Skip to content

Commit

Permalink
Extract llvmbc/dump assembly
Browse files Browse the repository at this point in the history
  • Loading branch information
@dtcxzyw
dtcxzyw committed Mar 10, 2023
1 parent 5a116c0 commit bd98b4b
Show file tree
Hide file tree
Showing 2 changed files with 82 additions and 36 deletions.
45 changes: 45 additions & 0 deletions binutils.py
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,45 @@
import subprocess
import os


def dump_asm(filename: str, objdump):
with open(filename+'.S', mode="w") as f:
subprocess.run([objdump, '--no-addresses',
'--symbolize-operands', '-S', filename], stdout=f)


def dump_bc(base, dis, count, data):
filename = "{}/seg{}.bc".format(base, count)
with open(filename, "wb") as f:
f.write(data)
subprocess.run([dis, "seg{}.bc".format(count)], cwd=base)


def extract_bc(filename: str, objcopy, dis):
bc_header = bytes([0x42, 0x43, 0xc0, 0xde])
basedir = filename+"_bc"
bc_filename = filename+".bc"

subprocess.run([objcopy, filename, "--dump-section",
".llvmbc={}".format(bc_filename)])

if not os.path.exists(bc_filename):
return

os.makedirs(basedir, exist_ok=True)

count = 0

with open(bc_filename, 'rb') as f:
data = f.read()

last_pos = 0
while True:
pos = data.find(bc_header, last_pos+4)
if pos == -1:
dump_bc(basedir, dis, count, data[last_pos:])
break
else:
dump_bc(basedir, dis, count, data[last_pos: pos])
last_pos = pos
count += 1
73 changes: 37 additions & 36 deletions report-generate.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -6,13 +6,15 @@
import html
import datetime
import statistics
import binutils

lhs = sys.argv[1]
rhs = sys.argv[2]
base_dir = sys.argv[3]
workflow_url = sys.argv[4]
binaries_src = base_dir+"/binaries/"
binaries_dst = base_dir+"/artifacts/binaries/"
llvm_path = base_dir+"/llvm-build/bin/"
os.makedirs(binaries_dst)

threshold_abs = 16
Expand All @@ -37,6 +39,9 @@ def parse(path):
def copy_binary(lhs, rhs):
if os.path.exists(lhs):
shutil.copyfile(lhs, rhs)
binutils.dump_asm(rhs, llvm_path+"llvm-objdump")
binutils.extract_bc(rhs, llvm_path+"llvm-objcopy",
llvm_path+"llvm-dis")


def dump_pretty_change_logs(report, change_logs_path):
Expand All @@ -58,7 +63,7 @@ def strip_name(name: str):
'test-suite :: ').removesuffix('.test')


def dump_diff(report, lhs_data, rhs_data, copy_binaries):
def dump_diff(report, lhs_data, rhs_data):
report.write('## Differences\n')
report.write(
'|Name|Baseline MD5|Current MD5|Baseline Size|Current Size|Ratio|\n')
Expand Down Expand Up @@ -89,9 +94,8 @@ def dump_diff(report, lhs_data, rhs_data, copy_binaries):
for name, lhs_hash, rhs_hash, lhs_size, rhs_size in diff_list:
report.write("|{}|{}|{}|{}|{}|{:.3f}|\n".format(
strip_name(name), lhs_hash, rhs_hash, lhs_size, rhs_size, rhs_size/lhs_size))
if copy_binaries:
copy_binary(binaries_src+lhs_hash, binaries_dst+lhs_hash)
copy_binary(binaries_src+rhs_hash, binaries_dst+rhs_hash)
copy_binary(binaries_src+lhs_hash, binaries_dst+lhs_hash)
copy_binary(binaries_src+rhs_hash, binaries_dst+rhs_hash)

if len(lhs_list) > 0:
gmean_lhs = statistics.geometric_mean(lhs_list)
Expand All @@ -115,7 +119,7 @@ def dump_diff(report, lhs_data, rhs_data, copy_binaries):
pr_comment.write(
'+ Patch SHA256: {}\n'.format(os.environ['PATCH_SHA256']))

dump_diff(pr_comment, lhs_data, rhs_data, True)
dump_diff(pr_comment, lhs_data, rhs_data)
else:
binary_bloating_list = []
for name in lhs_data.keys():
Expand All @@ -127,39 +131,36 @@ def dump_diff(report, lhs_data, rhs_data, copy_binaries):
binary_bloating_list.append(
(name, lhs_hash, rhs_hash, lhs_value, rhs_value))

change_logs_path = base_dir+"/artifacts/CHANGELOGS"
issue_report_path = base_dir+"/artifacts/issue_generated.md"
with open(issue_report_path, "w") as issue_report:
issue_report.write('---\n')
issue_report.write(
"title: Size Regressions Report {{ date | date('MMMM Do YYYY, h:mm:ss a') }}\n")
issue_report.write('labels: regression\n')
issue_report.write('---\n')
issue_report.write('## Metadata\n')
issue_report.write('+ Workflow URL: {}\n'.format(workflow_url))

dump_pretty_change_logs(issue_report, change_logs_path)

issue_report.write('## Regressions\n')
issue_report.write(
'|Name|Baseline MD5|Current MD5|Baseline Size|Current Size|Ratio|\n')
issue_report.write('|:--|:--:|:--:|--:|--:|--:|\n')

binary_bloating_list.sort(key=lambda x: x[4]/x[3], reverse=True)
if len(binary_bloating_list) > limit:
binary_bloating_list = binary_bloating_list[:limit]

for name, lhs_hash, rhs_hash, lhs_size, rhs_size in binary_bloating_list:
issue_report.write("|{}|{}|{}|{}|{}|{:.3f}|\n".format(strip_name(name), lhs_hash,
rhs_hash, lhs_size, rhs_size, rhs_size/lhs_size))

dump_diff(issue_report, lhs_data, rhs_data)

if len(binary_bloating_list) == 0:
print("No regressions")
exit(0)

else:
change_logs_path = base_dir+"/artifacts/CHANGELOGS"
issue_report_path = base_dir+"/artifacts/issue_generated.md"
with open(issue_report_path, "w") as issue_report:
issue_report.write('---\n')
issue_report.write(
"title: Size Regressions Report {{ date | date('MMMM Do YYYY, h:mm:ss a') }}\n")
issue_report.write('labels: regression\n')
issue_report.write('---\n')
issue_report.write('## Metadata\n')
issue_report.write('+ Workflow URL: {}\n'.format(workflow_url))

dump_pretty_change_logs(issue_report, change_logs_path)

issue_report.write('## Regressions\n')
issue_report.write(
'|Name|Baseline MD5|Current MD5|Baseline Size|Current Size|Ratio|\n')
issue_report.write('|:--|:--:|:--:|--:|--:|--:|\n')

binary_bloating_list.sort(key=lambda x: x[4]/x[3], reverse=True)
if len(binary_bloating_list) > limit:
binary_bloating_list = binary_bloating_list[:limit]

for name, lhs_hash, rhs_hash, lhs_size, rhs_size in binary_bloating_list:
issue_report.write("|{}|{}|{}|{}|{}|{:.3f}|\n".format(strip_name(name), lhs_hash,
rhs_hash, lhs_size, rhs_size, rhs_size/lhs_size))
copy_binary(binaries_src+lhs_hash, binaries_dst+lhs_hash)
copy_binary(binaries_src+rhs_hash, binaries_dst+rhs_hash)

dump_diff(issue_report, lhs_data, rhs_data, False)

exit(1)

0 comments on commit bd98b4b

Please sign in to comment.