blob: 79851ec8c855b66ee0ee009399370e44b3382e69 [file] [log] [blame]
#!/usr/bin/env python3
# Copyright (c) 2015-2021 The Khronos Group Inc.
# Copyright (c) 2015-2021 Valve Corporation
# Copyright (c) 2015-2021 LunarG, Inc.
# Copyright (c) 2015-2021 Google Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
# Author: Tobin Ehlis <tobine@google.com>
# Author: Dave Houlton <daveh@lunarg.com>
# Author: Shannon McPherson <shannon@lunarg.com>
import argparse
import common_codegen
import csv
import glob
import html
import json
import operator
import os
import platform
import re
import sys
import time
import unicodedata
import subprocess
from collections import defaultdict
from collections import OrderedDict
verbose_mode = False
txt_db = False
csv_db = False
html_db = False
txt_filename = "validation_error_database.txt"
csv_filename = "validation_error_database.csv"
html_filename = "validation_error_database.html"
header_filename = "vk_validation_error_messages.h"
vuid_prefixes = ['VUID-', 'UNASSIGNED-', 'kVUID_']
spirvtools_path = None # default is to not search for repo
# Hard-coded flags that could be command line args, if we decide that's useful
ignore_unassigned = True # These are not found in layer code unless they appear explicitly (most don't), so produce false positives
layer_source_files = [common_codegen.repo_relative(path) for path in [
'layers/buffer_validation.cpp',
'layers/core_validation.cpp',
'layers/descriptor_sets.cpp',
'layers/drawdispatch.cpp',
'layers/gpu_vuids.h',
'layers/parameter_validation_utils.cpp',
'layers/object_tracker_utils.cpp',
'layers/shader_module.cpp',
'layers/shader_validation.cpp',
'layers/stateless_validation.h',
'layers/synchronization_validation.cpp',
'layers/sync_vuid_maps.cpp',
'layers/generated/parameter_validation.cpp',
'layers/generated/object_tracker.cpp',
'layers/generated/spirv_validation_helper.cpp',
'layers/generated/command_validation.cpp',
]]
test_source_files = glob.glob(os.path.join(common_codegen.repo_relative('tests'), '*.cpp'))
unassigned_vuid_files = [common_codegen.repo_relative(path) for path in [
'layers/best_practices_error_enums.h',
'layers/stateless_validation.h',
'layers/core_validation_error_enums.h',
'layers/object_lifetime_validation.h'
]]
# These files should not change unless event there is a major refactoring in SPIR-V Tools
# Paths are relative from root of SPIR-V Tools repo
spirvtools_source_files = ["source/val/validation_state.cpp"]
spirvtools_test_files = ["test/val/*.cpp"]
def printHelp():
print ("Usage:")
print (" python vk_validation_stats.py <json_file>")
print (" [ -c ]")
print (" [ -todo ]")
print (" [ -vuid <vuid_name> ]")
print (" [ -unassigned ]")
print (" [ -spirvtools [ <path_to_spirv_tools_repo>] ]")
print (" [ -text [ <text_out_filename>] ]")
print (" [ -csv [ <csv_out_filename>] ]")
print (" [ -html [ <html_out_filename>] ]")
print (" [ -export_header ]")
print (" [ -summary ]")
print (" [ -verbose ]")
print (" [ -help ]")
print ("\n The vk_validation_stats script parses validation layer source files to")
print (" determine the set of valid usage checks and tests currently implemented,")
print (" and generates coverage values by comparing against the full set of valid")
print (" usage identifiers in the Vulkan-Headers registry file 'validusage.json'")
print ("\nArguments: ")
print (" <json-file> (required) registry file 'validusage.json'")
print (" -c report consistency warnings")
print (" -todo report unimplemented VUIDs")
print (" -vuid <vuid_name> report status of individual VUID <vuid_name>")
print (" -unassigned report unassigned VUIDs")
print (" -spirvtools [path] when pointed to root directory of SPIRV-Tools repo, will search")
print (" the repo for VUs that are implemented there")
print (" -text [filename] output the error database text to <text_database_filename>,")
print (" defaults to 'validation_error_database.txt'")
print (" -csv [filename] output the error database in csv to <csv_database_filename>,")
print (" defaults to 'validation_error_database.csv'")
print (" -html [filename] output the error database in html to <html_database_filename>,")
print (" defaults to 'validation_error_database.html'")
print (" -export_header export a new VUID error text header file to <%s>" % header_filename)
print (" -summary output summary of VUID coverage")
print (" -verbose show your work (to stdout)")
class ValidationJSON:
def __init__(self, filename):
self.filename = filename
self.explicit_vuids = set()
self.implicit_vuids = set()
self.all_vuids = set()
self.vuid_db = defaultdict(list) # Maps VUID string to list of json-data dicts
self.apiversion = ""
self.duplicate_vuids = set()
# A set of specific regular expression substitutions needed to clean up VUID text
self.regex_dict = {}
self.regex_dict[re.compile('<.*?>|&(amp;)+lt;|&(amp;)+gt;')] = ""
self.regex_dict[re.compile(r'\\\(codeSize \\over 4\\\)')] = "(codeSize/4)"
self.regex_dict[re.compile(r'\\\(\\lceil\{\\mathit\{rasterizationSamples} \\over 32}\\rceil\\\)')] = "(rasterizationSamples/32)"
self.regex_dict[re.compile(r'\\\(\\left\\lceil{\\frac{maxFramebufferWidth}{minFragmentDensityTexelSize_{width}}}\\right\\rceil\\\)')] = "the ceiling of maxFramebufferWidth/minFragmentDensityTexelSize.width"
self.regex_dict[re.compile(r'\\\(\\left\\lceil{\\frac{maxFramebufferHeight}{minFragmentDensityTexelSize_{height}}}\\right\\rceil\\\)')] = "the ceiling of maxFramebufferHeight/minFragmentDensityTexelSize.height"
self.regex_dict[re.compile(r'\\\(\\left\\lceil{\\frac{width}{maxFragmentDensityTexelSize_{width}}}\\right\\rceil\\\)')] = "the ceiling of width/maxFragmentDensityTexelSize.width"
self.regex_dict[re.compile(r'\\\(\\left\\lceil{\\frac{height}{maxFragmentDensityTexelSize_{height}}}\\right\\rceil\\\)')] = "the ceiling of height/maxFragmentDensityTexelSize.height"
self.regex_dict[re.compile(r'\\\(\\textrm\{codeSize} \\over 4\\\)')] = "(codeSize/4)"
# Regular expression for characters outside ascii range
self.unicode_regex = re.compile('[^\x00-\x7f]')
# Mapping from unicode char to ascii approximation
self.unicode_dict = {
'\u002b' : '+', # PLUS SIGN
'\u00b4' : "'", # ACUTE ACCENT
'\u200b' : '', # ZERO WIDTH SPACE
'\u2018' : "'", # LEFT SINGLE QUOTATION MARK
'\u2019' : "'", # RIGHT SINGLE QUOTATION MARK
'\u201c' : '"', # LEFT DOUBLE QUOTATION MARK
'\u201d' : '"', # RIGHT DOUBLE QUOTATION MARK
'\u2026' : '...',# HORIZONTAL ELLIPSIS
'\u2032' : "'", # PRIME
'\u2192' : '->', # RIGHTWARDS ARROW
}
def sanitize(self, text, location):
# Strip leading/trailing whitespace
text = text.strip()
# Apply regex text substitutions
for regex, replacement in self.regex_dict.items():
text = re.sub(regex, replacement, text)
# Un-escape html entity codes, ie &#XXXX;
text = html.unescape(text)
# Apply unicode substitutions
for unicode in self.unicode_regex.findall(text):
try:
# Replace known chars
text = text.replace(unicode, self.unicode_dict[unicode])
except KeyError:
# Strip and warn on unrecognized chars
text = text.replace(unicode, '')
name = unicodedata.name(unicode, 'UNKNOWN')
print('Warning: Unknown unicode character \\u{:04x} ({}) at {}'.format(ord(unicode), name, location))
return text
def read(self):
self.json_dict = {}
if os.path.isfile(self.filename):
json_file = open(self.filename, 'r', encoding='utf-8')
self.json_dict = json.load(json_file, object_pairs_hook=OrderedDict)
json_file.close()
if len(self.json_dict) == 0:
print("Error: Error loading validusage.json file <%s>" % self.filename)
sys.exit(-1)
try:
version = self.json_dict['version info']
validation = self.json_dict['validation']
self.apiversion = version['api version']
except:
print("Error: Failure parsing validusage.json object")
sys.exit(-1)
# Parse vuid from json into local databases
for apiname in validation.keys():
apidict = validation[apiname]
for ext in apidict.keys():
vlist = apidict[ext]
for ventry in vlist:
vuid_string = ventry['vuid']
if (vuid_string[-5:-1].isdecimal()):
self.explicit_vuids.add(vuid_string) # explicit end in 5 numeric chars
vtype = 'explicit'
else:
self.implicit_vuids.add(vuid_string) # otherwise, implicit
vtype = 'implicit'
vuid_text = self.sanitize(ventry['text'], vuid_string)
self.vuid_db[vuid_string].append({'api':apiname, 'ext':ext, 'type':vtype, 'text':vuid_text})
self.all_vuids = self.explicit_vuids | self.implicit_vuids
self.duplicate_vuids = set({v for v in self.vuid_db if len(self.vuid_db[v]) > 1})
if len(self.duplicate_vuids) > 0:
print("Warning: duplicate VUIDs found in validusage.json")
def buildKvuidDict():
kvuid_dict = {}
for uf in unassigned_vuid_files:
line_num = 0
with open(uf) as f:
for line in f:
line_num = line_num + 1
if True in [line.strip().startswith(comment) for comment in ['//', '/*']]:
continue
if 'kVUID_' in line:
kvuid_pos = line.find('kVUID_'); assert(kvuid_pos >= 0)
eq_pos = line.find('=', kvuid_pos)
if eq_pos >= 0:
kvuid = line[kvuid_pos:eq_pos].strip(' \t\n;"')
unassigned_str = line[eq_pos+1:].strip(' \t\n;"')
kvuid_dict[kvuid] = unassigned_str
return kvuid_dict
class ValidationSource:
def __init__(self, source_file_list):
self.source_files = source_file_list
self.vuid_count_dict = {} # dict of vuid values to the count of how much they're used, and location of where they're used
self.duplicated_checks = 0
self.explicit_vuids = set()
self.implicit_vuids = set()
self.unassigned_vuids = set()
self.all_vuids = set()
def parse(self, spirv_val):
kvuid_dict = buildKvuidDict()
if spirv_val and spirv_val.enabled:
self.source_files.extend(spirv_val.source_files)
# build self.vuid_count_dict
prepend = None
for sf in self.source_files:
spirv_file = True if spirv_val.enabled and sf.startswith(spirv_val.repo_path) else False
line_num = 0
with open(sf, encoding='utf-8') as f:
for line in f:
line_num = line_num + 1
if True in [line.strip().startswith(comment) for comment in ['//', '/*']]:
if 'VUID-' not in line or 'TODO:' in line:
continue
# Find vuid strings
if prepend is not None:
line = prepend[:-2] + line.lstrip().lstrip('"') # join lines skipping CR, whitespace and trailing/leading quote char
prepend = None
if any(prefix in line for prefix in vuid_prefixes):
# Replace the '(' of lines containing validation helper functions with ' ' to make them easier to parse
line = line.replace("(", " ")
line_list = line.split()
# A VUID string that has been broken by clang will start with a vuid prefix and end with -, and will be last in the list
broken_vuid = line_list[-1].strip('"')
if any(broken_vuid.startswith(prefix) for prefix in vuid_prefixes) and broken_vuid.endswith('-'):
prepend = line
continue
vuid_list = []
for str in line_list:
if any(prefix in str for prefix in vuid_prefixes):
vuid_list.append(str.strip(',);{}"*'))
for vuid in vuid_list:
if vuid.startswith('kVUID_'): vuid = kvuid_dict[vuid]
if vuid not in self.vuid_count_dict:
self.vuid_count_dict[vuid] = {}
self.vuid_count_dict[vuid]['count'] = 1
self.vuid_count_dict[vuid]['file_line'] = []
self.vuid_count_dict[vuid]['spirv'] = False # default
else:
if self.vuid_count_dict[vuid]['count'] == 1: # only count first time duplicated
self.duplicated_checks = self.duplicated_checks + 1
self.vuid_count_dict[vuid]['count'] = self.vuid_count_dict[vuid]['count'] + 1
self.vuid_count_dict[vuid]['file_line'].append('%s,%d' % (sf, line_num))
if spirv_file:
self.vuid_count_dict[vuid]['spirv'] = True
# Sort vuids by type
for vuid in self.vuid_count_dict.keys():
if (vuid.startswith('VUID-')):
if (vuid[-5:-1].isdecimal()):
self.explicit_vuids.add(vuid) # explicit end in 5 numeric chars
if self.vuid_count_dict[vuid]['spirv']:
spirv_val.source_explicit_vuids.add(vuid)
else:
self.implicit_vuids.add(vuid)
if self.vuid_count_dict[vuid]['spirv']:
spirv_val.source_implicit_vuids.add(vuid)
elif (vuid.startswith('UNASSIGNED-')):
self.unassigned_vuids.add(vuid)
else:
print("Unable to categorize VUID: %s" % vuid)
print("Confused while parsing VUIDs in layer source code - cannot proceed. (FIXME)")
exit(-1)
self.all_vuids = self.explicit_vuids | self.implicit_vuids | self.unassigned_vuids
if spirv_file:
spirv_val.source_all_vuids = spirv_val.source_explicit_vuids | spirv_val.source_implicit_vuids
# Class to parse the validation layer test source and store testnames
class ValidationTests:
def __init__(self, test_file_list, test_group_name=['VkLayerTest', 'VkPositiveLayerTest', 'VkWsiEnabledLayerTest', 'VkBestPracticesLayerTest']):
self.test_files = test_file_list
self.test_trigger_txt_list = []
for tg in test_group_name:
self.test_trigger_txt_list.append('TEST_F(%s' % tg)
self.explicit_vuids = set()
self.implicit_vuids = set()
self.unassigned_vuids = set()
self.all_vuids = set()
#self.test_to_vuids = {} # Map test name to VUIDs tested
self.vuid_to_tests = defaultdict(set) # Map VUIDs to set of test names where implemented
# Parse test files into internal data struct
def parse(self, spirv_val):
kvuid_dict = buildKvuidDict()
if spirv_val and spirv_val.enabled:
self.test_files.extend(spirv_val.test_files)
# For each test file, parse test names into set
grab_next_line = False # handle testname on separate line than wildcard
testname = ''
prepend = None
for test_file in self.test_files:
spirv_file = True if spirv_val.enabled and test_file.startswith(spirv_val.repo_path) else False
with open(test_file) as tf:
for line in tf:
if True in [line.strip().startswith(comment) for comment in ['//', '/*']]:
continue
# if line ends in a broken VUID string, fix that before proceeding
if prepend is not None:
line = prepend[:-2] + line.lstrip().lstrip('"') # join lines skipping CR, whitespace and trailing/leading quote char
prepend = None
if any(prefix in line for prefix in vuid_prefixes):
line_list = line.split()
# A VUID string that has been broken by clang will start with a vuid prefix and end with -, and will be last in the list
broken_vuid = line_list[-1].strip('"')
if any(broken_vuid.startswith(prefix) for prefix in vuid_prefixes) and broken_vuid.endswith('-'):
prepend = line
continue
if any(ttt in line for ttt in self.test_trigger_txt_list):
testname = line.split(',')[-1]
testname = testname.strip().strip(' {)')
if ('' == testname):
grab_next_line = True
continue
#self.test_to_vuids[testname] = []
if grab_next_line: # test name on its own line
grab_next_line = False
testname = testname.strip().strip(' {)')
#self.test_to_vuids[testname] = []
if any(prefix in line for prefix in vuid_prefixes):
line_list = re.split('[\s{}[\]()"]+',line)
for sub_str in line_list:
if any(prefix in sub_str for prefix in vuid_prefixes):
vuid_str = sub_str.strip(',);:"*')
if vuid_str.startswith('kVUID_'): vuid_str = kvuid_dict[vuid_str]
self.vuid_to_tests[vuid_str].add(testname)
#self.test_to_vuids[testname].append(vuid_str)
if (vuid_str.startswith('VUID-')):
if (vuid_str[-5:-1].isdecimal()):
self.explicit_vuids.add(vuid_str) # explicit end in 5 numeric chars
if spirv_file:
spirv_val.test_explicit_vuids.add(vuid_str)
else:
self.implicit_vuids.add(vuid_str)
if spirv_file:
spirv_val.test_implicit_vuids.add(vuid_str)
elif (vuid_str.startswith('UNASSIGNED-')):
self.unassigned_vuids.add(vuid_str)
else:
print("Unable to categorize VUID: %s" % vuid_str)
print("Confused while parsing VUIDs in test code - cannot proceed. (FIXME)")
exit(-1)
self.all_vuids = self.explicit_vuids | self.implicit_vuids | self.unassigned_vuids
# Class to do consistency checking
#
class Consistency:
def __init__(self, all_json, all_checks, all_tests):
self.valid = all_json
self.checks = all_checks
self.tests = all_tests
# Report undefined VUIDs in source code
def undef_vuids_in_layer_code(self):
undef_set = self.checks - self.valid
undef_set.discard('VUID-Undefined') # don't report Undefined
if ignore_unassigned:
unassigned = set({uv for uv in undef_set if uv.startswith('UNASSIGNED-')})
undef_set = undef_set - unassigned
if (len(undef_set) > 0):
print("\nFollowing VUIDs found in layer code are not defined in validusage.json (%d):" % len(undef_set))
undef = list(undef_set)
undef.sort()
for vuid in undef:
print(" %s" % vuid)
return False
return True
# Report undefined VUIDs in tests
def undef_vuids_in_tests(self):
undef_set = self.tests - self.valid
undef_set.discard('VUID-Undefined') # don't report Undefined
if ignore_unassigned:
unassigned = set({uv for uv in undef_set if uv.startswith('UNASSIGNED-')})
undef_set = undef_set - unassigned
if (len(undef_set) > 0):
ok = False
print("\nFollowing VUIDs found in layer tests are not defined in validusage.json (%d):" % len(undef_set))
undef = list(undef_set)
undef.sort()
for vuid in undef:
print(" %s" % vuid)
return False
return True
# Report vuids in tests that are not in source
def vuids_tested_not_checked(self):
undef_set = self.tests - self.checks
undef_set.discard('VUID-Undefined') # don't report Undefined
if ignore_unassigned:
unassigned = set()
for vuid in undef_set:
if vuid.startswith('UNASSIGNED-'):
unassigned.add(vuid)
undef_set = undef_set - unassigned
if (len(undef_set) > 0):
ok = False
print("\nFollowing VUIDs found in tests but are not checked in layer code (%d):" % len(undef_set))
undef = list(undef_set)
undef.sort()
for vuid in undef:
print(" %s" % vuid)
return False
return True
# TODO: Explicit checked VUIDs which have no test
# def explicit_vuids_checked_not_tested(self):
# Class to output database in various flavors
#
class OutputDatabase:
def __init__(self, val_json, val_source, val_tests, spirv_val):
self.vj = val_json
self.vs = val_source
self.vt = val_tests
self.sv = spirv_val
self.header_version = "/* THIS FILE IS GENERATED - DO NOT EDIT (scripts/vk_validation_stats.py) */"
self.header_version += "\n/* Vulkan specification version: %s */" % val_json.apiversion
self.header_preamble = """
/*
* Vulkan
*
* Copyright (c) 2016-2021 Google Inc.
* Copyright (c) 2016-2021 LunarG, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*
* Author: Tobin Ehlis <tobine@google.com>
* Author: Dave Houlton <daveh@lunarg.com>
*/
#pragma once
// Disable auto-formatting for generated file
// clang-format off
// Mapping from VUID string to the corresponding spec text
typedef struct _vuid_spec_text_pair {
const char * vuid;
const char * spec_text;
const char * url_id;
} vuid_spec_text_pair;
static const vuid_spec_text_pair vuid_spec_text[] = {
"""
self.header_postamble = """};
"""
def dump_txt(self, only_unimplemented = False):
print("\n Dumping database to text file: %s" % txt_filename)
with open (txt_filename, 'w') as txt:
txt.write("## VUID Database\n")
txt.write("## Format: VUID_NAME | CHECKED | SPIRV-TOOL | TEST | TYPE | API/STRUCT | EXTENSION | VUID_TEXT\n##\n")
vuid_list = list(self.vj.all_vuids)
vuid_list.sort()
for vuid in vuid_list:
db_list = self.vj.vuid_db[vuid]
db_list.sort(key=operator.itemgetter('ext')) # sort list to ease diffs of output file
for db_entry in db_list:
checked = 'N'
spirv = 'N'
if vuid in self.vs.all_vuids:
if only_unimplemented:
continue
else:
checked = 'Y'
if vuid in self.sv.source_all_vuids:
spirv = 'Y'
test = 'None'
if vuid in self.vt.vuid_to_tests:
test_list = list(self.vt.vuid_to_tests[vuid])
test_list.sort() # sort tests, for diff-ability
sep = ', '
test = sep.join(test_list)
txt.write("%s | %s | %s | %s | %s | %s | %s | %s\n" % (vuid, checked, test, spirv, db_entry['type'], db_entry['api'], db_entry['ext'], db_entry['text']))
def dump_csv(self, only_unimplemented = False):
print("\n Dumping database to csv file: %s" % csv_filename)
with open (csv_filename, 'w', newline='') as csvfile:
cw = csv.writer(csvfile)
cw.writerow(['VUID_NAME','CHECKED','SPIRV-TOOL', 'TEST','TYPE','API/STRUCT','EXTENSION','VUID_TEXT'])
vuid_list = list(self.vj.all_vuids)
vuid_list.sort()
for vuid in vuid_list:
for db_entry in self.vj.vuid_db[vuid]:
row = [vuid]
if vuid in self.vs.all_vuids:
if only_unimplemented:
continue
else:
row.append('Y') # checked
if vuid in self.sv.source_all_vuids:
row.append('Y') # spirv-tool
else:
row.append('N') # spirv-tool
else:
row.append('N') # checked
row.append('N') # spirv-tool
test = 'None'
if vuid in self.vt.vuid_to_tests:
sep = ', '
test = sep.join(self.vt.vuid_to_tests[vuid])
row.append(test)
row.append(db_entry['type'])
row.append(db_entry['api'])
row.append(db_entry['ext'])
row.append(db_entry['text'])
cw.writerow(row)
def dump_html(self, only_unimplemented = False):
print("\n Dumping database to html file: %s" % html_filename)
preamble = '<!DOCTYPE html>\n<html>\n<head>\n<style>\ntable, th, td {\n border: 1px solid black;\n border-collapse: collapse; \n}\n</style>\n<body>\n<h2>Valid Usage Database</h2>\n<font size="2" face="Arial">\n<table style="width:100%">\n'
headers = '<tr><th>VUID NAME</th><th>CHECKED</th><th>SPIRV-TOOL</th><th>TEST</th><th>TYPE</th><th>API/STRUCT</th><th>EXTENSION</th><th>VUID TEXT</th></tr>\n'
with open (html_filename, 'w') as hfile:
hfile.write(preamble)
hfile.write(headers)
vuid_list = list(self.vj.all_vuids)
vuid_list.sort()
for vuid in vuid_list:
for db_entry in self.vj.vuid_db[vuid]:
checked = '<span style="color:red;">N</span>'
spirv = ''
if vuid in self.vs.all_vuids:
if only_unimplemented:
continue
else:
checked = '<span style="color:limegreen;">Y</span>'
if vuid in self.sv.source_all_vuids:
spirv = 'Y'
hfile.write('<tr><th>%s</th>' % vuid)
hfile.write('<th>%s</th>' % checked)
hfile.write('<th>%s</th>' % spirv)
test = 'None'
if vuid in self.vt.vuid_to_tests:
sep = ', '
test = sep.join(self.vt.vuid_to_tests[vuid])
hfile.write('<th>%s</th>' % test)
hfile.write('<th>%s</th>' % db_entry['type'])
hfile.write('<th>%s</th>' % db_entry['api'])
hfile.write('<th>%s</th>' % db_entry['ext'])
hfile.write('<th>%s</th></tr>\n' % db_entry['text'])
hfile.write('</table>\n</body>\n</html>\n')
# make list of spec versions containing given VUID
@staticmethod
def make_vuid_spec_version_list(pattern, max_minor_version):
assert pattern
all_editions_list = []
for e in reversed(range(max_minor_version+1)):
all_editions_list.append({"version": e, "ext": True, "khr" : False})
all_editions_list.append({"version": e, "ext": False, "khr" : True})
all_editions_list.append({"version": e, "ext": False, "khr" : False})
if pattern == 'core':
return all_editions_list
# pattern is series of parentheses separated by plus
# each parentheses can be prepended by negation (!)
# each parentheses contains list of extensions or vk versions separated by either comma or plus
edition_list_out = []
for edition in all_editions_list:
resolved_pattern = True
raw_terms = re.split(r'\)\+', pattern)
for raw_term in raw_terms:
negated = raw_term.startswith('!')
term = raw_term.lstrip('!(').rstrip(')')
conjunction = '+' in term
disjunction = ',' in term
assert not (conjunction and disjunction)
if conjunction: features = term.split('+')
elif disjunction: features = term.split(',')
else: features = [term]
assert features
def isDefined(feature, edition):
def getVersion(f): return int(f.replace('VK_VERSION_1_', '', 1))
def isVersion(f): return f.startswith('VK_VERSION_') and feature != 'VK_VERSION_1_0' and getVersion(feature) < 1024
def isExtension(f): return f.startswith('VK_') and not isVersion(f)
def isKhr(f): return f.startswith('VK_KHR_')
assert isExtension(feature) or isVersion(feature)
if isVersion(feature) and getVersion(feature) <= edition['version']: return True
elif isExtension(feature) and edition['ext']: return True
elif isKhr(feature) and edition['khr']: return True
else: return False
if not negated and (conjunction or (not conjunction and not disjunction)): # all defined
resolved_term = True
for feature in features:
if not isDefined(feature, edition): resolved_term = False
elif negated and conjunction: # at least one not defined
resolved_term = False
for feature in features:
if not isDefined(feature, edition): resolved_term = True
elif not negated and disjunction: # at least one defined
resolved_term = False
for feature in features:
if isDefined(feature, edition): resolved_term = True
elif negated and (disjunction or (not conjunction and not disjunction)): # none defined
resolved_term = True
for feature in features:
if isDefined(feature, edition): resolved_term = False
resolved_pattern = resolved_pattern and resolved_term
if resolved_pattern: edition_list_out.append(edition)
return edition_list_out
def export_header(self):
if verbose_mode:
print("\n Exporting header file to: %s" % header_filename)
with open (header_filename, 'w', newline='\n') as hfile:
hfile.write(self.header_version)
hfile.write(self.header_preamble)
vuid_list = list(self.vj.all_vuids)
vuid_list.sort()
minor_version = int(self.vj.apiversion.split('.')[1])
for vuid in vuid_list:
db_entry = self.vj.vuid_db[vuid][0]
spec_list = self.make_vuid_spec_version_list(db_entry['ext'], minor_version)
if not spec_list: spec_url_id = 'default'
elif spec_list[0]['ext']: spec_url_id = '1.%s-extensions' % spec_list[0]['version']
elif spec_list[0]['khr']: spec_url_id = '1.%s-khr-extensions' % spec_list[0]['version']
else: spec_url_id = '1.%s' % spec_list[0]['version']
# Escape quotes and backslashes when generating C strings for source code
db_text = db_entry['text'].replace('\\', '\\\\').replace('"', '\\"')
hfile.write(' {"%s", "%s", "%s"},\n' % (vuid, db_text, spec_url_id))
# For multiply-defined VUIDs, include versions with extension appended
if len(self.vj.vuid_db[vuid]) > 1:
print('Warning: Found a duplicate VUID: %s' % vuid)
hfile.write(self.header_postamble)
class SpirvValidation:
def __init__(self, repo_path):
self.enabled = (repo_path != None)
self.repo_path = repo_path
self.version = 'unknown'
self.source_files = []
self.test_files = []
self.source_explicit_vuids = set()
self.source_implicit_vuids = set()
self.source_all_vuids = set()
self.test_explicit_vuids = set()
self.test_implicit_vuids = set()
def load(self, verbose):
if self.enabled == False:
return
# Get hash from git if available
try:
git_dir = os.path.join(self.repo_path, '.git')
process = subprocess.Popen(['git', '--git-dir='+git_dir ,'rev-parse', 'HEAD'], shell=False, stdout=subprocess.PIPE)
self.version = process.communicate()[0].strip().decode('utf-8')[:7]
if process.poll() != 0:
throw
elif verbose:
print('Found SPIR-V Tools version %s' % self.version)
except:
# leave as default
if verbose:
print('Could not find .git file for version of SPIR-V tools, marking as %s' % self.version)
# Find and parse files with VUIDs in source
for path in spirvtools_source_files:
self.source_files.extend(glob.glob(os.path.join(self.repo_path, path)))
for path in spirvtools_test_files:
self.test_files.extend(glob.glob(os.path.join(self.repo_path, path)))
def main(argv):
global verbose_mode
global txt_filename
global csv_filename
global html_filename
global spirvtools_path
run_consistency = False
report_unimplemented = False
report_unassigned = False
get_vuid_status = ''
txt_out = False
csv_out = False
html_out = False
header_out = False
show_summary = False
if (1 > len(argv)):
printHelp()
sys.exit()
# Parse script args
json_filename = argv[0]
i = 1
while (i < len(argv)):
arg = argv[i]
i = i + 1
if (arg == '-c'):
run_consistency = True
elif (arg == '-vuid'):
get_vuid_status = argv[i]
i = i + 1
elif (arg == '-todo'):
report_unimplemented = True
elif (arg == '-unassigned'):
report_unassigned = True
elif (arg == '-spirvtools'):
spirvtools_path = argv[i]
i = i + 1
elif (arg == '-text'):
txt_out = True
# Set filename if supplied, else use default
if i < len(argv) and not argv[i].startswith('-'):
txt_filename = argv[i]
i = i + 1
elif (arg == '-csv'):
csv_out = True
# Set filename if supplied, else use default
if i < len(argv) and not argv[i].startswith('-'):
csv_filename = argv[i]
i = i + 1
elif (arg == '-html'):
html_out = True
# Set filename if supplied, else use default
if i < len(argv) and not argv[i].startswith('-'):
html_filename = argv[i]
i = i + 1
elif (arg == '-export_header'):
header_out = True
elif (arg in ['-verbose']):
verbose_mode = True
elif (arg in ['-summary']):
show_summary = True
elif (arg in ['-help', '-h']):
printHelp()
sys.exit()
else:
print("Unrecognized argument: %s\n" % arg)
printHelp()
sys.exit()
result = 0 # Non-zero result indicates an error case
# Load in SPIRV-Tools if passed in
spirv_val = SpirvValidation(spirvtools_path)
spirv_val.load(verbose_mode)
# Parse validusage json
val_json = ValidationJSON(json_filename)
val_json.read()
exp_json = len(val_json.explicit_vuids)
imp_json = len(val_json.implicit_vuids)
all_json = len(val_json.all_vuids)
if verbose_mode:
print("Found %d unique error vuids in validusage.json file." % all_json)
print(" %d explicit" % exp_json)
print(" %d implicit" % imp_json)
if len(val_json.duplicate_vuids) > 0:
print("%d VUIDs appear in validusage.json more than once." % len(val_json.duplicate_vuids))
for vuid in val_json.duplicate_vuids:
print(" %s" % vuid)
for ext in val_json.vuid_db[vuid]:
print(" with extension: %s" % ext['ext'])
# Parse layer source files
val_source = ValidationSource(layer_source_files)
val_source.parse(spirv_val)
exp_checks = len(val_source.explicit_vuids)
imp_checks = len(val_source.implicit_vuids)
all_checks = len(val_source.vuid_count_dict.keys())
spirv_exp_checks = len(spirv_val.source_explicit_vuids) if spirv_val.enabled else 0
spirv_imp_checks = len(spirv_val.source_implicit_vuids) if spirv_val.enabled else 0
spirv_all_checks = (spirv_exp_checks + spirv_imp_checks) if spirv_val.enabled else 0
if verbose_mode:
print("Found %d unique vuid checks in layer source code." % all_checks)
print(" %d explicit" % exp_checks)
if spirv_val.enabled:
print(" SPIR-V Tool make up %d" % spirv_exp_checks)
print(" %d implicit" % imp_checks)
if spirv_val.enabled:
print(" SPIR-V Tool make up %d" % spirv_imp_checks)
print(" %d unassigned" % len(val_source.unassigned_vuids))
print(" %d checks are implemented more that once" % val_source.duplicated_checks)
# Parse test files
val_tests = ValidationTests(test_source_files)
val_tests.parse(spirv_val)
exp_tests = len(val_tests.explicit_vuids)
imp_tests = len(val_tests.implicit_vuids)
all_tests = len(val_tests.all_vuids)
spirv_exp_tests = len(spirv_val.test_explicit_vuids) if spirv_val.enabled else 0
spirv_imp_tests = len(spirv_val.test_implicit_vuids) if spirv_val.enabled else 0
spirv_all_tests = (spirv_exp_tests + spirv_imp_tests) if spirv_val.enabled else 0
if verbose_mode:
print("Found %d unique error vuids in test source code." % all_tests)
print(" %d explicit" % exp_tests)
if spirv_val.enabled:
print(" From SPIRV-Tools: %d" % spirv_exp_tests)
print(" %d implicit" % imp_tests)
if spirv_val.enabled:
print(" From SPIRV-Tools: %d" % spirv_imp_tests)
print(" %d unassigned" % len(val_tests.unassigned_vuids))
# Process stats
if show_summary:
if spirv_val.enabled:
print("\nValidation Statistics (using validusage.json version %s and SPIRV-Tools version %s)" % (val_json.apiversion, spirv_val.version))
else:
print("\nValidation Statistics (using validusage.json version %s)" % val_json.apiversion)
print(" VUIDs defined in JSON file: %04d explicit, %04d implicit, %04d total." % (exp_json, imp_json, all_json))
print(" VUIDs checked in layer code: %04d explicit, %04d implicit, %04d total." % (exp_checks, imp_checks, all_checks))
if spirv_val.enabled:
print(" From SPIRV-Tools: %04d explicit, %04d implicit, %04d total." % (spirv_exp_checks, spirv_imp_checks, spirv_all_checks))
print(" VUIDs tested in layer tests: %04d explicit, %04d implicit, %04d total." % (exp_tests, imp_tests, all_tests))
if spirv_val.enabled:
print(" From SPIRV-Tools: %04d explicit, %04d implicit, %04d total." % (spirv_exp_tests, spirv_imp_tests, spirv_all_tests))
print("\nVUID check coverage")
print(" Explicit VUIDs checked: %.1f%% (%d checked vs %d defined)" % ((100.0 * exp_checks / exp_json), exp_checks, exp_json))
print(" Implicit VUIDs checked: %.1f%% (%d checked vs %d defined)" % ((100.0 * imp_checks / imp_json), imp_checks, imp_json))
print(" Overall VUIDs checked: %.1f%% (%d checked vs %d defined)" % ((100.0 * all_checks / all_json), all_checks, all_json))
print("\nVUID test coverage")
print(" Explicit VUIDs tested: %.1f%% (%d tested vs %d checks)" % ((100.0 * exp_tests / exp_checks), exp_tests, exp_checks))
print(" Implicit VUIDs tested: %.1f%% (%d tested vs %d checks)" % ((100.0 * imp_tests / imp_checks), imp_tests, imp_checks))
print(" Overall VUIDs tested: %.1f%% (%d tested vs %d checks)" % ((100.0 * all_tests / all_checks), all_tests, all_checks))
# Report status of a single VUID
if len(get_vuid_status) > 1:
print("\n\nChecking status of <%s>" % get_vuid_status);
if get_vuid_status not in val_json.all_vuids and not get_vuid_status.startswith('UNASSIGNED-'):
print(' Not a valid VUID string.')
else:
if get_vuid_status in val_source.explicit_vuids:
print(' Implemented!')
line_list = val_source.vuid_count_dict[get_vuid_status]['file_line']
for line in line_list:
print(' => %s' % line)
elif get_vuid_status in val_source.implicit_vuids:
print(' Implemented! (Implicit)')
line_list = val_source.vuid_count_dict[get_vuid_status]['file_line']
for line in line_list:
print(' => %s' % line)
else:
print(' Not implemented.')
if get_vuid_status in val_tests.all_vuids:
print(' Has a test!')
test_list = val_tests.vuid_to_tests[get_vuid_status]
for test in test_list:
print(' => %s' % test)
else:
print(' Not tested.')
# Report unimplemented explicit VUIDs
if report_unimplemented:
unim_explicit = val_json.explicit_vuids - val_source.explicit_vuids
print("\n\n%d explicit VUID checks remain unimplemented:" % len(unim_explicit))
ulist = list(unim_explicit)
ulist.sort()
for vuid in ulist:
print(" => %s" % vuid)
# Report unassigned VUIDs
if report_unassigned:
# TODO: I do not really want VUIDs created for warnings though here
print("\n\n%d checks without a spec VUID:" % len(val_source.unassigned_vuids))
ulist = list(val_source.unassigned_vuids)
ulist.sort()
for vuid in ulist:
print(" => %s" % vuid)
line_list = val_source.vuid_count_dict[vuid]['file_line']
for line in line_list:
print(' => %s' % line)
print("\n%d tests without a spec VUID:" % len(val_source.unassigned_vuids))
ulist = list(val_tests.unassigned_vuids)
ulist.sort()
for vuid in ulist:
print(" => %s" % vuid)
test_list = val_tests.vuid_to_tests[vuid]
for test in test_list:
print(' => %s' % test)
# Consistency tests
if run_consistency:
print("\n\nRunning consistency tests...")
con = Consistency(val_json.all_vuids, val_source.all_vuids, val_tests.all_vuids)
ok = con.undef_vuids_in_layer_code()
ok &= con.undef_vuids_in_tests()
ok &= con.vuids_tested_not_checked()
if ok:
print(" OK! No inconsistencies found.")
# Output database in requested format(s)
db_out = OutputDatabase(val_json, val_source, val_tests, spirv_val)
if txt_out:
db_out.dump_txt(report_unimplemented)
if csv_out:
db_out.dump_csv(report_unimplemented)
if html_out:
db_out.dump_html(report_unimplemented)
if header_out:
db_out.export_header()
return result
if __name__ == "__main__":
sys.exit(main(sys.argv[1:]))