1 |
commit: 0b7eda500a0dcb98a67f33bf9ef25b202b358986 |
2 |
Author: Arfrever Frehtes Taifersar Arahesis <Arfrever <AT> Apache <DOT> Org> |
3 |
AuthorDate: Wed Aug 7 17:06:11 2019 +0000 |
4 |
Commit: Zac Medico <zmedico <AT> gentoo <DOT> org> |
5 |
CommitDate: Sun Aug 11 19:11:47 2019 +0000 |
6 |
URL: https://gitweb.gentoo.org/proj/portage.git/commit/?id=0b7eda50 |
7 |
|
8 |
dblink._collision_protect: Detect internal collisions. |
9 |
|
10 |
Implement detection of internal collisions (between files of the same package, |
11 |
located in separate directories in the installation image (${D}) corresponding |
12 |
to merged directories in the target filesystem (${ROOT})). |
13 |
|
14 |
This provides protection against overwriting some files when performing merging |
15 |
of files from ${D} to ${ROOT} in some filesystem layouts (such as /-merged layout |
16 |
or /usr-merged layout). |
17 |
|
18 |
Internal collisions between identical files are silently ignored. |
19 |
|
20 |
Bug: https://bugs.gentoo.org/690484 |
21 |
Signed-off-by: Arfrever Frehtes Taifersar Arahesis <Arfrever <AT> Apache.Org> |
22 |
Signed-off-by: Zac Medico <zmedico <AT> gentoo.org> |
23 |
|
24 |
lib/portage/dbapi/vartree.py | 55 ++++++++++++++++++-- |
25 |
lib/portage/util/_compare_files.py | 103 +++++++++++++++++++++++++++++++++++++ |
26 |
2 files changed, 154 insertions(+), 4 deletions(-) |
27 |
|
28 |
diff --git a/lib/portage/dbapi/vartree.py b/lib/portage/dbapi/vartree.py |
29 |
index e2fce7736..4f069474b 100644 |
30 |
--- a/lib/portage/dbapi/vartree.py |
31 |
+++ b/lib/portage/dbapi/vartree.py |
32 |
@@ -30,6 +30,7 @@ portage.proxy.lazyimport.lazyimport(globals(), |
33 |
'portage.util:apply_secpass_permissions,ConfigProtect,ensure_dirs,' + \ |
34 |
'writemsg,writemsg_level,write_atomic,atomic_ofstream,writedict,' + \ |
35 |
'grabdict,normalize_path,new_protect_filename', |
36 |
+ 'portage.util._compare_files:compare_files', |
37 |
'portage.util.digraph:digraph', |
38 |
'portage.util.env_update:env_update', |
39 |
'portage.util.install_mask:install_mask_dir,InstallMask', |
40 |
@@ -87,6 +88,7 @@ import io |
41 |
from itertools import chain |
42 |
import logging |
43 |
import os as _os |
44 |
+import operator |
45 |
import platform |
46 |
import pwd |
47 |
import re |
48 |
@@ -3418,6 +3420,8 @@ class dblink(object): |
49 |
|
50 |
os = _os_merge |
51 |
|
52 |
+ real_relative_paths = {} |
53 |
+ |
54 |
collision_ignore = [] |
55 |
for x in portage.util.shlex_split( |
56 |
self.settings.get("COLLISION_IGNORE", "")): |
57 |
@@ -3469,8 +3473,13 @@ class dblink(object): |
58 |
previous = current |
59 |
progress_shown = True |
60 |
|
61 |
- dest_path = normalize_path( |
62 |
- os.path.join(destroot, f.lstrip(os.path.sep))) |
63 |
+ dest_path = normalize_path(os.path.join(destroot, f.lstrip(os.path.sep))) |
64 |
+ |
65 |
+ # Relative path with symbolic links resolved only in parent directories |
66 |
+ real_relative_path = os.path.join(os.path.realpath(os.path.dirname(dest_path)), |
67 |
+ os.path.basename(dest_path))[len(destroot):] |
68 |
+ |
69 |
+ real_relative_paths.setdefault(real_relative_path, []).append(f.lstrip(os.path.sep)) |
70 |
|
71 |
parent = os.path.dirname(dest_path) |
72 |
if parent not in dirs: |
73 |
@@ -3556,9 +3565,24 @@ class dblink(object): |
74 |
break |
75 |
if stopmerge: |
76 |
collisions.append(f) |
77 |
+ |
78 |
+ internal_collisions = {} |
79 |
+ for real_relative_path, files in real_relative_paths.items(): |
80 |
+ # Detect internal collisions between non-identical files. |
81 |
+ if len(files) >= 2: |
82 |
+ files.sort() |
83 |
+ for i in range(len(files) - 1): |
84 |
+ file1 = normalize_path(os.path.join(srcroot, files[i])) |
85 |
+ file2 = normalize_path(os.path.join(srcroot, files[i+1])) |
86 |
+ # Compare files, ignoring differences in times. |
87 |
+ differences = compare_files(file1, file2, skipped_types=("atime", "mtime", "ctime")) |
88 |
+ if differences: |
89 |
+ internal_collisions.setdefault(real_relative_path, {})[(files[i], files[i+1])] = differences |
90 |
+ |
91 |
if progress_shown: |
92 |
showMessage(_("100% done\n")) |
93 |
- return collisions, dirs_ro, symlink_collisions, plib_collisions |
94 |
+ |
95 |
+ return collisions, internal_collisions, dirs_ro, symlink_collisions, plib_collisions |
96 |
|
97 |
def _lstat_inode_map(self, path_iter): |
98 |
""" |
99 |
@@ -4081,7 +4105,7 @@ class dblink(object): |
100 |
if blocker.exists(): |
101 |
blockers.append(blocker) |
102 |
|
103 |
- collisions, dirs_ro, symlink_collisions, plib_collisions = \ |
104 |
+ collisions, internal_collisions, dirs_ro, symlink_collisions, plib_collisions = \ |
105 |
self._collision_protect(srcroot, destroot, |
106 |
others_in_slot + blockers, filelist, linklist) |
107 |
|
108 |
@@ -4109,6 +4133,29 @@ class dblink(object): |
109 |
eerror(msg) |
110 |
return 1 |
111 |
|
112 |
+ if internal_collisions: |
113 |
+ msg = _("Package '%s' has internal collisions between non-identical files " |
114 |
+ "(located in separate directories in the installation image (${D}) " |
115 |
+ "corresponding to merged directories in the target " |
116 |
+ "filesystem (${ROOT})):") % self.settings.mycpv |
117 |
+ msg = textwrap.wrap(msg, 70) |
118 |
+ msg.append("") |
119 |
+ for k, v in sorted(internal_collisions.items(), key=operator.itemgetter(0)): |
120 |
+ msg.append("\t%s" % os.path.join(destroot, k.lstrip(os.path.sep))) |
121 |
+ for (file1, file2), differences in sorted(v.items()): |
122 |
+ msg.append("\t\t%s" % os.path.join(destroot, file1.lstrip(os.path.sep))) |
123 |
+ msg.append("\t\t%s" % os.path.join(destroot, file2.lstrip(os.path.sep))) |
124 |
+ msg.append("\t\t\tDifferences: %s" % ", ".join(differences)) |
125 |
+ msg.append("") |
126 |
+ self._elog("eerror", "preinst", msg) |
127 |
+ |
128 |
+ msg = _("Package '%s' NOT merged due to internal collisions " |
129 |
+ "between non-identical files.") % self.settings.mycpv |
130 |
+ msg += _(" If necessary, refer to your elog messages for the whole " |
131 |
+ "content of the above message.") |
132 |
+ eerror(textwrap.wrap(msg, 70)) |
133 |
+ return 1 |
134 |
+ |
135 |
if symlink_collisions: |
136 |
# Symlink collisions need to be distinguished from other types |
137 |
# of collisions, in order to avoid confusion (see bug #409359). |
138 |
|
139 |
diff --git a/lib/portage/util/_compare_files.py b/lib/portage/util/_compare_files.py |
140 |
new file mode 100644 |
141 |
index 000000000..bd993e501 |
142 |
--- /dev/null |
143 |
+++ b/lib/portage/util/_compare_files.py |
144 |
@@ -0,0 +1,103 @@ |
145 |
+# Copyright 2019 Gentoo Authors |
146 |
+# Distributed under the terms of the GNU General Public License v2 |
147 |
+ |
148 |
+__all__ = ["compare_files"] |
149 |
+ |
150 |
+import io |
151 |
+import os |
152 |
+import stat |
153 |
+import sys |
154 |
+ |
155 |
+from portage import _encodings |
156 |
+from portage import _unicode_encode |
157 |
+from portage.util._xattr import xattr |
158 |
+ |
159 |
+def compare_files(file1, file2, skipped_types=()): |
160 |
+ """ |
161 |
+ Compare metadata and contents of two files. |
162 |
+ |
163 |
+ @param file1: File 1 |
164 |
+ @type file1: str |
165 |
+ @param file2: File 2 |
166 |
+ @type file2: str |
167 |
+ @param skipped_types: Tuple of strings specifying types of properties excluded from comparison. |
168 |
+ Supported strings: type, mode, owner, group, device_number, xattr, atime, mtime, ctime, size, content |
169 |
+ @type skipped_types: tuple of str |
170 |
+ @rtype: tuple of str |
171 |
+ @return: Tuple of strings specifying types of properties different between compared files |
172 |
+ """ |
173 |
+ |
174 |
+ file1_stat = os.lstat(_unicode_encode(file1, encoding=_encodings["fs"], errors="strict")) |
175 |
+ file2_stat = os.lstat(_unicode_encode(file2, encoding=_encodings["fs"], errors="strict")) |
176 |
+ |
177 |
+ differences = [] |
178 |
+ |
179 |
+ if (file1_stat.st_dev, file1_stat.st_ino) == (file2_stat.st_dev, file2_stat.st_ino): |
180 |
+ return () |
181 |
+ |
182 |
+ if "type" not in skipped_types and stat.S_IFMT(file1_stat.st_mode) != stat.S_IFMT(file2_stat.st_mode): |
183 |
+ differences.append("type") |
184 |
+ if "mode" not in skipped_types and stat.S_IMODE(file1_stat.st_mode) != stat.S_IMODE(file2_stat.st_mode): |
185 |
+ differences.append("mode") |
186 |
+ if "owner" not in skipped_types and file1_stat.st_uid != file2_stat.st_uid: |
187 |
+ differences.append("owner") |
188 |
+ if "group" not in skipped_types and file1_stat.st_gid != file2_stat.st_gid: |
189 |
+ differences.append("group") |
190 |
+ if "device_number" not in skipped_types and file1_stat.st_rdev != file2_stat.st_rdev: |
191 |
+ differences.append("device_number") |
192 |
+ |
193 |
+ if "xattr" not in skipped_types and sorted(xattr.get_all(file1, nofollow=True)) != sorted(xattr.get_all(file2, nofollow=True)): |
194 |
+ differences.append("xattr") |
195 |
+ |
196 |
+ if sys.hexversion >= 0x3030000: |
197 |
+ if "atime" not in skipped_types and file1_stat.st_atime_ns != file2_stat.st_atime_ns: |
198 |
+ differences.append("atime") |
199 |
+ if "mtime" not in skipped_types and file1_stat.st_mtime_ns != file2_stat.st_mtime_ns: |
200 |
+ differences.append("mtime") |
201 |
+ if "ctime" not in skipped_types and file1_stat.st_ctime_ns != file2_stat.st_ctime_ns: |
202 |
+ differences.append("ctime") |
203 |
+ else: |
204 |
+ if "atime" not in skipped_types and file1_stat.st_atime != file2_stat.st_atime: |
205 |
+ differences.append("atime") |
206 |
+ if "mtime" not in skipped_types and file1_stat.st_mtime != file2_stat.st_mtime: |
207 |
+ differences.append("mtime") |
208 |
+ if "ctime" not in skipped_types and file1_stat.st_ctime != file2_stat.st_ctime: |
209 |
+ differences.append("ctime") |
210 |
+ |
211 |
+ if "type" in differences: |
212 |
+ pass |
213 |
+ elif file1_stat.st_size != file2_stat.st_size: |
214 |
+ if "size" not in skipped_types: |
215 |
+ differences.append("size") |
216 |
+ if "content" not in skipped_types: |
217 |
+ differences.append("content") |
218 |
+ else: |
219 |
+ if "content" not in skipped_types: |
220 |
+ if stat.S_ISLNK(file1_stat.st_mode): |
221 |
+ file1_stream = io.BytesIO(os.readlink(_unicode_encode(file1, |
222 |
+ encoding=_encodings["fs"], |
223 |
+ errors="strict"))) |
224 |
+ else: |
225 |
+ file1_stream = open(_unicode_encode(file1, |
226 |
+ encoding=_encodings["fs"], |
227 |
+ errors="strict"), "rb") |
228 |
+ if stat.S_ISLNK(file2_stat.st_mode): |
229 |
+ file2_stream = io.BytesIO(os.readlink(_unicode_encode(file2, |
230 |
+ encoding=_encodings["fs"], |
231 |
+ errors="strict"))) |
232 |
+ else: |
233 |
+ file2_stream = open(_unicode_encode(file2, |
234 |
+ encoding=_encodings["fs"], |
235 |
+ errors="strict"), "rb") |
236 |
+ while True: |
237 |
+ file1_content = file1_stream.read(4096) |
238 |
+ file2_content = file2_stream.read(4096) |
239 |
+ if file1_content != file2_content: |
240 |
+ differences.append("content") |
241 |
+ break |
242 |
+ if not file1_content or not file2_content: |
243 |
+ break |
244 |
+ file1_stream.close() |
245 |
+ file2_stream.close() |
246 |
+ |
247 |
+ return tuple(differences) |