llvm-project/compiler-rt/lib/hwasan/hwasan_report.cpp

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

782 lines
29 KiB
C++
Raw Normal View History

//===-- hwasan_report.cpp -------------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
// This file is a part of HWAddressSanitizer.
//
// Error reporting.
//===----------------------------------------------------------------------===//
#include "hwasan_report.h"
#include <dlfcn.h>
#include "hwasan.h"
#include "hwasan_allocator.h"
#include "hwasan_globals.h"
#include "hwasan_mapping.h"
#include "hwasan_thread.h"
#include "hwasan_thread_list.h"
#include "sanitizer_common/sanitizer_allocator_internal.h"
#include "sanitizer_common/sanitizer_common.h"
#include "sanitizer_common/sanitizer_flags.h"
#include "sanitizer_common/sanitizer_mutex.h"
#include "sanitizer_common/sanitizer_report_decorator.h"
#include "sanitizer_common/sanitizer_stackdepot.h"
#include "sanitizer_common/sanitizer_stacktrace_printer.h"
#include "sanitizer_common/sanitizer_symbolizer.h"
using namespace __sanitizer;
namespace __hwasan {
class ScopedReport {
public:
ScopedReport(bool fatal = false) : error_message_(1), fatal(fatal) {
Lock lock(&error_message_lock_);
error_message_ptr_ = fatal ? &error_message_ : nullptr;
++hwasan_report_count;
}
~ScopedReport() {
void (*report_cb)(const char *);
{
Lock lock(&error_message_lock_);
report_cb = error_report_callback_;
error_message_ptr_ = nullptr;
}
if (report_cb)
report_cb(error_message_.data());
if (fatal)
SetAbortMessage(error_message_.data());
if (common_flags()->print_module_map >= 2 ||
(fatal && common_flags()->print_module_map))
DumpProcessMap();
if (fatal)
Die();
}
static void MaybeAppendToErrorMessage(const char *msg) {
Lock lock(&error_message_lock_);
if (!error_message_ptr_)
return;
uptr len = internal_strlen(msg);
uptr old_size = error_message_ptr_->size();
error_message_ptr_->resize(old_size + len);
// overwrite old trailing '\0', keep new trailing '\0' untouched.
internal_memcpy(&(*error_message_ptr_)[old_size - 1], msg, len);
}
static void SetErrorReportCallback(void (*callback)(const char *)) {
Lock lock(&error_message_lock_);
error_report_callback_ = callback;
}
private:
ScopedErrorReportLock error_report_lock_;
InternalMmapVector<char> error_message_;
bool fatal;
static InternalMmapVector<char> *error_message_ptr_;
static Mutex error_message_lock_;
static void (*error_report_callback_)(const char *);
};
InternalMmapVector<char> *ScopedReport::error_message_ptr_;
Mutex ScopedReport::error_message_lock_;
void (*ScopedReport::error_report_callback_)(const char *);
// If there is an active ScopedReport, append to its error message.
void AppendToErrorMessageBuffer(const char *buffer) {
ScopedReport::MaybeAppendToErrorMessage(buffer);
}
static StackTrace GetStackTraceFromId(u32 id) {
CHECK(id);
StackTrace res = StackDepotGet(id);
CHECK(res.trace);
return res;
}
// A RAII object that holds a copy of the current thread stack ring buffer.
// The actual stack buffer may change while we are iterating over it (for
// example, Printf may call syslog() which can itself be built with hwasan).
class SavedStackAllocations {
public:
SavedStackAllocations(StackAllocationsRingBuffer *rb) {
uptr size = rb->size() * sizeof(uptr);
void *storage =
MmapAlignedOrDieOnFatalError(size, size * 2, "saved stack allocations");
new (&rb_) StackAllocationsRingBuffer(*rb, storage);
}
~SavedStackAllocations() {
StackAllocationsRingBuffer *rb = get();
UnmapOrDie(rb->StartOfStorage(), rb->size() * sizeof(uptr));
}
StackAllocationsRingBuffer *get() {
return (StackAllocationsRingBuffer *)&rb_;
}
private:
uptr rb_;
};
class Decorator: public __sanitizer::SanitizerCommonDecorator {
public:
Decorator() : SanitizerCommonDecorator() { }
const char *Access() { return Blue(); }
const char *Allocation() const { return Magenta(); }
const char *Origin() const { return Magenta(); }
const char *Name() const { return Green(); }
const char *Location() { return Green(); }
const char *Thread() { return Green(); }
};
static bool FindHeapAllocation(HeapAllocationsRingBuffer *rb, uptr tagged_addr,
HeapAllocationRecord *har, uptr *ring_index,
uptr *num_matching_addrs,
uptr *num_matching_addrs_4b) {
if (!rb) return false;
*num_matching_addrs = 0;
*num_matching_addrs_4b = 0;
for (uptr i = 0, size = rb->size(); i < size; i++) {
auto h = (*rb)[i];
if (h.tagged_addr <= tagged_addr &&
h.tagged_addr + h.requested_size > tagged_addr) {
*har = h;
*ring_index = i;
return true;
}
// Measure the number of heap ring buffer entries that would have matched
// if we had only one entry per address (e.g. if the ring buffer data was
// stored at the address itself). This will help us tune the allocator
// implementation for MTE.
if (UntagAddr(h.tagged_addr) <= UntagAddr(tagged_addr) &&
UntagAddr(h.tagged_addr) + h.requested_size > UntagAddr(tagged_addr)) {
++*num_matching_addrs;
}
// Measure the number of heap ring buffer entries that would have matched
// if we only had 4 tag bits, which is the case for MTE.
auto untag_4b = [](uptr p) {
return p & ((1ULL << 60) - 1);
};
if (untag_4b(h.tagged_addr) <= untag_4b(tagged_addr) &&
untag_4b(h.tagged_addr) + h.requested_size > untag_4b(tagged_addr)) {
++*num_matching_addrs_4b;
}
}
return false;
}
static void PrintStackAllocations(StackAllocationsRingBuffer *sa,
tag_t addr_tag, uptr untagged_addr) {
uptr frames = Min((uptr)flags()->stack_history_size, sa->size());
bool found_local = false;
for (uptr i = 0; i < frames; i++) {
const uptr *record_addr = &(*sa)[i];
uptr record = *record_addr;
if (!record)
break;
tag_t base_tag =
reinterpret_cast<uptr>(record_addr) >> kRecordAddrBaseTagShift;
uptr fp = (record >> kRecordFPShift) << kRecordFPLShift;
uptr pc_mask = (1ULL << kRecordFPShift) - 1;
uptr pc = record & pc_mask;
FrameInfo frame;
if (Symbolizer::GetOrInit()->SymbolizeFrame(pc, &frame)) {
for (LocalInfo &local : frame.locals) {
if (!local.has_frame_offset || !local.has_size || !local.has_tag_offset)
continue;
tag_t obj_tag = base_tag ^ local.tag_offset;
if (obj_tag != addr_tag)
continue;
// Calculate the offset from the object address to the faulting
// address. Because we only store bits 4-19 of FP (bits 0-3 are
// guaranteed to be zero), the calculation is performed mod 2^20 and may
// harmlessly underflow if the address mod 2^20 is below the object
// address.
uptr obj_offset =
(untagged_addr - fp - local.frame_offset) & (kRecordFPModulus - 1);
if (obj_offset >= local.size)
continue;
if (!found_local) {
Printf("Potentially referenced stack objects:\n");
found_local = true;
}
Printf(" %s in %s %s:%d\n", local.name, local.function_name,
local.decl_file, local.decl_line);
}
frame.Clear();
}
}
if (found_local)
return;
// We didn't find any locals. Most likely we don't have symbols, so dump
// the information that we have for offline analysis.
InternalScopedString frame_desc;
Printf("Previously allocated frames:\n");
for (uptr i = 0; i < frames; i++) {
const uptr *record_addr = &(*sa)[i];
uptr record = *record_addr;
if (!record)
break;
uptr pc_mask = (1ULL << 48) - 1;
uptr pc = record & pc_mask;
frame_desc.append(" record_addr:0x%zx record:0x%zx",
reinterpret_cast<uptr>(record_addr), record);
if (SymbolizedStack *frame = Symbolizer::GetOrInit()->SymbolizePC(pc)) {
RenderFrame(&frame_desc, " %F %L", 0, frame->info.address, &frame->info,
common_flags()->symbolize_vs_style,
common_flags()->strip_path_prefix);
frame->ClearAll();
}
Printf("%s\n", frame_desc.data());
frame_desc.clear();
}
}
hwasan: Improve precision of checks using short granule tags. A short granule is a granule of size between 1 and `TG-1` bytes. The size of a short granule is stored at the location in shadow memory where the granule's tag is normally stored, while the granule's actual tag is stored in the last byte of the granule. This means that in order to verify that a pointer tag matches a memory tag, HWASAN must check for two possibilities: * the pointer tag is equal to the memory tag in shadow memory, or * the shadow memory tag is actually a short granule size, the value being loaded is in bounds of the granule and the pointer tag is equal to the last byte of the granule. Pointer tags between 1 to `TG-1` are possible and are as likely as any other tag. This means that these tags in memory have two interpretations: the full tag interpretation (where the pointer tag is between 1 and `TG-1` and the last byte of the granule is ordinary data) and the short tag interpretation (where the pointer tag is stored in the granule). When HWASAN detects an error near a memory tag between 1 and `TG-1`, it will show both the memory tag and the last byte of the granule. Currently, it is up to the user to disambiguate the two possibilities. Because this functionality obsoletes the right aligned heap feature of the HWASAN memory allocator (and because we can no longer easily test it), the feature is removed. Also update the documentation to cover both short granule tags and outlined checks. Differential Revision: https://reviews.llvm.org/D63908 llvm-svn: 365551
2019-07-10 04:22:36 +08:00
// Returns true if tag == *tag_ptr, reading tags from short granules if
// necessary. This may return a false positive if tags 1-15 are used as a
// regular tag rather than a short granule marker.
static bool TagsEqual(tag_t tag, tag_t *tag_ptr) {
if (tag == *tag_ptr)
return true;
if (*tag_ptr == 0 || *tag_ptr > kShadowAlignment - 1)
return false;
uptr mem = ShadowToMem(reinterpret_cast<uptr>(tag_ptr));
tag_t inline_tag = *reinterpret_cast<tag_t *>(mem + kShadowAlignment - 1);
return tag == inline_tag;
}
// HWASan globals store the size of the global in the descriptor. In cases where
// we don't have a binary with symbols, we can't grab the size of the global
// from the debug info - but we might be able to retrieve it from the
// descriptor. Returns zero if the lookup failed.
static uptr GetGlobalSizeFromDescriptor(uptr ptr) {
// Find the ELF object that this global resides in.
Dl_info info;
if (dladdr(reinterpret_cast<void *>(ptr), &info) == 0)
return 0;
auto *ehdr = reinterpret_cast<const ElfW(Ehdr) *>(info.dli_fbase);
auto *phdr_begin = reinterpret_cast<const ElfW(Phdr) *>(
reinterpret_cast<const u8 *>(ehdr) + ehdr->e_phoff);
// Get the load bias. This is normally the same as the dli_fbase address on
// position-independent code, but can be different on non-PIE executables,
// binaries using LLD's partitioning feature, or binaries compiled with a
// linker script.
ElfW(Addr) load_bias = 0;
for (const auto &phdr :
ArrayRef<const ElfW(Phdr)>(phdr_begin, phdr_begin + ehdr->e_phnum)) {
if (phdr.p_type != PT_LOAD || phdr.p_offset != 0)
continue;
load_bias = reinterpret_cast<ElfW(Addr)>(ehdr) - phdr.p_vaddr;
break;
}
// Walk all globals in this ELF object, looking for the one we're interested
// in. Once we find it, we can stop iterating and return the size of the
// global we're interested in.
for (const hwasan_global &global :
HwasanGlobalsFor(load_bias, phdr_begin, ehdr->e_phnum))
if (global.addr() <= ptr && ptr < global.addr() + global.size())
return global.size();
return 0;
}
static void ShowHeapOrGlobalCandidate(uptr untagged_addr, tag_t *candidate,
tag_t *left, tag_t *right) {
Decorator d;
uptr mem = ShadowToMem(reinterpret_cast<uptr>(candidate));
HwasanChunkView chunk = FindHeapChunkByAddress(mem);
if (chunk.IsAllocated()) {
uptr offset;
const char *whence;
if (untagged_addr < chunk.End() && untagged_addr >= chunk.Beg()) {
offset = untagged_addr - chunk.Beg();
whence = "inside";
} else if (candidate == left) {
offset = untagged_addr - chunk.End();
whence = "to the right of";
} else {
offset = chunk.Beg() - untagged_addr;
whence = "to the left of";
}
Printf("%s", d.Error());
Printf("\nCause: heap-buffer-overflow\n");
Printf("%s", d.Default());
Printf("%s", d.Location());
Printf("%p is located %zd bytes %s %zd-byte region [%p,%p)\n",
untagged_addr, offset, whence, chunk.UsedSize(), chunk.Beg(),
chunk.End());
Printf("%s", d.Allocation());
Printf("allocated here:\n");
Printf("%s", d.Default());
GetStackTraceFromId(chunk.GetAllocStackId()).Print();
return;
}
// Check whether the address points into a loaded library. If so, this is
// most likely a global variable.
const char *module_name;
uptr module_address;
Symbolizer *sym = Symbolizer::GetOrInit();
if (sym->GetModuleNameAndOffsetForPC(mem, &module_name, &module_address)) {
Printf("%s", d.Error());
Printf("\nCause: global-overflow\n");
Printf("%s", d.Default());
DataInfo info;
Printf("%s", d.Location());
if (sym->SymbolizeData(mem, &info) && info.start) {
Printf(
"%p is located %zd bytes to the %s of %zd-byte global variable "
"%s [%p,%p) in %s\n",
untagged_addr,
candidate == left ? untagged_addr - (info.start + info.size)
: info.start - untagged_addr,
candidate == left ? "right" : "left", info.size, info.name,
info.start, info.start + info.size, module_name);
} else {
uptr size = GetGlobalSizeFromDescriptor(mem);
if (size == 0)
// We couldn't find the size of the global from the descriptors.
Printf(
"%p is located to the %s of a global variable in "
"\n #0 0x%x (%s+0x%x)\n",
untagged_addr, candidate == left ? "right" : "left", mem,
module_name, module_address);
else
Printf(
"%p is located to the %s of a %zd-byte global variable in "
"\n #0 0x%x (%s+0x%x)\n",
untagged_addr, candidate == left ? "right" : "left", size, mem,
module_name, module_address);
}
Printf("%s", d.Default());
}
}
void PrintAddressDescription(
uptr tagged_addr, uptr access_size,
StackAllocationsRingBuffer *current_stack_allocations) {
Decorator d;
int num_descriptions_printed = 0;
uptr untagged_addr = UntagAddr(tagged_addr);
if (MemIsShadow(untagged_addr)) {
Printf("%s%p is HWAsan shadow memory.\n%s", d.Location(), untagged_addr,
d.Default());
return;
}
// Print some very basic information about the address, if it's a heap.
HwasanChunkView chunk = FindHeapChunkByAddress(untagged_addr);
if (uptr beg = chunk.Beg()) {
uptr size = chunk.ActualSize();
Printf("%s[%p,%p) is a %s %s heap chunk; "
"size: %zd offset: %zd\n%s",
d.Location(),
beg, beg + size,
chunk.FromSmallHeap() ? "small" : "large",
chunk.IsAllocated() ? "allocated" : "unallocated",
size, untagged_addr - beg,
d.Default());
}
tag_t addr_tag = GetTagFromPointer(tagged_addr);
bool on_stack = false;
// Check stack first. If the address is on the stack of a live thread, we
// know it cannot be a heap / global overflow.
hwasanThreadList().VisitAllLiveThreads([&](Thread *t) {
if (t->AddrIsInStack(untagged_addr)) {
on_stack = true;
// TODO(fmayer): figure out how to distinguish use-after-return and
// stack-buffer-overflow.
Printf("%s", d.Error());
Printf("\nCause: stack tag-mismatch\n");
Printf("%s", d.Location());
Printf("Address %p is located in stack of thread T%zd\n", untagged_addr,
t->unique_id());
Printf("%s", d.Default());
t->Announce();
auto *sa = (t == GetCurrentThread() && current_stack_allocations)
? current_stack_allocations
: t->stack_allocations();
PrintStackAllocations(sa, addr_tag, untagged_addr);
num_descriptions_printed++;
}
});
// Check if this looks like a heap buffer overflow by scanning
// the shadow left and right and looking for the first adjacent
// object with a different memory tag. If that tag matches addr_tag,
// check the allocator if it has a live chunk there.
tag_t *tag_ptr = reinterpret_cast<tag_t*>(MemToShadow(untagged_addr));
hwasan: Improve precision of checks using short granule tags. A short granule is a granule of size between 1 and `TG-1` bytes. The size of a short granule is stored at the location in shadow memory where the granule's tag is normally stored, while the granule's actual tag is stored in the last byte of the granule. This means that in order to verify that a pointer tag matches a memory tag, HWASAN must check for two possibilities: * the pointer tag is equal to the memory tag in shadow memory, or * the shadow memory tag is actually a short granule size, the value being loaded is in bounds of the granule and the pointer tag is equal to the last byte of the granule. Pointer tags between 1 to `TG-1` are possible and are as likely as any other tag. This means that these tags in memory have two interpretations: the full tag interpretation (where the pointer tag is between 1 and `TG-1` and the last byte of the granule is ordinary data) and the short tag interpretation (where the pointer tag is stored in the granule). When HWASAN detects an error near a memory tag between 1 and `TG-1`, it will show both the memory tag and the last byte of the granule. Currently, it is up to the user to disambiguate the two possibilities. Because this functionality obsoletes the right aligned heap feature of the HWASAN memory allocator (and because we can no longer easily test it), the feature is removed. Also update the documentation to cover both short granule tags and outlined checks. Differential Revision: https://reviews.llvm.org/D63908 llvm-svn: 365551
2019-07-10 04:22:36 +08:00
tag_t *candidate = nullptr, *left = tag_ptr, *right = tag_ptr;
uptr candidate_distance = 0;
for (; candidate_distance < 1000; candidate_distance++) {
if (MemIsShadow(reinterpret_cast<uptr>(left)) &&
TagsEqual(addr_tag, left)) {
candidate = left;
hwasan: Improve precision of checks using short granule tags. A short granule is a granule of size between 1 and `TG-1` bytes. The size of a short granule is stored at the location in shadow memory where the granule's tag is normally stored, while the granule's actual tag is stored in the last byte of the granule. This means that in order to verify that a pointer tag matches a memory tag, HWASAN must check for two possibilities: * the pointer tag is equal to the memory tag in shadow memory, or * the shadow memory tag is actually a short granule size, the value being loaded is in bounds of the granule and the pointer tag is equal to the last byte of the granule. Pointer tags between 1 to `TG-1` are possible and are as likely as any other tag. This means that these tags in memory have two interpretations: the full tag interpretation (where the pointer tag is between 1 and `TG-1` and the last byte of the granule is ordinary data) and the short tag interpretation (where the pointer tag is stored in the granule). When HWASAN detects an error near a memory tag between 1 and `TG-1`, it will show both the memory tag and the last byte of the granule. Currently, it is up to the user to disambiguate the two possibilities. Because this functionality obsoletes the right aligned heap feature of the HWASAN memory allocator (and because we can no longer easily test it), the feature is removed. Also update the documentation to cover both short granule tags and outlined checks. Differential Revision: https://reviews.llvm.org/D63908 llvm-svn: 365551
2019-07-10 04:22:36 +08:00
break;
}
--left;
if (MemIsShadow(reinterpret_cast<uptr>(right)) &&
TagsEqual(addr_tag, right)) {
hwasan: Improve precision of checks using short granule tags. A short granule is a granule of size between 1 and `TG-1` bytes. The size of a short granule is stored at the location in shadow memory where the granule's tag is normally stored, while the granule's actual tag is stored in the last byte of the granule. This means that in order to verify that a pointer tag matches a memory tag, HWASAN must check for two possibilities: * the pointer tag is equal to the memory tag in shadow memory, or * the shadow memory tag is actually a short granule size, the value being loaded is in bounds of the granule and the pointer tag is equal to the last byte of the granule. Pointer tags between 1 to `TG-1` are possible and are as likely as any other tag. This means that these tags in memory have two interpretations: the full tag interpretation (where the pointer tag is between 1 and `TG-1` and the last byte of the granule is ordinary data) and the short tag interpretation (where the pointer tag is stored in the granule). When HWASAN detects an error near a memory tag between 1 and `TG-1`, it will show both the memory tag and the last byte of the granule. Currently, it is up to the user to disambiguate the two possibilities. Because this functionality obsoletes the right aligned heap feature of the HWASAN memory allocator (and because we can no longer easily test it), the feature is removed. Also update the documentation to cover both short granule tags and outlined checks. Differential Revision: https://reviews.llvm.org/D63908 llvm-svn: 365551
2019-07-10 04:22:36 +08:00
candidate = right;
break;
}
++right;
}
constexpr auto kCloseCandidateDistance = 1;
if (!on_stack && candidate && candidate_distance <= kCloseCandidateDistance) {
ShowHeapOrGlobalCandidate(untagged_addr, candidate, left, right);
num_descriptions_printed++;
}
hwasanThreadList().VisitAllLiveThreads([&](Thread *t) {
// Scan all threads' ring buffers to find if it's a heap-use-after-free.
HeapAllocationRecord har;
uptr ring_index, num_matching_addrs, num_matching_addrs_4b;
if (FindHeapAllocation(t->heap_allocations(), tagged_addr, &har,
&ring_index, &num_matching_addrs,
&num_matching_addrs_4b)) {
Printf("%s", d.Error());
Printf("\nCause: use-after-free\n");
Printf("%s", d.Location());
Printf("%p is located %zd bytes inside of %zd-byte region [%p,%p)\n",
untagged_addr, untagged_addr - UntagAddr(har.tagged_addr),
har.requested_size, UntagAddr(har.tagged_addr),
UntagAddr(har.tagged_addr) + har.requested_size);
Printf("%s", d.Allocation());
Printf("freed by thread T%zd here:\n", t->unique_id());
Printf("%s", d.Default());
GetStackTraceFromId(har.free_context_id).Print();
Printf("%s", d.Allocation());
Printf("previously allocated here:\n", t);
Printf("%s", d.Default());
GetStackTraceFromId(har.alloc_context_id).Print();
// Print a developer note: the index of this heap object
// in the thread's deallocation ring buffer.
Printf("hwasan_dev_note_heap_rb_distance: %zd %zd\n", ring_index + 1,
flags()->heap_history_size);
Printf("hwasan_dev_note_num_matching_addrs: %zd\n", num_matching_addrs);
Printf("hwasan_dev_note_num_matching_addrs_4b: %zd\n",
num_matching_addrs_4b);
t->Announce();
num_descriptions_printed++;
}
});
if (candidate && num_descriptions_printed == 0) {
ShowHeapOrGlobalCandidate(untagged_addr, candidate, left, right);
num_descriptions_printed++;
}
// Print the remaining threads, as an extra information, 1 line per thread.
hwasanThreadList().VisitAllLiveThreads([&](Thread *t) { t->Announce(); });
if (!num_descriptions_printed)
// We exhausted our possibilities. Bail out.
Printf("HWAddressSanitizer can not describe address in more detail.\n");
if (num_descriptions_printed > 1) {
Printf(
"There are %d potential causes, printed above in order "
"of likeliness.\n",
num_descriptions_printed);
}
}
void ReportStats() {}
hwasan: Improve precision of checks using short granule tags. A short granule is a granule of size between 1 and `TG-1` bytes. The size of a short granule is stored at the location in shadow memory where the granule's tag is normally stored, while the granule's actual tag is stored in the last byte of the granule. This means that in order to verify that a pointer tag matches a memory tag, HWASAN must check for two possibilities: * the pointer tag is equal to the memory tag in shadow memory, or * the shadow memory tag is actually a short granule size, the value being loaded is in bounds of the granule and the pointer tag is equal to the last byte of the granule. Pointer tags between 1 to `TG-1` are possible and are as likely as any other tag. This means that these tags in memory have two interpretations: the full tag interpretation (where the pointer tag is between 1 and `TG-1` and the last byte of the granule is ordinary data) and the short tag interpretation (where the pointer tag is stored in the granule). When HWASAN detects an error near a memory tag between 1 and `TG-1`, it will show both the memory tag and the last byte of the granule. Currently, it is up to the user to disambiguate the two possibilities. Because this functionality obsoletes the right aligned heap feature of the HWASAN memory allocator (and because we can no longer easily test it), the feature is removed. Also update the documentation to cover both short granule tags and outlined checks. Differential Revision: https://reviews.llvm.org/D63908 llvm-svn: 365551
2019-07-10 04:22:36 +08:00
static void PrintTagInfoAroundAddr(tag_t *tag_ptr, uptr num_rows,
void (*print_tag)(InternalScopedString &s,
tag_t *tag)) {
const uptr row_len = 16; // better be power of two.
tag_t *center_row_beg = reinterpret_cast<tag_t *>(
RoundDownTo(reinterpret_cast<uptr>(tag_ptr), row_len));
tag_t *beg_row = center_row_beg - row_len * (num_rows / 2);
tag_t *end_row = center_row_beg + row_len * ((num_rows + 1) / 2);
InternalScopedString s;
for (tag_t *row = beg_row; row < end_row; row += row_len) {
s.append("%s", row == center_row_beg ? "=>" : " ");
s.append("%p:", (void *)row);
for (uptr i = 0; i < row_len; i++) {
s.append("%s", row + i == tag_ptr ? "[" : " ");
hwasan: Improve precision of checks using short granule tags. A short granule is a granule of size between 1 and `TG-1` bytes. The size of a short granule is stored at the location in shadow memory where the granule's tag is normally stored, while the granule's actual tag is stored in the last byte of the granule. This means that in order to verify that a pointer tag matches a memory tag, HWASAN must check for two possibilities: * the pointer tag is equal to the memory tag in shadow memory, or * the shadow memory tag is actually a short granule size, the value being loaded is in bounds of the granule and the pointer tag is equal to the last byte of the granule. Pointer tags between 1 to `TG-1` are possible and are as likely as any other tag. This means that these tags in memory have two interpretations: the full tag interpretation (where the pointer tag is between 1 and `TG-1` and the last byte of the granule is ordinary data) and the short tag interpretation (where the pointer tag is stored in the granule). When HWASAN detects an error near a memory tag between 1 and `TG-1`, it will show both the memory tag and the last byte of the granule. Currently, it is up to the user to disambiguate the two possibilities. Because this functionality obsoletes the right aligned heap feature of the HWASAN memory allocator (and because we can no longer easily test it), the feature is removed. Also update the documentation to cover both short granule tags and outlined checks. Differential Revision: https://reviews.llvm.org/D63908 llvm-svn: 365551
2019-07-10 04:22:36 +08:00
print_tag(s, &row[i]);
s.append("%s", row + i == tag_ptr ? "]" : " ");
}
s.append("\n");
}
Printf("%s", s.data());
}
hwasan: Improve precision of checks using short granule tags. A short granule is a granule of size between 1 and `TG-1` bytes. The size of a short granule is stored at the location in shadow memory where the granule's tag is normally stored, while the granule's actual tag is stored in the last byte of the granule. This means that in order to verify that a pointer tag matches a memory tag, HWASAN must check for two possibilities: * the pointer tag is equal to the memory tag in shadow memory, or * the shadow memory tag is actually a short granule size, the value being loaded is in bounds of the granule and the pointer tag is equal to the last byte of the granule. Pointer tags between 1 to `TG-1` are possible and are as likely as any other tag. This means that these tags in memory have two interpretations: the full tag interpretation (where the pointer tag is between 1 and `TG-1` and the last byte of the granule is ordinary data) and the short tag interpretation (where the pointer tag is stored in the granule). When HWASAN detects an error near a memory tag between 1 and `TG-1`, it will show both the memory tag and the last byte of the granule. Currently, it is up to the user to disambiguate the two possibilities. Because this functionality obsoletes the right aligned heap feature of the HWASAN memory allocator (and because we can no longer easily test it), the feature is removed. Also update the documentation to cover both short granule tags and outlined checks. Differential Revision: https://reviews.llvm.org/D63908 llvm-svn: 365551
2019-07-10 04:22:36 +08:00
static void PrintTagsAroundAddr(tag_t *tag_ptr) {
Printf(
"Memory tags around the buggy address (one tag corresponds to %zd "
"bytes):\n", kShadowAlignment);
PrintTagInfoAroundAddr(tag_ptr, 17, [](InternalScopedString &s, tag_t *tag) {
s.append("%02x", *tag);
});
Printf(
"Tags for short granules around the buggy address (one tag corresponds "
"to %zd bytes):\n",
kShadowAlignment);
PrintTagInfoAroundAddr(tag_ptr, 3, [](InternalScopedString &s, tag_t *tag) {
if (*tag >= 1 && *tag <= kShadowAlignment) {
uptr granule_addr = ShadowToMem(reinterpret_cast<uptr>(tag));
s.append("%02x",
*reinterpret_cast<u8 *>(granule_addr + kShadowAlignment - 1));
} else {
s.append("..");
}
});
Printf(
"See "
"https://clang.llvm.org/docs/"
"HardwareAssistedAddressSanitizerDesign.html#short-granules for a "
"description of short granule tags\n");
}
uptr GetTopPc(StackTrace *stack) {
return stack->size ? StackTrace::GetPreviousInstructionPc(stack->trace[0])
: 0;
}
void ReportInvalidFree(StackTrace *stack, uptr tagged_addr) {
ScopedReport R(flags()->halt_on_error);
uptr untagged_addr = UntagAddr(tagged_addr);
tag_t ptr_tag = GetTagFromPointer(tagged_addr);
tag_t *tag_ptr = nullptr;
tag_t mem_tag = 0;
if (MemIsApp(untagged_addr)) {
tag_ptr = reinterpret_cast<tag_t *>(MemToShadow(untagged_addr));
if (MemIsShadow(reinterpret_cast<uptr>(tag_ptr)))
mem_tag = *tag_ptr;
else
tag_ptr = nullptr;
}
Decorator d;
Printf("%s", d.Error());
uptr pc = GetTopPc(stack);
const char *bug_type = "invalid-free";
const Thread *thread = GetCurrentThread();
if (thread) {
Report("ERROR: %s: %s on address %p at pc %p on thread T%zd\n",
SanitizerToolName, bug_type, untagged_addr, pc, thread->unique_id());
} else {
Report("ERROR: %s: %s on address %p at pc %p on unknown thread\n",
SanitizerToolName, bug_type, untagged_addr, pc);
}
Printf("%s", d.Access());
if (tag_ptr)
Printf("tags: %02x/%02x (ptr/mem)\n", ptr_tag, mem_tag);
Printf("%s", d.Default());
stack->Print();
PrintAddressDescription(tagged_addr, 0, nullptr);
if (tag_ptr)
PrintTagsAroundAddr(tag_ptr);
ReportErrorSummary(bug_type, stack);
}
void ReportTailOverwritten(StackTrace *stack, uptr tagged_addr, uptr orig_size,
hwasan: Improve precision of checks using short granule tags. A short granule is a granule of size between 1 and `TG-1` bytes. The size of a short granule is stored at the location in shadow memory where the granule's tag is normally stored, while the granule's actual tag is stored in the last byte of the granule. This means that in order to verify that a pointer tag matches a memory tag, HWASAN must check for two possibilities: * the pointer tag is equal to the memory tag in shadow memory, or * the shadow memory tag is actually a short granule size, the value being loaded is in bounds of the granule and the pointer tag is equal to the last byte of the granule. Pointer tags between 1 to `TG-1` are possible and are as likely as any other tag. This means that these tags in memory have two interpretations: the full tag interpretation (where the pointer tag is between 1 and `TG-1` and the last byte of the granule is ordinary data) and the short tag interpretation (where the pointer tag is stored in the granule). When HWASAN detects an error near a memory tag between 1 and `TG-1`, it will show both the memory tag and the last byte of the granule. Currently, it is up to the user to disambiguate the two possibilities. Because this functionality obsoletes the right aligned heap feature of the HWASAN memory allocator (and because we can no longer easily test it), the feature is removed. Also update the documentation to cover both short granule tags and outlined checks. Differential Revision: https://reviews.llvm.org/D63908 llvm-svn: 365551
2019-07-10 04:22:36 +08:00
const u8 *expected) {
uptr tail_size = kShadowAlignment - (orig_size % kShadowAlignment);
u8 actual_expected[kShadowAlignment];
internal_memcpy(actual_expected, expected, tail_size);
tag_t ptr_tag = GetTagFromPointer(tagged_addr);
// Short granule is stashed in the last byte of the magic string. To avoid
// confusion, make the expected magic string contain the short granule tag.
if (orig_size % kShadowAlignment != 0) {
actual_expected[tail_size - 1] = ptr_tag;
}
ScopedReport R(flags()->halt_on_error);
Decorator d;
uptr untagged_addr = UntagAddr(tagged_addr);
Printf("%s", d.Error());
const char *bug_type = "allocation-tail-overwritten";
Report("ERROR: %s: %s; heap object [%p,%p) of size %zd\n", SanitizerToolName,
bug_type, untagged_addr, untagged_addr + orig_size, orig_size);
Printf("\n%s", d.Default());
Printf(
"Stack of invalid access unknown. Issue detected at deallocation "
"time.\n");
Printf("%s", d.Allocation());
Printf("deallocated here:\n");
Printf("%s", d.Default());
stack->Print();
HwasanChunkView chunk = FindHeapChunkByAddress(untagged_addr);
if (chunk.Beg()) {
Printf("%s", d.Allocation());
Printf("allocated here:\n");
Printf("%s", d.Default());
GetStackTraceFromId(chunk.GetAllocStackId()).Print();
}
InternalScopedString s;
CHECK_GT(tail_size, 0U);
CHECK_LT(tail_size, kShadowAlignment);
u8 *tail = reinterpret_cast<u8*>(untagged_addr + orig_size);
s.append("Tail contains: ");
for (uptr i = 0; i < kShadowAlignment - tail_size; i++)
s.append(".. ");
for (uptr i = 0; i < tail_size; i++)
s.append("%02x ", tail[i]);
s.append("\n");
s.append("Expected: ");
for (uptr i = 0; i < kShadowAlignment - tail_size; i++)
s.append(".. ");
for (uptr i = 0; i < tail_size; i++) s.append("%02x ", actual_expected[i]);
s.append("\n");
s.append(" ");
for (uptr i = 0; i < kShadowAlignment - tail_size; i++)
s.append(" ");
for (uptr i = 0; i < tail_size; i++)
s.append("%s ", actual_expected[i] != tail[i] ? "^^" : " ");
s.append("\nThis error occurs when a buffer overflow overwrites memory\n"
"to the right of a heap object, but within the %zd-byte granule, e.g.\n"
" char *x = new char[20];\n"
" x[25] = 42;\n"
hwasan: Improve precision of checks using short granule tags. A short granule is a granule of size between 1 and `TG-1` bytes. The size of a short granule is stored at the location in shadow memory where the granule's tag is normally stored, while the granule's actual tag is stored in the last byte of the granule. This means that in order to verify that a pointer tag matches a memory tag, HWASAN must check for two possibilities: * the pointer tag is equal to the memory tag in shadow memory, or * the shadow memory tag is actually a short granule size, the value being loaded is in bounds of the granule and the pointer tag is equal to the last byte of the granule. Pointer tags between 1 to `TG-1` are possible and are as likely as any other tag. This means that these tags in memory have two interpretations: the full tag interpretation (where the pointer tag is between 1 and `TG-1` and the last byte of the granule is ordinary data) and the short tag interpretation (where the pointer tag is stored in the granule). When HWASAN detects an error near a memory tag between 1 and `TG-1`, it will show both the memory tag and the last byte of the granule. Currently, it is up to the user to disambiguate the two possibilities. Because this functionality obsoletes the right aligned heap feature of the HWASAN memory allocator (and because we can no longer easily test it), the feature is removed. Also update the documentation to cover both short granule tags and outlined checks. Differential Revision: https://reviews.llvm.org/D63908 llvm-svn: 365551
2019-07-10 04:22:36 +08:00
"%s does not detect such bugs in uninstrumented code at the time of write,"
"\nbut can detect them at the time of free/delete.\n"
"To disable this feature set HWASAN_OPTIONS=free_checks_tail_magic=0\n",
kShadowAlignment, SanitizerToolName);
Printf("%s", s.data());
GetCurrentThread()->Announce();
tag_t *tag_ptr = reinterpret_cast<tag_t*>(MemToShadow(untagged_addr));
PrintTagsAroundAddr(tag_ptr);
ReportErrorSummary(bug_type, stack);
}
void ReportTagMismatch(StackTrace *stack, uptr tagged_addr, uptr access_size,
[HWASan] Save + print registers when tag mismatch occurs in AArch64. Summary: This change change the instrumentation to allow users to view the registers at the point at which tag mismatch occured. Most of the heavy lifting is done in the runtime library, where we save the registers to the stack and emit unwind information. This allows us to reduce the overhead, as very little additional work needs to be done in each __hwasan_check instance. In this implementation, the fast path of __hwasan_check is unmodified. There are an additional 4 instructions (16B) emitted in the slow path in every __hwasan_check instance. This may increase binary size somewhat, but as most of the work is done in the runtime library, it's manageable. The failure trace now contains a list of registers at the point of which the failure occured, in a format similar to that of Android's tombstones. It currently has the following format: Registers where the failure occurred (pc 0x0055555561b4): x0 0000000000000014 x1 0000007ffffff6c0 x2 1100007ffffff6d0 x3 12000056ffffe025 x4 0000007fff800000 x5 0000000000000014 x6 0000007fff800000 x7 0000000000000001 x8 12000056ffffe020 x9 0200007700000000 x10 0200007700000000 x11 0000000000000000 x12 0000007fffffdde0 x13 0000000000000000 x14 02b65b01f7a97490 x15 0000000000000000 x16 0000007fb77376b8 x17 0000000000000012 x18 0000007fb7ed6000 x19 0000005555556078 x20 0000007ffffff768 x21 0000007ffffff778 x22 0000000000000001 x23 0000000000000000 x24 0000000000000000 x25 0000000000000000 x26 0000000000000000 x27 0000000000000000 x28 0000000000000000 x29 0000007ffffff6f0 x30 00000055555561b4 ... and prints after the dump of memory tags around the buggy address. Every register is saved exactly as it was at the point where the tag mismatch occurs, with the exception of x16/x17. These registers are used in the tag mismatch calculation as scratch registers during __hwasan_check, and cannot be saved without affecting the fast path. As these registers are designated as scratch registers for linking, there should be no important information in them that could aid in debugging. Reviewers: pcc, eugenis Reviewed By: pcc, eugenis Subscribers: srhines, kubamracek, mgorny, javed.absar, krytarowski, kristof.beyls, hiraditya, jdoerfert, llvm-commits, #sanitizers Tags: #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D58857 llvm-svn: 355738
2019-03-09 05:22:35 +08:00
bool is_store, bool fatal, uptr *registers_frame) {
ScopedReport R(fatal);
SavedStackAllocations current_stack_allocations(
GetCurrentThread()->stack_allocations());
Decorator d;
uptr untagged_addr = UntagAddr(tagged_addr);
// TODO: when possible, try to print heap-use-after-free, etc.
const char *bug_type = "tag-mismatch";
uptr pc = GetTopPc(stack);
Printf("%s", d.Error());
Report("ERROR: %s: %s on address %p at pc %p\n", SanitizerToolName, bug_type,
untagged_addr, pc);
Thread *t = GetCurrentThread();
sptr offset =
__hwasan_test_shadow(reinterpret_cast<void *>(tagged_addr), access_size);
CHECK(offset >= 0 && offset < static_cast<sptr>(access_size));
tag_t ptr_tag = GetTagFromPointer(tagged_addr);
tag_t *tag_ptr =
reinterpret_cast<tag_t *>(MemToShadow(untagged_addr + offset));
tag_t mem_tag = *tag_ptr;
Printf("%s", d.Access());
if (mem_tag && mem_tag < kShadowAlignment) {
tag_t *granule_ptr = reinterpret_cast<tag_t *>((untagged_addr + offset) &
~(kShadowAlignment - 1));
// If offset is 0, (untagged_addr + offset) is not aligned to granules.
// This is the offset of the leftmost accessed byte within the bad granule.
u8 in_granule_offset = (untagged_addr + offset) & (kShadowAlignment - 1);
tag_t short_tag = granule_ptr[kShadowAlignment - 1];
// The first mismatch was a short granule that matched the ptr_tag.
if (short_tag == ptr_tag) {
// If the access starts after the end of the short granule, then the first
// bad byte is the first byte of the access; otherwise it is the first
// byte past the end of the short granule
if (mem_tag > in_granule_offset) {
offset += mem_tag - in_granule_offset;
}
}
Printf(
"%s of size %zu at %p tags: %02x/%02x(%02x) (ptr/mem) in thread T%zd\n",
is_store ? "WRITE" : "READ", access_size, untagged_addr, ptr_tag,
mem_tag, short_tag, t->unique_id());
} else {
Printf("%s of size %zu at %p tags: %02x/%02x (ptr/mem) in thread T%zd\n",
is_store ? "WRITE" : "READ", access_size, untagged_addr, ptr_tag,
mem_tag, t->unique_id());
}
if (offset != 0)
Printf("Invalid access starting at offset %zu\n", offset);
Printf("%s", d.Default());
stack->Print();
PrintAddressDescription(tagged_addr, access_size,
current_stack_allocations.get());
t->Announce();
PrintTagsAroundAddr(tag_ptr);
[HWASan] Save + print registers when tag mismatch occurs in AArch64. Summary: This change change the instrumentation to allow users to view the registers at the point at which tag mismatch occured. Most of the heavy lifting is done in the runtime library, where we save the registers to the stack and emit unwind information. This allows us to reduce the overhead, as very little additional work needs to be done in each __hwasan_check instance. In this implementation, the fast path of __hwasan_check is unmodified. There are an additional 4 instructions (16B) emitted in the slow path in every __hwasan_check instance. This may increase binary size somewhat, but as most of the work is done in the runtime library, it's manageable. The failure trace now contains a list of registers at the point of which the failure occured, in a format similar to that of Android's tombstones. It currently has the following format: Registers where the failure occurred (pc 0x0055555561b4): x0 0000000000000014 x1 0000007ffffff6c0 x2 1100007ffffff6d0 x3 12000056ffffe025 x4 0000007fff800000 x5 0000000000000014 x6 0000007fff800000 x7 0000000000000001 x8 12000056ffffe020 x9 0200007700000000 x10 0200007700000000 x11 0000000000000000 x12 0000007fffffdde0 x13 0000000000000000 x14 02b65b01f7a97490 x15 0000000000000000 x16 0000007fb77376b8 x17 0000000000000012 x18 0000007fb7ed6000 x19 0000005555556078 x20 0000007ffffff768 x21 0000007ffffff778 x22 0000000000000001 x23 0000000000000000 x24 0000000000000000 x25 0000000000000000 x26 0000000000000000 x27 0000000000000000 x28 0000000000000000 x29 0000007ffffff6f0 x30 00000055555561b4 ... and prints after the dump of memory tags around the buggy address. Every register is saved exactly as it was at the point where the tag mismatch occurs, with the exception of x16/x17. These registers are used in the tag mismatch calculation as scratch registers during __hwasan_check, and cannot be saved without affecting the fast path. As these registers are designated as scratch registers for linking, there should be no important information in them that could aid in debugging. Reviewers: pcc, eugenis Reviewed By: pcc, eugenis Subscribers: srhines, kubamracek, mgorny, javed.absar, krytarowski, kristof.beyls, hiraditya, jdoerfert, llvm-commits, #sanitizers Tags: #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D58857 llvm-svn: 355738
2019-03-09 05:22:35 +08:00
if (registers_frame)
ReportRegisters(registers_frame, pc);
ReportErrorSummary(bug_type, stack);
}
[HWASan] Save + print registers when tag mismatch occurs in AArch64. Summary: This change change the instrumentation to allow users to view the registers at the point at which tag mismatch occured. Most of the heavy lifting is done in the runtime library, where we save the registers to the stack and emit unwind information. This allows us to reduce the overhead, as very little additional work needs to be done in each __hwasan_check instance. In this implementation, the fast path of __hwasan_check is unmodified. There are an additional 4 instructions (16B) emitted in the slow path in every __hwasan_check instance. This may increase binary size somewhat, but as most of the work is done in the runtime library, it's manageable. The failure trace now contains a list of registers at the point of which the failure occured, in a format similar to that of Android's tombstones. It currently has the following format: Registers where the failure occurred (pc 0x0055555561b4): x0 0000000000000014 x1 0000007ffffff6c0 x2 1100007ffffff6d0 x3 12000056ffffe025 x4 0000007fff800000 x5 0000000000000014 x6 0000007fff800000 x7 0000000000000001 x8 12000056ffffe020 x9 0200007700000000 x10 0200007700000000 x11 0000000000000000 x12 0000007fffffdde0 x13 0000000000000000 x14 02b65b01f7a97490 x15 0000000000000000 x16 0000007fb77376b8 x17 0000000000000012 x18 0000007fb7ed6000 x19 0000005555556078 x20 0000007ffffff768 x21 0000007ffffff778 x22 0000000000000001 x23 0000000000000000 x24 0000000000000000 x25 0000000000000000 x26 0000000000000000 x27 0000000000000000 x28 0000000000000000 x29 0000007ffffff6f0 x30 00000055555561b4 ... and prints after the dump of memory tags around the buggy address. Every register is saved exactly as it was at the point where the tag mismatch occurs, with the exception of x16/x17. These registers are used in the tag mismatch calculation as scratch registers during __hwasan_check, and cannot be saved without affecting the fast path. As these registers are designated as scratch registers for linking, there should be no important information in them that could aid in debugging. Reviewers: pcc, eugenis Reviewed By: pcc, eugenis Subscribers: srhines, kubamracek, mgorny, javed.absar, krytarowski, kristof.beyls, hiraditya, jdoerfert, llvm-commits, #sanitizers Tags: #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D58857 llvm-svn: 355738
2019-03-09 05:22:35 +08:00
// See the frame breakdown defined in __hwasan_tag_mismatch (from
// hwasan_tag_mismatch_aarch64.S).
void ReportRegisters(uptr *frame, uptr pc) {
Printf("Registers where the failure occurred (pc %p):\n", pc);
// We explicitly print a single line (4 registers/line) each iteration to
// reduce the amount of logcat error messages printed. Each Printf() will
// result in a new logcat line, irrespective of whether a newline is present,
// and so we wish to reduce the number of Printf() calls we have to make.
Printf(" x0 %016llx x1 %016llx x2 %016llx x3 %016llx\n",
frame[0], frame[1], frame[2], frame[3]);
Printf(" x4 %016llx x5 %016llx x6 %016llx x7 %016llx\n",
frame[4], frame[5], frame[6], frame[7]);
Printf(" x8 %016llx x9 %016llx x10 %016llx x11 %016llx\n",
frame[8], frame[9], frame[10], frame[11]);
Printf(" x12 %016llx x13 %016llx x14 %016llx x15 %016llx\n",
frame[12], frame[13], frame[14], frame[15]);
Printf(" x16 %016llx x17 %016llx x18 %016llx x19 %016llx\n",
frame[16], frame[17], frame[18], frame[19]);
Printf(" x20 %016llx x21 %016llx x22 %016llx x23 %016llx\n",
frame[20], frame[21], frame[22], frame[23]);
Printf(" x24 %016llx x25 %016llx x26 %016llx x27 %016llx\n",
frame[24], frame[25], frame[26], frame[27]);
// hwasan_check* reduces the stack pointer by 256, then __hwasan_tag_mismatch
// passes it to this function.
Printf(" x28 %016llx x29 %016llx x30 %016llx sp %016llx\n", frame[28],
frame[29], frame[30], reinterpret_cast<u8 *>(frame) + 256);
[HWASan] Save + print registers when tag mismatch occurs in AArch64. Summary: This change change the instrumentation to allow users to view the registers at the point at which tag mismatch occured. Most of the heavy lifting is done in the runtime library, where we save the registers to the stack and emit unwind information. This allows us to reduce the overhead, as very little additional work needs to be done in each __hwasan_check instance. In this implementation, the fast path of __hwasan_check is unmodified. There are an additional 4 instructions (16B) emitted in the slow path in every __hwasan_check instance. This may increase binary size somewhat, but as most of the work is done in the runtime library, it's manageable. The failure trace now contains a list of registers at the point of which the failure occured, in a format similar to that of Android's tombstones. It currently has the following format: Registers where the failure occurred (pc 0x0055555561b4): x0 0000000000000014 x1 0000007ffffff6c0 x2 1100007ffffff6d0 x3 12000056ffffe025 x4 0000007fff800000 x5 0000000000000014 x6 0000007fff800000 x7 0000000000000001 x8 12000056ffffe020 x9 0200007700000000 x10 0200007700000000 x11 0000000000000000 x12 0000007fffffdde0 x13 0000000000000000 x14 02b65b01f7a97490 x15 0000000000000000 x16 0000007fb77376b8 x17 0000000000000012 x18 0000007fb7ed6000 x19 0000005555556078 x20 0000007ffffff768 x21 0000007ffffff778 x22 0000000000000001 x23 0000000000000000 x24 0000000000000000 x25 0000000000000000 x26 0000000000000000 x27 0000000000000000 x28 0000000000000000 x29 0000007ffffff6f0 x30 00000055555561b4 ... and prints after the dump of memory tags around the buggy address. Every register is saved exactly as it was at the point where the tag mismatch occurs, with the exception of x16/x17. These registers are used in the tag mismatch calculation as scratch registers during __hwasan_check, and cannot be saved without affecting the fast path. As these registers are designated as scratch registers for linking, there should be no important information in them that could aid in debugging. Reviewers: pcc, eugenis Reviewed By: pcc, eugenis Subscribers: srhines, kubamracek, mgorny, javed.absar, krytarowski, kristof.beyls, hiraditya, jdoerfert, llvm-commits, #sanitizers Tags: #sanitizers, #llvm Differential Revision: https://reviews.llvm.org/D58857 llvm-svn: 355738
2019-03-09 05:22:35 +08:00
}
} // namespace __hwasan
void __hwasan_set_error_report_callback(void (*callback)(const char *)) {
__hwasan::ScopedReport::SetErrorReportCallback(callback);
}