From 98961815eda7c7ff44cfca3643d8b280f3873d47 Mon Sep 17 00:00:00 2001
From: Zac Medico <zmedico@gentoo.org>
Date: Tue, 31 Mar 2009 20:14:03 +0000
Subject: Add a new egencache --rsync option which enables a stat collision
 workaround for cases in which the content of a cache entry changes and
 neither the file mtime nor size changes (preventing rsync from detecting
 changes). See bug #139134. This option should only be needed for distribution
 via something like rsync, which relies on timestamps and file sizes to detect
 changes. It's not needed with git since that uses a more thorough mechanism
 which allows it to detect changed inode numbers (described in racy-git.txt in
 the git technical docs).

svn path=/main/trunk/; revision=13262
---
 pym/portage/cache/cache_errors.py | 22 ++++++++++++++++++
 pym/portage/cache/metadata.py     | 49 ++++++++++++++++++++++++++++++++-------
 2 files changed, 63 insertions(+), 8 deletions(-)

(limited to 'pym')

diff --git a/pym/portage/cache/cache_errors.py b/pym/portage/cache/cache_errors.py
index f63e5994b..e1e8eead0 100644
--- a/pym/portage/cache/cache_errors.py
+++ b/pym/portage/cache/cache_errors.py
@@ -39,3 +39,25 @@ class ReadOnlyRestriction(CacheError):
 		self.info = info
 	def __str__(self):
 		return "cache is non-modifiable"+str(self.info)
+
+class StatCollision(CacheError):
+	"""
+	If the content of a cache entry changes and neither the file mtime nor
+	size changes, it will prevent rsync from detecting changes. Cache backends
+	may raise this exception from _setitem() if they detect this type of stat
+	collision. See bug #139134.
+	"""
+	def __init__(self, key, filename, mtime, size):
+		self.key = key
+		self.filename = filename
+		self.mtime = mtime
+		self.size = size
+
+	def __str__(self):
+		return "%s has stat collision with size %s and mtime %s" % \
+			(self.key, self.size, self.mtime)
+
+	def __repr__(self):
+		return "portage.cache.cache_errors.StatCollision(%s)" % \
+			(', '.join((repr(self.key), repr(self.filename),
+			repr(self.mtime), repr(self.size))),)
diff --git a/pym/portage/cache/metadata.py b/pym/portage/cache/metadata.py
index 5222223c6..a8be01095 100644
--- a/pym/portage/cache/metadata.py
+++ b/pym/portage/cache/metadata.py
@@ -3,7 +3,7 @@
 # License: GPL2
 # $Id$
 
-import errno, os, re
+import errno, os, re, sys
 from portage.cache import cache_errors, flat_hash
 import portage.eclass_cache
 from portage.cache.template import reconstruct_eclasses
@@ -30,6 +30,7 @@ class database(flat_hash.database):
 		super(database, self).__init__(location, *args, **config)
 		self.location = os.path.join(loc, "metadata","cache")
 		self.ec = portage.eclass_cache.cache(loc)
+		self.raise_stat_collision = False
 
 	def _parse_data(self, data, cpv):
 		_hashed_re_match = self._hashed_re.match
@@ -73,31 +74,63 @@ class database(flat_hash.database):
 			values = ProtectedDict(values)
 			values["INHERITED"] = ' '.join(sorted(values["_eclasses_"]))
 
+		new_content = []
+		for k in self.auxdbkey_order:
+			new_content.append(unicode(values.get(k, ''), errors='replace'))
+			new_content.append(u'\n')
+		for i in xrange(magic_line_count - len(self.auxdbkey_order)):
+			new_content.append(u'\n')
+		new_content = u''.join(new_content)
+		new_content = new_content.encode(
+			sys.getdefaultencoding(), 'backslashreplace')
+
+		new_fp = os.path.join(self.location, cpv)
+		try:
+			f = open(new_fp, 'rb')
+		except EnvironmentError:
+			pass
+		else:
+			try:
+				try:
+					existing_st = os.fstat(f.fileno())
+					existing_content = f.read()
+				finally:
+					f.close()
+			except EnvironmentError:
+				pass
+			else:
+				existing_mtime = long(existing_st.st_mtime)
+				if values['_mtime_'] == existing_mtime and \
+					existing_content == new_content:
+					return
+
+				if self.raise_stat_collision and \
+					values['_mtime_'] == existing_mtime and \
+					len(new_content) == existing_st.st_size:
+					raise cache_errors.StatCollision(cpv, new_fp,
+						existing_mtime, existing_st.st_size)
+
 		s = cpv.rfind("/")
 		fp = os.path.join(self.location,cpv[:s],
 			".update.%i.%s" % (os.getpid(), cpv[s+1:]))
 		try:
-			myf = open(fp, "w")
+			myf = open(fp, 'wb')
 		except EnvironmentError, e:
 			if errno.ENOENT == e.errno:
 				try:
 					self._ensure_dirs(cpv)
-					myf = open(fp, "w")
+					myf = open(fp, 'wb')
 				except EnvironmentError, e:
 					raise cache_errors.CacheCorruption(cpv, e)
 			else:
 				raise cache_errors.CacheCorruption(cpv, e)
 
 		try:
-			for k in self.auxdbkey_order:
-				myf.write(values.get(k, "") + "\n")
-			for i in xrange(magic_line_count - len(self.auxdbkey_order)):
-				myf.write("\n")
+			myf.write(new_content)
 		finally:
 			myf.close()
 		self._ensure_access(fp, mtime=values["_mtime_"])
 
-		new_fp = os.path.join(self.location, cpv)
 		try:
 			os.rename(fp, new_fp)
 		except EnvironmentError, e:
-- 
cgit v1.2.3-1-g7c22