misc/link_maps{,_portage}: compare reading /var/db/pkg directly and via portage

2 files changed, 199 insertions, 45 deletions

diff --git a/misc/link_maps b/misc/link_maps
index a6adacd..3e4e3cb 100755
--- a/misc/link_maps
+++ b/misc/link_maps
@@ -168,57 +168,27 @@ def main():
 
 	object_reverse_linkings = get_object_reverse_linkings( object_linkings )
 
-	""" Print out all ELF objects and their PaX flags
-	for elf in object_needed:
-		try:
-			flags = pax.getflags(elf)[0]
-			if flags:
-				print("%s %s" % (flags, elf))
-			else:
-				print("NONE: %s" % elf)
-		except pax.error:
-			print("CANT: %s" % elf)
-
-	"""
-
-	""" Print out all sonames and their library paths
-	for soname in sorted(soname2library):
-		elf = soname2library[soname]
-		print("%s : %s" % ( soname, elf ))
-	"""
-
-
-	""" Print out all ELF objects and the NEEDED sonames and full library paths
-	for elf in object_needed:
-		sonames = object_needed[elf]
+	""" Print out all ELF objects and the NEEDED sonames and full library paths """
+	for elf in object_linkings:
+		sonames = object_linkings[elf]
 		print("%s" % elf)
-		for soname in sorted(object_needed[elf]):
+		for soname in sorted(object_linkings[elf]):
 			try:
 				print("\t%s\t=> %s" % (soname, soname2library[soname]))
 			except KeyError:
-				print("\t%s\t=> ****" % soname)
+				print("%s\t=>%s " % (soname, '***'))
+		print("\n\n")
+
+	""" Print out all ELF objects and the NEEDED sonames and full library paths """
+	for soname in object_reverse_linkings:
+		try:
+			print("%s\t=> %s" % (soname, soname2library[soname]))
+		except KeyError:
+			print("%s\t=>%s " % (soname, '***'))
+		for elf in sorted(object_reverse_linkings[soname]):
+			print("\t%s" % elf)
 		print("\n\n")
-	"""
 
-	""" Print out all the soname to soname NEEDED
-	for soname in soname_needed:
-		print("%s" % soname)
-		for s in soname_needed[soname]:
-			print("\t%s" % s )
-		print('')
-	"""
-
-
-	""" Print out all the soname to soname linkings
-	for soname in soname_linkings:
-		print("%s => %s" % (soname, soname2library[soname]))
-		for s in soname_linkings[soname]:
-			if s in soname2library:
-				print("\t%s => %s" % (s, soname2library[s]))
-			else:
-				print("\t%s => ****" %s )
-		print('')
-	"""
 
 if __name__ == '__main__':
 	main()
diff --git a/misc/link_maps_portage b/misc/link_maps_portage
new file mode 100755
index 0000000..bf5b447
--- /dev/null
+++ b/misc/link_maps_portage
@@ -0,0 +1,184 @@
+#!/usr/bin/env python
+
+#
+# Note: This alternative way of doing revdep-pax only
+# works on Gentoo systems where NEEDED.ELF.2 all the
+# information we need generated by scanelf during emerge.
+#
+# See /usr/lib/portage/bin/misc-functions.sh ~line 520
+# echo "${arch:3};${obj};${soname};${rpath};${needed}" >> "${PORTAGE_BUILDDIR}"/build-info/NEEDED.ELF.2
+#
+
+import os
+import sys
+import re
+import pax
+import portage
+
+
+"""
+Return object_needed dictionary which has structure
+
+	{ full_path_to_ELF_object : [ soname1, soname2, ... ], ... }
+
+Here the sonames were obtained from the ELF object by scanelf -nm
+(like readelf -d) during emerge.
+"""
+def get_object_needed():
+
+	vardb = portage.db[portage.root]["vartree"].dbapi
+
+	object_needed = {}
+
+	for pkg in vardb.cpv_all():
+		needs = vardb.aux_get(pkg, ['NEEDED.ELF.2'])[0].strip()
+		if not needs:			#skip empty lines
+			continue
+		lines = re.split('\n', needs)
+		for line in lines:
+			link = re.split(';', line)
+			elf = link[1]
+			sonames = re.split(',', link[4])
+			object_needed[elf] = sonames
+
+	return object_needed
+
+
+"""
+Return library2soname dictionary which has structure
+
+	{ full_path_to_library : soname, ... }
+
+and its inverse which has structure
+
+	{ soname : full_path_to_library, ... }
+"""
+def get_libraries():
+
+	vardb = portage.db[portage.root]["vartree"].dbapi
+
+	library2soname = {}
+	soname2library = {}
+
+	for pkg in vardb.cpv_all():
+		needs = vardb.aux_get(pkg, ['NEEDED.ELF.2'])[0].strip()
+		if not needs:			#skip empty lines
+			continue
+		lines = re.split('\n', needs)
+		for line in lines:
+			link = re.split(';', line)
+			elf = link[1]
+			soname = link[2]
+			if soname:				#no soname => executable
+				library2soname[elf] = soname
+				soname2library[soname] = elf
+
+	return ( library2soname, soname2library )
+
+
+"""
+Return get_soname_needed dictionary which has structure:
+
+	{ soname : [ soname1, soname2, ... ], .... }
+
+Here the soname1, soname2,... were obtained from soname's corresponding
+ELF object by scanelf -n during emerge.
+"""
+def get_soname_needed( object_needed, library2soname ):
+
+	soname_needed = {}
+
+	for elf in object_needed:
+		try:
+			soname = library2soname[elf]
+			soname_needed[soname] = object_needed[elf]
+			#soname_needed[soname] = copy(object_needed[elf]) #copy the list
+		except KeyError:
+			continue  # no soname, its probably an executable
+
+	return soname_needed
+
+
+def get_soname_linkings( soname_needed, soname2library ):
+	for soname in soname_needed:
+		while True:
+			count = 0
+			for s in soname_needed[soname]:
+				try:
+					for sf in soname_needed[s]:
+						if sf in soname_needed[soname]:		# Skip it if we already included it
+							continue
+						if not sf in soname2library:		#Skip if its a vdso
+							continue
+						# This appends to the object_needed and soname_needed list.
+						#  No copy was done so its the same list in memory for both.
+						soname_needed[soname].append(sf)
+						count = 1
+				except KeyError:
+					continue
+
+			if count == 0:
+				break
+	return
+
+
+def get_object_reverse_linkings( object_linkings ):
+	object_reverse_linkings = {}
+
+	for elf in object_linkings:
+		for soname in object_linkings[elf]:
+			object_reverse_linkings.setdefault(soname,[]).append(elf)
+
+	return object_reverse_linkings
+
+
+def main():
+
+	# Run as root to be able to real all files
+	uid = os.getuid()
+	if uid != 0:
+		print('RUN AS ROOT: cannot read all flags')
+		sys.exit(0)
+
+	object_needed = get_object_needed()
+	( library2soname, soname2library ) = get_libraries()
+	soname_needed = get_soname_needed( object_needed, library2soname )
+
+	# After the appending to needed in get_soname_linkings(), forward_needed
+	# and soname_needed have been extended through the entire chain of linking.
+	# If we want to keep only the object_needed and soname_needed, then do
+	# a copy before calling get_soname_linkings().
+	get_soname_linkings( soname_needed, soname2library )
+
+	object_linkings = object_needed
+	object_needed = None
+
+	soname_linkings = soname_needed
+	soname_needed = None
+
+	object_reverse_linkings = get_object_reverse_linkings( object_linkings )
+
+	""" Print out all ELF objects and the NEEDED sonames and full library paths """
+	for elf in object_linkings:
+		sonames = object_linkings[elf]
+		print("%s" % elf)
+		for soname in sorted(object_linkings[elf]):
+			try:
+				print("\t%s\t=> %s" % (soname, soname2library[soname]))
+			except KeyError:
+				print("%s\t=>%s " % (soname, '***'))
+		print("\n\n")
+
+	""" Print out all ELF objects and the NEEDED sonames and full library paths """
+	for soname in object_reverse_linkings:
+		try:
+			print("%s\t=> %s" % (soname, soname2library[soname]))
+		except KeyError:
+			print("%s\t=>%s " % (soname, '***'))
+		for elf in sorted(object_reverse_linkings[soname]):
+			print("\t%s" % elf)
+		print("\n\n")
+
+
+if __name__ == '__main__':
+	main()