shadow 1 |
38/// +--------------------+ 0x500000000000
39/// | invalid |
40/// +--------------------+ 0x400000000000
41/// | origin 3 |
42/// +--------------------+ 0x300000000000
43/// | shadow 3 |
44/// +--------------------+ 0x200000000000
45/// | origin 2 |
46/// +--------------------+ 0x110000000000
47/// | invalid |
48/// +--------------------+ 0x100000000000
49/// | shadow 2 |
50/// +--------------------+ 0x010000000000
51/// | application 1 |
52/// +--------------------+ 0x000000000000
53///
54/// MEM_TO_SHADOW(mem) = mem ^ 0x500000000000
55/// SHADOW_TO_ORIGIN(shadow) = shadow + 0x100000000000
56///
57/// For more information, please refer to the design document:
58/// http://clang.llvm.org/docs/DataFlowSanitizerDesign.html
59//
60//===----------------------------------------------------------------------===//
61
62#include "llvm/Transforms/Instrumentation/DataFlowSanitizer.h"
63#include "llvm/ADT/DenseMap.h"
64#include "llvm/ADT/DenseSet.h"
65#include "llvm/ADT/DepthFirstIterator.h"
66#include "llvm/ADT/SmallPtrSet.h"
67#include "llvm/ADT/SmallVector.h"
68#include "llvm/ADT/StringRef.h"
69#include "llvm/ADT/StringSet.h"
70#include "llvm/ADT/iterator.h"
71#include "llvm/Analysis/DomTreeUpdater.h"
72#include "llvm/Analysis/GlobalsModRef.h"
73#include "llvm/Analysis/TargetLibraryInfo.h"
74#include "llvm/Analysis/ValueTracking.h"
75#include "llvm/IR/Argument.h"
76#include "llvm/IR/AttributeMask.h"
77#include "llvm/IR/Attributes.h"
78#include "llvm/IR/BasicBlock.h"
79#include "llvm/IR/Constant.h"
80#include "llvm/IR/Constants.h"
81#include "llvm/IR/DataLayout.h"
82#include "llvm/IR/DerivedTypes.h"
83#include "llvm/IR/Dominators.h"
84#include "llvm/IR/Function.h"
85#include "llvm/IR/GlobalAlias.h"
86#include "llvm/IR/GlobalValue.h"
87#include "llvm/IR/GlobalVariable.h"
88#include "llvm/IR/IRBuilder.h"
89#include "llvm/IR/InstVisitor.h"
90#include "llvm/IR/InstrTypes.h"
91#include "llvm/IR/Instruction.h"
92#include "llvm/IR/Instructions.h"
93#include "llvm/IR/IntrinsicInst.h"
94#include "llvm/IR/MDBuilder.h"
95#include "llvm/IR/Module.h"
96#include "llvm/IR/PassManager.h"
97#include "llvm/IR/Type.h"
98#include "llvm/IR/User.h"
99#include "llvm/IR/Value.h"
100#include "llvm/Support/Alignment.h"
101#include "llvm/Support/Casting.h"
102#include "llvm/Support/CommandLine.h"
103#include "llvm/Support/ErrorHandling.h"
104#include "llvm/Support/SpecialCaseList.h"
105#include "llvm/Support/VirtualFileSystem.h"
106#include "llvm/TargetParser/Triple.h"
107#include "llvm/Transforms/Instrumentation.h"
108#include "llvm/Transforms/Utils/BasicBlockUtils.h"
109#include "llvm/Transforms/Utils/Local.h"
110#include <algorithm>
111#include <cassert>
112#include <cstddef>
113#include <cstdint>
114#include <memory>
115#include <set>
116#include <string>
117#include <utility>
118#include <vector>
119
120using namespace llvm;
121
122// This must be consistent with ShadowWidthBits.
123static const Align ShadowTLSAlignment = Align(2);
124
125static const Align MinOriginAlignment = Align(4);
126
127// The size of TLS variables. These constants must be kept in sync with the ones
128// in dfsan.cpp.
129static const unsigned ArgTLSSize = 800;
130static const unsigned RetvalTLSSize = 800;
131
132// The -dfsan-preserve-alignment flag controls whether this pass assumes that
133// alignment requirements provided by the input IR are correct. For example,
134// if the input IR contains a load with alignment 8, this flag will cause
135// the shadow load to have alignment 16. This flag is disabled by default as
136// we have unfortunately encountered too much code (including Clang itself;
137// see PR14291) which performs misaligned access.
138static cl::opt<bool> ClPreserveAlignment(
139 "dfsan-preserve-alignment",
140 cl::desc("respect alignment requirements provided by input IR"), cl::Hidden,
141 cl::init(Val: false));
142
143// The ABI list files control how shadow parameters are passed. The pass treats
144// every function labelled "uninstrumented" in the ABI list file as conforming
145// to the "native" (i.e. unsanitized) ABI. Unless the ABI list contains
146// additional annotations for those functions, a call to one of those functions
147// will produce a warning message, as the labelling behaviour of the function is
148// unknown. The other supported annotations for uninstrumented functions are
149// "functional" and "discard", which are described below under
150// DataFlowSanitizer::WrapperKind.
151// Functions will often be labelled with both "uninstrumented" and one of
152// "functional" or "discard". This will leave the function unchanged by this
153// pass, and create a wrapper function that will call the original.
154//
155// Instrumented functions can also be annotated as "force_zero_labels", which
156// will make all shadow and return values set zero labels.
157// Functions should never be labelled with both "force_zero_labels" and
158// "uninstrumented" or any of the unistrumented wrapper kinds.
159static cl::list<std::string> ClABIListFiles(
160 "dfsan-abilist",
161 cl::desc("File listing native ABI functions and how the pass treats them"),
162 cl::Hidden);
163
164// Controls whether the pass includes or ignores the labels of pointers in load
165// instructions.
166static cl::opt<bool> ClCombinePointerLabelsOnLoad(
167 "dfsan-combine-pointer-labels-on-load",
168 cl::desc("Combine the label of the pointer with the label of the data when "
169 "loading from memory."),
170 cl::Hidden, cl::init(Val: true));
171
172// Controls whether the pass includes or ignores the labels of pointers in
173// stores instructions.
174static cl::opt<bool> ClCombinePointerLabelsOnStore(
175 "dfsan-combine-pointer-labels-on-store",
176 cl::desc("Combine the label of the pointer with the label of the data when "
177 "storing in memory."),
178 cl::Hidden, cl::init(Val: false));
179
180// Controls whether the pass propagates labels of offsets in GEP instructions.
181static cl::opt<bool> ClCombineOffsetLabelsOnGEP(
182 "dfsan-combine-offset-labels-on-gep",
183 cl::desc(
184 "Combine the label of the offset with the label of the pointer when "
185 "doing pointer arithmetic."),
186 cl::Hidden, cl::init(Val: true));
187
188static cl::list<std::string> ClCombineTaintLookupTables(
189 "dfsan-combine-taint-lookup-table",
190 cl::desc(
191 "When dfsan-combine-offset-labels-on-gep and/or "
192 "dfsan-combine-pointer-labels-on-load are false, this flag can "
193 "be used to re-enable combining offset and/or pointer taint when "
194 "loading specific constant global variables (i.e. lookup tables)."),
195 cl::Hidden);
196
197static cl::opt<bool> ClDebugNonzeroLabels(
198 "dfsan-debug-nonzero-labels",
199 cl::desc("Insert calls to __dfsan_nonzero_label on observing a parameter, "
200 "load or return with a nonzero label"),
201 cl::Hidden);
202
203// Experimental feature that inserts callbacks for certain data events.
204// Currently callbacks are only inserted for loads, stores, memory transfers
205// (i.e. memcpy and memmove), and comparisons.
206//
207// If this flag is set to true, the user must provide definitions for the
208// following callback functions:
209// void __dfsan_load_callback(dfsan_label Label, void* addr);
210// void __dfsan_store_callback(dfsan_label Label, void* addr);
211// void __dfsan_mem_transfer_callback(dfsan_label *Start, size_t Len);
212// void __dfsan_cmp_callback(dfsan_label CombinedLabel);
213static cl::opt<bool> ClEventCallbacks(
214 "dfsan-event-callbacks",
215 cl::desc("Insert calls to __dfsan_*_callback functions on data events."),
216 cl::Hidden, cl::init(Val: false));
217
218// Experimental feature that inserts callbacks for conditionals, including:
219// conditional branch, switch, select.
220// This must be true for dfsan_set_conditional_callback() to have effect.
221static cl::opt<bool> ClConditionalCallbacks(
222 "dfsan-conditional-callbacks",
223 cl::desc("Insert calls to callback functions on conditionals."), cl::Hidden,
224 cl::init(Val: false));
225
226// Experimental feature that inserts callbacks for data reaching a function,
227// either via function arguments and loads.
228// This must be true for dfsan_set_reaches_function_callback() to have effect.
229static cl::opt<bool> ClReachesFunctionCallbacks(
230 "dfsan-reaches-function-callbacks",
231 cl::desc("Insert calls to callback functions on data reaching a function."),
232 cl::Hidden, cl::init(Val: false));
233
234// Controls whether the pass tracks the control flow of select instructions.
235static cl::opt<bool> ClTrackSelectControlFlow(
236 "dfsan-track-select-control-flow",
237 cl::desc("Propagate labels from condition values of select instructions "
238 "to results."),
239 cl::Hidden, cl::init(Val: true));
240
241// TODO: This default value follows MSan. DFSan may use a different value.
242static cl::opt<int> ClInstrumentWithCallThreshold(
243 "dfsan-instrument-with-call-threshold",
244 cl::desc("If the function being instrumented requires more than "
245 "this number of origin stores, use callbacks instead of "
246 "inline checks (-1 means never use callbacks)."),
247 cl::Hidden, cl::init(Val: 3500));
248
249// Controls how to track origins.
250// * 0: do not track origins.
251// * 1: track origins at memory store operations.
252// * 2: track origins at memory load and store operations.
253// TODO: track callsites.
254static cl::opt<int> ClTrackOrigins("dfsan-track-origins",
255 cl::desc("Track origins of labels"),
256 cl::Hidden, cl::init(Val: 0));
257
258static cl::opt<bool> ClIgnorePersonalityRoutine(
259 "dfsan-ignore-personality-routine",
260 cl::desc("If a personality routine is marked uninstrumented from the ABI "
261 "list, do not create a wrapper for it."),
262 cl::Hidden, cl::init(Val: false));
263
264static StringRef getGlobalTypeString(const GlobalValue &G) {
265 // Types of GlobalVariables are always pointer types.
266 Type *GType = G.getValueType();
267 // For now we support excluding struct types only.
268 if (StructType *SGType = dyn_cast<StructType>(Val: GType)) {
269 if (!SGType->isLiteral())
270 return SGType->getName();
271 }
272 return "<unknown type>";
273}
274
275namespace {
276
277// Memory map parameters used in application-to-shadow address calculation.
278// Offset = (Addr & ~AndMask) ^ XorMask
279// Shadow = ShadowBase + Offset
280// Origin = (OriginBase + Offset) & ~3ULL
281struct MemoryMapParams {
282 uint64_t AndMask;
283 uint64_t XorMask;
284 uint64_t ShadowBase;
285 uint64_t OriginBase;
286};
287
288} // end anonymous namespace
289
290// NOLINTBEGIN(readability-identifier-naming)
291// aarch64 Linux
292const MemoryMapParams Linux_AArch64_MemoryMapParams = {
293 .AndMask: 0, // AndMask (not used)
294 .XorMask: 0x0B00000000000, // XorMask
295 .ShadowBase: 0, // ShadowBase (not used)
296 .OriginBase: 0x0200000000000, // OriginBase
297};
298
299// x86_64 Linux
300const MemoryMapParams Linux_X86_64_MemoryMapParams = {
301 .AndMask: 0, // AndMask (not used)
302 .XorMask: 0x500000000000, // XorMask
303 .ShadowBase: 0, // ShadowBase (not used)
304 .OriginBase: 0x100000000000, // OriginBase
305};
306// NOLINTEND(readability-identifier-naming)
307
308// loongarch64 Linux
309const MemoryMapParams Linux_LoongArch64_MemoryMapParams = {
310 .AndMask: 0, // AndMask (not used)
311 .XorMask: 0x500000000000, // XorMask
312 .ShadowBase: 0, // ShadowBase (not used)
313 .OriginBase: 0x100000000000, // OriginBase
314};
315
316namespace {
317
318class DFSanABIList {
319 std::unique_ptr<SpecialCaseList> SCL;
320
321public:
322 DFSanABIList() = default;
323
324 void set(std::unique_ptr<SpecialCaseList> List) { SCL = std::move(List); }
325
326 /// Returns whether either this function or its source file are listed in the
327 /// given category.
328 bool isIn(const Function &F, StringRef Category) const {
329 return isIn(M: *F.getParent(), Category) ||
330 SCL->inSection(Section: "dataflow", Prefix: "fun", Query: F.getName(), Category);
331 }
332
333 /// Returns whether this global alias is listed in the given category.
334 ///
335 /// If GA aliases a function, the alias's name is matched as a function name
336 /// would be. Similarly, aliases of globals are matched like globals.
337 bool isIn(const GlobalAlias &GA, StringRef Category) const {
338 if (isIn(M: *GA.getParent(), Category))
339 return true;
340
341 if (isa<FunctionType>(Val: GA.getValueType()))
342 return SCL->inSection(Section: "dataflow", Prefix: "fun", Query: GA.getName(), Category);
343
344 return SCL->inSection(Section: "dataflow", Prefix: "global", Query: GA.getName(), Category) ||
345 SCL->inSection(Section: "dataflow", Prefix: "type", Query: getGlobalTypeString(G: GA),
346 Category);
347 }
348
349 /// Returns whether this module is listed in the given category.
350 bool isIn(const Module &M, StringRef Category) const {
351 return SCL->inSection(Section: "dataflow", Prefix: "src", Query: M.getModuleIdentifier(), Category);
352 }
353};
354
355/// TransformedFunction is used to express the result of transforming one
356/// function type into another. This struct is immutable. It holds metadata
357/// useful for updating calls of the old function to the new type.
358struct TransformedFunction {
359 TransformedFunction(FunctionType *OriginalType, FunctionType *TransformedType,
360 const std::vector<unsigned> &ArgumentIndexMapping)
361 : OriginalType(OriginalType), TransformedType(TransformedType),
362 ArgumentIndexMapping(ArgumentIndexMapping) {}
363
364 // Disallow copies.
365 TransformedFunction(const TransformedFunction &) = delete;
366 TransformedFunction &operator=(const TransformedFunction &) = delete;
367
368 // Allow moves.
369 TransformedFunction(TransformedFunction &&) = default;
370 TransformedFunction &operator=(TransformedFunction &&) = default;
371
372 /// Type of the function before the transformation.
373 FunctionType *OriginalType;
374
375 /// Type of the function after the transformation.
376 FunctionType *TransformedType;
377
378 /// Transforming a function may change the position of arguments. This
379 /// member records the mapping from each argument's old position to its new
380 /// position. Argument positions are zero-indexed. If the transformation
381 /// from F to F' made the first argument of F into the third argument of F',
382 /// then ArgumentIndexMapping[0] will equal 2.
383 std::vector<unsigned> ArgumentIndexMapping;
384};
385
386/// Given function attributes from a call site for the original function,
387/// return function attributes appropriate for a call to the transformed
388/// function.
389AttributeList
390transformFunctionAttributes(const TransformedFunction &TransformedFunction,
391 LLVMContext &Ctx, AttributeList CallSiteAttrs) {
392
393 // Construct a vector of AttributeSet for each function argument.
394 std::vector<llvm::AttributeSet> ArgumentAttributes(
395 TransformedFunction.TransformedType->getNumParams());
396
397 // Copy attributes from the parameter of the original function to the
398 // transformed version. 'ArgumentIndexMapping' holds the mapping from
399 // old argument position to new.
400 for (unsigned I = 0, IE = TransformedFunction.ArgumentIndexMapping.size();
401 I < IE; ++I) {
402 unsigned TransformedIndex = TransformedFunction.ArgumentIndexMapping[I];
403 ArgumentAttributes[TransformedIndex] = CallSiteAttrs.getParamAttrs(ArgNo: I);
404 }
405
406 // Copy annotations on varargs arguments.
407 for (unsigned I = TransformedFunction.OriginalType->getNumParams(),
408 IE = CallSiteAttrs.getNumAttrSets();
409 I < IE; ++I) {
410 ArgumentAttributes.push_back(x: CallSiteAttrs.getParamAttrs(ArgNo: I));
411 }
412
413 return AttributeList::get(C&: Ctx, FnAttrs: CallSiteAttrs.getFnAttrs(),
414 RetAttrs: CallSiteAttrs.getRetAttrs(),
415 ArgAttrs: llvm::ArrayRef(ArgumentAttributes));
416}
417
418class DataFlowSanitizer {
419 friend struct DFSanFunction;
420 friend class DFSanVisitor;
421
422 enum { ShadowWidthBits = 8, ShadowWidthBytes = ShadowWidthBits / 8 };
423
424 enum { OriginWidthBits = 32, OriginWidthBytes = OriginWidthBits / 8 };
425
426 /// How should calls to uninstrumented functions be handled?
427 enum WrapperKind {
428 /// This function is present in an uninstrumented form but we don't know
429 /// how it should be handled. Print a warning and call the function anyway.
430 /// Don't label the return value.
431 WK_Warning,
432
433 /// This function does not write to (user-accessible) memory, and its return
434 /// value is unlabelled.
435 WK_Discard,
436
437 /// This function does not write to (user-accessible) memory, and the label
438 /// of its return value is the union of the label of its arguments.
439 WK_Functional,
440
441 /// Instead of calling the function, a custom wrapper __dfsw_F is called,
442 /// where F is the name of the function. This function may wrap the
443 /// original function or provide its own implementation. WK_Custom uses an
444 /// extra pointer argument to return the shadow. This allows the wrapped
445 /// form of the function type to be expressed in C.
446 WK_Custom
447 };
448
449 Module *Mod;
450 LLVMContext *Ctx;
451 Type *Int8Ptr;
452 IntegerType *OriginTy;
453 PointerType *OriginPtrTy;
454 ConstantInt *ZeroOrigin;
455 /// The shadow type for all primitive types and vector types.
456 IntegerType *PrimitiveShadowTy;
457 PointerType *PrimitiveShadowPtrTy;
458 IntegerType *IntptrTy;
459 ConstantInt *ZeroPrimitiveShadow;
460 Constant *ArgTLS;
461 ArrayType *ArgOriginTLSTy;
462 Constant *ArgOriginTLS;
463 Constant *RetvalTLS;
464 Constant *RetvalOriginTLS;
465 FunctionType *DFSanUnionLoadFnTy;
466 FunctionType *DFSanLoadLabelAndOriginFnTy;
467 FunctionType *DFSanUnimplementedFnTy;
468 FunctionType *DFSanWrapperExternWeakNullFnTy;
469 FunctionType *DFSanSetLabelFnTy;
470 FunctionType *DFSanNonzeroLabelFnTy;
471 FunctionType *DFSanVarargWrapperFnTy;
472 FunctionType *DFSanConditionalCallbackFnTy;
473 FunctionType *DFSanConditionalCallbackOriginFnTy;
474 FunctionType *DFSanReachesFunctionCallbackFnTy;
475 FunctionType *DFSanReachesFunctionCallbackOriginFnTy;
476 FunctionType *DFSanCmpCallbackFnTy;
477 FunctionType *DFSanLoadStoreCallbackFnTy;
478 FunctionType *DFSanMemTransferCallbackFnTy;
479 FunctionType *DFSanChainOriginFnTy;
480 FunctionType *DFSanChainOriginIfTaintedFnTy;
481 FunctionType *DFSanMemOriginTransferFnTy;
482 FunctionType *DFSanMemShadowOriginTransferFnTy;
483 FunctionType *DFSanMemShadowOriginConditionalExchangeFnTy;
484 FunctionType *DFSanMaybeStoreOriginFnTy;
485 FunctionCallee DFSanUnionLoadFn;
486 FunctionCallee DFSanLoadLabelAndOriginFn;
487 FunctionCallee DFSanUnimplementedFn;
488 FunctionCallee DFSanWrapperExternWeakNullFn;
489 FunctionCallee DFSanSetLabelFn;
490 FunctionCallee DFSanNonzeroLabelFn;
491 FunctionCallee DFSanVarargWrapperFn;
492 FunctionCallee DFSanLoadCallbackFn;
493 FunctionCallee DFSanStoreCallbackFn;
494 FunctionCallee DFSanMemTransferCallbackFn;
495 FunctionCallee DFSanConditionalCallbackFn;
496 FunctionCallee DFSanConditionalCallbackOriginFn;
497 FunctionCallee DFSanReachesFunctionCallbackFn;
498 FunctionCallee DFSanReachesFunctionCallbackOriginFn;
499 FunctionCallee DFSanCmpCallbackFn;
500 FunctionCallee DFSanChainOriginFn;
501 FunctionCallee DFSanChainOriginIfTaintedFn;
502 FunctionCallee DFSanMemOriginTransferFn;
503 FunctionCallee DFSanMemShadowOriginTransferFn;
504 FunctionCallee DFSanMemShadowOriginConditionalExchangeFn;
505 FunctionCallee DFSanMaybeStoreOriginFn;
506 SmallPtrSet<Value *, 16> DFSanRuntimeFunctions;
507 MDNode *ColdCallWeights;
508 MDNode *OriginStoreWeights;
509 DFSanABIList ABIList;
510 DenseMap<Value *, Function *> UnwrappedFnMap;
511 AttributeMask ReadOnlyNoneAttrs;
512 StringSet<> CombineTaintLookupTableNames;
513
514 /// Memory map parameters used in calculation mapping application addresses
515 /// to shadow addresses and origin addresses.
516 const MemoryMapParams *MapParams;
517
518 Value *getShadowOffset(Value *Addr, IRBuilder<> &IRB);
519 Value *getShadowAddress(Value *Addr, BasicBlock::iterator Pos);
520 Value *getShadowAddress(Value *Addr, BasicBlock::iterator Pos,
521 Value *ShadowOffset);
522 std::pair<Value *, Value *> getShadowOriginAddress(Value *Addr,
523 Align InstAlignment,
524 BasicBlock::iterator Pos);
525 bool isInstrumented(const Function *F);
526 bool isInstrumented(const GlobalAlias *GA);
527 bool isForceZeroLabels(const Function *F);
528 TransformedFunction getCustomFunctionType(FunctionType *T);
529 WrapperKind getWrapperKind(Function *F);
530 void addGlobalNameSuffix(GlobalValue *GV);
531 void buildExternWeakCheckIfNeeded(IRBuilder<> &IRB, Function *F);
532 Function *buildWrapperFunction(Function *F, StringRef NewFName,
533 GlobalValue::LinkageTypes NewFLink,
534 FunctionType *NewFT);
535 void initializeCallbackFunctions(Module &M);
536 void initializeRuntimeFunctions(Module &M);
537 bool initializeModule(Module &M);
538
539 /// Advances \p OriginAddr to point to the next 32-bit origin and then loads
540 /// from it. Returns the origin's loaded value.
541 Value *loadNextOrigin(BasicBlock::iterator Pos, Align OriginAlign,
542 Value **OriginAddr);
543
544 /// Returns whether the given load byte size is amenable to inlined
545 /// optimization patterns.
546 bool hasLoadSizeForFastPath(uint64_t Size);
547
548 /// Returns whether the pass tracks origins. Supports only TLS ABI mode.
549 bool shouldTrackOrigins();
550
551 /// Returns a zero constant with the shadow type of OrigTy.
552 ///
553 /// getZeroShadow({T1,T2,...}) = {getZeroShadow(T1),getZeroShadow(T2,...}
554 /// getZeroShadow([n x T]) = [n x getZeroShadow(T)]
555 /// getZeroShadow(other type) = i16(0)
556 Constant *getZeroShadow(Type *OrigTy);
557 /// Returns a zero constant with the shadow type of V's type.
558 Constant *getZeroShadow(Value *V);
559
560 /// Checks if V is a zero shadow.
561 bool isZeroShadow(Value *V);
562
563 /// Returns the shadow type of OrigTy.
564 ///
565 /// getShadowTy({T1,T2,...}) = {getShadowTy(T1),getShadowTy(T2),...}
566 /// getShadowTy([n x T]) = [n x getShadowTy(T)]
567 /// getShadowTy(other type) = i16
568 Type *getShadowTy(Type *OrigTy);
569 /// Returns the shadow type of V's type.
570 Type *getShadowTy(Value *V);
571
572 const uint64_t NumOfElementsInArgOrgTLS = ArgTLSSize / OriginWidthBytes;
573
574public:
575 DataFlowSanitizer(const std::vector<std::string> &ABIListFiles);
576
577 bool runImpl(Module &M,
578 llvm::function_ref<TargetLibraryInfo &(Function &)> GetTLI);
579};
580
581struct DFSanFunction {
582 DataFlowSanitizer &DFS;
583 Function *F;
584 DominatorTree DT;
585 bool IsNativeABI;
586 bool IsForceZeroLabels;
587 TargetLibraryInfo &TLI;
588 AllocaInst *LabelReturnAlloca = nullptr;
589 AllocaInst *OriginReturnAlloca = nullptr;
590 DenseMap<Value *, Value *> ValShadowMap;
591 DenseMap<Value *, Value *> ValOriginMap;
592 DenseMap<AllocaInst *, AllocaInst *> AllocaShadowMap;
593 DenseMap<AllocaInst *, AllocaInst *> AllocaOriginMap;
594
595 struct PHIFixupElement {
596 PHINode *Phi;
597 PHINode *ShadowPhi;
598 PHINode *OriginPhi;
599 };
600 std::vector<PHIFixupElement> PHIFixups;
601
602 DenseSet<Instruction *> SkipInsts;
603 std::vector<Value *> NonZeroChecks;
604
605 struct CachedShadow {
606 BasicBlock *Block; // The block where Shadow is defined.
607 Value *Shadow;
608 };
609 /// Maps a value to its latest shadow value in terms of domination tree.
610 DenseMap<std::pair<Value *, Value *>, CachedShadow> CachedShadows;
611 /// Maps a value to its latest collapsed shadow value it was converted to in
612 /// terms of domination tree. When ClDebugNonzeroLabels is on, this cache is
613 /// used at a post process where CFG blocks are split. So it does not cache
614 /// BasicBlock like CachedShadows, but uses domination between values.
615 DenseMap<Value *, Value *> CachedCollapsedShadows;
616 DenseMap<Value *, std::set<Value *>> ShadowElements;
617
618 DFSanFunction(DataFlowSanitizer &DFS, Function *F, bool IsNativeABI,
619 bool IsForceZeroLabels, TargetLibraryInfo &TLI)
620 : DFS(DFS), F(F), IsNativeABI(IsNativeABI),
621 IsForceZeroLabels(IsForceZeroLabels), TLI(TLI) {
622 DT.recalculate(Func&: *F);
623 }
624
625 /// Computes the shadow address for a given function argument.
626 ///
627 /// Shadow = ArgTLS+ArgOffset.
628 Value *getArgTLS(Type *T, unsigned ArgOffset, IRBuilder<> &IRB);
629
630 /// Computes the shadow address for a return value.
631 Value *getRetvalTLS(Type *T, IRBuilder<> &IRB);
632
633 /// Computes the origin address for a given function argument.
634 ///
635 /// Origin = ArgOriginTLS[ArgNo].
636 Value *getArgOriginTLS(unsigned ArgNo, IRBuilder<> &IRB);
637
638 /// Computes the origin address for a return value.
639 Value *getRetvalOriginTLS();
640
641 Value *getOrigin(Value *V);
642 void setOrigin(Instruction *I, Value *Origin);
643 /// Generates IR to compute the origin of the last operand with a taint label.
644 Value *combineOperandOrigins(Instruction *Inst);
645 /// Before the instruction Pos, generates IR to compute the last origin with a
646 /// taint label. Labels and origins are from vectors Shadows and Origins
647 /// correspondingly. The generated IR is like
648 /// Sn-1 != Zero ? On-1: ... S2 != Zero ? O2: S1 != Zero ? O1: O0
649 /// When Zero is nullptr, it uses ZeroPrimitiveShadow. Otherwise it can be
650 /// zeros with other bitwidths.
651 Value *combineOrigins(const std::vector<Value *> &Shadows,
652 const std::vector<Value *> &Origins,
653 BasicBlock::iterator Pos, ConstantInt *Zero = nullptr);
654
655 Value *getShadow(Value *V);
656 void setShadow(Instruction *I, Value *Shadow);
657 /// Generates IR to compute the union of the two given shadows, inserting it
658 /// before Pos. The combined value is with primitive type.
659 Value *combineShadows(Value *V1, Value *V2, BasicBlock::iterator Pos);
660 /// Combines the shadow values of V1 and V2, then converts the combined value
661 /// with primitive type into a shadow value with the original type T.
662 Value *combineShadowsThenConvert(Type *T, Value *V1, Value *V2,
663 BasicBlock::iterator Pos);
664 Value *combineOperandShadows(Instruction *Inst);
665
666 /// Generates IR to load shadow and origin corresponding to bytes [\p
667 /// Addr, \p Addr + \p Size), where addr has alignment \p
668 /// InstAlignment, and take the union of each of those shadows. The returned
669 /// shadow always has primitive type.
670 ///
671 /// When tracking loads is enabled, the returned origin is a chain at the
672 /// current stack if the returned shadow is tainted.
673 std::pair<Value *, Value *> loadShadowOrigin(Value *Addr, uint64_t Size,
674 Align InstAlignment,
675 BasicBlock::iterator Pos);
676
677 void storePrimitiveShadowOrigin(Value *Addr, uint64_t Size,
678 Align InstAlignment, Value *PrimitiveShadow,
679 Value *Origin, BasicBlock::iterator Pos);
680 /// Applies PrimitiveShadow to all primitive subtypes of T, returning
681 /// the expanded shadow value.
682 ///
683 /// EFP({T1,T2, ...}, PS) = {EFP(T1,PS),EFP(T2,PS),...}
684 /// EFP([n x T], PS) = [n x EFP(T,PS)]
685 /// EFP(other types, PS) = PS
686 Value *expandFromPrimitiveShadow(Type *T, Value *PrimitiveShadow,
687 BasicBlock::iterator Pos);
688 /// Collapses Shadow into a single primitive shadow value, unioning all
689 /// primitive shadow values in the process. Returns the final primitive
690 /// shadow value.
691 ///
692 /// CTP({V1,V2, ...}) = UNION(CFP(V1,PS),CFP(V2,PS),...)
693 /// CTP([V1,V2,...]) = UNION(CFP(V1,PS),CFP(V2,PS),...)
694 /// CTP(other types, PS) = PS
695 Value *collapseToPrimitiveShadow(Value *Shadow, BasicBlock::iterator Pos);
696
697 void storeZeroPrimitiveShadow(Value *Addr, uint64_t Size, Align ShadowAlign,
698 BasicBlock::iterator Pos);
699
700 Align getShadowAlign(Align InstAlignment);
701
702 // If ClConditionalCallbacks is enabled, insert a callback after a given
703 // branch instruction using the given conditional expression.
704 void addConditionalCallbacksIfEnabled(Instruction &I, Value *Condition);
705
706 // If ClReachesFunctionCallbacks is enabled, insert a callback for each
707 // argument and load instruction.
708 void addReachesFunctionCallbacksIfEnabled(IRBuilder<> &IRB, Instruction &I,
709 Value *Data);
710
711 bool isLookupTableConstant(Value *P);
712
713private:
714 /// Collapses the shadow with aggregate type into a single primitive shadow
715 /// value.
716 template <class AggregateType>
717 Value *collapseAggregateShadow(AggregateType *AT, Value *Shadow,
718 IRBuilder<> &IRB);
719
720 Value *collapseToPrimitiveShadow(Value *Shadow, IRBuilder<> &IRB);
721
722 /// Returns the shadow value of an argument A.
723 Value *getShadowForTLSArgument(Argument *A);
724
725 /// The fast path of loading shadows.
726 std::pair<Value *, Value *>
727 loadShadowFast(Value *ShadowAddr, Value *OriginAddr, uint64_t Size,
728 Align ShadowAlign, Align OriginAlign, Value *FirstOrigin,
729 BasicBlock::iterator Pos);
730
731 Align getOriginAlign(Align InstAlignment);
732
733 /// Because 4 contiguous bytes share one 4-byte origin, the most accurate load
734 /// is __dfsan_load_label_and_origin. This function returns the union of all
735 /// labels and the origin of the first taint label. However this is an
736 /// additional call with many instructions. To ensure common cases are fast,
737 /// checks if it is possible to load labels and origins without using the
738 /// callback function.
739 ///
740 /// When enabling tracking load instructions, we always use
741 /// __dfsan_load_label_and_origin to reduce code size.
742 bool useCallbackLoadLabelAndOrigin(uint64_t Size, Align InstAlignment);
743
744 /// Returns a chain at the current stack with previous origin V.
745 Value *updateOrigin(Value *V, IRBuilder<> &IRB);
746
747 /// Returns a chain at the current stack with previous origin V if Shadow is
748 /// tainted.
749 Value *updateOriginIfTainted(Value *Shadow, Value *Origin, IRBuilder<> &IRB);
750
751 /// Creates an Intptr = Origin | Origin << 32 if Intptr's size is 64. Returns
752 /// Origin otherwise.
753 Value *originToIntptr(IRBuilder<> &IRB, Value *Origin);
754
755 /// Stores Origin into the address range [StoreOriginAddr, StoreOriginAddr +
756 /// Size).
757 void paintOrigin(IRBuilder<> &IRB, Value *Origin, Value *StoreOriginAddr,
758 uint64_t StoreOriginSize, Align Alignment);
759
760 /// Stores Origin in terms of its Shadow value.
761 /// * Do not write origins for zero shadows because we do not trace origins
762 /// for untainted sinks.
763 /// * Use __dfsan_maybe_store_origin if there are too many origin store
764 /// instrumentations.
765 void storeOrigin(BasicBlock::iterator Pos, Value *Addr, uint64_t Size,
766 Value *Shadow, Value *Origin, Value *StoreOriginAddr,
767 Align InstAlignment);
768
769 /// Convert a scalar value to an i1 by comparing with 0.
770 Value *convertToBool(Value *V, IRBuilder<> &IRB, const Twine &Name = "");
771
772 bool shouldInstrumentWithCall();
773
774 /// Generates IR to load shadow and origin corresponding to bytes [\p
775 /// Addr, \p Addr + \p Size), where addr has alignment \p
776 /// InstAlignment, and take the union of each of those shadows. The returned
777 /// shadow always has primitive type.
778 std::pair<Value *, Value *>
779 loadShadowOriginSansLoadTracking(Value *Addr, uint64_t Size,
780 Align InstAlignment,
781 BasicBlock::iterator Pos);
782 int NumOriginStores = 0;
783};
784
785class DFSanVisitor : public InstVisitor<DFSanVisitor> {
786public:
787 DFSanFunction &DFSF;
788
789 DFSanVisitor(DFSanFunction &DFSF) : DFSF(DFSF) {}
790
791 const DataLayout &getDataLayout() const {
792 return DFSF.F->getDataLayout();
793 }
794
795 // Combines shadow values and origins for all of I's operands.
796 void visitInstOperands(Instruction &I);
797
798 void visitUnaryOperator(UnaryOperator &UO);
799 void visitBinaryOperator(BinaryOperator &BO);
800 void visitBitCastInst(BitCastInst &BCI);
801 void visitCastInst(CastInst &CI);
802 void visitCmpInst(CmpInst &CI);
803 void visitLandingPadInst(LandingPadInst &LPI);
804 void visitGetElementPtrInst(GetElementPtrInst &GEPI);
805 void visitLoadInst(LoadInst &LI);
806 void visitStoreInst(StoreInst &SI);
807 void visitAtomicRMWInst(AtomicRMWInst &I);
808 void visitAtomicCmpXchgInst(AtomicCmpXchgInst &I);
809 void visitReturnInst(ReturnInst &RI);
810 void visitLibAtomicLoad(CallBase &CB);
811 void visitLibAtomicStore(CallBase &CB);
812 void visitLibAtomicExchange(CallBase &CB);
813 void visitLibAtomicCompareExchange(CallBase &CB);
814 void visitCallBase(CallBase &CB);
815 void visitPHINode(PHINode &PN);
816 void visitExtractElementInst(ExtractElementInst &I);
817 void visitInsertElementInst(InsertElementInst &I);
818 void visitShuffleVectorInst(ShuffleVectorInst &I);
819 void visitExtractValueInst(ExtractValueInst &I);
820 void visitInsertValueInst(InsertValueInst &I);
821 void visitAllocaInst(AllocaInst &I);
822 void visitSelectInst(SelectInst &I);
823 void visitMemSetInst(MemSetInst &I);
824 void visitMemTransferInst(MemTransferInst &I);
825 void visitBranchInst(BranchInst &BR);
826 void visitSwitchInst(SwitchInst &SW);
827
828private:
829 void visitCASOrRMW(Align InstAlignment, Instruction &I);
830
831 // Returns false when this is an invoke of a custom function.
832 bool visitWrappedCallBase(Function &F, CallBase &CB);
833
834 // Combines origins for all of I's operands.
835 void visitInstOperandOrigins(Instruction &I);
836
837 void addShadowArguments(Function &F, CallBase &CB, std::vector<Value *> &Args,
838 IRBuilder<> &IRB);
839
840 void addOriginArguments(Function &F, CallBase &CB, std::vector<Value *> &Args,
841 IRBuilder<> &IRB);
842
843 Value *makeAddAcquireOrderingTable(IRBuilder<> &IRB);
844 Value *makeAddReleaseOrderingTable(IRBuilder<> &IRB);
845};
846
847bool LibAtomicFunction(const Function &F) {
848 // This is a bit of a hack because TargetLibraryInfo is a function pass.
849 // The DFSan pass would need to be refactored to be function pass oriented
850 // (like MSan is) in order to fit together nicely with TargetLibraryInfo.
851 // We need this check to prevent them from being instrumented, or wrapped.
852 // Match on name and number of arguments.
853 if (!F.hasName() || F.isVarArg())
854 return false;
855 switch (F.arg_size()) {
856 case 4:
857 return F.getName() == "__atomic_load" || F.getName() == "__atomic_store";
858 case 5:
859 return F.getName() == "__atomic_exchange";
860 case 6:
861 return F.getName() == "__atomic_compare_exchange";
862 default:
863 return false;
864 }
865}
866
867} // end anonymous namespace
868
869DataFlowSanitizer::DataFlowSanitizer(
870 const std::vector<std::string> &ABIListFiles) {
871 std::vector<std::string> AllABIListFiles(std::move(ABIListFiles));
872 llvm::append_range(C&: AllABIListFiles, R&: ClABIListFiles);
873 // FIXME: should we propagate vfs::FileSystem to this constructor?
874 ABIList.set(
875 SpecialCaseList::createOrDie(Paths: AllABIListFiles, FS&: *vfs::getRealFileSystem()));
876
877 for (StringRef v : ClCombineTaintLookupTables)
878 CombineTaintLookupTableNames.insert(key: v);
879}
880
881TransformedFunction DataFlowSanitizer::getCustomFunctionType(FunctionType *T) {
882 SmallVector<Type *, 4> ArgTypes;
883
884 // Some parameters of the custom function being constructed are
885 // parameters of T. Record the mapping from parameters of T to
886 // parameters of the custom function, so that parameter attributes
887 // at call sites can be updated.
888 std::vector<unsigned> ArgumentIndexMapping;
889 for (unsigned I = 0, E = T->getNumParams(); I != E; ++I) {
890 Type *ParamType = T->getParamType(i: I);
891 ArgumentIndexMapping.push_back(x: ArgTypes.size());
892 ArgTypes.push_back(Elt: ParamType);
893 }
894 for (unsigned I = 0, E = T->getNumParams(); I != E; ++I)
895 ArgTypes.push_back(Elt: PrimitiveShadowTy);
896 if (T->isVarArg())
897 ArgTypes.push_back(Elt: PrimitiveShadowPtrTy);
898 Type *RetType = T->getReturnType();
899 if (!RetType->isVoidTy())
900 ArgTypes.push_back(Elt: PrimitiveShadowPtrTy);
901
902 if (shouldTrackOrigins()) {
903 for (unsigned I = 0, E = T->getNumParams(); I != E; ++I)
904 ArgTypes.push_back(Elt: OriginTy);
905 if (T->isVarArg())
906 ArgTypes.push_back(Elt: OriginPtrTy);
907 if (!RetType->isVoidTy())
908 ArgTypes.push_back(Elt: OriginPtrTy);
909 }
910
911 return TransformedFunction(
912 T, FunctionType::get(Result: T->getReturnType(), Params: ArgTypes, isVarArg: T->isVarArg()),
913 ArgumentIndexMapping);
914}
915
916bool DataFlowSanitizer::isZeroShadow(Value *V) {
917 Type *T = V->getType();
918 if (!isa<ArrayType>(Val: T) && !isa<StructType>(Val: T)) {
919 if (const ConstantInt *CI = dyn_cast<ConstantInt>(Val: V))
920 return CI->isZero();
921 return false;
922 }
923
924 return isa<ConstantAggregateZero>(Val: V);
925}
926
927bool DataFlowSanitizer::hasLoadSizeForFastPath(uint64_t Size) {
928 uint64_t ShadowSize = Size * ShadowWidthBytes;
929 return ShadowSize % 8 == 0 || ShadowSize == 4;
930}
931
932bool DataFlowSanitizer::shouldTrackOrigins() {
933 static const bool ShouldTrackOrigins = ClTrackOrigins;
934 return ShouldTrackOrigins;
935}
936
937Constant *DataFlowSanitizer::getZeroShadow(Type *OrigTy) {
938 if (!isa<ArrayType>(Val: OrigTy) && !isa<StructType>(Val: OrigTy))
939 return ZeroPrimitiveShadow;
940 Type *ShadowTy = getShadowTy(OrigTy);
941 return ConstantAggregateZero::get(Ty: ShadowTy);
942}
943
944Constant *DataFlowSanitizer::getZeroShadow(Value *V) {
945 return getZeroShadow(OrigTy: V->getType());
946}
947
948static Value *expandFromPrimitiveShadowRecursive(
949 Value *Shadow, SmallVector<unsigned, 4> &Indices, Type *SubShadowTy,
950 Value *PrimitiveShadow, IRBuilder<> &IRB) {
951 if (!isa<ArrayType>(Val: SubShadowTy) && !isa<StructType>(Val: SubShadowTy))
952 return IRB.CreateInsertValue(Agg: Shadow, Val: PrimitiveShadow, Idxs: Indices);
953
954 if (ArrayType *AT = dyn_cast<ArrayType>(Val: SubShadowTy)) {
955 for (unsigned Idx = 0; Idx < AT->getNumElements(); Idx++) {
956 Indices.push_back(Elt: Idx);
957 Shadow = expandFromPrimitiveShadowRecursive(
958 Shadow, Indices, SubShadowTy: AT->getElementType(), PrimitiveShadow, IRB);
959 Indices.pop_back();
960 }
961 return Shadow;
962 }
963
964 if (StructType *ST = dyn_cast<StructType>(Val: SubShadowTy)) {
965 for (unsigned Idx = 0; Idx < ST->getNumElements(); Idx++) {
966 Indices.push_back(Elt: Idx);
967 Shadow = expandFromPrimitiveShadowRecursive(
968 Shadow, Indices, SubShadowTy: ST->getElementType(N: Idx), PrimitiveShadow, IRB);
969 Indices.pop_back();
970 }
971 return Shadow;
972 }
973 llvm_unreachable("Unexpected shadow type");
974}
975
976bool DFSanFunction::shouldInstrumentWithCall() {
977 return ClInstrumentWithCallThreshold >= 0 &&
978 NumOriginStores >= ClInstrumentWithCallThreshold;
979}
980
981Value *DFSanFunction::expandFromPrimitiveShadow(Type *T, Value *PrimitiveShadow,
982 BasicBlock::iterator Pos) {
983 Type *ShadowTy = DFS.getShadowTy(OrigTy: T);
984
985 if (!isa<ArrayType>(Val: ShadowTy) && !isa<StructType>(Val: ShadowTy))
986 return PrimitiveShadow;
987
988 if (DFS.isZeroShadow(V: PrimitiveShadow))
989 return DFS.getZeroShadow(OrigTy: ShadowTy);
990
991 IRBuilder<> IRB(Pos->getParent(), Pos);
992 SmallVector<unsigned, 4> Indices;
993 Value *Shadow = UndefValue::get(T: ShadowTy);
994 Shadow = expandFromPrimitiveShadowRecursive(Shadow, Indices, SubShadowTy: ShadowTy,
995 PrimitiveShadow, IRB);
996
997 // Caches the primitive shadow value that built the shadow value.
998 CachedCollapsedShadows[Shadow] = PrimitiveShadow;
999 return Shadow;
1000}
1001
1002template <class AggregateType>
1003Value *DFSanFunction::collapseAggregateShadow(AggregateType *AT, Value *Shadow,
1004 IRBuilder<> &IRB) {
1005 if (!AT->getNumElements())
1006 return DFS.ZeroPrimitiveShadow;
1007
1008 Value *FirstItem = IRB.CreateExtractValue(Agg: Shadow, Idxs: 0);
1009 Value *Aggregator = collapseToPrimitiveShadow(Shadow: FirstItem, IRB);
1010
1011 for (unsigned Idx = 1; Idx < AT->getNumElements(); Idx++) {
1012 Value *ShadowItem = IRB.CreateExtractValue(Agg: Shadow, Idxs: Idx);
1013 Value *ShadowInner = collapseToPrimitiveShadow(Shadow: ShadowItem, IRB);
1014 Aggregator = IRB.CreateOr(LHS: Aggregator, RHS: ShadowInner);
1015 }
1016 return Aggregator;
1017}
1018
1019Value *DFSanFunction::collapseToPrimitiveShadow(Value *Shadow,
1020 IRBuilder<> &IRB) {
1021 Type *ShadowTy = Shadow->getType();
1022 if (!isa<ArrayType>(Val: ShadowTy) && !isa<StructType>(Val: ShadowTy))
1023 return Shadow;
1024 if (ArrayType *AT = dyn_cast<ArrayType>(Val: ShadowTy))
1025 return collapseAggregateShadow<>(AT, Shadow, IRB);
1026 if (StructType *ST = dyn_cast<StructType>(Val: ShadowTy))
1027 return collapseAggregateShadow<>(AT: ST, Shadow, IRB);
1028 llvm_unreachable("Unexpected shadow type");
1029}
1030
1031Value *DFSanFunction::collapseToPrimitiveShadow(Value *Shadow,
1032 BasicBlock::iterator Pos) {
1033 Type *ShadowTy = Shadow->getType();
1034 if (!isa<ArrayType>(Val: ShadowTy) && !isa<StructType>(Val: ShadowTy))
1035 return Shadow;
1036
1037 // Checks if the cached collapsed shadow value dominates Pos.
1038 Value *&CS = CachedCollapsedShadows[Shadow];
1039 if (CS && DT.dominates(Def: CS, User: Pos))
1040 return CS;
1041
1042 IRBuilder<> IRB(Pos->getParent(), Pos);
1043 Value *PrimitiveShadow = collapseToPrimitiveShadow(Shadow, IRB);
1044 // Caches the converted primitive shadow value.
1045 CS = PrimitiveShadow;
1046 return PrimitiveShadow;
1047}
1048
1049void DFSanFunction::addConditionalCallbacksIfEnabled(Instruction &I,
1050 Value *Condition) {
1051 if (!ClConditionalCallbacks) {
1052 return;
1053 }
1054 IRBuilder<> IRB(&I);
1055 Value *CondShadow = getShadow(V: Condition);
1056 CallInst *CI;
1057 if (DFS.shouldTrackOrigins()) {
1058 Value *CondOrigin = getOrigin(V: Condition);
1059 CI = IRB.CreateCall(Callee: DFS.DFSanConditionalCallbackOriginFn,
1060 Args: {CondShadow, CondOrigin});
1061 } else {
1062 CI = IRB.CreateCall(Callee: DFS.DFSanConditionalCallbackFn, Args: {CondShadow});
1063 }
1064 CI->addParamAttr(ArgNo: 0, Kind: Attribute::ZExt);
1065}
1066
1067void DFSanFunction::addReachesFunctionCallbacksIfEnabled(IRBuilder<> &IRB,
1068 Instruction &I,
1069 Value *Data) {
1070 if (!ClReachesFunctionCallbacks) {
1071 return;
1072 }
1073 const DebugLoc &dbgloc = I.getDebugLoc();
1074 Value *DataShadow = collapseToPrimitiveShadow(Shadow: getShadow(V: Data), IRB);
1075 ConstantInt *CILine;
1076 llvm::Value *FilePathPtr;
1077
1078 if (dbgloc.get() == nullptr) {
1079 CILine = llvm::ConstantInt::get(Context&: I.getContext(), V: llvm::APInt(32, 0));
1080 FilePathPtr = IRB.CreateGlobalStringPtr(
1081 Str: I.getFunction()->getParent()->getSourceFileName());
1082 } else {
1083 CILine = llvm::ConstantInt::get(Context&: I.getContext(),
1084 V: llvm::APInt(32, dbgloc.getLine()));
1085 FilePathPtr =
1086 IRB.CreateGlobalStringPtr(Str: dbgloc->getFilename());
1087 }
1088
1089 llvm::Value *FunctionNamePtr =
1090 IRB.CreateGlobalStringPtr(Str: I.getFunction()->getName());
1091
1092 CallInst *CB;
1093 std::vector<Value *> args;
1094
1095 if (DFS.shouldTrackOrigins()) {
1096 Value *DataOrigin = getOrigin(V: Data);
1097 args = { DataShadow, DataOrigin, FilePathPtr, CILine, FunctionNamePtr };
1098 CB = IRB.CreateCall(Callee: DFS.DFSanReachesFunctionCallbackOriginFn, Args: args);
1099 } else {
1100 args = { DataShadow, FilePathPtr, CILine, FunctionNamePtr };
1101 CB = IRB.CreateCall(Callee: DFS.DFSanReachesFunctionCallbackFn, Args: args);
1102 }
1103 CB->addParamAttr(ArgNo: 0, Kind: Attribute::ZExt);
1104 CB->setDebugLoc(dbgloc);
1105}
1106
1107Type *DataFlowSanitizer::getShadowTy(Type *OrigTy) {
1108 if (!OrigTy->isSized())
1109 return PrimitiveShadowTy;
1110 if (isa<IntegerType>(Val: OrigTy))
1111 return PrimitiveShadowTy;
1112 if (isa<VectorType>(Val: OrigTy))
1113 return PrimitiveShadowTy;
1114 if (ArrayType *AT = dyn_cast<ArrayType>(Val: OrigTy))
1115 return ArrayType::get(ElementType: getShadowTy(OrigTy: AT->getElementType()),
1116 NumElements: AT->getNumElements());
1117 if (StructType *ST = dyn_cast<StructType>(Val: OrigTy)) {
1118 SmallVector<Type *, 4> Elements;
1119 for (unsigned I = 0, N = ST->getNumElements(); I < N; ++I)
1120 Elements.push_back(Elt: getShadowTy(OrigTy: ST->getElementType(N: I)));
1121 return StructType::get(Context&: *Ctx, Elements);
1122 }
1123 return PrimitiveShadowTy;
1124}
1125
1126Type *DataFlowSanitizer::getShadowTy(Value *V) {
1127 return getShadowTy(OrigTy: V->getType());
1128}
1129
1130bool DataFlowSanitizer::initializeModule(Module &M) {
1131 Triple TargetTriple(M.getTargetTriple());
1132 const DataLayout &DL = M.getDataLayout();
1133
1134 if (TargetTriple.getOS() != Triple::Linux)
1135 report_fatal_error(reason: "unsupported operating system");
1136 switch (TargetTriple.getArch()) {
1137 case Triple::aarch64:
1138 MapParams = &Linux_AArch64_MemoryMapParams;
1139 break;
1140 case Triple::x86_64:
1141 MapParams = &Linux_X86_64_MemoryMapParams;
1142 break;
1143 case Triple::loongarch64:
1144 MapParams = &Linux_LoongArch64_MemoryMapParams;
1145 break;
1146 default:
1147 report_fatal_error(reason: "unsupported architecture");
1148 }
1149
1150 Mod = &M;
1151 Ctx = &M.getContext();
1152 Int8Ptr = PointerType::getUnqual(C&: *Ctx);
1153 OriginTy = IntegerType::get(C&: *Ctx, NumBits: OriginWidthBits);
1154 OriginPtrTy = PointerType::getUnqual(ElementType: OriginTy);
1155 PrimitiveShadowTy = IntegerType::get(C&: *Ctx, NumBits: ShadowWidthBits);
1156 PrimitiveShadowPtrTy = PointerType::getUnqual(ElementType: PrimitiveShadowTy);
1157 IntptrTy = DL.getIntPtrType(C&: *Ctx);
1158 ZeroPrimitiveShadow = ConstantInt::getSigned(Ty: PrimitiveShadowTy, V: 0);
1159 ZeroOrigin = ConstantInt::getSigned(Ty: OriginTy, V: 0);
1160
1161 Type *DFSanUnionLoadArgs[2] = {PrimitiveShadowPtrTy, IntptrTy};
1162 DFSanUnionLoadFnTy = FunctionType::get(Result: PrimitiveShadowTy, Params: DFSanUnionLoadArgs,
1163 /*isVarArg=*/false);
1164 Type *DFSanLoadLabelAndOriginArgs[2] = {Int8Ptr, IntptrTy};
1165 DFSanLoadLabelAndOriginFnTy =
1166 FunctionType::get(Result: IntegerType::get(C&: *Ctx, NumBits: 64), Params: DFSanLoadLabelAndOriginArgs,
1167 /*isVarArg=*/false);
1168 DFSanUnimplementedFnTy = FunctionType::get(
1169 Result: Type::getVoidTy(C&: *Ctx), Params: PointerType::getUnqual(C&: *Ctx), /*isVarArg=*/false);
1170 Type *DFSanWrapperExternWeakNullArgs[2] = {Int8Ptr, Int8Ptr};
1171 DFSanWrapperExternWeakNullFnTy =
1172 FunctionType::get(Result: Type::getVoidTy(C&: *Ctx), Params: DFSanWrapperExternWeakNullArgs,
1173 /*isVarArg=*/false);
1174 Type *DFSanSetLabelArgs[4] = {PrimitiveShadowTy, OriginTy,
1175 PointerType::getUnqual(C&: *Ctx), IntptrTy};
1176 DFSanSetLabelFnTy = FunctionType::get(Result: Type::getVoidTy(C&: *Ctx),
1177 Params: DFSanSetLabelArgs, /*isVarArg=*/false);
1178 DFSanNonzeroLabelFnTy = FunctionType::get(Result: Type::getVoidTy(C&: *Ctx), Params: std::nullopt,
1179 /*isVarArg=*/false);
1180 DFSanVarargWrapperFnTy = FunctionType::get(
1181 Result: Type::getVoidTy(C&: *Ctx), Params: PointerType::getUnqual(C&: *Ctx), /*isVarArg=*/false);
1182 DFSanConditionalCallbackFnTy =
1183 FunctionType::get(Result: Type::getVoidTy(C&: *Ctx), Params: PrimitiveShadowTy,
1184 /*isVarArg=*/false);
1185 Type *DFSanConditionalCallbackOriginArgs[2] = {PrimitiveShadowTy, OriginTy};
1186 DFSanConditionalCallbackOriginFnTy = FunctionType::get(
1187 Result: Type::getVoidTy(C&: *Ctx), Params: DFSanConditionalCallbackOriginArgs,
1188 /*isVarArg=*/false);
1189 Type *DFSanReachesFunctionCallbackArgs[4] = {PrimitiveShadowTy, Int8Ptr,
1190 OriginTy, Int8Ptr};
1191 DFSanReachesFunctionCallbackFnTy =
1192 FunctionType::get(Result: Type::getVoidTy(C&: *Ctx), Params: DFSanReachesFunctionCallbackArgs,
1193 /*isVarArg=*/false);
1194 Type *DFSanReachesFunctionCallbackOriginArgs[5] = {
1195 PrimitiveShadowTy, OriginTy, Int8Ptr, OriginTy, Int8Ptr};
1196 DFSanReachesFunctionCallbackOriginFnTy = FunctionType::get(
1197 Result: Type::getVoidTy(C&: *Ctx), Params: DFSanReachesFunctionCallbackOriginArgs,
1198 /*isVarArg=*/false);
1199 DFSanCmpCallbackFnTy =
1200 FunctionType::get(Result: Type::getVoidTy(C&: *Ctx), Params: PrimitiveShadowTy,
1201 /*isVarArg=*/false);
1202 DFSanChainOriginFnTy =
1203 FunctionType::get(Result: OriginTy, Params: OriginTy, /*isVarArg=*/false);
1204 Type *DFSanChainOriginIfTaintedArgs[2] = {PrimitiveShadowTy, OriginTy};
1205 DFSanChainOriginIfTaintedFnTy = FunctionType::get(
1206 Result: OriginTy, Params: DFSanChainOriginIfTaintedArgs, /*isVarArg=*/false);
1207 Type *DFSanMaybeStoreOriginArgs[4] = {IntegerType::get(C&: *Ctx, NumBits: ShadowWidthBits),
1208 Int8Ptr, IntptrTy, OriginTy};
1209 DFSanMaybeStoreOriginFnTy = FunctionType::get(
1210 Result: Type::getVoidTy(C&: *Ctx), Params: DFSanMaybeStoreOriginArgs, /*isVarArg=*/false);
1211 Type *DFSanMemOriginTransferArgs[3] = {Int8Ptr, Int8Ptr, IntptrTy};
1212 DFSanMemOriginTransferFnTy = FunctionType::get(
1213 Result: Type::getVoidTy(C&: *Ctx), Params: DFSanMemOriginTransferArgs, /*isVarArg=*/false);
1214 Type *DFSanMemShadowOriginTransferArgs[3] = {Int8Ptr, Int8Ptr, IntptrTy};
1215 DFSanMemShadowOriginTransferFnTy =
1216 FunctionType::get(Result: Type::getVoidTy(C&: *Ctx), Params: DFSanMemShadowOriginTransferArgs,
1217 /*isVarArg=*/false);
1218 Type *DFSanMemShadowOriginConditionalExchangeArgs[5] = {
1219 IntegerType::get(C&: *Ctx, NumBits: 8), Int8Ptr, Int8Ptr, Int8Ptr, IntptrTy};
1220 DFSanMemShadowOriginConditionalExchangeFnTy = FunctionType::get(
1221 Result: Type::getVoidTy(C&: *Ctx), Params: DFSanMemShadowOriginConditionalExchangeArgs,
1222 /*isVarArg=*/false);
1223 Type *DFSanLoadStoreCallbackArgs[2] = {PrimitiveShadowTy, Int8Ptr};
1224 DFSanLoadStoreCallbackFnTy =
1225 FunctionType::get(Result: Type::getVoidTy(C&: *Ctx), Params: DFSanLoadStoreCallbackArgs,
1226 /*isVarArg=*/false);
1227 Type *DFSanMemTransferCallbackArgs[2] = {PrimitiveShadowPtrTy, IntptrTy};
1228 DFSanMemTransferCallbackFnTy =
1229 FunctionType::get(Result: Type::getVoidTy(C&: *Ctx), Params: DFSanMemTransferCallbackArgs,
1230 /*isVarArg=*/false);
1231
1232 ColdCallWeights = MDBuilder(*Ctx).createUnlikelyBranchWeights();
1233 OriginStoreWeights = MDBuilder(*Ctx).createUnlikelyBranchWeights();
1234 return true;
1235}
1236
1237bool DataFlowSanitizer::isInstrumented(const Function *F) {
1238 return !ABIList.isIn(F: *F, Category: "uninstrumented");
1239}
1240
1241bool DataFlowSanitizer::isInstrumented(const GlobalAlias *GA) {
1242 return !ABIList.isIn(GA: *GA, Category: "uninstrumented");
1243}
1244
1245bool DataFlowSanitizer::isForceZeroLabels(const Function *F) {
1246 return ABIList.isIn(F: *F, Category: "force_zero_labels");
1247}
1248
1249DataFlowSanitizer::WrapperKind DataFlowSanitizer::getWrapperKind(Function *F) {
1250 if (ABIList.isIn(F: *F, Category: "functional"))
1251 return WK_Functional;
1252 if (ABIList.isIn(F: *F, Category: "discard"))
1253 return WK_Discard;
1254 if (ABIList.isIn(F: *F, Category: "custom"))
1255 return WK_Custom;
1256
1257 return WK_Warning;
1258}
1259
1260void DataFlowSanitizer::addGlobalNameSuffix(GlobalValue *GV) {
1261 std::string GVName = std::string(GV->getName()), Suffix = ".dfsan";
1262 GV->setName(GVName + Suffix);
1263
1264 // Try to change the name of the function in module inline asm. We only do
1265 // this for specific asm directives, currently only ".symver", to try to avoid
1266 // corrupting asm which happens to contain the symbol name as a substring.
1267 // Note that the substitution for .symver assumes that the versioned symbol
1268 // also has an instrumented name.
1269 std::string Asm = GV->getParent()->getModuleInlineAsm();
1270 std::string SearchStr = ".symver " + GVName + ",";
1271 size_t Pos = Asm.find(str: SearchStr);
1272 if (Pos != std::string::npos) {
1273 Asm.replace(pos: Pos, n: SearchStr.size(), str: ".symver " + GVName + Suffix + ",");
1274 Pos = Asm.find(c: '@');
1275
1276 if (Pos == std::string::npos)
1277 report_fatal_error(reason: Twine("unsupported .symver: ", Asm));
1278
1279 Asm.replace(pos: Pos, n: 1, str: Suffix + "@");
1280 GV->getParent()->setModuleInlineAsm(Asm);
1281 }
1282}
1283
1284void DataFlowSanitizer::buildExternWeakCheckIfNeeded(IRBuilder<> &IRB,
1285 Function *F) {
1286 // If the function we are wrapping was ExternWeak, it may be null.
1287 // The original code before calling this wrapper may have checked for null,
1288 // but replacing with a known-to-not-be-null wrapper can break this check.
1289 // When replacing uses of the extern weak function with the wrapper we try
1290 // to avoid replacing uses in conditionals, but this is not perfect.
1291 // In the case where we fail, and accidentally optimize out a null check
1292 // for a extern weak function, add a check here to help identify the issue.
1293 if (GlobalValue::isExternalWeakLinkage(Linkage: F->getLinkage())) {
1294 std::vector<Value *> Args;
1295 Args.push_back(x: F);
1296 Args.push_back(x: IRB.CreateGlobalStringPtr(Str: F->getName()));
1297 IRB.CreateCall(Callee: DFSanWrapperExternWeakNullFn, Args);
1298 }
1299}
1300
1301Function *
1302DataFlowSanitizer::buildWrapperFunction(Function *F, StringRef NewFName,
1303 GlobalValue::LinkageTypes NewFLink,
1304 FunctionType *NewFT) {
1305 FunctionType *FT = F->getFunctionType();
1306 Function *NewF = Function::Create(Ty: NewFT, Linkage: NewFLink, AddrSpace: F->getAddressSpace(),
1307 N: NewFName, M: F->getParent());
1308 NewF->copyAttributesFrom(Src: F);
1309 NewF->removeRetAttrs(
1310 Attrs: AttributeFuncs::typeIncompatible(Ty: NewFT->getReturnType()));
1311
1312 BasicBlock *BB = BasicBlock::Create(Context&: *Ctx, Name: "entry", Parent: NewF);
1313 if (F->isVarArg()) {
1314 NewF->removeFnAttr(Kind: "split-stack");
1315 CallInst::Create(Func: DFSanVarargWrapperFn,
1316 Args: IRBuilder<>(BB).CreateGlobalStringPtr(Str: F->getName()), NameStr: "",
1317 InsertBefore: BB);
1318 new UnreachableInst(*Ctx, BB);
1319 } else {
1320 auto ArgIt = pointer_iterator<Argument *>(NewF->arg_begin());
1321 std::vector<Value *> Args(ArgIt, ArgIt + FT->getNumParams());
1322
1323 CallInst *CI = CallInst::Create(Func: F, Args, NameStr: "", InsertBefore: BB);
1324 if (FT->getReturnType()->isVoidTy())
1325 ReturnInst::Create(C&: *Ctx, InsertAtEnd: BB);
1326 else
1327 ReturnInst::Create(C&: *Ctx, retVal: CI, InsertBefore: BB);
1328 }
1329
1330 return NewF;
1331}
1332
1333// Initialize DataFlowSanitizer runtime functions and declare them in the module
1334void DataFlowSanitizer::initializeRuntimeFunctions(Module &M) {
1335 LLVMContext &C = M.getContext();
1336 {
1337 AttributeList AL;
1338 AL = AL.addFnAttribute(C, Kind: Attribute::NoUnwind);
1339 AL = AL.addFnAttribute(
1340 C, Attr: Attribute::getWithMemoryEffects(Context&: C, ME: MemoryEffects::readOnly()));
1341 AL = AL.addRetAttribute(C, Kind: Attribute::ZExt);
1342 DFSanUnionLoadFn =
1343 Mod->getOrInsertFunction(Name: "__dfsan_union_load", T: DFSanUnionLoadFnTy, AttributeList: AL);
1344 }
1345 {
1346 AttributeList AL;
1347 AL = AL.addFnAttribute(C, Kind: Attribute::NoUnwind);
1348 AL = AL.addFnAttribute(
1349 C, Attr: Attribute::getWithMemoryEffects(Context&: C, ME: MemoryEffects::readOnly()));
1350 AL = AL.addRetAttribute(C, Kind: Attribute::ZExt);
1351 DFSanLoadLabelAndOriginFn = Mod->getOrInsertFunction(
1352 Name: "__dfsan_load_label_and_origin", T: DFSanLoadLabelAndOriginFnTy, AttributeList: AL);
1353 }
1354 DFSanUnimplementedFn =
1355 Mod->getOrInsertFunction(Name: "__dfsan_unimplemented", T: DFSanUnimplementedFnTy);
1356 DFSanWrapperExternWeakNullFn = Mod->getOrInsertFunction(
1357 Name: "__dfsan_wrapper_extern_weak_null", T: DFSanWrapperExternWeakNullFnTy);
1358 {
1359 AttributeList AL;
1360 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 0, Kind: Attribute::ZExt);
1361 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 1, Kind: Attribute::ZExt);
1362 DFSanSetLabelFn =
1363 Mod->getOrInsertFunction(Name: "__dfsan_set_label", T: DFSanSetLabelFnTy, AttributeList: AL);
1364 }
1365 DFSanNonzeroLabelFn =
1366 Mod->getOrInsertFunction(Name: "__dfsan_nonzero_label", T: DFSanNonzeroLabelFnTy);
1367 DFSanVarargWrapperFn = Mod->getOrInsertFunction(Name: "__dfsan_vararg_wrapper",
1368 T: DFSanVarargWrapperFnTy);
1369 {
1370 AttributeList AL;
1371 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 0, Kind: Attribute::ZExt);
1372 AL = AL.addRetAttribute(C&: M.getContext(), Kind: Attribute::ZExt);
1373 DFSanChainOriginFn = Mod->getOrInsertFunction(Name: "__dfsan_chain_origin",
1374 T: DFSanChainOriginFnTy, AttributeList: AL);
1375 }
1376 {
1377 AttributeList AL;
1378 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 0, Kind: Attribute::ZExt);
1379 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 1, Kind: Attribute::ZExt);
1380 AL = AL.addRetAttribute(C&: M.getContext(), Kind: Attribute::ZExt);
1381 DFSanChainOriginIfTaintedFn = Mod->getOrInsertFunction(
1382 Name: "__dfsan_chain_origin_if_tainted", T: DFSanChainOriginIfTaintedFnTy, AttributeList: AL);
1383 }
1384 DFSanMemOriginTransferFn = Mod->getOrInsertFunction(
1385 Name: "__dfsan_mem_origin_transfer", T: DFSanMemOriginTransferFnTy);
1386
1387 DFSanMemShadowOriginTransferFn = Mod->getOrInsertFunction(
1388 Name: "__dfsan_mem_shadow_origin_transfer", T: DFSanMemShadowOriginTransferFnTy);
1389
1390 DFSanMemShadowOriginConditionalExchangeFn =
1391 Mod->getOrInsertFunction(Name: "__dfsan_mem_shadow_origin_conditional_exchange",
1392 T: DFSanMemShadowOriginConditionalExchangeFnTy);
1393
1394 {
1395 AttributeList AL;
1396 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 0, Kind: Attribute::ZExt);
1397 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 3, Kind: Attribute::ZExt);
1398 DFSanMaybeStoreOriginFn = Mod->getOrInsertFunction(
1399 Name: "__dfsan_maybe_store_origin", T: DFSanMaybeStoreOriginFnTy, AttributeList: AL);
1400 }
1401
1402 DFSanRuntimeFunctions.insert(
1403 Ptr: DFSanUnionLoadFn.getCallee()->stripPointerCasts());
1404 DFSanRuntimeFunctions.insert(
1405 Ptr: DFSanLoadLabelAndOriginFn.getCallee()->stripPointerCasts());
1406 DFSanRuntimeFunctions.insert(
1407 Ptr: DFSanUnimplementedFn.getCallee()->stripPointerCasts());
1408 DFSanRuntimeFunctions.insert(
1409 Ptr: DFSanWrapperExternWeakNullFn.getCallee()->stripPointerCasts());
1410 DFSanRuntimeFunctions.insert(
1411 Ptr: DFSanSetLabelFn.getCallee()->stripPointerCasts());
1412 DFSanRuntimeFunctions.insert(
1413 Ptr: DFSanNonzeroLabelFn.getCallee()->stripPointerCasts());
1414 DFSanRuntimeFunctions.insert(
1415 Ptr: DFSanVarargWrapperFn.getCallee()->stripPointerCasts());
1416 DFSanRuntimeFunctions.insert(
1417 Ptr: DFSanLoadCallbackFn.getCallee()->stripPointerCasts());
1418 DFSanRuntimeFunctions.insert(
1419 Ptr: DFSanStoreCallbackFn.getCallee()->stripPointerCasts());
1420 DFSanRuntimeFunctions.insert(
1421 Ptr: DFSanMemTransferCallbackFn.getCallee()->stripPointerCasts());
1422 DFSanRuntimeFunctions.insert(
1423 Ptr: DFSanConditionalCallbackFn.getCallee()->stripPointerCasts());
1424 DFSanRuntimeFunctions.insert(
1425 Ptr: DFSanConditionalCallbackOriginFn.getCallee()->stripPointerCasts());
1426 DFSanRuntimeFunctions.insert(
1427 Ptr: DFSanReachesFunctionCallbackFn.getCallee()->stripPointerCasts());
1428 DFSanRuntimeFunctions.insert(
1429 Ptr: DFSanReachesFunctionCallbackOriginFn.getCallee()->stripPointerCasts());
1430 DFSanRuntimeFunctions.insert(
1431 Ptr: DFSanCmpCallbackFn.getCallee()->stripPointerCasts());
1432 DFSanRuntimeFunctions.insert(
1433 Ptr: DFSanChainOriginFn.getCallee()->stripPointerCasts());
1434 DFSanRuntimeFunctions.insert(
1435 Ptr: DFSanChainOriginIfTaintedFn.getCallee()->stripPointerCasts());
1436 DFSanRuntimeFunctions.insert(
1437 Ptr: DFSanMemOriginTransferFn.getCallee()->stripPointerCasts());
1438 DFSanRuntimeFunctions.insert(
1439 Ptr: DFSanMemShadowOriginTransferFn.getCallee()->stripPointerCasts());
1440 DFSanRuntimeFunctions.insert(
1441 Ptr: DFSanMemShadowOriginConditionalExchangeFn.getCallee()
1442 ->stripPointerCasts());
1443 DFSanRuntimeFunctions.insert(
1444 Ptr: DFSanMaybeStoreOriginFn.getCallee()->stripPointerCasts());
1445}
1446
1447// Initializes event callback functions and declare them in the module
1448void DataFlowSanitizer::initializeCallbackFunctions(Module &M) {
1449 {
1450 AttributeList AL;
1451 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 0, Kind: Attribute::ZExt);
1452 DFSanLoadCallbackFn = Mod->getOrInsertFunction(
1453 Name: "__dfsan_load_callback", T: DFSanLoadStoreCallbackFnTy, AttributeList: AL);
1454 }
1455 {
1456 AttributeList AL;
1457 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 0, Kind: Attribute::ZExt);
1458 DFSanStoreCallbackFn = Mod->getOrInsertFunction(
1459 Name: "__dfsan_store_callback", T: DFSanLoadStoreCallbackFnTy, AttributeList: AL);
1460 }
1461 DFSanMemTransferCallbackFn = Mod->getOrInsertFunction(
1462 Name: "__dfsan_mem_transfer_callback", T: DFSanMemTransferCallbackFnTy);
1463 {
1464 AttributeList AL;
1465 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 0, Kind: Attribute::ZExt);
1466 DFSanCmpCallbackFn = Mod->getOrInsertFunction(Name: "__dfsan_cmp_callback",
1467 T: DFSanCmpCallbackFnTy, AttributeList: AL);
1468 }
1469 {
1470 AttributeList AL;
1471 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 0, Kind: Attribute::ZExt);
1472 DFSanConditionalCallbackFn = Mod->getOrInsertFunction(
1473 Name: "__dfsan_conditional_callback", T: DFSanConditionalCallbackFnTy, AttributeList: AL);
1474 }
1475 {
1476 AttributeList AL;
1477 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 0, Kind: Attribute::ZExt);
1478 DFSanConditionalCallbackOriginFn =
1479 Mod->getOrInsertFunction(Name: "__dfsan_conditional_callback_origin",
1480 T: DFSanConditionalCallbackOriginFnTy, AttributeList: AL);
1481 }
1482 {
1483 AttributeList AL;
1484 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 0, Kind: Attribute::ZExt);
1485 DFSanReachesFunctionCallbackFn =
1486 Mod->getOrInsertFunction(Name: "__dfsan_reaches_function_callback",
1487 T: DFSanReachesFunctionCallbackFnTy, AttributeList: AL);
1488 }
1489 {
1490 AttributeList AL;
1491 AL = AL.addParamAttribute(C&: M.getContext(), ArgNo: 0, Kind: Attribute::ZExt);
1492 DFSanReachesFunctionCallbackOriginFn =
1493 Mod->getOrInsertFunction(Name: "__dfsan_reaches_function_callback_origin",
1494 T: DFSanReachesFunctionCallbackOriginFnTy, AttributeList: AL);
1495 }
1496}
1497
1498bool DataFlowSanitizer::runImpl(
1499 Module &M, llvm::function_ref<TargetLibraryInfo &(Function &)> GetTLI) {
1500 initializeModule(M);
1501
1502 if (ABIList.isIn(M, Category: "skip"))
1503 return false;
1504
1505 const unsigned InitialGlobalSize = M.global_size();
1506 const unsigned InitialModuleSize = M.size();
1507
1508 bool Changed = false;
1509
1510 auto GetOrInsertGlobal = [this, &Changed](StringRef Name,
1511 Type *Ty) -> Constant * {
1512 Constant *C = Mod->getOrInsertGlobal(Name, Ty);
1513 if (GlobalVariable *G = dyn_cast<GlobalVariable>(Val: C)) {
1514 Changed |= G->getThreadLocalMode() != GlobalVariable::InitialExecTLSModel;
1515 G->setThreadLocalMode(GlobalVariable::InitialExecTLSModel);
1516 }
1517 return C;
1518 };
1519
1520 // These globals must be kept in sync with the ones in dfsan.cpp.
1521 ArgTLS =
1522 GetOrInsertGlobal("__dfsan_arg_tls",
1523 ArrayType::get(ElementType: Type::getInt64Ty(C&: *Ctx), NumElements: ArgTLSSize / 8));
1524 RetvalTLS = GetOrInsertGlobal(
1525 "__dfsan_retval_tls",
1526 ArrayType::get(ElementType: Type::getInt64Ty(C&: *Ctx), NumElements: RetvalTLSSize / 8));
1527 ArgOriginTLSTy = ArrayType::get(ElementType: OriginTy, NumElements: NumOfElementsInArgOrgTLS);
1528 ArgOriginTLS = GetOrInsertGlobal("__dfsan_arg_origin_tls", ArgOriginTLSTy);
1529 RetvalOriginTLS = GetOrInsertGlobal("__dfsan_retval_origin_tls", OriginTy);
1530
1531 (void)Mod->getOrInsertGlobal(Name: "__dfsan_track_origins", Ty: OriginTy, CreateGlobalCallback: [&] {
1532 Changed = true;
1533 return new GlobalVariable(
1534 M, OriginTy, true, GlobalValue::WeakODRLinkage,
1535 ConstantInt::getSigned(Ty: OriginTy,
1536 V: shouldTrackOrigins() ? ClTrackOrigins : 0),
1537 "__dfsan_track_origins");
1538 });
1539
1540 initializeCallbackFunctions(M);
1541 initializeRuntimeFunctions(M);
1542
1543 std::vector<Function *> FnsToInstrument;
1544 SmallPtrSet<Function *, 2> FnsWithNativeABI;
1545 SmallPtrSet<Function *, 2> FnsWithForceZeroLabel;
1546 SmallPtrSet<Constant *, 1> PersonalityFns;
1547 for (Function &F : M)
1548 if (!F.isIntrinsic() && !DFSanRuntimeFunctions.contains(Ptr: &F) &&
1549 !LibAtomicFunction(F) &&
1550 !F.hasFnAttribute(Kind: Attribute::DisableSanitizerInstrumentation)) {
1551 FnsToInstrument.push_back(x: &F);
1552 if (F.hasPersonalityFn())
1553 PersonalityFns.insert(Ptr: F.getPersonalityFn()->stripPointerCasts());
1554 }
1555
1556 if (ClIgnorePersonalityRoutine) {
1557 for (auto *C : PersonalityFns) {
1558 assert(isa<Function>(C) && "Personality routine is not a function!");
1559 Function *F = cast<Function>(Val: C);
1560 if (!isInstrumented(F))
1561 llvm::erase(C&: FnsToInstrument, V: F);
1562 }
1563 }
1564
1565 // Give function aliases prefixes when necessary, and build wrappers where the
1566 // instrumentedness is inconsistent.
1567 for (GlobalAlias &GA : llvm::make_early_inc_range(Range: M.aliases())) {
1568 // Don't stop on weak. We assume people aren't playing games with the
1569 // instrumentedness of overridden weak aliases.
1570 auto *F = dyn_cast<Function>(Val: GA.getAliaseeObject());
1571 if (!F)
1572 continue;
1573
1574 bool GAInst = isInstrumented(GA: &GA), FInst = isInstrumented(F);
1575 if (GAInst && FInst) {
1576 addGlobalNameSuffix(GV: &GA);
1577 } else if (GAInst != FInst) {
1578 // Non-instrumented alias of an instrumented function, or vice versa.
1579 // Replace the alias with a native-ABI wrapper of the aliasee. The pass
1580 // below will take care of instrumenting it.
1581 Function *NewF =
1582 buildWrapperFunction(F, NewFName: "", NewFLink: GA.getLinkage(), NewFT: F->getFunctionType());
1583 GA.replaceAllUsesWith(V: NewF);
1584 NewF->takeName(V: &GA);
1585 GA.eraseFromParent();
1586 FnsToInstrument.push_back(x: NewF);
1587 }
1588 }
1589
1590 // TODO: This could be more precise.
1591 ReadOnlyNoneAttrs.addAttribute(Val: Attribute::Memory);
1592
1593 // First, change the ABI of every function in the module. ABI-listed
1594 // functions keep their original ABI and get a wrapper function.
1595 for (std::vector<Function *>::iterator FI = FnsToInstrument.begin(),
1596 FE = FnsToInstrument.end();
1597 FI != FE; ++FI) {
1598 Function &F = **FI;
1599 FunctionType *FT = F.getFunctionType();
1600
1601 bool IsZeroArgsVoidRet = (FT->getNumParams() == 0 && !FT->isVarArg() &&
1602 FT->getReturnType()->isVoidTy());
1603
1604 if (isInstrumented(F: &F)) {
1605 if (isForceZeroLabels(F: &F))
1606 FnsWithForceZeroLabel.insert(Ptr: &F);
1607
1608 // Instrumented functions get a '.dfsan' suffix. This allows us to more
1609 // easily identify cases of mismatching ABIs. This naming scheme is
1610 // mangling-compatible (see Itanium ABI), using a vendor-specific suffix.
1611 addGlobalNameSuffix(GV: &F);
1612 } else if (!IsZeroArgsVoidRet || getWrapperKind(F: &F) == WK_Custom) {
1613 // Build a wrapper function for F. The wrapper simply calls F, and is
1614 // added to FnsToInstrument so that any instrumentation according to its
1615 // WrapperKind is done in the second pass below.
1616
1617 // If the function being wrapped has local linkage, then preserve the
1618 // function's linkage in the wrapper function.
1619 GlobalValue::LinkageTypes WrapperLinkage =
1620 F.hasLocalLinkage() ? F.getLinkage()
1621 : GlobalValue::LinkOnceODRLinkage;
1622
1623 Function *NewF = buildWrapperFunction(
1624 F: &F,
1625 NewFName: (shouldTrackOrigins() ? std::string("dfso$") : std::string("dfsw$")) +
1626 std::string(F.getName()),
1627 NewFLink: WrapperLinkage, NewFT: FT);
1628 NewF->removeFnAttrs(Attrs: ReadOnlyNoneAttrs);
1629
1630 // Extern weak functions can sometimes be null at execution time.
1631 // Code will sometimes check if an extern weak function is null.
1632 // This could look something like:
1633 // declare extern_weak i8 @my_func(i8)
1634 // br i1 icmp ne (i8 (i8)* @my_func, i8 (i8)* null), label %use_my_func,
1635 // label %avoid_my_func
1636 // The @"dfsw$my_func" wrapper is never null, so if we replace this use
1637 // in the comparison, the icmp will simplify to false and we have
1638 // accidentally optimized away a null check that is necessary.
1639 // This can lead to a crash when the null extern_weak my_func is called.
1640 //
1641 // To prevent (the most common pattern of) this problem,
1642 // do not replace uses in comparisons with the wrapper.
1643 // We definitely want to replace uses in call instructions.
1644 // Other uses (e.g. store the function address somewhere) might be
1645 // called or compared or both - this case may not be handled correctly.
1646 // We will default to replacing with wrapper in cases we are unsure.
1647 auto IsNotCmpUse = [](Use &U) -> bool {
1648 User *Usr = U.getUser();
1649 if (ConstantExpr *CE = dyn_cast<ConstantExpr>(Val: Usr)) {
1650 // This is the most common case for icmp ne null
1651 if (CE->getOpcode() == Instruction::ICmp) {
1652 return false;
1653 }
1654 }
1655 if (Instruction *I = dyn_cast<Instruction>(Val: Usr)) {
1656 if (I->getOpcode() == Instruction::ICmp) {
1657 return false;
1658 }
1659 }
1660 return true;
1661 };
1662 F.replaceUsesWithIf(New: NewF, ShouldReplace: IsNotCmpUse);
1663
1664 UnwrappedFnMap[NewF] = &F;
1665 *FI = NewF;
1666
1667 if (!F.isDeclaration()) {
1668 // This function is probably defining an interposition of an
1669 // uninstrumented function and hence needs to keep the original ABI.
1670 // But any functions it may call need to use the instrumented ABI, so
1671 // we instrument it in a mode which preserves the original ABI.
1672 FnsWithNativeABI.insert(Ptr: &F);
1673
1674 // This code needs to rebuild the iterators, as they may be invalidated
1675 // by the push_back, taking care that the new range does not include
1676 // any functions added by this code.
1677 size_t N = FI - FnsToInstrument.begin(),
1678 Count = FE - FnsToInstrument.begin();
1679 FnsToInstrument.push_back(x: &F);
1680 FI = FnsToInstrument.begin() + N;
1681 FE = FnsToInstrument.begin() + Count;
1682 }
1683 // Hopefully, nobody will try to indirectly call a vararg
1684 // function... yet.
1685 } else if (FT->isVarArg()) {
1686 UnwrappedFnMap[&F] = &F;
1687 *FI = nullptr;
1688 }
1689 }
1690
1691 for (Function *F : FnsToInstrument) {
1692 if (!F || F->isDeclaration())
1693 continue;
1694
1695 removeUnreachableBlocks(F&: *F);
1696
1697 DFSanFunction DFSF(*this, F, FnsWithNativeABI.count(Ptr: F),
1698 FnsWithForceZeroLabel.count(Ptr: F), GetTLI(*F));
1699
1700 if (ClReachesFunctionCallbacks) {
1701 // Add callback for arguments reaching this function.
1702 for (auto &FArg : F->args()) {
1703 Instruction *Next = &F->getEntryBlock().front();
1704 Value *FArgShadow = DFSF.getShadow(V: &FArg);
1705 if (isZeroShadow(V: FArgShadow))
1706 continue;
1707 if (Instruction *FArgShadowInst = dyn_cast<Instruction>(Val: FArgShadow)) {
1708 Next = FArgShadowInst->getNextNode();
1709 }
1710 if (shouldTrackOrigins()) {
1711 if (Instruction *Origin =
1712 dyn_cast<Instruction>(Val: DFSF.getOrigin(V: &FArg))) {
1713 // Ensure IRB insertion point is after loads for shadow and origin.
1714 Instruction *OriginNext = Origin->getNextNode();
1715 if (Next->comesBefore(Other: OriginNext)) {
1716 Next = OriginNext;
1717 }
1718 }
1719 }
1720 IRBuilder<> IRB(Next);
1721 DFSF.addReachesFunctionCallbacksIfEnabled(IRB, I&: *Next, Data: &FArg);
1722 }
1723 }
1724
1725 // DFSanVisitor may create new basic blocks, which confuses df_iterator.
1726 // Build a copy of the list before iterating over it.
1727 SmallVector<BasicBlock *, 4> BBList(depth_first(G: &F->getEntryBlock()));
1728
1729 for (BasicBlock *BB : BBList) {
1730 Instruction *Inst = &BB->front();
1731 while (true) {
1732 // DFSanVisitor may split the current basic block, changing the current
1733 // instruction's next pointer and moving the next instruction to the
1734 // tail block from which we should continue.
1735 Instruction *Next = Inst->getNextNode();
1736 // DFSanVisitor may delete Inst, so keep track of whether it was a
1737 // terminator.
1738 bool IsTerminator = Inst->isTerminator();
1739 if (!DFSF.SkipInsts.count(V: Inst))
1740 DFSanVisitor(DFSF).visit(I: Inst);
1741 if (IsTerminator)
1742 break;
1743 Inst = Next;
1744 }
1745 }
1746
1747 // We will not necessarily be able to compute the shadow for every phi node
1748 // until we have visited every block. Therefore, the code that handles phi
1749 // nodes adds them to the PHIFixups list so that they can be properly
1750 // handled here.
1751 for (DFSanFunction::PHIFixupElement &P : DFSF.PHIFixups) {
1752 for (unsigned Val = 0, N = P.Phi->getNumIncomingValues(); Val != N;
1753 ++Val) {
1754 P.ShadowPhi->setIncomingValue(
1755 i: Val, V: DFSF.getShadow(V: P.Phi->getIncomingValue(i: Val)));
1756 if (P.OriginPhi)
1757 P.OriginPhi->setIncomingValue(
1758 i: Val, V: DFSF.getOrigin(V: P.Phi->getIncomingValue(i: Val)));
1759 }
1760 }
1761
1762 // -dfsan-debug-nonzero-labels will split the CFG in all kinds of crazy
1763 // places (i.e. instructions in basic blocks we haven't even begun visiting
1764 // yet). To make our life easier, do this work in a pass after the main
1765 // instrumentation.
1766 if (ClDebugNonzeroLabels) {
1767 for (Value *V : DFSF.NonZeroChecks) {
1768 BasicBlock::iterator Pos;
1769 if (Instruction *I = dyn_cast<Instruction>(Val: V))
1770 Pos = std::next(x: I->getIterator());
1771 else
1772 Pos = DFSF.F->getEntryBlock().begin();
1773 while (isa<PHINode>(Val: Pos) || isa<AllocaInst>(Val: Pos))
1774 Pos = std::next(x: Pos->getIterator());
1775 IRBuilder<> IRB(Pos->getParent(), Pos);
1776 Value *PrimitiveShadow = DFSF.collapseToPrimitiveShadow(Shadow: V, Pos);
1777 Value *Ne =
1778 IRB.CreateICmpNE(LHS: PrimitiveShadow, RHS: DFSF.DFS.ZeroPrimitiveShadow);
1779 BranchInst *BI = cast<BranchInst>(Val: SplitBlockAndInsertIfThen(
1780 Cond: Ne, SplitBefore: Pos, /*Unreachable=*/false, BranchWeights: ColdCallWeights));
1781 IRBuilder<> ThenIRB(BI);
1782 ThenIRB.CreateCall(Callee: DFSF.DFS.DFSanNonzeroLabelFn, Args: {});
1783 }
1784 }
1785 }
1786
1787 return Changed || !FnsToInstrument.empty() ||
1788 M.global_size() != InitialGlobalSize || M.size() != InitialModuleSize;
1789}
1790
1791Value *DFSanFunction::getArgTLS(Type *T, unsigned ArgOffset, IRBuilder<> &IRB) {
1792 Value *Base = IRB.CreatePointerCast(V: DFS.ArgTLS, DestTy: DFS.IntptrTy);
1793 if (ArgOffset)
1794 Base = IRB.CreateAdd(LHS: Base, RHS: ConstantInt::get(Ty: DFS.IntptrTy, V: ArgOffset));
1795 return IRB.CreateIntToPtr(V: Base, DestTy: PointerType::get(ElementType: DFS.getShadowTy(OrigTy: T), AddressSpace: 0),
1796 Name: "_dfsarg");
1797}
1798
1799Value *DFSanFunction::getRetvalTLS(Type *T, IRBuilder<> &IRB) {
1800 return IRB.CreatePointerCast(
1801 V: DFS.RetvalTLS, DestTy: PointerType::get(ElementType: DFS.getShadowTy(OrigTy: T), AddressSpace: 0), Name: "_dfsret");
1802}
1803
1804Value *DFSanFunction::getRetvalOriginTLS() { return DFS.RetvalOriginTLS; }
1805
1806Value *DFSanFunction::getArgOriginTLS(unsigned ArgNo, IRBuilder<> &IRB) {
1807 return IRB.CreateConstInBoundsGEP2_64(Ty: DFS.ArgOriginTLSTy, Ptr: DFS.ArgOriginTLS, Idx0: 0,
1808 Idx1: ArgNo, Name: "_dfsarg_o");
1809}
1810
1811Value *DFSanFunction::getOrigin(Value *V) {
1812 assert(DFS.shouldTrackOrigins());
1813 if (!isa<Argument>(Val: V) && !isa<Instruction>(Val: V))
1814 return DFS.ZeroOrigin;
1815 Value *&Origin = ValOriginMap[V];
1816 if (!Origin) {
1817 if (Argument *A = dyn_cast<Argument>(Val: V)) {
1818 if (IsNativeABI)
1819 return DFS.ZeroOrigin;
1820 if (A->getArgNo() < DFS.NumOfElementsInArgOrgTLS) {
1821 Instruction *ArgOriginTLSPos = &*F->getEntryBlock().begin();
1822 IRBuilder<> IRB(ArgOriginTLSPos);
1823 Value *ArgOriginPtr = getArgOriginTLS(ArgNo: A->getArgNo(), IRB);
1824 Origin = IRB.CreateLoad(Ty: DFS.OriginTy, Ptr: ArgOriginPtr);
1825 } else {
1826 // Overflow
1827 Origin = DFS.ZeroOrigin;
1828 }
1829 } else {
1830 Origin = DFS.ZeroOrigin;
1831 }
1832 }
1833 return Origin;
1834}
1835
1836void DFSanFunction::setOrigin(Instruction *I, Value *Origin) {
1837 if (!DFS.shouldTrackOrigins())
1838 return;
1839 assert(!ValOriginMap.count(I));
1840 assert(Origin->getType() == DFS.OriginTy);
1841 ValOriginMap[I] = Origin;
1842}
1843
1844Value *DFSanFunction::getShadowForTLSArgument(Argument *A) {
1845 unsigned ArgOffset = 0;
1846 const DataLayout &DL = F->getDataLayout();
1847 for (auto &FArg : F->args()) {
1848 if (!FArg.getType()->isSized()) {
1849 if (A == &FArg)
1850 break;
1851 continue;
1852 }
1853
1854 unsigned Size = DL.getTypeAllocSize(Ty: DFS.getShadowTy(V: &FArg));
1855 if (A != &FArg) {
1856 ArgOffset += alignTo(Size, A: ShadowTLSAlignment);
1857 if (ArgOffset > ArgTLSSize)
1858 break; // ArgTLS overflows, uses a zero shadow.
1859 continue;
1860 }
1861
1862 if (ArgOffset + Size > ArgTLSSize)
1863 break; // ArgTLS overflows, uses a zero shadow.
1864
1865 Instruction *ArgTLSPos = &*F->getEntryBlock().begin();
1866 IRBuilder<> IRB(ArgTLSPos);
1867 Value *ArgShadowPtr = getArgTLS(T: FArg.getType(), ArgOffset, IRB);
1868 return IRB.CreateAlignedLoad(Ty: DFS.getShadowTy(V: &FArg), Ptr: ArgShadowPtr,
1869 Align: ShadowTLSAlignment);
1870 }
1871
1872 return DFS.getZeroShadow(V: A);
1873}
1874
1875Value *DFSanFunction::getShadow(Value *V) {
1876 if (!isa<Argument>(Val: V) && !isa<Instruction>(Val: V))
1877 return DFS.getZeroShadow(V);
1878 if (IsForceZeroLabels)
1879 return DFS.getZeroShadow(V);
1880 Value *&Shadow = ValShadowMap[V];
1881 if (!Shadow) {
1882 if (Argument *A = dyn_cast<Argument>(Val: V)) {
1883 if (IsNativeABI)
1884 return DFS.getZeroShadow(V);
1885 Shadow = getShadowForTLSArgument(A);
1886 NonZeroChecks.push_back(x: Shadow);
1887 } else {
1888 Shadow = DFS.getZeroShadow(V);
1889 }
1890 }
1891 return Shadow;
1892}
1893
1894void DFSanFunction::setShadow(Instruction *I, Value *Shadow) {
1895 assert(!ValShadowMap.count(I));
1896 ValShadowMap[I] = Shadow;
1897}
1898
1899/// Compute the integer shadow offset that corresponds to a given
1900/// application address.
1901///
1902/// Offset = (Addr & ~AndMask) ^ XorMask
1903Value *DataFlowSanitizer::getShadowOffset(Value *Addr, IRBuilder<> &IRB) {
1904 assert(Addr != RetvalTLS && "Reinstrumenting?");
1905 Value *OffsetLong = IRB.CreatePointerCast(V: Addr, DestTy: IntptrTy);
1906
1907 uint64_t AndMask = MapParams->AndMask;
1908 if (AndMask)
1909 OffsetLong =
1910 IRB.CreateAnd(LHS: OffsetLong, RHS: ConstantInt::get(Ty: IntptrTy, V: ~AndMask));
1911
1912 uint64_t XorMask = MapParams->XorMask;
1913 if (XorMask)
1914 OffsetLong = IRB.CreateXor(LHS: OffsetLong, RHS: ConstantInt::get(Ty: IntptrTy, V: XorMask));
1915 return OffsetLong;
1916}
1917
1918std::pair<Value *, Value *>
1919DataFlowSanitizer::getShadowOriginAddress(Value *Addr, Align InstAlignment,
1920 BasicBlock::iterator Pos) {
1921 // Returns ((Addr & shadow_mask) + origin_base - shadow_base) & ~4UL
1922 IRBuilder<> IRB(Pos->getParent(), Pos);
1923 Value *ShadowOffset = getShadowOffset(Addr, IRB);
1924 Value *ShadowLong = ShadowOffset;
1925 uint64_t ShadowBase = MapParams->ShadowBase;
1926 if (ShadowBase != 0) {
1927 ShadowLong =
1928 IRB.CreateAdd(LHS: ShadowLong, RHS: ConstantInt::get(Ty: IntptrTy, V: ShadowBase));
1929 }
1930 IntegerType *ShadowTy = IntegerType::get(C&: *Ctx, NumBits: ShadowWidthBits);
1931 Value *ShadowPtr =
1932 IRB.CreateIntToPtr(V: ShadowLong, DestTy: PointerType::get(ElementType: ShadowTy, AddressSpace: 0));
1933 Value *OriginPtr = nullptr;
1934 if (shouldTrackOrigins()) {
1935 Value *OriginLong = ShadowOffset;
1936 uint64_t OriginBase = MapParams->OriginBase;
1937 if (OriginBase != 0)
1938 OriginLong =
1939 IRB.CreateAdd(LHS: OriginLong, RHS: ConstantInt::get(Ty: IntptrTy, V: OriginBase));
1940 const Align Alignment = llvm::assumeAligned(Value: InstAlignment.value());
1941 // When alignment is >= 4, Addr must be aligned to 4, otherwise it is UB.
1942 // So Mask is unnecessary.
1943 if (Alignment < MinOriginAlignment) {
1944 uint64_t Mask = MinOriginAlignment.value() - 1;
1945 OriginLong = IRB.CreateAnd(LHS: OriginLong, RHS: ConstantInt::get(Ty: IntptrTy, V: ~Mask));
1946 }
1947 OriginPtr = IRB.CreateIntToPtr(V: OriginLong, DestTy: OriginPtrTy);
1948 }
1949 return std::make_pair(x&: ShadowPtr, y&: OriginPtr);
1950}
1951
1952Value *DataFlowSanitizer::getShadowAddress(Value *Addr,
1953 BasicBlock::iterator Pos,
1954 Value *ShadowOffset) {
1955 IRBuilder<> IRB(Pos->getParent(), Pos);
1956 return IRB.CreateIntToPtr(V: ShadowOffset, DestTy: PrimitiveShadowPtrTy);
1957}
1958
1959Value *DataFlowSanitizer::getShadowAddress(Value *Addr,
1960 BasicBlock::iterator Pos) {
1961 IRBuilder<> IRB(Pos->getParent(), Pos);
1962 Value *ShadowOffset = getShadowOffset(Addr, IRB);
1963 return getShadowAddress(Addr, Pos, ShadowOffset);
1964}
1965
1966Value *DFSanFunction::combineShadowsThenConvert(Type *T, Value *V1, Value *V2,
1967 BasicBlock::iterator Pos) {
1968 Value *PrimitiveValue = combineShadows(V1, V2, Pos);
1969 return expandFromPrimitiveShadow(T, PrimitiveShadow: PrimitiveValue, Pos);
1970}
1971
1972// Generates IR to compute the union of the two given shadows, inserting it
1973// before Pos. The combined value is with primitive type.
1974Value *DFSanFunction::combineShadows(Value *V1, Value *V2,
1975 BasicBlock::iterator Pos) {
1976 if (DFS.isZeroShadow(V: V1))
1977 return collapseToPrimitiveShadow(Shadow: V2, Pos);
1978 if (DFS.isZeroShadow(V: V2))
1979 return collapseToPrimitiveShadow(Shadow: V1, Pos);
1980 if (V1 == V2)
1981 return collapseToPrimitiveShadow(Shadow: V1, Pos);
1982
1983 auto V1Elems = ShadowElements.find(Val: V1);
1984 auto V2Elems = ShadowElements.find(Val: V2);
1985 if (V1Elems != ShadowElements.end() && V2Elems != ShadowElements.end()) {
1986 if (std::includes(first1: V1Elems->second.begin(), last1: V1Elems->second.end(),
1987 first2: V2Elems->second.begin(), last2: V2Elems->second.end())) {
1988 return collapseToPrimitiveShadow(Shadow: V1, Pos);
1989 }
1990 if (std::includes(first1: V2Elems->second.begin(), last1: V2Elems->second.end(),
1991 first2: V1Elems->second.begin(), last2: V1Elems->second.end())) {
1992 return collapseToPrimitiveShadow(Shadow: V2, Pos);
1993 }
1994 } else if (V1Elems != ShadowElements.end()) {
1995 if (V1Elems->second.count(x: V2))
1996 return collapseToPrimitiveShadow(Shadow: V1, Pos);
1997 } else if (V2Elems != ShadowElements.end()) {
1998 if (V2Elems->second.count(x: V1))
1999 return collapseToPrimitiveShadow(Shadow: V2, Pos);
2000 }
2001
2002 auto Key = std::make_pair(x&: V1, y&: V2);
2003 if (V1 > V2)
2004 std::swap(a&: Key.first, b&: Key.second);
2005 CachedShadow &CCS = CachedShadows[Key];
2006 if (CCS.Block && DT.dominates(A: CCS.Block, B: Pos->getParent()))
2007 return CCS.Shadow;
2008
2009 // Converts inputs shadows to shadows with primitive types.
2010 Value *PV1 = collapseToPrimitiveShadow(Shadow: V1, Pos);
2011 Value *PV2 = collapseToPrimitiveShadow(Shadow: V2, Pos);
2012
2013 IRBuilder<> IRB(Pos->getParent(), Pos);
2014 CCS.Block = Pos->getParent();
2015 CCS.Shadow = IRB.CreateOr(LHS: PV1, RHS: PV2);
2016
2017 std::set<Value *> UnionElems;
2018 if (V1Elems != ShadowElements.end()) {
2019 UnionElems = V1Elems->second;
2020 } else {
2021 UnionElems.insert(x: V1);
2022 }
2023 if (V2Elems != ShadowElements.end()) {
2024 UnionElems.insert(first: V2Elems->second.begin(), last: V2Elems->second.end());
2025 } else {
2026 UnionElems.insert(x: V2);
2027 }
2028 ShadowElements[CCS.Shadow] = std::move(UnionElems);
2029
2030 return CCS.Shadow;
2031}
2032
2033// A convenience function which folds the shadows of each of the operands
2034// of the provided instruction Inst, inserting the IR before Inst. Returns
2035// the computed union Value.
2036Value *DFSanFunction::combineOperandShadows(Instruction *Inst) {
2037 if (Inst->getNumOperands() == 0)
2038 return DFS.getZeroShadow(V: Inst);
2039
2040 Value *Shadow = getShadow(V: Inst->getOperand(i: 0));
2041 for (unsigned I = 1, N = Inst->getNumOperands(); I < N; ++I)
2042 Shadow = combineShadows(V1: Shadow, V2: getShadow(V: Inst->getOperand(i: I)),
2043 Pos: Inst->getIterator());
2044
2045 return expandFromPrimitiveShadow(T: Inst->getType(), PrimitiveShadow: Shadow,
2046 Pos: Inst->getIterator());
2047}
2048
2049void DFSanVisitor::visitInstOperands(Instruction &I) {
2050 Value *CombinedShadow = DFSF.combineOperandShadows(Inst: &I);
2051 DFSF.setShadow(I: &I, Shadow: CombinedShadow);
2052 visitInstOperandOrigins(I);
2053}
2054
2055Value *DFSanFunction::combineOrigins(const std::vector<Value *> &Shadows,
2056 const std::vector<Value *> &Origins,
2057 BasicBlock::iterator Pos,
2058 ConstantInt *Zero) {
2059 assert(Shadows.size() == Origins.size());
2060 size_t Size = Origins.size();
2061 if (Size == 0)
2062 return DFS.ZeroOrigin;
2063 Value *Origin = nullptr;
2064 if (!Zero)
2065 Zero = DFS.ZeroPrimitiveShadow;
2066 for (size_t I = 0; I != Size; ++I) {
2067 Value *OpOrigin = Origins[I];
2068 Constant *ConstOpOrigin = dyn_cast<Constant>(Val: OpOrigin);
2069 if (ConstOpOrigin && ConstOpOrigin->isNullValue())
2070 continue;
2071 if (!Origin) {
2072 Origin = OpOrigin;
2073 continue;
2074 }
2075 Value *OpShadow = Shadows[I];
2076 Value *PrimitiveShadow = collapseToPrimitiveShadow(Shadow: OpShadow, Pos);
2077 IRBuilder<> IRB(Pos->getParent(), Pos);
2078 Value *Cond = IRB.CreateICmpNE(LHS: PrimitiveShadow, RHS: Zero);
2079 Origin = IRB.CreateSelect(C: Cond, True: OpOrigin, False: Origin);
2080 }
2081 return Origin ? Origin : DFS.ZeroOrigin;
2082}
2083
2084Value *DFSanFunction::combineOperandOrigins(Instruction *Inst) {
2085 size_t Size = Inst->getNumOperands();
2086 std::vector<Value *> Shadows(Size);
2087 std::vector<Value *> Origins(Size);
2088 for (unsigned I = 0; I != Size; ++I) {
2089 Shadows[I] = getShadow(V: Inst->getOperand(i: I));
2090 Origins[I] = getOrigin(V: Inst->getOperand(i: I));
2091 }
2092 return combineOrigins(Shadows, Origins, Pos: Inst->getIterator());
2093}
2094
2095void DFSanVisitor::visitInstOperandOrigins(Instruction &I) {
2096 if (!DFSF.DFS.shouldTrackOrigins())
2097 return;
2098 Value *CombinedOrigin = DFSF.combineOperandOrigins(Inst: &I);
2099 DFSF.setOrigin(I: &I, Origin: CombinedOrigin);
2100}
2101
2102Align DFSanFunction::getShadowAlign(Align InstAlignment) {
2103 const Align Alignment = ClPreserveAlignment ? InstAlignment : Align(1);
2104 return Align(Alignment.value() * DFS.ShadowWidthBytes);
2105}
2106
2107Align DFSanFunction::getOriginAlign(Align InstAlignment) {
2108 const Align Alignment = llvm::assumeAligned(Value: InstAlignment.value());
2109 return Align(std::max(a: MinOriginAlignment, b: Alignment));
2110}
2111
2112bool DFSanFunction::isLookupTableConstant(Value *P) {
2113 if (GlobalVariable *GV = dyn_cast<GlobalVariable>(Val: P->stripPointerCasts()))
2114 if (GV->isConstant() && GV->hasName())
2115 return DFS.CombineTaintLookupTableNames.count(Key: GV->getName());
2116
2117 return false;
2118}
2119
2120bool DFSanFunction::useCallbackLoadLabelAndOrigin(uint64_t Size,
2121 Align InstAlignment) {
2122 // When enabling tracking load instructions, we always use
2123 // __dfsan_load_label_and_origin to reduce code size.
2124 if (ClTrackOrigins == 2)
2125 return true;
2126
2127 assert(Size != 0);
2128 // * if Size == 1, it is sufficient to load its origin aligned at 4.
2129 // * if Size == 2, we assume most cases Addr % 2 == 0, so it is sufficient to
2130 // load its origin aligned at 4. If not, although origins may be lost, it
2131 // should not happen very often.
2132 // * if align >= 4, Addr must be aligned to 4, otherwise it is UB. When
2133 // Size % 4 == 0, it is more efficient to load origins without callbacks.
2134 // * Otherwise we use __dfsan_load_label_and_origin.
2135 // This should ensure that common cases run efficiently.
2136 if (Size <= 2)
2137 return false;
2138
2139 const Align Alignment = llvm::assumeAligned(Value: InstAlignment.value());
2140 return Alignment < MinOriginAlignment || !DFS.hasLoadSizeForFastPath(Size);
2141}
2142
2143Value *DataFlowSanitizer::loadNextOrigin(BasicBlock::iterator Pos,
2144 Align OriginAlign,
2145 Value **OriginAddr) {
2146 IRBuilder<> IRB(Pos->getParent(), Pos);
2147 *OriginAddr =
2148 IRB.CreateGEP(Ty: OriginTy, Ptr: *OriginAddr, IdxList: ConstantInt::get(Ty: IntptrTy, V: 1));
2149 return IRB.CreateAlignedLoad(Ty: OriginTy, Ptr: *OriginAddr, Align: OriginAlign);
2150}
2151
2152std::pair<Value *, Value *> DFSanFunction::loadShadowFast(
2153 Value *ShadowAddr, Value *OriginAddr, uint64_t Size, Align ShadowAlign,
2154 Align OriginAlign, Value *FirstOrigin, BasicBlock::iterator Pos) {
2155 const bool ShouldTrackOrigins = DFS.shouldTrackOrigins();
2156 const uint64_t ShadowSize = Size * DFS.ShadowWidthBytes;
2157
2158 assert(Size >= 4 && "Not large enough load size for fast path!");
2159
2160 // Used for origin tracking.
2161 std::vector<Value *> Shadows;
2162 std::vector<Value *> Origins;
2163
2164 // Load instructions in LLVM can have arbitrary byte sizes (e.g., 3, 12, 20)
2165 // but this function is only used in a subset of cases that make it possible
2166 // to optimize the instrumentation.
2167 //
2168 // Specifically, when the shadow size in bytes (i.e., loaded bytes x shadow
2169 // per byte) is either:
2170 // - a multiple of 8 (common)
2171 // - equal to 4 (only for load32)
2172 //
2173 // For the second case, we can fit the wide shadow in a 32-bit integer. In all
2174 // other cases, we use a 64-bit integer to hold the wide shadow.
2175 Type *WideShadowTy =
2176 ShadowSize == 4 ? Type::getInt32Ty(C&: *DFS.Ctx) : Type::getInt64Ty(C&: *DFS.Ctx);
2177
2178 IRBuilder<> IRB(Pos->getParent(), Pos);
2179 Value *CombinedWideShadow =
2180 IRB.CreateAlignedLoad(Ty: WideShadowTy, Ptr: ShadowAddr, Align: ShadowAlign);
2181
2182 unsigned WideShadowBitWidth = WideShadowTy->getIntegerBitWidth();
2183 const uint64_t BytesPerWideShadow = WideShadowBitWidth / DFS.ShadowWidthBits;
2184
2185 auto AppendWideShadowAndOrigin = [&](Value *WideShadow, Value *Origin) {
2186 if (BytesPerWideShadow > 4) {
2187 assert(BytesPerWideShadow == 8);
2188 // The wide shadow relates to two origin pointers: one for the first four
2189 // application bytes, and one for the latest four. We use a left shift to
2190 // get just the shadow bytes that correspond to the first origin pointer,
2191 // and then the entire shadow for the second origin pointer (which will be
2192 // chosen by combineOrigins() iff the least-significant half of the wide
2193 // shadow was empty but the other half was not).
2194 Value *WideShadowLo = IRB.CreateShl(
2195 LHS: WideShadow, RHS: ConstantInt::get(Ty: WideShadowTy, V: WideShadowBitWidth / 2));
2196 Shadows.push_back(x: WideShadow);
2197 Origins.push_back(x: DFS.loadNextOrigin(Pos, OriginAlign, OriginAddr: &OriginAddr));
2198
2199 Shadows.push_back(x: WideShadowLo);
2200 Origins.push_back(x: Origin);
2201 } else {
2202 Shadows.push_back(x: WideShadow);
2203 Origins.push_back(x: Origin);
2204 }
2205 };
2206
2207 if (ShouldTrackOrigins)
2208 AppendWideShadowAndOrigin(CombinedWideShadow, FirstOrigin);
2209
2210 // First OR all the WideShadows (i.e., 64bit or 32bit shadow chunks) linearly;
2211 // then OR individual shadows within the combined WideShadow by binary ORing.
2212 // This is fewer instructions than ORing shadows individually, since it
2213 // needs logN shift/or instructions (N being the bytes of the combined wide
2214 // shadow).
2215 for (uint64_t ByteOfs = BytesPerWideShadow; ByteOfs < Size;
2216 ByteOfs += BytesPerWideShadow) {
2217 ShadowAddr = IRB.CreateGEP(Ty: WideShadowTy, Ptr: ShadowAddr,
2218 IdxList: ConstantInt::get(Ty: DFS.IntptrTy, V: 1));
2219 Value *NextWideShadow =
2220 IRB.CreateAlignedLoad(Ty: WideShadowTy, Ptr: ShadowAddr, Align: ShadowAlign);
2221 CombinedWideShadow = IRB.CreateOr(LHS: CombinedWideShadow, RHS: NextWideShadow);
2222 if (ShouldTrackOrigins) {
2223 Value *NextOrigin = DFS.loadNextOrigin(Pos, OriginAlign, OriginAddr: &OriginAddr);
2224 AppendWideShadowAndOrigin(NextWideShadow, NextOrigin);
2225 }
2226 }
2227 for (unsigned Width = WideShadowBitWidth / 2; Width >= DFS.ShadowWidthBits;
2228 Width >>= 1) {
2229 Value *ShrShadow = IRB.CreateLShr(LHS: CombinedWideShadow, RHS: Width);
2230 CombinedWideShadow = IRB.CreateOr(LHS: CombinedWideShadow, RHS: ShrShadow);
2231 }
2232 return {IRB.CreateTrunc(V: CombinedWideShadow, DestTy: DFS.PrimitiveShadowTy),
2233 ShouldTrackOrigins
2234 ? combineOrigins(Shadows, Origins, Pos,
2235 Zero: ConstantInt::getSigned(Ty: IRB.getInt64Ty(), V: 0))
2236 : DFS.ZeroOrigin};
2237}
2238
2239std::pair<Value *, Value *> DFSanFunction::loadShadowOriginSansLoadTracking(
2240 Value *Addr, uint64_t Size, Align InstAlignment, BasicBlock::iterator Pos) {
2241 const bool ShouldTrackOrigins = DFS.shouldTrackOrigins();
2242
2243 // Non-escaped loads.
2244 if (AllocaInst *AI = dyn_cast<AllocaInst>(Val: Addr)) {
2245 const auto SI = AllocaShadowMap.find(Val: AI);
2246 if (SI != AllocaShadowMap.end()) {
2247 IRBuilder<> IRB(Pos->getParent(), Pos);
2248 Value *ShadowLI = IRB.CreateLoad(Ty: DFS.PrimitiveShadowTy, Ptr: SI->second);
2249 const auto OI = AllocaOriginMap.find(Val: AI);
2250 assert(!ShouldTrackOrigins || OI != AllocaOriginMap.end());
2251 return {ShadowLI, ShouldTrackOrigins
2252 ? IRB.CreateLoad(Ty: DFS.OriginTy, Ptr: OI->second)
2253 : nullptr};
2254 }
2255 }
2256
2257 // Load from constant addresses.
2258 SmallVector<const Value *, 2> Objs;
2259 getUnderlyingObjects(V: Addr, Objects&: Objs);
2260 bool AllConstants = true;
2261 for (const Value *Obj : Objs) {
2262 if (isa<Function>(Val: Obj) || isa<BlockAddress>(Val: Obj))
2263 continue;
2264 if (isa<GlobalVariable>(Val: Obj) && cast<GlobalVariable>(Val: Obj)->isConstant())
2265 continue;
2266
2267 AllConstants = false;
2268 break;
2269 }
2270 if (AllConstants)
2271 return {DFS.ZeroPrimitiveShadow,
2272 ShouldTrackOrigins ? DFS.ZeroOrigin : nullptr};
2273
2274 if (Size == 0)
2275 return {DFS.ZeroPrimitiveShadow,
2276 ShouldTrackOrigins ? DFS.ZeroOrigin : nullptr};
2277
2278 // Use callback to load if this is not an optimizable case for origin
2279 // tracking.
2280 if (ShouldTrackOrigins &&
2281 useCallbackLoadLabelAndOrigin(Size, InstAlignment)) {
2282 IRBuilder<> IRB(Pos->getParent(), Pos);
2283 CallInst *Call =
2284 IRB.CreateCall(Callee: DFS.DFSanLoadLabelAndOriginFn,
2285 Args: {Addr, ConstantInt::get(Ty: DFS.IntptrTy, V: Size)});
2286 Call->addRetAttr(Kind: Attribute::ZExt);
2287 return {IRB.CreateTrunc(V: IRB.CreateLShr(LHS: Call, RHS: DFS.OriginWidthBits),
2288 DestTy: DFS.PrimitiveShadowTy),
2289 IRB.CreateTrunc(V: Call, DestTy: DFS.OriginTy)};
2290 }
2291
2292 // Other cases that support loading shadows or origins in a fast way.
2293 Value *ShadowAddr, *OriginAddr;
2294 std::tie(args&: ShadowAddr, args&: OriginAddr) =
2295 DFS.getShadowOriginAddress(Addr, InstAlignment, Pos);
2296
2297 const Align ShadowAlign = getShadowAlign(InstAlignment);
2298 const Align OriginAlign = getOriginAlign(InstAlignment);
2299 Value *Origin = nullptr;
2300 if (ShouldTrackOrigins) {
2301 IRBuilder<> IRB(Pos->getParent(), Pos);
2302 Origin = IRB.CreateAlignedLoad(Ty: DFS.OriginTy, Ptr: OriginAddr, Align: OriginAlign);
2303 }
2304
2305 // When the byte size is small enough, we can load the shadow directly with
2306 // just a few instructions.
2307 switch (Size) {
2308 case 1: {
2309 LoadInst *LI = new LoadInst(DFS.PrimitiveShadowTy, ShadowAddr, "", Pos);
2310 LI->setAlignment(ShadowAlign);
2311 return {LI, Origin};
2312 }
2313 case 2: {
2314 IRBuilder<> IRB(Pos->getParent(), Pos);
2315 Value *ShadowAddr1 = IRB.CreateGEP(Ty: DFS.PrimitiveShadowTy, Ptr: ShadowAddr,
2316 IdxList: ConstantInt::get(Ty: DFS.IntptrTy, V: 1));
2317 Value *Load =
2318 IRB.CreateAlignedLoad(Ty: DFS.PrimitiveShadowTy, Ptr: ShadowAddr, Align: ShadowAlign);
2319 Value *Load1 =
2320 IRB.CreateAlignedLoad(Ty: DFS.PrimitiveShadowTy, Ptr: ShadowAddr1, Align: ShadowAlign);
2321 return {combineShadows(V1: Load, V2: Load1, Pos), Origin};
2322 }
2323 }
2324 bool HasSizeForFastPath = DFS.hasLoadSizeForFastPath(Size);
2325
2326 if (HasSizeForFastPath)
2327 return loadShadowFast(ShadowAddr, OriginAddr, Size, ShadowAlign,
2328 OriginAlign, FirstOrigin: Origin, Pos);
2329
2330 IRBuilder<> IRB(Pos->getParent(), Pos);
2331 CallInst *FallbackCall = IRB.CreateCall(
2332 Callee: DFS.DFSanUnionLoadFn, Args: {ShadowAddr, ConstantInt::get(Ty: DFS.IntptrTy, V: Size)});
2333 FallbackCall->addRetAttr(Kind: Attribute::ZExt);
2334 return {FallbackCall, Origin};
2335}
2336
2337std::pair<Value *, Value *>
2338DFSanFunction::loadShadowOrigin(Value *Addr, uint64_t Size, Align InstAlignment,
2339 BasicBlock::iterator Pos) {
2340 Value *PrimitiveShadow, *Origin;
2341 std::tie(args&: PrimitiveShadow, args&: Origin) =
2342 loadShadowOriginSansLoadTracking(Addr, Size, InstAlignment, Pos);
2343 if (DFS.shouldTrackOrigins()) {
2344 if (ClTrackOrigins == 2) {
2345 IRBuilder<> IRB(Pos->getParent(), Pos);
2346 auto *ConstantShadow = dyn_cast<Constant>(Val: PrimitiveShadow);
2347 if (!ConstantShadow || !ConstantShadow->isZeroValue())
2348 Origin = updateOriginIfTainted(Shadow: PrimitiveShadow, Origin, IRB);
2349 }
2350 }
2351 return {PrimitiveShadow, Origin};
2352}
2353
2354static AtomicOrdering addAcquireOrdering(AtomicOrdering AO) {
2355 switch (AO) {
2356 case AtomicOrdering::NotAtomic:
2357 return AtomicOrdering::NotAtomic;
2358 case AtomicOrdering::Unordered:
2359 case AtomicOrdering::Monotonic:
2360 case AtomicOrdering::Acquire:
2361 return AtomicOrdering::Acquire;
2362 case AtomicOrdering::Release:
2363 case AtomicOrdering::AcquireRelease:
2364 return AtomicOrdering::AcquireRelease;
2365 case AtomicOrdering::SequentiallyConsistent:
2366 return AtomicOrdering::SequentiallyConsistent;
2367 }
2368 llvm_unreachable("Unknown ordering");
2369}
2370
2371Value *StripPointerGEPsAndCasts(Value *V) {
2372 if (!V->getType()->isPointerTy())
2373 return V;
2374
2375 // DFSan pass should be running on valid IR, but we'll
2376 // keep a seen set to ensure there are no issues.
2377 SmallPtrSet<const Value *, 4> Visited;
2378 Visited.insert(Ptr: V);
2379 do {
2380 if (auto *GEP = dyn_cast<GEPOperator>(Val: V)) {
2381 V = GEP->getPointerOperand();
2382 } else if (Operator::getOpcode(V) == Instruction::BitCast) {
2383 V = cast<Operator>(Val: V)->getOperand(i: 0);
2384 if (!V->getType()->isPointerTy())
2385 return V;
2386 } else if (isa<GlobalAlias>(Val: V)) {
2387 V = cast<GlobalAlias>(Val: V)->getAliasee();
2388 }
2389 } while (Visited.insert(Ptr: V).second);
2390
2391 return V;
2392}
2393
2394void DFSanVisitor::visitLoadInst(LoadInst &LI) {
2395 auto &DL = LI.getDataLayout();
2396 uint64_t Size = DL.getTypeStoreSize(Ty: LI.getType());
2397 if (Size == 0) {
2398 DFSF.setShadow(I: &LI, Shadow: DFSF.DFS.getZeroShadow(V: &LI));
2399 DFSF.setOrigin(I: &LI, Origin: DFSF.DFS.ZeroOrigin);
2400 return;
2401 }
2402
2403 // When an application load is atomic, increase atomic ordering between
2404 // atomic application loads and stores to ensure happen-before order; load
2405 // shadow data after application data; store zero shadow data before
2406 // application data. This ensure shadow loads return either labels of the
2407 // initial application data or zeros.
2408 if (LI.isAtomic())
2409 LI.setOrdering(addAcquireOrdering(AO: LI.getOrdering()));
2410
2411 BasicBlock::iterator AfterLi = std::next(x: LI.getIterator());
2412 BasicBlock::iterator Pos = LI.getIterator();
2413 if (LI.isAtomic())
2414 Pos = std::next(x: Pos);
2415
2416 std::vector<Value *> Shadows;
2417 std::vector<Value *> Origins;
2418 Value *PrimitiveShadow, *Origin;
2419 std::tie(args&: PrimitiveShadow, args&: Origin) =
2420 DFSF.loadShadowOrigin(Addr: LI.getPointerOperand(), Size, InstAlignment: LI.getAlign(), Pos);
2421 const bool ShouldTrackOrigins = DFSF.DFS.shouldTrackOrigins();
2422 if (ShouldTrackOrigins) {
2423 Shadows.push_back(x: PrimitiveShadow);
2424 Origins.push_back(x: Origin);
2425 }
2426 if (ClCombinePointerLabelsOnLoad ||
2427 DFSF.isLookupTableConstant(
2428 P: StripPointerGEPsAndCasts(V: LI.getPointerOperand()))) {
2429 Value *PtrShadow = DFSF.getShadow(V: LI.getPointerOperand());
2430 PrimitiveShadow = DFSF.combineShadows(V1: PrimitiveShadow, V2: PtrShadow, Pos);
2431 if (ShouldTrackOrigins) {
2432 Shadows.push_back(x: PtrShadow);
2433 Origins.push_back(x: DFSF.getOrigin(V: LI.getPointerOperand()));
2434 }
2435 }
2436 if (!DFSF.DFS.isZeroShadow(V: PrimitiveShadow))
2437 DFSF.NonZeroChecks.push_back(x: PrimitiveShadow);
2438
2439 Value *Shadow =
2440 DFSF.expandFromPrimitiveShadow(T: LI.getType(), PrimitiveShadow, Pos);
2441 DFSF.setShadow(I: &LI, Shadow);
2442
2443 if (ShouldTrackOrigins) {
2444 DFSF.setOrigin(I: &LI, Origin: DFSF.combineOrigins(Shadows, Origins, Pos));
2445 }
2446
2447 if (ClEventCallbacks) {
2448 IRBuilder<> IRB(Pos->getParent(), Pos);
2449 Value *Addr = LI.getPointerOperand();
2450 CallInst *CI =
2451 IRB.CreateCall(Callee: DFSF.DFS.DFSanLoadCallbackFn, Args: {PrimitiveShadow, Addr});
2452 CI->addParamAttr(ArgNo: 0, Kind: Attribute::ZExt);
2453 }
2454
2455 IRBuilder<> IRB(AfterLi->getParent(), AfterLi);
2456 DFSF.addReachesFunctionCallbacksIfEnabled(IRB, I&: LI, Data: &LI);
2457}
2458
2459Value *DFSanFunction::updateOriginIfTainted(Value *Shadow, Value *Origin,
2460 IRBuilder<> &IRB) {
2461 assert(DFS.shouldTrackOrigins());
2462 return IRB.CreateCall(Callee: DFS.DFSanChainOriginIfTaintedFn, Args: {Shadow, Origin});
2463}
2464
2465Value *DFSanFunction::updateOrigin(Value *V, IRBuilder<> &IRB) {
2466 if (!DFS.shouldTrackOrigins())
2467 return V;
2468 return IRB.CreateCall(Callee: DFS.DFSanChainOriginFn, Args: V);
2469}
2470
2471Value *DFSanFunction::originToIntptr(IRBuilder<> &IRB, Value *Origin) {
2472 const unsigned OriginSize = DataFlowSanitizer::OriginWidthBytes;
2473 const DataLayout &DL = F->getDataLayout();
2474 unsigned IntptrSize = DL.getTypeStoreSize(Ty: DFS.IntptrTy);
2475 if (IntptrSize == OriginSize)
2476 return Origin;
2477 assert(IntptrSize == OriginSize * 2);
2478 Origin = IRB.CreateIntCast(V: Origin, DestTy: DFS.IntptrTy, /* isSigned */ false);
2479 return IRB.CreateOr(LHS: Origin, RHS: IRB.CreateShl(LHS: Origin, RHS: OriginSize * 8));
2480}
2481
2482void DFSanFunction::paintOrigin(IRBuilder<> &IRB, Value *Origin,
2483 Value *StoreOriginAddr,
2484 uint64_t StoreOriginSize, Align Alignment) {
2485 const unsigned OriginSize = DataFlowSanitizer::OriginWidthBytes;
2486 const DataLayout &DL = F->getDataLayout();
2487 const Align IntptrAlignment = DL.getABITypeAlign(Ty: DFS.IntptrTy);
2488 unsigned IntptrSize = DL.getTypeStoreSize(Ty: DFS.IntptrTy);
2489 assert(IntptrAlignment >= MinOriginAlignment);
2490 assert(IntptrSize >= OriginSize);
2491
2492 unsigned Ofs = 0;
2493 Align CurrentAlignment = Alignment;
2494 if (Alignment >= IntptrAlignment && IntptrSize > OriginSize) {
2495 Value *IntptrOrigin = originToIntptr(IRB, Origin);
2496 Value *IntptrStoreOriginPtr = IRB.CreatePointerCast(
2497 V: StoreOriginAddr, DestTy: PointerType::get(ElementType: DFS.IntptrTy, AddressSpace: 0));
2498 for (unsigned I = 0; I < StoreOriginSize / IntptrSize; ++I) {
2499 Value *Ptr =
2500 I ? IRB.CreateConstGEP1_32(Ty: DFS.IntptrTy, Ptr: IntptrStoreOriginPtr, Idx0: I)
2501 : IntptrStoreOriginPtr;
2502 IRB.CreateAlignedStore(Val: IntptrOrigin, Ptr, Align: CurrentAlignment);
2503 Ofs += IntptrSize / OriginSize;
2504 CurrentAlignment = IntptrAlignment;
2505 }
2506 }
2507
2508 for (unsigned I = Ofs; I < (StoreOriginSize + OriginSize - 1) / OriginSize;
2509 ++I) {
2510 Value *GEP = I ? IRB.CreateConstGEP1_32(Ty: DFS.OriginTy, Ptr: StoreOriginAddr, Idx0: I)
2511 : StoreOriginAddr;
2512 IRB.CreateAlignedStore(Val: Origin, Ptr: GEP, Align: CurrentAlignment);
2513 CurrentAlignment = MinOriginAlignment;
2514 }
2515}
2516
2517Value *DFSanFunction::convertToBool(Value *V, IRBuilder<> &IRB,
2518 const Twine &Name) {
2519 Type *VTy = V->getType();
2520 assert(VTy->isIntegerTy());
2521 if (VTy->getIntegerBitWidth() == 1)
2522 // Just converting a bool to a bool, so do nothing.
2523 return V;
2524 return IRB.CreateICmpNE(LHS: V, RHS: ConstantInt::get(Ty: VTy, V: 0), Name);
2525}
2526
2527void DFSanFunction::storeOrigin(BasicBlock::iterator Pos, Value *Addr,
2528 uint64_t Size, Value *Shadow, Value *Origin,
2529 Value *StoreOriginAddr, Align InstAlignment) {
2530 // Do not write origins for zero shadows because we do not trace origins for
2531 // untainted sinks.
2532 const Align OriginAlignment = getOriginAlign(InstAlignment);
2533 Value *CollapsedShadow = collapseToPrimitiveShadow(Shadow, Pos);
2534 IRBuilder<> IRB(Pos->getParent(), Pos);
2535 if (auto *ConstantShadow = dyn_cast<Constant>(Val: CollapsedShadow)) {
2536 if (!ConstantShadow->isZeroValue())
2537 paintOrigin(IRB, Origin: updateOrigin(V: Origin, IRB), StoreOriginAddr, StoreOriginSize: Size,
2538 Alignment: OriginAlignment);
2539 return;
2540 }
2541
2542 if (shouldInstrumentWithCall()) {
2543 IRB.CreateCall(
2544 Callee: DFS.DFSanMaybeStoreOriginFn,
2545 Args: {CollapsedShadow, Addr, ConstantInt::get(Ty: DFS.IntptrTy, V: Size), Origin});
2546 } else {
2547 Value *Cmp = convertToBool(V: CollapsedShadow, IRB, Name: "_dfscmp");
2548 DomTreeUpdater DTU(DT, DomTreeUpdater::UpdateStrategy::Lazy);
2549 Instruction *CheckTerm = SplitBlockAndInsertIfThen(
2550 Cond: Cmp, SplitBefore: &*IRB.GetInsertPoint(), Unreachable: false, BranchWeights: DFS.OriginStoreWeights, DTU: &DTU);
2551 IRBuilder<> IRBNew(CheckTerm);
2552 paintOrigin(IRB&: IRBNew, Origin: updateOrigin(V: Origin, IRB&: IRBNew), StoreOriginAddr, StoreOriginSize: Size,
2553 Alignment: OriginAlignment);
2554 ++NumOriginStores;
2555 }
2556}
2557
2558void DFSanFunction::storeZeroPrimitiveShadow(Value *Addr, uint64_t Size,
2559 Align ShadowAlign,
2560 BasicBlock::iterator Pos) {
2561 IRBuilder<> IRB(Pos->getParent(), Pos);
2562 IntegerType *ShadowTy =
2563 IntegerType::get(C&: *DFS.Ctx, NumBits: Size * DFS.ShadowWidthBits);
2564 Value *ExtZeroShadow = ConstantInt::get(Ty: ShadowTy, V: 0);
2565 Value *ShadowAddr = DFS.getShadowAddress(Addr, Pos);
2566 IRB.CreateAlignedStore(Val: ExtZeroShadow, Ptr: ShadowAddr, Align: ShadowAlign);
2567 // Do not write origins for 0 shadows because we do not trace origins for
2568 // untainted sinks.
2569}
2570
2571void DFSanFunction::storePrimitiveShadowOrigin(Value *Addr, uint64_t Size,
2572 Align InstAlignment,
2573 Value *PrimitiveShadow,
2574 Value *Origin,
2575 BasicBlock::iterator Pos) {
2576 const bool ShouldTrackOrigins = DFS.shouldTrackOrigins() && Origin;
2577
2578 if (AllocaInst *AI = dyn_cast<AllocaInst>(Val: Addr)) {
2579 const auto SI = AllocaShadowMap.find(Val: AI);
2580 if (SI != AllocaShadowMap.end()) {
2581 IRBuilder<> IRB(Pos->getParent(), Pos);
2582 IRB.CreateStore(Val: PrimitiveShadow, Ptr: SI->second);
2583
2584 // Do not write origins for 0 shadows because we do not trace origins for
2585 // untainted sinks.
2586 if (ShouldTrackOrigins && !DFS.isZeroShadow(V: PrimitiveShadow)) {
2587 const auto OI = AllocaOriginMap.find(Val: AI);
2588 assert(OI != AllocaOriginMap.end() && Origin);
2589 IRB.CreateStore(Val: Origin, Ptr: OI->second);
2590 }
2591 return;
2592 }
2593 }
2594
2595 const Align ShadowAlign = getShadowAlign(InstAlignment);
2596 if (DFS.isZeroShadow(V: PrimitiveShadow)) {
2597 storeZeroPrimitiveShadow(Addr, Size, ShadowAlign, Pos);
2598 return;
2599 }
2600
2601 IRBuilder<> IRB(Pos->getParent(), Pos);
2602 Value *ShadowAddr, *OriginAddr;
2603 std::tie(args&: ShadowAddr, args&: OriginAddr) =
2604 DFS.getShadowOriginAddress(Addr, InstAlignment, Pos);
2605
2606 const unsigned ShadowVecSize = 8;
2607 assert(ShadowVecSize * DFS.ShadowWidthBits <= 128 &&
2608 "Shadow vector is too large!");
2609
2610 uint64_t Offset = 0;
2611 uint64_t LeftSize = Size;
2612 if (LeftSize >= ShadowVecSize) {
2613 auto *ShadowVecTy =
2614 FixedVectorType::get(ElementType: DFS.PrimitiveShadowTy, NumElts: ShadowVecSize);
2615 Value *ShadowVec = PoisonValue::get(T: ShadowVecTy);
2616 for (unsigned I = 0; I != ShadowVecSize; ++I) {
2617 ShadowVec = IRB.CreateInsertElement(
2618 Vec: ShadowVec, NewElt: PrimitiveShadow,
2619 Idx: ConstantInt::get(Ty: Type::getInt32Ty(C&: *DFS.Ctx), V: I));
2620 }
2621 do {
2622 Value *CurShadowVecAddr =
2623 IRB.CreateConstGEP1_32(Ty: ShadowVecTy, Ptr: ShadowAddr, Idx0: Offset);
2624 IRB.CreateAlignedStore(Val: ShadowVec, Ptr: CurShadowVecAddr, Align: ShadowAlign);
2625 LeftSize -= ShadowVecSize;
2626 ++Offset;
2627 } while (LeftSize >= ShadowVecSize);
2628 Offset *= ShadowVecSize;
2629 }
2630 while (LeftSize > 0) {
2631 Value *CurShadowAddr =
2632 IRB.CreateConstGEP1_32(Ty: DFS.PrimitiveShadowTy, Ptr: ShadowAddr, Idx0: Offset);
2633 IRB.CreateAlignedStore(Val: PrimitiveShadow, Ptr: CurShadowAddr, Align: ShadowAlign);
2634 --LeftSize;
2635 ++Offset;
2636 }
2637
2638 if (ShouldTrackOrigins) {
2639 storeOrigin(Pos, Addr, Size, Shadow: PrimitiveShadow, Origin, StoreOriginAddr: OriginAddr,
2640 InstAlignment);
2641 }
2642}
2643
2644static AtomicOrdering addReleaseOrdering(AtomicOrdering AO) {
2645 switch (AO) {
2646 case AtomicOrdering::NotAtomic:
2647 return AtomicOrdering::NotAtomic;
2648 case AtomicOrdering::Unordered:
2649 case AtomicOrdering::Monotonic:
2650 case AtomicOrdering::Release:
2651 return AtomicOrdering::Release;
2652 case AtomicOrdering::Acquire:
2653 case AtomicOrdering::AcquireRelease:
2654 return AtomicOrdering::AcquireRelease;
2655 case AtomicOrdering::SequentiallyConsistent:
2656 return AtomicOrdering::SequentiallyConsistent;
2657 }
2658 llvm_unreachable("Unknown ordering");
2659}
2660
2661void DFSanVisitor::visitStoreInst(StoreInst &SI) {
2662 auto &DL = SI.getDataLayout();
2663 Value *Val = SI.getValueOperand();
2664 uint64_t Size = DL.getTypeStoreSize(Ty: Val->getType());
2665 if (Size == 0)
2666 return;
2667
2668 // When an application store is atomic, increase atomic ordering between
2669 // atomic application loads and stores to ensure happen-before order; load
2670 // shadow data after application data; store zero shadow data before
2671 // application data. This ensure shadow loads return either labels of the
2672 // initial application data or zeros.
2673 if (SI.isAtomic())
2674 SI.setOrdering(addReleaseOrdering(AO: SI.getOrdering()));
2675
2676 const bool ShouldTrackOrigins =
2677 DFSF.DFS.shouldTrackOrigins() && !SI.isAtomic();
2678 std::vector<Value *> Shadows;
2679 std::vector<Value *> Origins;
2680
2681 Value *Shadow =
2682 SI.isAtomic() ? DFSF.DFS.getZeroShadow(V: Val) : DFSF.getShadow(V: Val);
2683
2684 if (ShouldTrackOrigins) {
2685 Shadows.push_back(x: Shadow);
2686 Origins.push_back(x: DFSF.getOrigin(V: Val));
2687 }
2688
2689 Value *PrimitiveShadow;
2690 if (ClCombinePointerLabelsOnStore) {
2691 Value *PtrShadow = DFSF.getShadow(V: SI.getPointerOperand());
2692 if (ShouldTrackOrigins) {
2693 Shadows.push_back(x: PtrShadow);
2694 Origins.push_back(x: DFSF.getOrigin(V: SI.getPointerOperand()));
2695 }
2696 PrimitiveShadow = DFSF.combineShadows(V1: Shadow, V2: PtrShadow, Pos: SI.getIterator());
2697 } else {
2698 PrimitiveShadow = DFSF.collapseToPrimitiveShadow(Shadow, Pos: SI.getIterator());
2699 }
2700 Value *Origin = nullptr;
2701 if (ShouldTrackOrigins)
2702 Origin = DFSF.combineOrigins(Shadows, Origins, Pos: SI.getIterator());
2703 DFSF.storePrimitiveShadowOrigin(Addr: SI.getPointerOperand(), Size, InstAlignment: SI.getAlign(),
2704 PrimitiveShadow, Origin, Pos: SI.getIterator());
2705 if (ClEventCallbacks) {
2706 IRBuilder<> IRB(&SI);
2707 Value *Addr = SI.getPointerOperand();
2708 CallInst *CI =
2709 IRB.CreateCall(Callee: DFSF.DFS.DFSanStoreCallbackFn, Args: {PrimitiveShadow, Addr});
2710 CI->addParamAttr(ArgNo: 0, Kind: Attribute::ZExt);
2711 }
2712}
2713
2714void DFSanVisitor::visitCASOrRMW(Align InstAlignment, Instruction &I) {
2715 assert(isa<AtomicRMWInst>(I) || isa<AtomicCmpXchgInst>(I));
2716
2717 Value *Val = I.getOperand(i: 1);
2718 const auto &DL = I.getDataLayout();
2719 uint64_t Size = DL.getTypeStoreSize(Ty: Val->getType());
2720 if (Size == 0)
2721 return;
2722
2723 // Conservatively set data at stored addresses and return with zero shadow to
2724 // prevent shadow data races.
2725 IRBuilder<> IRB(&I);
2726 Value *Addr = I.getOperand(i: 0);
2727 const Align ShadowAlign = DFSF.getShadowAlign(InstAlignment);
2728 DFSF.storeZeroPrimitiveShadow(Addr, Size, ShadowAlign, Pos: I.getIterator());
2729 DFSF.setShadow(I: &I, Shadow: DFSF.DFS.getZeroShadow(V: &I));
2730 DFSF.setOrigin(I: &I, Origin: DFSF.DFS.ZeroOrigin);
2731}
2732
2733void DFSanVisitor::visitAtomicRMWInst(AtomicRMWInst &I) {
2734 visitCASOrRMW(InstAlignment: I.getAlign(), I);
2735 // TODO: The ordering change follows MSan. It is possible not to change
2736 // ordering because we always set and use 0 shadows.
2737 I.setOrdering(addReleaseOrdering(AO: I.getOrdering()));
2738}
2739
2740void DFSanVisitor::visitAtomicCmpXchgInst(AtomicCmpXchgInst &I) {
2741 visitCASOrRMW(InstAlignment: I.getAlign(), I);
2742 // TODO: The ordering change follows MSan. It is possible not to change
2743 // ordering because we always set and use 0 shadows.
2744 I.setSuccessOrdering(addReleaseOrdering(AO: I.getSuccessOrdering()));
2745}
2746
2747void DFSanVisitor::visitUnaryOperator(UnaryOperator &UO) {
2748 visitInstOperands(I&: UO);
2749}
2750
2751void DFSanVisitor::visitBinaryOperator(BinaryOperator &BO) {
2752 visitInstOperands(I&: BO);
2753}
2754
2755void DFSanVisitor::visitBitCastInst(BitCastInst &BCI) {
2756 // Special case: if this is the bitcast (there is exactly 1 allowed) between
2757 // a musttail call and a ret, don't instrument. New instructions are not
2758 // allowed after a musttail call.
2759 if (auto *CI = dyn_cast<CallInst>(Val: BCI.getOperand(i_nocapture: 0)))
2760 if (CI->isMustTailCall())
2761 return;
2762 visitInstOperands(I&: BCI);
2763}
2764
2765void DFSanVisitor::visitCastInst(CastInst &CI) { visitInstOperands(I&: CI); }
2766
2767void DFSanVisitor::visitCmpInst(CmpInst &CI) {
2768 visitInstOperands(I&: CI);
2769 if (ClEventCallbacks) {
2770 IRBuilder<> IRB(&CI);
2771 Value *CombinedShadow = DFSF.getShadow(V: &CI);
2772 CallInst *CallI =
2773 IRB.CreateCall(Callee: DFSF.DFS.DFSanCmpCallbackFn, Args: CombinedShadow);
2774 CallI->addParamAttr(ArgNo: 0, Kind: Attribute::ZExt);
2775 }
2776}
2777
2778void DFSanVisitor::visitLandingPadInst(LandingPadInst &LPI) {
2779 // We do not need to track data through LandingPadInst.
2780 //
2781 // For the C++ exceptions, if a value is thrown, this value will be stored
2782 // in a memory location provided by __cxa_allocate_exception(...) (on the
2783 // throw side) or __cxa_begin_catch(...) (on the catch side).
2784 // This memory will have a shadow, so with the loads and stores we will be
2785 // able to propagate labels on data thrown through exceptions, without any
2786 // special handling of the LandingPadInst.
2787 //
2788 // The second element in the pair result of the LandingPadInst is a
2789 // register value, but it is for a type ID and should never be tainted.
2790 DFSF.setShadow(I: &LPI, Shadow: DFSF.DFS.getZeroShadow(V: &LPI));
2791 DFSF.setOrigin(I: &LPI, Origin: DFSF.DFS.ZeroOrigin);
2792}
2793
2794void DFSanVisitor::visitGetElementPtrInst(GetElementPtrInst &GEPI) {
2795 if (ClCombineOffsetLabelsOnGEP ||
2796 DFSF.isLookupTableConstant(
2797 P: StripPointerGEPsAndCasts(V: GEPI.getPointerOperand()))) {
2798 visitInstOperands(I&: GEPI);
2799 return;
2800 }
2801
2802 // Only propagate shadow/origin of base pointer value but ignore those of
2803 // offset operands.
2804 Value *BasePointer = GEPI.getPointerOperand();
2805 DFSF.setShadow(I: &GEPI, Shadow: DFSF.getShadow(V: BasePointer));
2806 if (DFSF.DFS.shouldTrackOrigins())
2807 DFSF.setOrigin(I: &GEPI, Origin: DFSF.getOrigin(V: BasePointer));
2808}
2809
2810void DFSanVisitor::visitExtractElementInst(ExtractElementInst &I) {
2811 visitInstOperands(I);
2812}
2813
2814void DFSanVisitor::visitInsertElementInst(InsertElementInst &I) {
2815 visitInstOperands(I);
2816}
2817
2818void DFSanVisitor::visitShuffleVectorInst(ShuffleVectorInst &I) {
2819 visitInstOperands(I);
2820}
2821
2822void DFSanVisitor::visitExtractValueInst(ExtractValueInst &I) {
2823 IRBuilder<> IRB(&I);
2824 Value *Agg = I.getAggregateOperand();
2825 Value *AggShadow = DFSF.getShadow(V: Agg);
2826 Value *ResShadow = IRB.CreateExtractValue(Agg: AggShadow, Idxs: I.getIndices());
2827 DFSF.setShadow(I: &I, Shadow: ResShadow);
2828 visitInstOperandOrigins(I);
2829}
2830
2831void DFSanVisitor::visitInsertValueInst(InsertValueInst &I) {
2832 IRBuilder<> IRB(&I);
2833 Value *AggShadow = DFSF.getShadow(V: I.getAggregateOperand());
2834 Value *InsShadow = DFSF.getShadow(V: I.getInsertedValueOperand());
2835 Value *Res = IRB.CreateInsertValue(Agg: AggShadow, Val: InsShadow, Idxs: I.getIndices());
2836 DFSF.setShadow(I: &I, Shadow: Res);
2837 visitInstOperandOrigins(I);
2838}
2839
2840void DFSanVisitor::visitAllocaInst(AllocaInst &I) {
2841 bool AllLoadsStores = true;
2842 for (User *U : I.users()) {
2843 if (isa<LoadInst>(Val: U))
2844 continue;
2845
2846 if (StoreInst *SI = dyn_cast<StoreInst>(Val: U)) {
2847 if (SI->getPointerOperand() == &I)
2848 continue;
2849 }
2850
2851 AllLoadsStores = false;
2852 break;
2853 }
2854 if (AllLoadsStores) {
2855 IRBuilder<> IRB(&I);
2856 DFSF.AllocaShadowMap[&I] = IRB.CreateAlloca(Ty: DFSF.DFS.PrimitiveShadowTy);
2857 if (DFSF.DFS.shouldTrackOrigins()) {
2858 DFSF.AllocaOriginMap[&I] =
2859 IRB.CreateAlloca(Ty: DFSF.DFS.OriginTy, ArraySize: nullptr, Name: "_dfsa");
2860 }
2861 }
2862 DFSF.setShadow(I: &I, Shadow: DFSF.DFS.ZeroPrimitiveShadow);
2863 DFSF.setOrigin(I: &I, Origin: DFSF.DFS.ZeroOrigin);
2864}
2865
2866void DFSanVisitor::visitSelectInst(SelectInst &I) {
2867 Value *CondShadow = DFSF.getShadow(V: I.getCondition());
2868 Value *TrueShadow = DFSF.getShadow(V: I.getTrueValue());
2869 Value *FalseShadow = DFSF.getShadow(V: I.getFalseValue());
2870 Value *ShadowSel = nullptr;
2871 const bool ShouldTrackOrigins = DFSF.DFS.shouldTrackOrigins();
2872 std::vector<Value *> Shadows;
2873 std::vector<Value *> Origins;
2874 Value *TrueOrigin =
2875 ShouldTrackOrigins ? DFSF.getOrigin(V: I.getTrueValue()) : nullptr;
2876 Value *FalseOrigin =
2877 ShouldTrackOrigins ? DFSF.getOrigin(V: I.getFalseValue()) : nullptr;
2878
2879 DFSF.addConditionalCallbacksIfEnabled(I, Condition: I.getCondition());
2880
2881 if (isa<VectorType>(Val: I.getCondition()->getType())) {
2882 ShadowSel = DFSF.combineShadowsThenConvert(T: I.getType(), V1: TrueShadow,
2883 V2: FalseShadow, Pos: I.getIterator());
2884 if (ShouldTrackOrigins) {
2885 Shadows.push_back(x: TrueShadow);
2886 Shadows.push_back(x: FalseShadow);
2887 Origins.push_back(x: TrueOrigin);
2888 Origins.push_back(x: FalseOrigin);
2889 }
2890 } else {
2891 if (TrueShadow == FalseShadow) {
2892 ShadowSel = TrueShadow;
2893 if (ShouldTrackOrigins) {
2894 Shadows.push_back(x: TrueShadow);
2895 Origins.push_back(x: TrueOrigin);
2896 }
2897 } else {
2898 ShadowSel = SelectInst::Create(C: I.getCondition(), S1: TrueShadow, S2: FalseShadow,
2899 NameStr: "", InsertBefore: I.getIterator());
2900 if (ShouldTrackOrigins) {
2901 Shadows.push_back(x: ShadowSel);
2902 Origins.push_back(x: SelectInst::Create(C: I.getCondition(), S1: TrueOrigin,
2903 S2: FalseOrigin, NameStr: "", InsertBefore: I.getIterator()));
2904 }
2905 }
2906 }
2907 DFSF.setShadow(I: &I, Shadow: ClTrackSelectControlFlow ? DFSF.combineShadowsThenConvert(
2908 T: I.getType(), V1: CondShadow,
2909 V2: ShadowSel, Pos: I.getIterator())
2910 : ShadowSel);
2911 if (ShouldTrackOrigins) {
2912 if (ClTrackSelectControlFlow) {
2913 Shadows.push_back(x: CondShadow);
2914 Origins.push_back(x: DFSF.getOrigin(V: I.getCondition()));
2915 }
2916 DFSF.setOrigin(I: &I, Origin: DFSF.combineOrigins(Shadows, Origins, Pos: I.getIterator()));
2917 }
2918}
2919
2920void DFSanVisitor::visitMemSetInst(MemSetInst &I) {
2921 IRBuilder<> IRB(&I);
2922 Value *ValShadow = DFSF.getShadow(V: I.getValue());
2923 Value *ValOrigin = DFSF.DFS.shouldTrackOrigins()
2924 ? DFSF.getOrigin(V: I.getValue())
2925 : DFSF.DFS.ZeroOrigin;
2926 IRB.CreateCall(Callee: DFSF.DFS.DFSanSetLabelFn,
2927 Args: {ValShadow, ValOrigin, I.getDest(),
2928 IRB.CreateZExtOrTrunc(V: I.getLength(), DestTy: DFSF.DFS.IntptrTy)});
2929}
2930
2931void DFSanVisitor::visitMemTransferInst(MemTransferInst &I) {
2932 IRBuilder<> IRB(&I);
2933
2934 // CopyOrMoveOrigin transfers origins by refering to their shadows. So we
2935 // need to move origins before moving shadows.
2936 if (DFSF.DFS.shouldTrackOrigins()) {
2937 IRB.CreateCall(
2938 Callee: DFSF.DFS.DFSanMemOriginTransferFn,
2939 Args: {I.getArgOperand(i: 0), I.getArgOperand(i: 1),
2940 IRB.CreateIntCast(V: I.getArgOperand(i: 2), DestTy: DFSF.DFS.IntptrTy, isSigned: false)});
2941 }
2942
2943 Value *DestShadow = DFSF.DFS.getShadowAddress(Addr: I.getDest(), Pos: I.getIterator());
2944 Value *SrcShadow = DFSF.DFS.getShadowAddress(Addr: I.getSource(), Pos: I.getIterator());
2945 Value *LenShadow =
2946 IRB.CreateMul(LHS: I.getLength(), RHS: ConstantInt::get(Ty: I.getLength()->getType(),
2947 V: DFSF.DFS.ShadowWidthBytes));
2948 auto *MTI = cast<MemTransferInst>(
2949 Val: IRB.CreateCall(FTy: I.getFunctionType(), Callee: I.getCalledOperand(),
2950 Args: {DestShadow, SrcShadow, LenShadow, I.getVolatileCst()}));
2951 MTI->setDestAlignment(DFSF.getShadowAlign(InstAlignment: I.getDestAlign().valueOrOne()));
2952 MTI->setSourceAlignment(DFSF.getShadowAlign(InstAlignment: I.getSourceAlign().valueOrOne()));
2953 if (ClEventCallbacks) {
2954 IRB.CreateCall(
2955 Callee: DFSF.DFS.DFSanMemTransferCallbackFn,
2956 Args: {DestShadow, IRB.CreateZExtOrTrunc(V: I.getLength(), DestTy: DFSF.DFS.IntptrTy)});
2957 }
2958}
2959
2960void DFSanVisitor::visitBranchInst(BranchInst &BR) {
2961 if (!BR.isConditional())
2962 return;
2963
2964 DFSF.addConditionalCallbacksIfEnabled(I&: BR, Condition: BR.getCondition());
2965}
2966
2967void DFSanVisitor::visitSwitchInst(SwitchInst &SW) {
2968 DFSF.addConditionalCallbacksIfEnabled(I&: SW, Condition: SW.getCondition());
2969}
2970
2971static bool isAMustTailRetVal(Value *RetVal) {
2972 // Tail call may have a bitcast between return.
2973 if (auto *I = dyn_cast<BitCastInst>(Val: RetVal)) {
2974 RetVal = I->getOperand(i_nocapture: 0);
2975 }
2976 if (auto *I = dyn_cast<CallInst>(Val: RetVal)) {
2977 return I->isMustTailCall();
2978 }
2979 return false;
2980}
2981
2982void DFSanVisitor::visitReturnInst(ReturnInst &RI) {
2983 if (!DFSF.IsNativeABI && RI.getReturnValue()) {
2984 // Don't emit the instrumentation for musttail call returns.
2985 if (isAMustTailRetVal(RetVal: RI.getReturnValue()))
2986 return;
2987
2988 Value *S = DFSF.getShadow(V: RI.getReturnValue());
2989 IRBuilder<> IRB(&RI);
2990 Type *RT = DFSF.F->getFunctionType()->getReturnType();
2991 unsigned Size = getDataLayout().getTypeAllocSize(Ty: DFSF.DFS.getShadowTy(OrigTy: RT));
2992 if (Size <= RetvalTLSSize) {
2993 // If the size overflows, stores nothing. At callsite, oversized return
2994 // shadows are set to zero.
2995 IRB.CreateAlignedStore(Val: S, Ptr: DFSF.getRetvalTLS(T: RT, IRB), Align: ShadowTLSAlignment);
2996 }
2997 if (DFSF.DFS.shouldTrackOrigins()) {
2998 Value *O = DFSF.getOrigin(V: RI.getReturnValue());
2999 IRB.CreateStore(Val: O, Ptr: DFSF.getRetvalOriginTLS());
3000 }
3001 }
3002}
3003
3004void DFSanVisitor::addShadowArguments(Function &F, CallBase &CB,
3005 std::vector<Value *> &Args,
3006 IRBuilder<> &IRB) {
3007 FunctionType *FT = F.getFunctionType();
3008
3009 auto *I = CB.arg_begin();
3010
3011 // Adds non-variable argument shadows.
3012 for (unsigned N = FT->getNumParams(); N != 0; ++I, --N)
3013 Args.push_back(
3014 x: DFSF.collapseToPrimitiveShadow(Shadow: DFSF.getShadow(V: *I), Pos: CB.getIterator()));
3015
3016 // Adds variable argument shadows.
3017 if (FT->isVarArg()) {
3018 auto *LabelVATy = ArrayType::get(ElementType: DFSF.DFS.PrimitiveShadowTy,
3019 NumElements: CB.arg_size() - FT->getNumParams());
3020 auto *LabelVAAlloca =
3021 new AllocaInst(LabelVATy, getDataLayout().getAllocaAddrSpace(),
3022 "labelva", DFSF.F->getEntryBlock().begin());
3023
3024 for (unsigned N = 0; I != CB.arg_end(); ++I, ++N) {
3025 auto *LabelVAPtr = IRB.CreateStructGEP(Ty: LabelVATy, Ptr: LabelVAAlloca, Idx: N);
3026 IRB.CreateStore(
3027 Val: DFSF.collapseToPrimitiveShadow(Shadow: DFSF.getShadow(V: *I), Pos: CB.getIterator()),
3028 Ptr: LabelVAPtr);
3029 }
3030
3031 Args.push_back(x: IRB.CreateStructGEP(Ty: LabelVATy, Ptr: LabelVAAlloca, Idx: 0));
3032 }
3033
3034 // Adds the return value shadow.
3035 if (!FT->getReturnType()->isVoidTy()) {
3036 if (!DFSF.LabelReturnAlloca) {
3037 DFSF.LabelReturnAlloca = new AllocaInst(
3038 DFSF.DFS.PrimitiveShadowTy, getDataLayout().getAllocaAddrSpace(),
3039 "labelreturn", DFSF.F->getEntryBlock().begin());
3040 }
3041 Args.push_back(x: DFSF.LabelReturnAlloca);
3042 }
3043}
3044
3045void DFSanVisitor::addOriginArguments(Function &F, CallBase &CB,
3046 std::vector<Value *> &Args,
3047 IRBuilder<> &IRB) {
3048 FunctionType *FT = F.getFunctionType();
3049
3050 auto *I = CB.arg_begin();
3051
3052 // Add non-variable argument origins.
3053 for (unsigned N = FT->getNumParams(); N != 0; ++I, --N)
3054 Args.push_back(x: DFSF.getOrigin(V: *I));
3055
3056 // Add variable argument origins.
3057 if (FT->isVarArg()) {
3058 auto *OriginVATy =
3059 ArrayType::get(ElementType: DFSF.DFS.OriginTy, NumElements: CB.arg_size() - FT->getNumParams());
3060 auto *OriginVAAlloca =
3061 new AllocaInst(OriginVATy, getDataLayout().getAllocaAddrSpace(),
3062 "originva", DFSF.F->getEntryBlock().begin());
3063
3064 for (unsigned N = 0; I != CB.arg_end(); ++I, ++N) {
3065 auto *OriginVAPtr = IRB.CreateStructGEP(Ty: OriginVATy, Ptr: OriginVAAlloca, Idx: N);
3066 IRB.CreateStore(Val: DFSF.getOrigin(V: *I), Ptr: OriginVAPtr);
3067 }
3068
3069 Args.push_back(x: IRB.CreateStructGEP(Ty: OriginVATy, Ptr: OriginVAAlloca, Idx: 0));
3070 }
3071
3072 // Add the return value origin.
3073 if (!FT->getReturnType()->isVoidTy()) {
3074 if (!DFSF.OriginReturnAlloca) {
3075 DFSF.OriginReturnAlloca = new AllocaInst(
3076 DFSF.DFS.OriginTy, getDataLayout().getAllocaAddrSpace(),
3077 "originreturn", DFSF.F->getEntryBlock().begin());
3078 }
3079 Args.push_back(x: DFSF.OriginReturnAlloca);
3080 }
3081}
3082
3083bool DFSanVisitor::visitWrappedCallBase(Function &F, CallBase &CB) {
3084 IRBuilder<> IRB(&CB);
3085 switch (DFSF.DFS.getWrapperKind(F: &F)) {
3086 case DataFlowSanitizer::WK_Warning:
3087 CB.setCalledFunction(&F);
3088 IRB.CreateCall(Callee: DFSF.DFS.DFSanUnimplementedFn,
3089 Args: IRB.CreateGlobalStringPtr(Str: F.getName()));
3090 DFSF.DFS.buildExternWeakCheckIfNeeded(IRB, F: &F);
3091 DFSF.setShadow(I: &CB, Shadow: DFSF.DFS.getZeroShadow(V: &CB));
3092 DFSF.setOrigin(I: &CB, Origin: DFSF.DFS.ZeroOrigin);
3093 return true;
3094 case DataFlowSanitizer::WK_Discard:
3095 CB.setCalledFunction(&F);
3096 DFSF.DFS.buildExternWeakCheckIfNeeded(IRB, F: &F);
3097 DFSF.setShadow(I: &CB, Shadow: DFSF.DFS.getZeroShadow(V: &CB));
3098 DFSF.setOrigin(I: &CB, Origin: DFSF.DFS.ZeroOrigin);
3099 return true;
3100 case DataFlowSanitizer::WK_Functional:
3101 CB.setCalledFunction(&F);
3102 DFSF.DFS.buildExternWeakCheckIfNeeded(IRB, F: &F);
3103 visitInstOperands(I&: CB);
3104 return true;
3105 case DataFlowSanitizer::WK_Custom:
3106 // Don't try to handle invokes of custom functions, it's too complicated.
3107 // Instead, invoke the dfsw$ wrapper, which will in turn call the __dfsw_
3108 // wrapper.
3109 CallInst *CI = dyn_cast<CallInst>(Val: &CB);
3110 if (!CI)
3111 return false;
3112
3113 const bool ShouldTrackOrigins = DFSF.DFS.shouldTrackOrigins();
3114 FunctionType *FT = F.getFunctionType();
3115 TransformedFunction CustomFn = DFSF.DFS.getCustomFunctionType(T: FT);
3116 std::string CustomFName = ShouldTrackOrigins ? "__dfso_" : "__dfsw_";
3117 CustomFName += F.getName();
3118 FunctionCallee CustomF = DFSF.DFS.Mod->getOrInsertFunction(
3119 Name: CustomFName, T: CustomFn.TransformedType);
3120 if (Function *CustomFn = dyn_cast<Function>(Val: CustomF.getCallee())) {
3121 CustomFn->copyAttributesFrom(Src: &F);
3122
3123 // Custom functions returning non-void will write to the return label.
3124 if (!FT->getReturnType()->isVoidTy()) {
3125 CustomFn->removeFnAttrs(Attrs: DFSF.DFS.ReadOnlyNoneAttrs);
3126 }
3127 }
3128
3129 std::vector<Value *> Args;
3130
3131 // Adds non-variable arguments.
3132 auto *I = CB.arg_begin();
3133 for (unsigned N = FT->getNumParams(); N != 0; ++I, --N) {
3134 Args.push_back(x: *I);
3135 }
3136
3137 // Adds shadow arguments.
3138 const unsigned ShadowArgStart = Args.size();
3139 addShadowArguments(F, CB, Args, IRB);
3140
3141 // Adds origin arguments.
3142 const unsigned OriginArgStart = Args.size();
3143 if (ShouldTrackOrigins)
3144 addOriginArguments(F, CB, Args, IRB);
3145
3146 // Adds variable arguments.
3147 append_range(C&: Args, R: drop_begin(RangeOrContainer: CB.args(), N: FT->getNumParams()));
3148
3149 CallInst *CustomCI = IRB.CreateCall(Callee: CustomF, Args);
3150 CustomCI->setCallingConv(CI->getCallingConv());
3151 CustomCI->setAttributes(transformFunctionAttributes(
3152 TransformedFunction: CustomFn, Ctx&: CI->getContext(), CallSiteAttrs: CI->getAttributes()));
3153
3154 // Update the parameter attributes of the custom call instruction to
3155 // zero extend the shadow parameters. This is required for targets
3156 // which consider PrimitiveShadowTy an illegal type.
3157 for (unsigned N = 0; N < FT->getNumParams(); N++) {
3158 const unsigned ArgNo = ShadowArgStart + N;
3159 if (CustomCI->getArgOperand(i: ArgNo)->getType() ==
3160 DFSF.DFS.PrimitiveShadowTy)
3161 CustomCI->addParamAttr(ArgNo, Kind: Attribute::ZExt);
3162 if (ShouldTrackOrigins) {
3163 const unsigned OriginArgNo = OriginArgStart + N;
3164 if (CustomCI->getArgOperand(i: OriginArgNo)->getType() ==
3165 DFSF.DFS.OriginTy)
3166 CustomCI->addParamAttr(ArgNo: OriginArgNo, Kind: Attribute::ZExt);
3167 }
3168 }
3169
3170 // Loads the return value shadow and origin.
3171 if (!FT->getReturnType()->isVoidTy()) {
3172 LoadInst *LabelLoad =
3173 IRB.CreateLoad(Ty: DFSF.DFS.PrimitiveShadowTy, Ptr: DFSF.LabelReturnAlloca);
3174 DFSF.setShadow(I: CustomCI,
3175 Shadow: DFSF.expandFromPrimitiveShadow(
3176 T: FT->getReturnType(), PrimitiveShadow: LabelLoad, Pos: CB.getIterator()));
3177 if (ShouldTrackOrigins) {
3178 LoadInst *OriginLoad =
3179 IRB.CreateLoad(Ty: DFSF.DFS.OriginTy, Ptr: DFSF.OriginReturnAlloca);
3180 DFSF.setOrigin(I: CustomCI, Origin: OriginLoad);
3181 }
3182 }
3183
3184 CI->replaceAllUsesWith(V: CustomCI);
3185 CI->eraseFromParent();
3186 return true;
3187 }
3188 return false;
3189}
3190
3191Value *DFSanVisitor::makeAddAcquireOrderingTable(IRBuilder<> &IRB) {
3192 constexpr int NumOrderings = (int)AtomicOrderingCABI::seq_cst + 1;
3193 uint32_t OrderingTable[NumOrderings] = {};
3194
3195 OrderingTable[(int)AtomicOrderingCABI::relaxed] =
3196 OrderingTable[(int)AtomicOrderingCABI::acquire] =
3197 OrderingTable[(int)AtomicOrderingCABI::consume] =
3198 (int)AtomicOrderingCABI::acquire;
3199 OrderingTable[(int)AtomicOrderingCABI::release] =
3200 OrderingTable[(int)AtomicOrderingCABI::acq_rel] =
3201 (int)AtomicOrderingCABI::acq_rel;
3202 OrderingTable[(int)AtomicOrderingCABI::seq_cst] =
3203 (int)AtomicOrderingCABI::seq_cst;
3204
3205 return ConstantDataVector::get(Context&: IRB.getContext(), Elts: OrderingTable);
3206}
3207
3208void DFSanVisitor::visitLibAtomicLoad(CallBase &CB) {
3209 // Since we use getNextNode here, we can't have CB terminate the BB.
3210 assert(isa<CallInst>(CB));
3211
3212 IRBuilder<> IRB(&CB);
3213 Value *Size = CB.getArgOperand(i: 0);
3214 Value *SrcPtr = CB.getArgOperand(i: 1);
3215 Value *DstPtr = CB.getArgOperand(i: 2);
3216 Value *Ordering = CB.getArgOperand(i: 3);
3217 // Convert the call to have at least Acquire ordering to make sure
3218 // the shadow operations aren't reordered before it.
3219 Value *NewOrdering =
3220 IRB.CreateExtractElement(Vec: makeAddAcquireOrderingTable(IRB), Idx: Ordering);
3221 CB.setArgOperand(i: 3, v: NewOrdering);
3222
3223 IRBuilder<> NextIRB(CB.getNextNode());
3224 NextIRB.SetCurrentDebugLocation(CB.getDebugLoc());
3225
3226 // TODO: Support ClCombinePointerLabelsOnLoad
3227 // TODO: Support ClEventCallbacks
3228
3229 NextIRB.CreateCall(
3230 Callee: DFSF.DFS.DFSanMemShadowOriginTransferFn,
3231 Args: {DstPtr, SrcPtr, NextIRB.CreateIntCast(V: Size, DestTy: DFSF.DFS.IntptrTy, isSigned: false)});
3232}
3233
3234Value *DFSanVisitor::makeAddReleaseOrderingTable(IRBuilder<> &IRB) {
3235 constexpr int NumOrderings = (int)AtomicOrderingCABI::seq_cst + 1;
3236 uint32_t OrderingTable[NumOrderings] = {};
3237
3238 OrderingTable[(int)AtomicOrderingCABI::relaxed] =
3239 OrderingTable[(int)AtomicOrderingCABI::release] =
3240 (int)AtomicOrderingCABI::release;
3241 OrderingTable[(int)AtomicOrderingCABI::consume] =
3242 OrderingTable[(int)AtomicOrderingCABI::acquire] =
3243 OrderingTable[(int)AtomicOrderingCABI::acq_rel] =
3244 (int)AtomicOrderingCABI::acq_rel;
3245 OrderingTable[(int)AtomicOrderingCABI::seq_cst] =
3246 (int)AtomicOrderingCABI::seq_cst;
3247
3248 return ConstantDataVector::get(Context&: IRB.getContext(), Elts: OrderingTable);
3249}
3250
3251void DFSanVisitor::visitLibAtomicStore(CallBase &CB) {
3252 IRBuilder<> IRB(&CB);
3253 Value *Size = CB.getArgOperand(i: 0);
3254 Value *SrcPtr = CB.getArgOperand(i: 1);
3255 Value *DstPtr = CB.getArgOperand(i: 2);
3256 Value *Ordering = CB.getArgOperand(i: 3);
3257 // Convert the call to have at least Release ordering to make sure
3258 // the shadow operations aren't reordered after it.
3259 Value *NewOrdering =
3260 IRB.CreateExtractElement(Vec: makeAddReleaseOrderingTable(IRB), Idx: Ordering);
3261 CB.setArgOperand(i: 3, v: NewOrdering);
3262
3263 // TODO: Support ClCombinePointerLabelsOnStore
3264 // TODO: Support ClEventCallbacks
3265
3266 IRB.CreateCall(
3267 Callee: DFSF.DFS.DFSanMemShadowOriginTransferFn,
3268 Args: {DstPtr, SrcPtr, IRB.CreateIntCast(V: Size, DestTy: DFSF.DFS.IntptrTy, isSigned: false)});
3269}
3270
3271void DFSanVisitor::visitLibAtomicExchange(CallBase &CB) {
3272 // void __atomic_exchange(size_t size, void *ptr, void *val, void *ret, int
3273 // ordering)
3274 IRBuilder<> IRB(&CB);
3275 Value *Size = CB.getArgOperand(i: 0);
3276 Value *TargetPtr = CB.getArgOperand(i: 1);
3277 Value *SrcPtr = CB.getArgOperand(i: 2);
3278 Value *DstPtr = CB.getArgOperand(i: 3);
3279
3280 // This operation is not atomic for the shadow and origin memory.
3281 // This could result in DFSan false positives or false negatives.
3282 // For now we will assume these operations are rare, and
3283 // the additional complexity to address this is not warrented.
3284
3285 // Current Target to Dest
3286 IRB.CreateCall(
3287 Callee: DFSF.DFS.DFSanMemShadowOriginTransferFn,
3288 Args: {DstPtr, TargetPtr, IRB.CreateIntCast(V: Size, DestTy: DFSF.DFS.IntptrTy, isSigned: false)});
3289
3290 // Current Src to Target (overriding)
3291 IRB.CreateCall(
3292 Callee: DFSF.DFS.DFSanMemShadowOriginTransferFn,
3293 Args: {TargetPtr, SrcPtr, IRB.CreateIntCast(V: Size, DestTy: DFSF.DFS.IntptrTy, isSigned: false)});
3294}
3295
3296void DFSanVisitor::visitLibAtomicCompareExchange(CallBase &CB) {
3297 // bool __atomic_compare_exchange(size_t size, void *ptr, void *expected, void
3298 // *desired, int success_order, int failure_order)
3299 Value *Size = CB.getArgOperand(i: 0);
3300 Value *TargetPtr = CB.getArgOperand(i: 1);
3301 Value *ExpectedPtr = CB.getArgOperand(i: 2);
3302 Value *DesiredPtr = CB.getArgOperand(i: 3);
3303
3304 // This operation is not atomic for the shadow and origin memory.
3305 // This could result in DFSan false positives or false negatives.
3306 // For now we will assume these operations are rare, and
3307 // the additional complexity to address this is not warrented.
3308
3309 IRBuilder<> NextIRB(CB.getNextNode());
3310 NextIRB.SetCurrentDebugLocation(CB.getDebugLoc());
3311
3312 DFSF.setShadow(I: &CB, Shadow: DFSF.DFS.getZeroShadow(V: &CB));
3313
3314 // If original call returned true, copy Desired to Target.
3315 // If original call returned false, copy Target to Expected.
3316 NextIRB.CreateCall(Callee: DFSF.DFS.DFSanMemShadowOriginConditionalExchangeFn,
3317 Args: {NextIRB.CreateIntCast(V: &CB, DestTy: NextIRB.getInt8Ty(), isSigned: false),
3318 TargetPtr, ExpectedPtr, DesiredPtr,
3319 NextIRB.CreateIntCast(V: Size, DestTy: DFSF.DFS.IntptrTy, isSigned: false)});
3320}
3321
3322void DFSanVisitor::visitCallBase(CallBase &CB) {
3323 Function *F = CB.getCalledFunction();
3324 if ((F && F->isIntrinsic()) || CB.isInlineAsm()) {
3325 visitInstOperands(I&: CB);
3326 return;
3327 }
3328
3329 // Calls to this function are synthesized in wrappers, and we shouldn't
3330 // instrument them.
3331 if (F == DFSF.DFS.DFSanVarargWrapperFn.getCallee()->stripPointerCasts())
3332 return;
3333
3334 LibFunc LF;
3335 if (DFSF.TLI.getLibFunc(CB, F&: LF)) {
3336 // libatomic.a functions need to have special handling because there isn't
3337 // a good way to intercept them or compile the library with
3338 // instrumentation.
3339 switch (LF) {
3340 case LibFunc_atomic_load:
3341 if (!isa<CallInst>(Val: CB)) {
3342 llvm::errs() << "DFSAN -- cannot instrument invoke of libatomic load. "
3343 "Ignoring!\n";
3344 break;
3345 }
3346 visitLibAtomicLoad(CB);
3347 return;
3348 case LibFunc_atomic_store:
3349 visitLibAtomicStore(CB);
3350 return;
3351 default:
3352 break;
3353 }
3354 }
3355
3356 // TODO: These are not supported by TLI? They are not in the enum.
3357 if (F && F->hasName() && !F->isVarArg()) {
3358 if (F->getName() == "__atomic_exchange") {
3359 visitLibAtomicExchange(CB);
3360 return;
3361 }
3362 if (F->getName() == "__atomic_compare_exchange") {
3363 visitLibAtomicCompareExchange(CB);
3364 return;
3365 }
3366 }
3367
3368 DenseMap<Value *, Function *>::iterator UnwrappedFnIt =
3369 DFSF.DFS.UnwrappedFnMap.find(Val: CB.getCalledOperand());
3370 if (UnwrappedFnIt != DFSF.DFS.UnwrappedFnMap.end())
3371 if (visitWrappedCallBase(F&: *UnwrappedFnIt->second, CB))
3372 return;
3373
3374 IRBuilder<> IRB(&CB);
3375
3376 const bool ShouldTrackOrigins = DFSF.DFS.shouldTrackOrigins();
3377 FunctionType *FT = CB.getFunctionType();
3378 const DataLayout &DL = getDataLayout();
3379
3380 // Stores argument shadows.
3381 unsigned ArgOffset = 0;
3382 for (unsigned I = 0, N = FT->getNumParams(); I != N; ++I) {
3383 if (ShouldTrackOrigins) {
3384 // Ignore overflowed origins
3385 Value *ArgShadow = DFSF.getShadow(V: CB.getArgOperand(i: I));
3386 if (I < DFSF.DFS.NumOfElementsInArgOrgTLS &&
3387 !DFSF.DFS.isZeroShadow(V: ArgShadow))
3388 IRB.CreateStore(Val: DFSF.getOrigin(V: CB.getArgOperand(i: I)),
3389 Ptr: DFSF.getArgOriginTLS(ArgNo: I, IRB));
3390 }
3391
3392 unsigned Size =
3393 DL.getTypeAllocSize(Ty: DFSF.DFS.getShadowTy(OrigTy: FT->getParamType(i: I)));
3394 // Stop storing if arguments' size overflows. Inside a function, arguments
3395 // after overflow have zero shadow values.
3396 if (ArgOffset + Size > ArgTLSSize)
3397 break;
3398 IRB.CreateAlignedStore(Val: DFSF.getShadow(V: CB.getArgOperand(i: I)),
3399 Ptr: DFSF.getArgTLS(T: FT->getParamType(i: I), ArgOffset, IRB),
3400 Align: ShadowTLSAlignment);
3401 ArgOffset += alignTo(Size, A: ShadowTLSAlignment);
3402 }
3403
3404 Instruction *Next = nullptr;
3405 if (!CB.getType()->isVoidTy()) {
3406 if (InvokeInst *II = dyn_cast<InvokeInst>(Val: &CB)) {
3407 if (II->getNormalDest()->getSinglePredecessor()) {
3408 Next = &II->getNormalDest()->front();
3409 } else {
3410 BasicBlock *NewBB =
3411 SplitEdge(From: II->getParent(), To: II->getNormalDest(), DT: &DFSF.DT);
3412 Next = &NewBB->front();
3413 }
3414 } else {
3415 assert(CB.getIterator() != CB.getParent()->end());
3416 Next = CB.getNextNode();
3417 }
3418
3419 // Don't emit the epilogue for musttail call returns.
3420 if (isa<CallInst>(Val: CB) && cast<CallInst>(Val&: CB).isMustTailCall())
3421 return;
3422
3423 // Loads the return value shadow.
3424 IRBuilder<> NextIRB(Next);
3425 unsigned Size = DL.getTypeAllocSize(Ty: DFSF.DFS.getShadowTy(V: &CB));
3426 if (Size > RetvalTLSSize) {
3427 // Set overflowed return shadow to be zero.
3428 DFSF.setShadow(I: &CB, Shadow: DFSF.DFS.getZeroShadow(V: &CB));
3429 } else {
3430 LoadInst *LI = NextIRB.CreateAlignedLoad(
3431 Ty: DFSF.DFS.getShadowTy(V: &CB), Ptr: DFSF.getRetvalTLS(T: CB.getType(), IRB&: NextIRB),
3432 Align: ShadowTLSAlignment, Name: "_dfsret");
3433 DFSF.SkipInsts.insert(V: LI);
3434 DFSF.setShadow(I: &CB, Shadow: LI);
3435 DFSF.NonZeroChecks.push_back(x: LI);
3436 }
3437
3438 if (ShouldTrackOrigins) {
3439 LoadInst *LI = NextIRB.CreateLoad(Ty: DFSF.DFS.OriginTy,
3440 Ptr: DFSF.getRetvalOriginTLS(), Name: "_dfsret_o");
3441 DFSF.SkipInsts.insert(V: LI);
3442 DFSF.setOrigin(I: &CB, Origin: LI);
3443 }
3444
3445 DFSF.addReachesFunctionCallbacksIfEnabled(IRB&: NextIRB, I&: CB, Data: &CB);
3446 }
3447}
3448
3449void DFSanVisitor::visitPHINode(PHINode &PN) {
3450 Type *ShadowTy = DFSF.DFS.getShadowTy(V: &PN);
3451 PHINode *ShadowPN = PHINode::Create(Ty: ShadowTy, NumReservedValues: PN.getNumIncomingValues(), NameStr: "",
3452 InsertBefore: PN.getIterator());
3453
3454 // Give the shadow phi node valid predecessors to fool SplitEdge into working.
3455 Value *UndefShadow = UndefValue::get(T: ShadowTy);
3456 for (BasicBlock *BB : PN.blocks())
3457 ShadowPN->addIncoming(V: UndefShadow, BB);
3458
3459 DFSF.setShadow(I: &PN, Shadow: ShadowPN);
3460
3461 PHINode *OriginPN = nullptr;
3462 if (DFSF.DFS.shouldTrackOrigins()) {
3463 OriginPN = PHINode::Create(Ty: DFSF.DFS.OriginTy, NumReservedValues: PN.getNumIncomingValues(), NameStr: "",
3464 InsertBefore: PN.getIterator());
3465 Value *UndefOrigin = UndefValue::get(T: DFSF.DFS.OriginTy);
3466 for (BasicBlock *BB : PN.blocks())
3467 OriginPN->addIncoming(V: UndefOrigin, BB);
3468 DFSF.setOrigin(I: &PN, Origin: OriginPN);
3469 }
3470
3471 DFSF.PHIFixups.push_back(x: {.Phi: &PN, .ShadowPhi: ShadowPN, .OriginPhi: OriginPN});
3472}
3473
3474PreservedAnalyses DataFlowSanitizerPass::run(Module &M,
3475 ModuleAnalysisManager &AM) {
3476 auto GetTLI = [&](Function &F) -> TargetLibraryInfo & {
3477 auto &FAM =
3478 AM.getResult<FunctionAnalysisManagerModuleProxy>(IR&: M).getManager();
3479 return FAM.getResult<TargetLibraryAnalysis>(IR&: F);
3480 };
3481 if (!DataFlowSanitizer(ABIListFiles).runImpl(M, GetTLI))
3482 return PreservedAnalyses::all();
3483
3484 PreservedAnalyses PA = PreservedAnalyses::none();
3485 // GlobalsAA is considered stateless and does not get invalidated unless
3486 // explicitly invalidated; PreservedAnalyses::none() is not enough. Sanitizers
3487 // make changes that require GlobalsAA to be invalidated.
3488 PA.abandon<GlobalsAA>();
3489 return PA;
3490}
3491