maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/clang/test/CodeGen/catch-nullptr-and-nonzero-o...

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

422 lines
38 KiB
C
Raw Normal View History

[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// RUN: %clang_cc1 -x c -emit-llvm %s -o - -triple x86_64-linux-gnu | FileCheck %s --check-prefixes=CHECK,CHECK-NOSANITIZE
// RUN: %clang_cc1 -x c -fsanitize=pointer-overflow -fno-sanitize-recover=pointer-overflow -emit-llvm %s -o - -triple x86_64-linux-gnu | FileCheck %s -implicit-check-not="call void @__ubsan_handle_pointer_overflow" --check-prefixes=CHECK,CHECK-SANITIZE,CHECK-SANITIZE-ANYRECOVER,CHECK-SANITIZE-NORECOVER,CHECK-SANITIZE-UNREACHABLE,CHECK-SANITIZE-C,CHECK-SANITIZE-ANYRECOVER-C,CHECK-SANITIZE-NORECOVER-C,CHECK-SANITIZE-UNREACHABLE-C
// RUN: %clang_cc1 -x c -fsanitize=pointer-overflow -fsanitize-recover=pointer-overflow -emit-llvm %s -o - -triple x86_64-linux-gnu | FileCheck %s -implicit-check-not="call void @__ubsan_handle_pointer_overflow" --check-prefixes=CHECK,CHECK-SANITIZE,CHECK-SANITIZE-ANYRECOVER,CHECK-SANITIZE-RECOVER,CHECK-SANITIZE-C,CHECK-SANITIZE-ANYRECOVER-C,CHECK-SANITIZE-RECOVER-C
// RUN: %clang_cc1 -x c -fsanitize=pointer-overflow -fsanitize-trap=pointer-overflow -emit-llvm %s -o - -triple x86_64-linux-gnu | FileCheck %s -implicit-check-not="call void @__ubsan_handle_pointer_overflow" --check-prefixes=CHECK,CHECK-SANITIZE,CHECK-SANITIZE-TRAP,CHECK-SANITIZE-UNREACHABLE,CHECK-SANITIZE-C,CHECK-SANITIZE-TRAP-C,CHECK-SANITIZE-UNREACHABLE-C
// RUN: %clang_cc1 -x c++ -emit-llvm %s -o - -triple x86_64-linux-gnu | FileCheck %s --check-prefixes=CHECK,CHECK-NOSANITIZE
// RUN: %clang_cc1 -x c++ -fsanitize=pointer-overflow -fno-sanitize-recover=pointer-overflow -emit-llvm %s -o - -triple x86_64-linux-gnu | FileCheck %s -implicit-check-not="call void @__ubsan_handle_pointer_overflow" --check-prefixes=CHECK,CHECK-SANITIZE,CHECK-SANITIZE-ANYRECOVER,CHECK-SANITIZE-NORECOVER,CHECK-SANITIZE-UNREACHABLE,CHECK-SANITIZE-CPP,CHECK-SANITIZE-ANYRECOVER-CPP,CHECK-SANITIZE-NORECOVER-CPP,CHECK-SANITIZE-UNREACHABLE-CPP
// RUN: %clang_cc1 -x c++ -fsanitize=pointer-overflow -fsanitize-recover=pointer-overflow -emit-llvm %s -o - -triple x86_64-linux-gnu | FileCheck %s -implicit-check-not="call void @__ubsan_handle_pointer_overflow" --check-prefixes=CHECK,CHECK-SANITIZE,CHECK-SANITIZE-ANYRECOVER,CHECK-SANITIZE-RECOVER,CHECK-SANITIZE-CPP,CHECK-SANITIZE-ANYRECOVER-CPP,CHECK-SANITIZE-RECOVER-CPP
// RUN: %clang_cc1 -x c++ -fsanitize=pointer-overflow -fsanitize-trap=pointer-overflow -emit-llvm %s -o - -triple x86_64-linux-gnu | FileCheck %s -implicit-check-not="call void @__ubsan_handle_pointer_overflow" --check-prefixes=CHECK,CHECK-SANITIZE,CHECK-SANITIZE-TRAP,CHECK-SANITIZE-UNREACHABLE,CHECK-SANITIZE-CPP,CHECK-SANITIZE-TRAP-CPP,CHECK-SANITIZE-UNREACHABLE-CPP
// In C++/LLVM IR, if the base pointer evaluates to a null pointer value,
// the only valid pointer this inbounds GEP can produce is also a null pointer.
// Likewise, if we have non-zero base pointer, we can not get null pointer
// as a result, so the offset can not be -int(BasePtr).
// So in other words, the offset can not change "null status" of the pointer,
// as in if the pointer was null, it can not become non-null, and vice versa,
// if it was non-null, it can not become null.
// In C, however, offsetting null pointer is completely undefined, even by 0.
// CHECK-SANITIZE-ANYRECOVER-DAG: @[[LINE_100:.*]] = {{.*}}, i32 100, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-C-DAG: @[[LINE_200:.*]] = {{.*}}, i32 200, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-DAG: @[[LINE_300:.*]] = {{.*}}, i32 300, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-DAG: @[[LINE_400:.*]] = {{.*}}, i32 400, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-DAG: @[[LINE_500:.*]] = {{.*}}, i32 500, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-C-DAG: @[[LINE_600:.*]] = {{.*}}, i32 600, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-DAG: @[[LINE_700:.*]] = {{.*}}, i32 700, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-DAG: @[[LINE_800:.*]] = {{.*}}, i32 800, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-DAG: @[[LINE_900:.*]] = {{.*}}, i32 900, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-C-DAG: @[[LINE_1000:.*]] = {{.*}}, i32 1000, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-DAG: @[[LINE_1100:.*]] = {{.*}}, i32 1100, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-DAG: @[[LINE_1200:.*]] = {{.*}}, i32 1200, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-DAG: @[[LINE_1300:.*]] = {{.*}}, i32 1300, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-C-DAG: @[[LINE_1400:.*]] = {{.*}}, i32 1400, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-DAG: @[[LINE_1500:.*]] = {{.*}}, i32 1500, i32 15 } }
// CHECK-SANITIZE-ANYRECOVER-DAG: @[[LINE_1600:.*]] = {{.*}}, i32 1600, i32 15 } }
#ifdef __cplusplus
extern "C" {
#endif
char *var_var(char *base, unsigned long offset) {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @var_var(i8* %[[BASE:.*]], i64 %[[OFFSET:.*]])
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-NEXT: %[[BASE_ADDR:.*]] = alloca i8*, align 8
// CHECK-NEXT: %[[OFFSET_ADDR:.*]] = alloca i64, align 8
// CHECK-NEXT: store i8* %[[BASE]], i8** %[[BASE_ADDR]], align 8
// CHECK-NEXT: store i64 %[[OFFSET]], i64* %[[OFFSET_ADDR]], align 8
// CHECK-NEXT: %[[BASE_RELOADED:.*]] = load i8*, i8** %[[BASE_ADDR]], align 8
// CHECK-NEXT: %[[OFFSET_RELOADED:.*]] = load i64, i64* %[[OFFSET_ADDR]], align 8
// CHECK-NEXT: %[[ADD_PTR:.*]] = getelementptr inbounds i8, i8* %[[BASE_RELOADED]], i64 %[[OFFSET_RELOADED]]
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET_AGGREGATE:.*]] = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 1, i64 %[[OFFSET_RELOADED]]), !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET_OVERFLOWED:.*]] = extractvalue { i64, i1 } %[[COMPUTED_OFFSET_AGGREGATE]], 1, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET:.*]] = extractvalue { i64, i1 } %[[COMPUTED_OFFSET_AGGREGATE]], 0, !nosanitize
// CHECK-SANITIZE-NEXT: %[[BASE_RELOADED_INT:.*]] = ptrtoint i8* %[[BASE_RELOADED]] to i64, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP:.*]] = add i64 %[[BASE_RELOADED_INT]], %[[COMPUTED_OFFSET]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[BASE_IS_NOT_NULLPTR:.*]] = icmp ne i8* %[[BASE_RELOADED]], null, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP_IS_NOT_NULL:.*]] = icmp ne i64 %[[COMPUTED_GEP]], 0, !nosanitize
// CHECK-SANITIZE-C-NEXT: %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL:.*]] = and i1 %[[BASE_IS_NOT_NULLPTR]], %[[COMPUTED_GEP_IS_NOT_NULL]], !nosanitize
// CHECK-SANITIZE-CPP-NEXT: %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL:.*]] = icmp eq i1 %[[BASE_IS_NOT_NULLPTR]], %[[COMPUTED_GEP_IS_NOT_NULL]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET_DID_NOT_OVERFLOW:.*]] = xor i1 %[[COMPUTED_OFFSET_OVERFLOWED]], true, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP_IS_UGE_BASE:.*]] = icmp uge i64 %[[COMPUTED_GEP]], %[[BASE_RELOADED_INT]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[GEP_DID_NOT_OVERFLOW:.*]] = and i1 %[[COMPUTED_GEP_IS_UGE_BASE]], %[[COMPUTED_OFFSET_DID_NOT_OVERFLOW]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[GEP_IS_OKAY:.*]] = and i1 %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL]], %[[GEP_DID_NOT_OVERFLOW]], !nosanitize
// CHECK-SANITIZE-NEXT: br i1 %[[GEP_IS_OKAY]], label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_100]] to i8*), i64 %[[BASE_RELOADED_INT]], i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-RECOVER-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_100]] to i8*), i64 %[[BASE_RELOADED_INT]], i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE: [[CONT]]:
// CHECK-NEXT: ret i8* %[[ADD_PTR]]
#line 100
return base + offset;
}
char *var_zero(char *base) {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @var_zero(i8* %[[BASE:.*]])
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-NEXT: %[[BASE_ADDR:.*]] = alloca i8*, align 8
// CHECK-NEXT: store i8* %[[BASE]], i8** %[[BASE_ADDR]], align 8
// CHECK-NEXT: %[[BASE_RELOADED:.*]] = load i8*, i8** %[[BASE_ADDR]], align 8
// CHECK-NEXT: %[[ADD_PTR:.*]] = getelementptr inbounds i8, i8* %[[BASE_RELOADED]], i64 0
// CHECK-SANITIZE-C-NEXT: %[[BASE_RELOADED_INT:.*]] = ptrtoint i8* %[[BASE_RELOADED]] to i64, !nosanitize
// CHECK-SANITIZE-C-NEXT: %[[COMPUTED_GEP:.*]] = add i64 %[[BASE_RELOADED_INT]], 0, !nosanitize
// CHECK-SANITIZE-C-NEXT: %[[BASE_IS_NOT_NULLPTR:.*]] = icmp ne i8* %[[BASE_RELOADED]], null, !nosanitize
// CHECK-SANITIZE-C-NEXT: %[[COMPUTED_GEP_IS_NOT_NULL:.*]] = icmp ne i64 %[[COMPUTED_GEP]], 0, !nosanitize
// CHECK-SANITIZE-C-NEXT: %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL:.*]] = and i1 %[[BASE_IS_NOT_NULLPTR]], %[[COMPUTED_GEP_IS_NOT_NULL]], !nosanitize
// CHECK-SANITIZE-C-NEXT: %[[COMPUTED_GEP_IS_UGE_BASE:.*]] = icmp uge i64 %[[COMPUTED_GEP]], %[[BASE_RELOADED_INT]], !nosanitize
// CHECK-SANITIZE-C-NEXT: %[[GEP_IS_OKAY:.*]] = and i1 %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL]], %[[COMPUTED_GEP_IS_UGE_BASE]], !nosanitize
// CHECK-SANITIZE-C-NEXT: br i1 %[[GEP_IS_OKAY]], label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE-C: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-C-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_200]] to i8*), i64 %[[BASE_RELOADED_INT]], i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-RECOVER-C-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_200]] to i8*), i64 %[[BASE_RELOADED_INT]], i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-TRAP-C-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-C-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE-C: [[CONT]]:
// CHECK-NEXT: ret i8* %[[ADD_PTR]]
static const unsigned long offset = 0;
#line 200
return base + offset;
}
char *var_one(char *base) {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @var_one(i8* %[[BASE:.*]])
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-NEXT: %[[BASE_ADDR:.*]] = alloca i8*, align 8
// CHECK-NEXT: store i8* %[[BASE]], i8** %[[BASE_ADDR]], align 8
// CHECK-NEXT: %[[BASE_RELOADED:.*]] = load i8*, i8** %[[BASE_ADDR]], align 8
// CHECK-NEXT: %[[ADD_PTR:.*]] = getelementptr inbounds i8, i8* %[[BASE_RELOADED]], i64 1
// CHECK-SANITIZE-NEXT: %[[BASE_RELOADED_INT:.*]] = ptrtoint i8* %[[BASE_RELOADED]] to i64, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP:.*]] = add i64 %[[BASE_RELOADED_INT]], 1, !nosanitize
// CHECK-SANITIZE-NEXT: %[[BASE_IS_NOT_NULLPTR:.*]] = icmp ne i8* %[[BASE_RELOADED]], null, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP_IS_NOT_NULL:.*]] = icmp ne i64 %[[COMPUTED_GEP]], 0, !nosanitize
// CHECK-SANITIZE-C-NEXT: %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL:.*]] = and i1 %[[BASE_IS_NOT_NULLPTR]], %[[COMPUTED_GEP_IS_NOT_NULL]], !nosanitize
// CHECK-SANITIZE-CPP-NEXT: %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL:.*]] = icmp eq i1 %[[BASE_IS_NOT_NULLPTR]], %[[COMPUTED_GEP_IS_NOT_NULL]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP_IS_UGE_BASE:.*]] = icmp uge i64 %[[COMPUTED_GEP]], %[[BASE_RELOADED_INT]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[GEP_IS_OKAY:.*]] = and i1 %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL]], %[[COMPUTED_GEP_IS_UGE_BASE]], !nosanitize
// CHECK-SANITIZE-NEXT: br i1 %[[GEP_IS_OKAY]], label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_300]] to i8*), i64 %[[BASE_RELOADED_INT]], i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-RECOVER-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_300]] to i8*), i64 %[[BASE_RELOADED_INT]], i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE: [[CONT]]:
// CHECK-NEXT: ret i8* %[[ADD_PTR]]
static const unsigned long offset = 1;
#line 300
return base + offset;
}
char *var_allones(char *base) {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @var_allones(i8* %[[BASE:.*]])
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-NEXT: %[[BASE_ADDR:.*]] = alloca i8*, align 8
// CHECK-NEXT: store i8* %[[BASE]], i8** %[[BASE_ADDR]], align 8
// CHECK-NEXT: %[[BASE_RELOADED:.*]] = load i8*, i8** %[[BASE_ADDR]], align 8
// CHECK-NEXT: %[[ADD_PTR:.*]] = getelementptr inbounds i8, i8* %[[BASE_RELOADED]], i64 -1
// CHECK-SANITIZE-NEXT: %[[BASE_RELOADED_INT:.*]] = ptrtoint i8* %[[BASE_RELOADED]] to i64, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP:.*]] = add i64 %[[BASE_RELOADED_INT]], -1, !nosanitize
// CHECK-SANITIZE-NEXT: %[[BASE_IS_NOT_NULLPTR:.*]] = icmp ne i8* %[[BASE_RELOADED]], null, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP_IS_NOT_NULL:.*]] = icmp ne i64 %[[COMPUTED_GEP]], 0, !nosanitize
// CHECK-SANITIZE-C-NEXT: %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL:.*]] = and i1 %[[BASE_IS_NOT_NULLPTR]], %[[COMPUTED_GEP_IS_NOT_NULL]], !nosanitize
// CHECK-SANITIZE-CPP-NEXT: %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL:.*]] = icmp eq i1 %[[BASE_IS_NOT_NULLPTR]], %[[COMPUTED_GEP_IS_NOT_NULL]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP_IS_UGE_BASE:.*]] = icmp uge i64 %[[COMPUTED_GEP]], %[[BASE_RELOADED_INT]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[GEP_IS_OKAY:.*]] = and i1 %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL]], %[[COMPUTED_GEP_IS_UGE_BASE]], !nosanitize
// CHECK-SANITIZE-NEXT: br i1 %[[GEP_IS_OKAY]], label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_400]] to i8*), i64 %[[BASE_RELOADED_INT]], i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-RECOVER-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_400]] to i8*), i64 %[[BASE_RELOADED_INT]], i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE: [[CONT]]:
// CHECK-NEXT: ret i8* %[[ADD_PTR]]
static const unsigned long offset = -1;
#line 400
return base + offset;
}
//------------------------------------------------------------------------------
char *nullptr_var(unsigned long offset) {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @nullptr_var(i64 %[[OFFSET:.*]])
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-NEXT: %[[OFFSET_ADDR:.*]] = alloca i64, align 8
// CHECK-NEXT: store i64 %[[OFFSET]], i64* %[[OFFSET_ADDR]], align 8
// CHECK-NEXT: %[[OFFSET_RELOADED:.*]] = load i64, i64* %[[OFFSET_ADDR]], align 8
// CHECK-NEXT: %[[ADD_PTR:.*]] = getelementptr inbounds i8, i8* null, i64 %[[OFFSET_RELOADED]]
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET_AGGREGATE:.*]] = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 1, i64 %[[OFFSET_RELOADED]]), !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET_OVERFLOWED:.*]] = extractvalue { i64, i1 } %[[COMPUTED_OFFSET_AGGREGATE]], 1, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET:.*]] = extractvalue { i64, i1 } %[[COMPUTED_OFFSET_AGGREGATE]], 0, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP:.*]] = add i64 0, %[[COMPUTED_OFFSET]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP_IS_NOT_NULL:.*]] = icmp ne i64 %[[COMPUTED_GEP]], 0, !nosanitize
// CHECK-SANITIZE-C-NEXT: %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL:.*]] = and i1 false, %[[COMPUTED_GEP_IS_NOT_NULL]], !nosanitize
// CHECK-SANITIZE-CPP-NEXT: %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL:.*]] = icmp eq i1 false, %[[COMPUTED_GEP_IS_NOT_NULL]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET_DID_NOT_OVERFLOW:.*]] = xor i1 %[[COMPUTED_OFFSET_OVERFLOWED]], true, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP_IS_UGE_BASE:.*]] = icmp uge i64 %[[COMPUTED_GEP]], 0, !nosanitize
// CHECK-SANITIZE-NEXT: %[[GEP_DID_NOT_OVERFLOW:.*]] = and i1 %[[COMPUTED_GEP_IS_UGE_BASE]], %[[COMPUTED_OFFSET_DID_NOT_OVERFLOW]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[GEP_IS_OKAY:.*]] = and i1 %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL]], %[[GEP_DID_NOT_OVERFLOW]], !nosanitize
// CHECK-SANITIZE-NEXT: br i1 %[[GEP_IS_OKAY]], label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_500]] to i8*), i64 0, i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-RECOVER-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_500]] to i8*), i64 0, i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE: [[CONT]]:
// CHECK-NEXT: ret i8* %[[ADD_PTR]]
static char *const base = (char *)0;
#line 500
return base + offset;
}
char *nullptr_zero() {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @nullptr_zero()
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-SANITIZE-C-NEXT: br i1 false, label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE-C: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-C-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_600]] to i8*), i64 0, i64 0)
// CHECK-SANITIZE-RECOVER-C-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_600]] to i8*), i64 0, i64 0)
// CHECK-SANITIZE-TRAP-C-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-C-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE-C: [[CONT]]:
// CHECK-NEXT: ret i8* null
static char *const base = (char *)0;
static const unsigned long offset = 0;
#line 600
return base + offset;
}
char *nullptr_one_BAD() {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @nullptr_one_BAD()
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-SANITIZE-C-NEXT: br i1 false, label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE-CPP-NEXT: br i1 icmp eq (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* null, i64 1) to i64), i64 0), label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_700]] to i8*), i64 0, i64 ptrtoint (i8* getelementptr inbounds (i8, i8* null, i64 1) to i64))
// CHECK-SANITIZE-RECOVER-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_700]] to i8*), i64 0, i64 ptrtoint (i8* getelementptr inbounds (i8, i8* null, i64 1) to i64))
// CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE: [[CONT]]:
// CHECK-NEXT: ret i8* getelementptr inbounds (i8, i8* null, i64 1)
static char *const base = (char *)0;
static const unsigned long offset = 1;
#line 700
return base + offset;
}
char *nullptr_allones_BAD() {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @nullptr_allones_BAD()
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-SANITIZE-C-NEXT: br i1 false, label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE-CPP-NEXT: br i1 icmp eq (i64 mul (i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64), i64 -1), i64 0), label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_800]] to i8*), i64 0, i64 mul (i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64), i64 -1))
// CHECK-SANITIZE-RECOVER-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_800]] to i8*), i64 0, i64 mul (i64 ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64), i64 -1))
// CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE: [[CONT]]:
// CHECK-NEXT: ret i8* getelementptr inbounds (i8, i8* null, i64 -1)
static char *const base = (char *)0;
static const unsigned long offset = -1;
#line 800
return base + offset;
}
//------------------------------------------------------------------------------
char *one_var(unsigned long offset) {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @one_var(i64 %[[OFFSET:.*]])
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-NEXT: %[[OFFSET_ADDR:.*]] = alloca i64, align 8
// CHECK-NEXT: store i64 %[[OFFSET]], i64* %[[OFFSET_ADDR]], align 8
// CHECK-NEXT: %[[OFFSET_RELOADED:.*]] = load i64, i64* %[[OFFSET_ADDR]], align 8
// CHECK-NEXT: %[[ADD_PTR:.*]] = getelementptr inbounds i8, i8* inttoptr (i64 1 to i8*), i64 %[[OFFSET_RELOADED]]
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET_AGGREGATE:.*]] = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 1, i64 %[[OFFSET_RELOADED]]), !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET_OVERFLOWED:.*]] = extractvalue { i64, i1 } %[[COMPUTED_OFFSET_AGGREGATE]], 1, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET:.*]] = extractvalue { i64, i1 } %[[COMPUTED_OFFSET_AGGREGATE]], 0, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP:.*]] = add i64 1, %[[COMPUTED_OFFSET]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP_IS_NOT_NULL:.*]] = icmp ne i64 %[[COMPUTED_GEP]], 0, !nosanitize
// CHECK-SANITIZE-C-NEXT: %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL:.*]] = and i1 icmp ne (i8* inttoptr (i64 1 to i8*), i8* null), %[[COMPUTED_GEP_IS_NOT_NULL]], !nosanitize
// CHECK-SANITIZE-CPP-NEXT: %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL:.*]] = icmp eq i1 icmp ne (i8* inttoptr (i64 1 to i8*), i8* null), %[[COMPUTED_GEP_IS_NOT_NULL]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET_DID_NOT_OVERFLOW:.*]] = xor i1 %[[COMPUTED_OFFSET_OVERFLOWED]], true, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP_IS_UGE_BASE:.*]] = icmp uge i64 %[[COMPUTED_GEP]], 1, !nosanitize
// CHECK-SANITIZE-NEXT: %[[GEP_DID_NOT_OVERFLOW:.*]] = and i1 %[[COMPUTED_GEP_IS_UGE_BASE]], %[[COMPUTED_OFFSET_DID_NOT_OVERFLOW]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[GEP_IS_OKAY:.*]] = and i1 %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL]], %[[GEP_DID_NOT_OVERFLOW]], !nosanitize
// CHECK-SANITIZE-NEXT: br i1 %[[GEP_IS_OKAY]], label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_900]] to i8*), i64 1, i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-RECOVER-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_900]] to i8*), i64 1, i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE: [[CONT]]:
// CHECK-NEXT: ret i8* %[[ADD_PTR]]
static char *const base = (char *)1;
#line 900
return base + offset;
}
char *one_zero() {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @one_zero()
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-SANITIZE-C-NEXT: br i1 icmp ne (i8* inttoptr (i64 1 to i8*), i8* null), label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE-C: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-C-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_1000]] to i8*), i64 1, i64 1)
// CHECK-SANITIZE-RECOVER-C-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_1000]] to i8*), i64 1, i64 1)
// CHECK-SANITIZE-TRAP-C-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-C-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE-C: [[CONT]]:
// CHECK-NEXT: ret i8* inttoptr (i64 1 to i8*)
static char *const base = (char *)1;
static const unsigned long offset = 0;
#line 1000
return base + offset;
}
char *one_one_OK() {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @one_one_OK()
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-SANITIZE-C-NEXT: br i1 and (i1 icmp ne (i8* inttoptr (i64 1 to i8*), i8* null), i1 icmp ne (i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 1 to i8*), i64 1) to i64), i64 1), i64 1), i64 0)), label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE-CPP-NEXT: br i1 xor (i1 icmp eq (i8* inttoptr (i64 1 to i8*), i8* null), i1 icmp ne (i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 1 to i8*), i64 1) to i64), i64 1), i64 1), i64 0)), label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_1100]] to i8*), i64 1, i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 1 to i8*), i64 1) to i64), i64 1), i64 1))
// CHECK-SANITIZE-RECOVER-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_1100]] to i8*), i64 1, i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 1 to i8*), i64 1) to i64), i64 1), i64 1))
// CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE: [[CONT]]:
// CHECK-NEXT: ret i8* getelementptr inbounds (i8, i8* inttoptr (i64 1 to i8*), i64 1)
static char *const base = (char *)1;
static const unsigned long offset = 1;
#line 1100
return base + offset;
}
char *one_allones_BAD() {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @one_allones_BAD()
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-SANITIZE-C-NEXT: br i1 and (i1 icmp ne (i8* inttoptr (i64 1 to i8*), i8* null), i1 icmp ne (i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 1 to i8*), i64 -1) to i64), i64 1), i64 1), i64 0)), label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE-CPP-NEXT: br i1 xor (i1 icmp eq (i8* inttoptr (i64 1 to i8*), i8* null), i1 icmp ne (i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 1 to i8*), i64 -1) to i64), i64 1), i64 1), i64 0)), label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_1200]] to i8*), i64 1, i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 1 to i8*), i64 -1) to i64), i64 1), i64 1))
// CHECK-SANITIZE-RECOVER-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_1200]] to i8*), i64 1, i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 1 to i8*), i64 -1) to i64), i64 1), i64 1))
// CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE: [[CONT]]:
// CHECK-NEXT: ret i8* getelementptr inbounds (i8, i8* inttoptr (i64 1 to i8*), i64 -1)
static char *const base = (char *)1;
static const unsigned long offset = -1;
#line 1200
return base + offset;
}
//------------------------------------------------------------------------------
char *allones_var(unsigned long offset) {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @allones_var(i64 %[[OFFSET:.*]])
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-NEXT: %[[OFFSET_ADDR:.*]] = alloca i64, align 8
// CHECK-NEXT: store i64 %[[OFFSET]], i64* %[[OFFSET_ADDR]], align 8
// CHECK-NEXT: %[[OFFSET_RELOADED:.*]] = load i64, i64* %[[OFFSET_ADDR]], align 8
// CHECK-NEXT: %[[ADD_PTR:.*]] = getelementptr inbounds i8, i8* inttoptr (i64 -1 to i8*), i64 %[[OFFSET_RELOADED]]
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET_AGGREGATE:.*]] = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 1, i64 %[[OFFSET_RELOADED]]), !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET_OVERFLOWED:.*]] = extractvalue { i64, i1 } %[[COMPUTED_OFFSET_AGGREGATE]], 1, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET:.*]] = extractvalue { i64, i1 } %[[COMPUTED_OFFSET_AGGREGATE]], 0, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP:.*]] = add i64 -1, %[[COMPUTED_OFFSET]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP_IS_NOT_NULL:.*]] = icmp ne i64 %[[COMPUTED_GEP]], 0, !nosanitize
// CHECK-SANITIZE-C-NEXT: %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL:.*]] = and i1 icmp ne (i8* inttoptr (i64 -1 to i8*), i8* null), %[[COMPUTED_GEP_IS_NOT_NULL]], !nosanitize
// CHECK-SANITIZE-CPP-NEXT: %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL:.*]] = icmp eq i1 icmp ne (i8* inttoptr (i64 -1 to i8*), i8* null), %[[COMPUTED_GEP_IS_NOT_NULL]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_OFFSET_DID_NOT_OVERFLOW:.*]] = xor i1 %[[COMPUTED_OFFSET_OVERFLOWED]], true, !nosanitize
// CHECK-SANITIZE-NEXT: %[[COMPUTED_GEP_IS_UGE_BASE:.*]] = icmp uge i64 %[[COMPUTED_GEP]], -1, !nosanitize
// CHECK-SANITIZE-NEXT: %[[GEP_DID_NOT_OVERFLOW:.*]] = and i1 %[[COMPUTED_GEP_IS_UGE_BASE]], %[[COMPUTED_OFFSET_DID_NOT_OVERFLOW]], !nosanitize
// CHECK-SANITIZE-NEXT: %[[GEP_IS_OKAY:.*]] = and i1 %[[BOTH_POINTERS_ARE_NULL_OR_BOTH_ARE_NONNULL]], %[[GEP_DID_NOT_OVERFLOW]], !nosanitize
// CHECK-SANITIZE-NEXT: br i1 %[[GEP_IS_OKAY]], label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_1300]] to i8*), i64 -1, i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-RECOVER-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_1300]] to i8*), i64 -1, i64 %[[COMPUTED_GEP]])
// CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE: [[CONT]]:
// CHECK-NEXT: ret i8* %[[ADD_PTR]]
static char *const base = (char *)-1;
#line 1300
return base + offset;
}
char *allones_zero_OK() {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @allones_zero_OK()
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-SANITIZE-C-NEXT: br i1 icmp ne (i8* inttoptr (i64 -1 to i8*), i8* null), label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE-C: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-C-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_1400]] to i8*), i64 -1, i64 -1)
// CHECK-SANITIZE-RECOVER-C-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_1400]] to i8*), i64 -1, i64 -1)
// CHECK-SANITIZE-TRAP-C-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-C-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE-C: [[CONT]]:
// CHECK-NEXT: ret i8* inttoptr (i64 -1 to i8*)
static char *const base = (char *)-1;
static const unsigned long offset = 0;
#line 1400
return base + offset;
}
char *allones_one_BAD() {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @allones_one_BAD()
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-SANITIZE-C-NEXT: br i1 and (i1 icmp ne (i8* inttoptr (i64 -1 to i8*), i8* null), i1 icmp ne (i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 -1 to i8*), i64 1) to i64), i64 -1), i64 -1), i64 0)), label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE-CPP-NEXT: br i1 xor (i1 icmp eq (i8* inttoptr (i64 -1 to i8*), i8* null), i1 icmp ne (i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 -1 to i8*), i64 1) to i64), i64 -1), i64 -1), i64 0)), label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_1500]] to i8*), i64 -1, i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 -1 to i8*), i64 1) to i64), i64 -1), i64 -1))
// CHECK-SANITIZE-RECOVER-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_1500]] to i8*), i64 -1, i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 -1 to i8*), i64 1) to i64), i64 -1), i64 -1))
// CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE: [[CONT]]:
// CHECK-NEXT: ret i8* getelementptr inbounds (i8, i8* inttoptr (i64 -1 to i8*), i64 1)
static char *const base = (char *)-1;
static const unsigned long offset = 1;
#line 1500
return base + offset;
}
char *allones_allones_OK() {
Revert "[CodeGenModule] Assume dso_local for -fpic -fno-semantic-interposition" This reverts commit 789a46f2d742e11edaade28cb59a0f4d2a1d770e. Accidentally committed.
2020-02-04 02:09:39 +08:00
// CHECK: define i8* @allones_allones_OK()
[UBSan][clang][compiler-rt] Applying non-zero offset to nullptr is undefined behaviour Summary: Quote from http://eel.is/c++draft/expr.add#4: ``` 4 When an expression J that has integral type is added to or subtracted from an expression P of pointer type, the result has the type of P. (4.1) If P evaluates to a null pointer value and J evaluates to 0, the result is a null pointer value. (4.2) Otherwise, if P points to an array element i of an array object x with n elements ([dcl.array]), the expressions P + J and J + P (where J has the value j) point to the (possibly-hypothetical) array element i+j of x if 0≤i+j≤n and the expression P - J points to the (possibly-hypothetical) array element i−j of x if 0≤i−j≤n. (4.3) Otherwise, the behavior is undefined. ``` Therefore, as per the standard, applying non-zero offset to `nullptr` (or making non-`nullptr` a `nullptr`, by subtracting pointer's integral value from the pointer itself) is undefined behavior. (*if* `nullptr` is not defined, i.e. e.g. `-fno-delete-null-pointer-checks` was *not* specified.) To make things more fun, in C (6.5.6p8), applying *any* offset to null pointer is undefined, although Clang front-end pessimizes the code by not lowering that info, so this UB is "harmless". Since rL369789 (D66608 `[InstCombine] icmp eq/ne (gep inbounds P, Idx..), null -> icmp eq/ne P, null`) LLVM middle-end uses those guarantees for transformations. If the source contains such UB's, said code may now be miscompiled. Such miscompilations were already observed: * https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20190826/687838.html * https://github.com/google/filament/pull/1566 Surprisingly, UBSan does not catch those issues ... until now. This diff teaches UBSan about these UB's. `getelementpointer inbounds` is a pretty frequent instruction, so this does have a measurable impact on performance; I've addressed most of the obvious missing folds (and thus decreased the performance impact by ~5%), and then re-performed some performance measurements using my [[ https://github.com/darktable-org/rawspeed | RawSpeed ]] benchmark: (all measurements done with LLVM ToT, the sanitizer never fired.) * no sanitization vs. existing check: average `+21.62%` slowdown * existing check vs. check after this patch: average `22.04%` slowdown * no sanitization vs. this patch: average `48.42%` slowdown Reviewers: vsk, filcab, rsmith, aaron.ballman, vitalybuka, rjmccall, #sanitizers Reviewed By: rsmith Subscribers: kristof.beyls, nickdesaulniers, nikic, ychen, dtzWill, xbolva00, dberris, arphaman, rupprecht, reames, regehr, llvm-commits, cfe-commits Tags: #clang, #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D67122 llvm-svn: 374293
2019-10-10 17:25:02 +08:00
// CHECK-NEXT: [[ENTRY:.*]]:
// CHECK-SANITIZE-C-NEXT: br i1 and (i1 icmp ne (i8* inttoptr (i64 -1 to i8*), i8* null), i1 icmp ne (i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 -1 to i8*), i64 -1) to i64), i64 -1), i64 -1), i64 0)), label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE-CPP-NEXT: br i1 xor (i1 icmp eq (i8* inttoptr (i64 -1 to i8*), i8* null), i1 icmp ne (i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 -1 to i8*), i64 -1) to i64), i64 -1), i64 -1), i64 0)), label %[[CONT:.*]], label %[[HANDLER_POINTER_OVERFLOW:[^,]+]],{{.*}} !nosanitize
// CHECK-SANITIZE: [[HANDLER_POINTER_OVERFLOW]]:
// CHECK-SANITIZE-NORECOVER-NEXT: call void @__ubsan_handle_pointer_overflow_abort(i8* bitcast ({ {{{.*}}} }* @[[LINE_1600]] to i8*), i64 -1, i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 -1 to i8*), i64 -1) to i64), i64 -1), i64 -1))
// CHECK-SANITIZE-RECOVER-NEXT: call void @__ubsan_handle_pointer_overflow(i8* bitcast ({ {{{.*}}} }* @[[LINE_1600]] to i8*), i64 -1, i64 add (i64 sub (i64 ptrtoint (i8* getelementptr inbounds (i8, i8* inttoptr (i64 -1 to i8*), i64 -1) to i64), i64 -1), i64 -1))
// CHECK-SANITIZE-TRAP-NEXT: call void @llvm.trap(){{.*}}, !nosanitize
// CHECK-SANITIZE-UNREACHABLE-NEXT: unreachable, !nosanitize
// CHECK-SANITIZE: [[CONT]]:
// CHECK-NEXT: ret i8* getelementptr inbounds (i8, i8* inttoptr (i64 -1 to i8*), i64 -1)
static char *const base = (char *)-1;
static const unsigned long offset = -1;
#line 1600
return base + offset;
}
#ifdef __cplusplus
}
#endif