Re: [edk2-devel] [PATCH V4] BaseTools: Optimize GenerateByteArrayValue and CollectPlatformGuids APIs

["Feng, Bob C" <bob.c.feng@intel.com>]

Reviewed-by: Bob Feng <bob.c.feng@intel.com>

-----Original Message-----
From: S, Ashraf Ali <ashraf.ali.s@intel.com> 
Sent: Thursday, February 8, 2024 1:21 AM
To: devel@edk2.groups.io
Cc: S, Ashraf Ali <ashraf.ali.s@intel.com>; Chen, Christine <yuwei.chen@intel.com>; Rebecca Cran <rebecca@bsdio.com>; Liming Gao <gaoliming@byosoft.com.cn>; Feng, Bob C <bob.c.feng@intel.com>; Chan, Amy <amy.chan@intel.com>; Chaganty, Rangasai V <rangasai.v.chaganty@intel.com>; Solanki, Digant H <digant.h.solanki@intel.com>
Subject: [PATCH V4] BaseTools: Optimize GenerateByteArrayValue and CollectPlatformGuids APIs

During the Incremental build GenerateByteArrayValue used to generate the ByteArrayValue even when there is no change in the PCD/VPDs. which is time consuming API based on the number of PCD/VPDs and SKU IDs.

The optimization is that GenerateByteArrayValue is used to store the StructuredPcdsData in a JSON file for each of the arch. and during the Incremental build this API will check, if there is any change in the Structured PCD/VPDs then rest of the flow remains the same.
if there is no change then it will return the provious build data.

Flow:
during the 1st build StructuredPcdsData.json is not exists, StructuredPcdsData will be dumped to json file. and it will copy the output.txt as well.
Note: as the output.txt are different for different Arch, so it will be stored in the Arch folder.
During the Incremental build check if there is any change in Structured PCD/VPD. if there is a change in Structured VPD/PCD then recreate the StructuredPcdsData.json, and rest of the flow remains same.
if there is no change in VPD/PCD read the output.txt and return the data

Unit Test:
Test1: Modified the Structured Pcds default from DEC file. current flow is executing.
Test2: Override the default value of the PCD from DEC file. current flow is executing.
Test3: Modified/Override the PCD from DSC file. current flow executing
Test4: Modified/Override the FDF from DSC file. current flow executing
Test5: update the default value from Command Line.current flow executing
Test6: Build without change in PCD in DSC, FDF, DEC and Command Line the proposed changes will be executing, and the return data remains the same with and without the changes.
Test7: Build with and without modified the include headers of Structured PCDs. if there is any change in those Structured PCD header then current flow will be executed.

With these changes it's helping to save around ~2.5min to ~3.5min of Incremental build time in my build environment.

Sample PR: https://github.com/tianocore/edk2-basetools/pull/113

Cc: Yuwei Chen <yuwei.chen@intel.com>
Cc: Rebecca Cran <rebecca@bsdio.com>
Cc: Liming Gao <gaoliming@byosoft.com.cn>
Cc: Bob Feng <bob.c.feng@intel.com>
Cc: Amy Chan <amy.chan@intel.com>
Cc: Sai Chaganty <rangasai.v.chaganty@intel.com>
Cc: Digant H Solanki <digant.h.solanki@intel.com>
Signed-off-by: Ashraf Ali S <ashraf.ali.s@intel.com>
---
 .../Source/Python/AutoGen/WorkspaceAutoGen.py |  16 +-
 .../Source/Python/Workspace/DscBuildData.py   | 181 +++++++++++++-----
 2 files changed, 139 insertions(+), 58 deletions(-)

diff --git a/BaseTools/Source/Python/AutoGen/WorkspaceAutoGen.py b/BaseTools/Source/Python/AutoGen/WorkspaceAutoGen.py
index 160e3a3cd3..eec9280c8e 100644
--- a/BaseTools/Source/Python/AutoGen/WorkspaceAutoGen.py
+++ b/BaseTools/Source/Python/AutoGen/WorkspaceAutoGen.py
@@ -160,22 +160,18 @@ class WorkspaceAutoGen(AutoGen):
 
     def CollectPlatformGuids(self):
         oriInfList = []
-        oriPkgSet = set()
-        PlatformPkg = set()
+        pkgSet = set()
         for Arch in self.ArchList:
             Platform = self.BuildDatabase[self.MetaFile, Arch, self.BuildTarget, self.ToolChain]
             oriInfList = Platform.Modules
             for ModuleFile in oriInfList:
                 ModuleData = self.BuildDatabase[ModuleFile, Platform._Arch, Platform._Target, Platform._Toolchain]
-                oriPkgSet.update(ModuleData.Packages)
-                for Pkg in oriPkgSet:
-                    Guids = Pkg.Guids
-                    GlobalData.gGuidDict.update(Guids)
+                pkgSet.update(ModuleData.Packages)
             if Platform.Packages:
-                PlatformPkg.update(Platform.Packages)
-                for Pkg in PlatformPkg:
-                    Guids = Pkg.Guids
-                    GlobalData.gGuidDict.update(Guids)
+                pkgSet.update(Platform.Packages)
+        for Pkg in pkgSet:
+            Guids = Pkg.Guids
+            GlobalData.gGuidDict.update(Guids)
 
     @cached_property
     def FdfProfile(self):
diff --git a/BaseTools/Source/Python/Workspace/DscBuildData.py b/BaseTools/Source/Python/Workspace/DscBuildData.py
index 4768099343..ddcd157c73 100644
--- a/BaseTools/Source/Python/Workspace/DscBuildData.py
+++ b/BaseTools/Source/Python/Workspace/DscBuildData.py
@@ -37,6 +37,8 @@ from functools import reduce  from Common.Misc import SaveFileOnChange  from Workspace.BuildClassObject import PlatformBuildClassObject, StructurePcd, PcdClassObject, ModuleBuildClassObject  from collections import OrderedDict, defaultdict
+import json
+import shutil
 
 def _IsFieldValueAnArray (Value):
     Value = Value.strip()
@@ -56,6 +58,7 @@ def _IsFieldValueAnArray (Value):
 
 PcdValueInitName = 'PcdValueInit'
 PcdValueCommonName = 'PcdValueCommon'
+StructuredPcdsDataName = 'StructuredPcdsData.json'
 
 PcdMainCHeader = '''
 /**
@@ -2750,6 +2753,63 @@ class DscBuildData(PlatformBuildClassObject):
                 ccflags.add(item)
             i +=1
         return ccflags
+
+    def GetStructurePcdSet (self, OutputValueFile):
+        if not os.path.isfile(OutputValueFile):
+            EdkLogger.error("GetStructurePcdSet", FILE_NOT_FOUND, "Output.txt doesn't exist", ExtraData=OutputValueFile)
+            return []
+        File = open (OutputValueFile, 'r')
+        FileBuffer = File.readlines()
+        File.close()
+
+        #start update structure pcd final value
+        StructurePcdSet = []
+        for Pcd in FileBuffer:
+            PcdValue = Pcd.split ('|')
+            PcdInfo = PcdValue[0].split ('.')
+            StructurePcdSet.append((PcdInfo[0], PcdInfo[1], PcdInfo[2], PcdInfo[3], PcdValue[2].strip()))
+        return StructurePcdSet
+
+    def GetBuildOptionsValueList(self):
+        CC_FLAGS = LinuxCFLAGS
+        if sys.platform == "win32":
+            CC_FLAGS = WindowsCFLAGS
+        BuildOptions = OrderedDict()
+        for Options in self.BuildOptions:
+            if Options[2] != EDKII_NAME:
+                continue
+            Family = Options[0]
+            if Family and Family != self.ToolChainFamily:
+                continue
+            Target, Tag, Arch, Tool, Attr = Options[1].split("_")
+            if Tool != 'CC':
+                continue
+            if Attr != "FLAGS":
+                continue
+            if Target == TAB_STAR or Target == self._Target:
+                if Tag == TAB_STAR or Tag == self._Toolchain:
+                    if 'COMMON' not in BuildOptions:
+                        BuildOptions['COMMON'] = set()
+                    if Arch == TAB_STAR:
+                        BuildOptions['COMMON']|= self.ParseCCFlags(self.BuildOptions[Options])
+                    if Arch in self.SupArchList:
+                        if Arch not in BuildOptions:
+                            BuildOptions[Arch] = set()
+                        BuildOptions[Arch] |= 
+ self.ParseCCFlags(self.BuildOptions[Options])
+
+        if BuildOptions:
+            ArchBuildOptions = {arch:flags for arch,flags in BuildOptions.items() if arch != 'COMMON'}
+            if len(ArchBuildOptions.keys()) == 1:
+                BuildOptions['COMMON'] |= (list(ArchBuildOptions.values())[0])
+            elif len(ArchBuildOptions.keys()) > 1:
+                CommonBuildOptions = reduce(lambda x,y: x&y, ArchBuildOptions.values())
+                BuildOptions['COMMON'] |= CommonBuildOptions
+            ValueList = [item for item in BuildOptions['COMMON'] if item.startswith((r"/U","-U"))]
+            ValueList.extend([item for item in BuildOptions['COMMON'] if item.startswith((r"/D", "-D"))])
+            CC_FLAGS += " ".join(ValueList)
+        return CC_FLAGS
+
+
     def GenerateByteArrayValue (self, StructuredPcds):
         #
         # Generate/Compile/Run C application to determine if there are any flexible array members @@ -2757,6 +2817,66 @@ class DscBuildData(PlatformBuildClassObject):
         if not StructuredPcds:
             return
 
+        StructuredPcdsData = {}
+        StoredStructuredPcdObjectPaths = {}
+        SkipPcdValueInit = False
+
+        CC_FLAGS = self.GetBuildOptionsValueList()
+
+        for PcdName in StructuredPcds:
+            Pcd = StructuredPcds[PcdName]
+            TokenSpaceGuidCName = Pcd.TokenSpaceGuidCName
+            TokenCName = Pcd.TokenCName
+
+            # Create a key using TokenSpaceGuidCName and TokenCName
+            StructuredPcdsData[f"{TokenSpaceGuidCName}_{TokenCName}"] = {
+                "DefaultValueFromDec": Pcd.DefaultValueFromDec,
+                "DefaultValues": Pcd.DefaultValues,
+                "PcdFieldValueFromComm": Pcd.PcdFieldValueFromComm,
+                "PcdFieldValueFromFdf": Pcd.PcdFieldValueFromFdf,
+                "DefaultFromDSC": Pcd.DefaultFromDSC,
+                "PcdFiledValueFromDscComponent": Pcd.PcdFiledValueFromDscComponent
+            }
+
+        # Store the CC Flags
+        StructuredPcdsData["CC_FLAGS"] = CC_FLAGS
+        #
+        # If the output path doesn't exists then create it
+        #
+        if not os.path.exists(self.OutputPath):
+            os.makedirs(self.OutputPath)
+
+        StructuredPcdsDataPath = os.path.join(self.OutputPath, self._Arch, StructuredPcdsDataName)
+        PcdRecordOutputValueFile = os.path.join(self.OutputPath, 
+ self._Arch, 'Output.txt')
+
+        if not os.path.exists(os.path.dirname(StructuredPcdsDataPath)):
+            os.makedirs(os.path.dirname(StructuredPcdsDataPath))
+        #
+        # Check if the StructuredPcdsData.json exists or not
+        # if exits then it might be a incremental build then check if the StructuredPcdsData has been changed or not.
+        # if changed then proceed further, if not changed then return the stored data from earlier build
+        #
+        if os.path.isfile(StructuredPcdsDataPath):
+            with open(StructuredPcdsDataPath, 'r') as file:
+                StoredStructuredPcdsData = json.load(file)
+                # OBJECTS will have the modified time, which needs to be checked later
+                StoredStructuredPcdObjectPaths = 
+ StoredStructuredPcdsData.pop("OBJECTS", {})
+
+                if StructuredPcdsData == StoredStructuredPcdsData:
+                    SkipPcdValueInit = True
+                    for filename, file_mtime in StoredStructuredPcdObjectPaths.items():
+                        f_mtime = os.path.getmtime(filename)
+                        #
+                        # check if the include_file are modified or not,
+                        # if modified then generate the PcdValueInit
+                        #
+                        if f_mtime != file_mtime:
+                            SkipPcdValueInit = False
+                            break
+
+        if SkipPcdValueInit:
+            return self.GetStructurePcdSet(PcdRecordOutputValueFile)
+
         InitByteValue = ""
         CApp = PcdMainCHeader
 
@@ -2832,8 +2952,6 @@ class DscBuildData(PlatformBuildClassObject):
 
         CApp = CApp + PcdMainCEntry + '\n'
 
-        if not os.path.exists(self.OutputPath):
-            os.makedirs(self.OutputPath)
         CAppBaseFileName = os.path.join(self.OutputPath, PcdValueInitName)
         SaveFileOnChange(CAppBaseFileName + '.c', CApp, False)
 
@@ -2890,42 +3008,6 @@ class DscBuildData(PlatformBuildClassObject):
                         IncSearchList.append(inc)
         MakeApp = MakeApp + '\n'
 
-        CC_FLAGS = LinuxCFLAGS
-        if sys.platform == "win32":
-            CC_FLAGS = WindowsCFLAGS
-        BuildOptions = OrderedDict()
-        for Options in self.BuildOptions:
-            if Options[2] != EDKII_NAME:
-                continue
-            Family = Options[0]
-            if Family and Family != self.ToolChainFamily:
-                continue
-            Target, Tag, Arch, Tool, Attr = Options[1].split("_")
-            if Tool != 'CC':
-                continue
-            if Attr != "FLAGS":
-                continue
-            if Target == TAB_STAR or Target == self._Target:
-                if Tag == TAB_STAR or Tag == self._Toolchain:
-                    if 'COMMON' not in BuildOptions:
-                        BuildOptions['COMMON'] = set()
-                    if Arch == TAB_STAR:
-                        BuildOptions['COMMON']|= self.ParseCCFlags(self.BuildOptions[Options])
-                    if Arch in self.SupArchList:
-                        if Arch not in BuildOptions:
-                            BuildOptions[Arch] = set()
-                        BuildOptions[Arch] |= self.ParseCCFlags(self.BuildOptions[Options])
-
-        if BuildOptions:
-            ArchBuildOptions = {arch:flags for arch,flags in BuildOptions.items() if arch != 'COMMON'}
-            if len(ArchBuildOptions.keys()) == 1:
-                BuildOptions['COMMON'] |= (list(ArchBuildOptions.values())[0])
-            elif len(ArchBuildOptions.keys()) > 1:
-                CommonBuildOptions = reduce(lambda x,y: x&y, ArchBuildOptions.values())
-                BuildOptions['COMMON'] |= CommonBuildOptions
-            ValueList = [item for item in BuildOptions['COMMON'] if item.startswith((r"/U","-U"))]
-            ValueList.extend([item for item in BuildOptions['COMMON'] if item.startswith((r"/D", "-D"))])
-            CC_FLAGS += " ".join(ValueList)
         MakeApp += CC_FLAGS
 
         if sys.platform == "win32":
@@ -2946,7 +3028,9 @@ class DscBuildData(PlatformBuildClassObject):
         SearchPathList.append(os.path.normpath(mws.join(GlobalData.gGlobalDefines["EDK_TOOLS_PATH"], "BaseTools/Source/C/Common")))
         SearchPathList.extend(str(item) for item in IncSearchList)
         IncFileList = GetDependencyList(IncludeFileFullPaths, SearchPathList)
+        StructuredPcdsData["OBJECTS"] = {}
         for include_file in IncFileList:
+            StructuredPcdsData["OBJECTS"][include_file] = 
+ os.path.getmtime(include_file)
             MakeApp += "$(OBJECTS) : %s\n" % include_file
         if sys.platform == "win32":
             PcdValueCommonPath = os.path.normpath(mws.join(GlobalData.gGlobalDefines["EDK_TOOLS_PATH"], "Source\C\Common\PcdValueCommon.c"))
@@ -3042,17 +3126,18 @@ class DscBuildData(PlatformBuildClassObject):
             if returncode != 0:
                 EdkLogger.warn('Build', COMMAND_FAILURE, 'Can not collect output from command: %s\n%s\n%s\n' % (Command, StdOut, StdErr))
 
-        #start update structure pcd final value
-        File = open (OutputValueFile, 'r')
-        FileBuffer = File.readlines()
-        File.close()
+        #
+        # In 1st build create the StructuredPcdsData.json
+        # update the record as PCD Input has been changed if its incremental build
+        #
+        with open(StructuredPcdsDataPath, 'w') as file:
+            json.dump(StructuredPcdsData, file, indent=2)
 
-        StructurePcdSet = []
-        for Pcd in FileBuffer:
-            PcdValue = Pcd.split ('|')
-            PcdInfo = PcdValue[0].split ('.')
-            StructurePcdSet.append((PcdInfo[0], PcdInfo[1], PcdInfo[2], PcdInfo[3], PcdValue[2].strip()))
-        return StructurePcdSet
+        # Copy update output file for each Arch
+        shutil.copyfile(OutputValueFile, PcdRecordOutputValueFile)
+
+        #start update structure pcd final value
+        return self.GetStructurePcdSet(OutputValueFile)
 
     @staticmethod
     def NeedUpdateOutput(OutputFile, ValueCFile, StructureInput):
--
2.39.1.windows.1



-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#115295): https://edk2.groups.io/g/devel/message/115295
Mute This Topic: https://groups.io/mt/104222804/7686176
Group Owner: devel+owner@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [rebecca@openfw.io]
-=-=-=-=-=-=-=-=-=-=-=-