This is an archive of the discontinued LLVM Phabricator instance.

Differential D123644

[HWASan] symbolize: use buildid index for locals.
ClosedPublic

Authored by fmayer on Apr 12 2022, 4:26 PM.

Details

Reviewers
eugenis
Commits
rGffd656a2fe5d: [HWASan] symbolize: use buildid index for locals.

Diff Detail

Unit TestsFailed

TimeTest
1,400 msx64 debian > Clang.utils/update_cc_test_checks::check-globals.test
640 msx64 debian > Clang.utils/update_cc_test_checks::global-hex-value-regex.test
610 msx64 debian > Clang.utils/update_cc_test_checks::global-value-regex.test
910 msx64 debian > SanitizerCommon-asan-x86_64-Linux.Linux::decorate_proc_maps.cpp
690 msx64 debian > SanitizerCommon-lsan-x86_64-Linux.Linux::decorate_proc_maps.cpp
View Full Test Results (45 Failed)

Event Timeline

fmayer created this revision.Apr 12 2022, 4:26 PM
Herald added a project: Restricted Project. · View Herald TranscriptApr 12 2022, 4:26 PM
Herald added a subscriber: arphaman. · View Herald Transcript
fmayer requested review of this revision.Apr 12 2022, 4:26 PM
Herald added a project: Restricted Project. · View Herald TranscriptApr 12 2022, 4:26 PM
Herald added a subscriber: Restricted Project. · View Herald Transcript
fmayer added a reviewer: eugenis.Apr 12 2022, 4:26 PM
kstoimenov added a subscriber: kstoimenov.Apr 12 2022, 4:37 PM
kstoimenov added inline comments.
compiler-rt/lib/hwasan/scripts/hwasan_symbolize
262

Consider splitting this into multiple lines?

fmayer updated this revision to Diff 422364.Apr 12 2022, 4:39 PM

oformat

fmayer marked an inline comment as done.Apr 12 2022, 4:39 PM
Harbormaster completed remote builds in B159347: Diff 422364.Apr 12 2022, 4:57 PM
fmayer updated this revision to Diff 422541.Apr 13 2022, 9:41 AM

format

fmayer updated this revision to Diff 422543.Apr 13 2022, 9:44 AM

format2

Harbormaster completed remote builds in B159464: Diff 422543.Apr 13 2022, 9:58 AM
eugenis accepted this revision.Apr 13 2022, 5:10 PM

LGTM

compiler-rt/lib/hwasan/scripts/hwasan_symbolize
261

Please update the comment.

This revision is now accepted and ready to land.Apr 13 2022, 5:10 PM
fmayer updated this revision to Diff 422717.Apr 13 2022, 7:07 PM

comment

fmayer marked an inline comment as done.Apr 13 2022, 7:08 PM
This revision was landed with ongoing or failed builds.Apr 13 2022, 7:09 PM
Closed by commit rGffd656a2fe5d: [HWASan] symbolize: use buildid index for locals. (authored by fmayer). · Explain Why
This revision was automatically updated to reflect the committed changes.
fmayer added a commit: rGffd656a2fe5d: [HWASan] symbolize: use buildid index for locals..
Harbormaster completed remote builds in B159593: Diff 422717.Apr 13 2022, 7:29 PM

Revision Contents

PathSize
compiler-rt/
lib/
hwasan/
scripts/
14 lines

Diff 422364

compiler-rt/lib/hwasan/scripts/hwasan_symbolize

Show First 20 Lines • Show All 140 Lines • ▼ Show 20 Linesclass Symbolizer:
def __process_source_path(self, file_name): def __process_source_path(self, file_name):
for path_to_cut in self.__paths_to_cut: for path_to_cut in self.__paths_to_cut:
file_name = re.sub(".*" + path_to_cut, "", file_name) file_name = re.sub(".*" + path_to_cut, "", file_name)
file_name = re.sub(".*hwasan_[a-z_]*.(cc|h):[0-9]*", "[hwasan_rtl]", file_name) file_name = re.sub(".*hwasan_[a-z_]*.(cc|h):[0-9]*", "[hwasan_rtl]", file_name)
file_name = re.sub(".*asan_[a-z_]*.(cc|h):[0-9]*", "[asan_rtl]", file_name) file_name = re.sub(".*asan_[a-z_]*.(cc|h):[0-9]*", "[asan_rtl]", file_name)
file_name = re.sub(".*crtstuff.c:0", "???:0", file_name) file_name = re.sub(".*crtstuff.c:0", "???:0", file_name)
return file_name return file_name
def __process_binary_name(self, name, buildid=None): def __process_binary_name(self, name, buildid):
if name.startswith('/'): if name.startswith('/'):
name = name[1:] name = name[1:]
if buildid is not None and buildid in self.__index: if buildid is not None and buildid in self.__index:
return self.__index[buildid] return self.__index[buildid]
for p in self.__binary_prefixes: for p in self.__binary_prefixes:
full_path = os.path.join(p, name) full_path = os.path.join(p, name)
if os.path.exists(full_path): if os.path.exists(full_path):
return full_path return full_path
apex_prefix = "apex/com.android." apex_prefix = "apex/com.android."
if name.startswith(apex_prefix): if name.startswith(apex_prefix):
full_path = os.path.join(p, "apex/com.google.android." + name[len(apex_prefix):]) full_path = os.path.join(p, "apex/com.google.android." + name[len(apex_prefix):])
if os.path.exists(full_path): if os.path.exists(full_path):
return full_path return full_path
# Try stripping extra path components as the last resort. # Try stripping extra path components as the last resort.
for p in self.__binary_prefixes: for p in self.__binary_prefixes:
full_path = os.path.join(p, os.path.basename(name)) full_path = os.path.join(p, os.path.basename(name))
if os.path.exists(full_path): if os.path.exists(full_path):
return full_path return full_path
if name not in self.__warnings: if name not in self.__warnings:
print("Could not find symbols for", name, file=sys.stderr) print("Could not find symbols for", name, file=sys.stderr)
self.__warnings.add(name) self.__warnings.add(name)
return None return None
def iter_locals(self, binary, addr): def iter_locals(self, binary, addr, buildid):
self.__open_pipe() self.__open_pipe()
p = self.__pipe p = self.__pipe
binary = self.__process_binary_name(binary) binary = self.__process_binary_name(binary, buildid)
if not binary: if not binary:
return return
self.__write("FRAME %s %s" % (binary, addr)) self.__write("FRAME %s %s" % (binary, addr))
try: try:
while True: while True:
function_name = self.__read() function_name = self.__read()
local_name = self.__read() local_name = self.__read()
file_line = self.__read() file_line = self.__read()
Show All 29 Lines for p in self.__binary_prefixes:
for fn in fnames: for fn in fnames:
filename = os.path.join(dname, fn) filename = os.path.join(dname, fn)
bid = get_buildid(filename) bid = get_buildid(filename)
if bid is not None: if bid is not None:
self.__index[bid] = filename self.__index[bid] = filename
def symbolize_line(line, symbolizer_path): def symbolize_line(line, symbolizer_path):
#0 0x7f6e35cf2e45 (/blah/foo.so+0x11fe45) #0 0x7f6e35cf2e45 (/blah/foo.so+0x11fe45)
match = re.match(r'^(.*?)#([0-9]+)( *)(0x[0-9a-f]*) *\((.*)\+(0x[0-9a-f]+)\)(?:\s*\(BuildId: ([0-9a-f]+)\))?', line, re.UNICODE) match = re.match(r'^(.*?)#([0-9]+)( *)(0x[0-9a-f]*) *\((.*)\+(0x[0-9a-f]+)\)'
'(?:\s*\(BuildId: ([0-9a-f]+)\))?', line, re.UNICODE)
if match: if match:
frameno = match.group(2) frameno = match.group(2)
binary = match.group(5) binary = match.group(5)
addr = int(match.group(6), 16) addr = int(match.group(6), 16)
buildid = match.group(7) buildid = match.group(7)
frames = list(symbolizer.iter_call_stack(binary, buildid, addr)) frames = list(symbolizer.iter_call_stack(binary, buildid, addr))
Show All 20 Lines
def process_stack_history(line, symbolizer, ignore_tags=False): def process_stack_history(line, symbolizer, ignore_tags=False):
if last_access_address is None or last_access_tag is None: if last_access_address is None or last_access_tag is None:
return return
if re.match(r'Previously allocated frames:', line, re.UNICODE): if re.match(r'Previously allocated frames:', line, re.UNICODE):
return True return True
pc_mask = (1 << 48) - 1 pc_mask = (1 << 48) - 1
fp_mask = (1 << 20) - 1 fp_mask = (1 << 20) - 1
# record_addr:0x1234ABCD record:0x1234ABCD (/path/to/binary+0x1234ABCD) # record_addr:0x1234ABCD record:0x1234ABCD (/path/to/binary+0x1234ABCD)
eugenisUnsubmitted
Done
ReplyInline Actions

Please update the comment.

eugenis: Please update the comment.
match = re.match(r'^(.*?)record_addr:(0x[0-9a-f]+) +record:(0x[0-9a-f]+) +\((.*)\+(0x[0-9a-f]+)\)', line, re.UNICODE) match = re.match(r'^(.*?)record_addr:(0x[0-9a-f]+) +record:(0x[0-9a-f]+) +\((.*)\+(0x[0-9a-f]+)\)(?:\s*\(BuildId: ([0-9a-f]+)\))?', line, re.UNICODE)
kstoimenovUnsubmitted
Done
ReplyInline Actions

Consider splitting this into multiple lines?

kstoimenov: Consider splitting this into multiple lines?
if match: if match:
record_addr = int(match.group(2), 16) record_addr = int(match.group(2), 16)
record = int(match.group(3), 16) record = int(match.group(3), 16)
binary = match.group(4) binary = match.group(4)
addr = int(match.group(5), 16) addr = int(match.group(5), 16)
buildid = match.group(6)
base_tag = (record_addr >> 3) & 0xFF base_tag = (record_addr >> 3) & 0xFF
fp = (record >> 48) << 4 fp = (record >> 48) << 4
pc = record & pc_mask pc = record & pc_mask
for local in symbolizer.iter_locals(binary, addr): for local in symbolizer.iter_locals(binary, addr, buildid):
frame_offset = local[3] frame_offset = local[3]
size = local[4] size = local[4]
if frame_offset is None or size is None: if frame_offset is None or size is None:
continue continue
obj_offset = (last_access_address - fp - frame_offset) & fp_mask obj_offset = (last_access_address - fp - frame_offset) & fp_mask
if obj_offset >= size: if obj_offset >= size:
continue continue
tag_offset = local[5] tag_offset = local[5]
▲ Show 20 Lines • Show All 113 Lines • Show Last 20 Lines