[96] in athena10
Corruption bug affecting debathenificator
daemon@ATHENA.MIT.EDU (ghudson@MIT.EDU)
Tue Feb 26 17:00:11 2008
Date: Tue, 26 Feb 2008 16:59:28 -0500 (EST)
From: ghudson@MIT.EDU
Message-Id: <200802262159.m1QLxSc7014383@outgoing.mit.edu>
To: athena10@mit.edu
MIME-version: 1.0
Content-type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit
Anders reports having previously seen bit differences between the orig
tarball from the cwd (in AFS) and in $tmpdir (on local disk) while
running debathenify-krb5. I have now reproduced this problem on
linux-build-10.mit.edu.
AFS is an obvious culprit, but not the only suspect. Since I have to
stop for the evening soon, I'm dumping a bit of data here. I've seen
several different results for successive runs, each with about equal
frequency:
1. Everything appears to work. There may be corruption but it
doesn't cause a build error.
2. gzip fails and the xxd|diff displays output.
3. The binary package build fails due to a syntax error. Some of
the file corruptions I've seen are:
* "_" replaced by "^" in src/kdc/kdc_preauth.c line 2176
* "lock" replacedby "lobk" in src/util/profile/prof_file.c line 523
In at least the _ to ^ case, the change was represented in the
diff, which may indicate that the corruption is not happening
during the copy to /tmp. (If it were happening during the copy
to /tmp then it wouldn't show up in the diff and the binary build
would have succeeded. I think.)
A common theme is that it's always an error in a single byte, always a
subtraction by 1, and (at least when xxd gives output) always in the
same place within a 1K block.
I plan to add some more instrumentation to the script to isolate where
the corruption is happening. If it is in AFS then Derrick is willing
to help if given useful tcpdump or fstrace output.
----- Case 2 failure output, run 1 -----
+ schroot -r -c gutsy-amd64-sbuild-cd39ec57-0a74-4782-b971-2ecfc708e6fc -- debuild -S -sa -us -uc -i -I.svn
I: [gutsy-amd64-sbuild-cd39ec57-0a74-4782-b971-2ecfc708e6fc chroot] Running command: “debuild -S -sa -us -uc -i -I.svn”
fakeroot debian/rules clean
dh_testdir
dh_testroot
rm -rf build
dh_clean build-stamp configure-stamp
dpkg-source -i -I.svn -b krb5-1.6.dfsg.1
dpkg-source: building krb5 using existing krb5_1.6.dfsg.1.orig.tar.gz
gzip: stdin: invalid compressed data--crc error
gzip: stdin: invalid compressed data--length error
tar: Skipping to next header
tar: Error exit delayed from previous errors
dpkg-source: failure: tar -xkf - gave error exit status 2
debuild: fatal error at line 1247:
dpkg-source -i -I.svn -b krb5-1.6.dfsg.1 failed
+ schroot -e -c gutsy-amd64-sbuild-cd39ec57-0a74-4782-b971-2ecfc708e6fc
--- /dev/fd/63 2008-02-26 16:20:28.780700157 -0500
+++ /dev/fd/62 2008-02-26 16:20:28.790699004 -0500
@@ -759175,7 +759175,7 @@
0b95860: 64da 0fc9 8891 5207 b00a c55d 477c 9e24 d.....R....]G|.$
0b95870: ecc2 eb97 2969 6698 040f 7d18 07b0 93f5 ....)if...}.....
0b95880: 4709 d7f4 1947 7976 a388 3d88 564a 63a4 G....Gyv..=.VJc.
-0b95890: e329 4e44 1c52 e849 35ca 84a6 d17e 7683 .)ND.R.I5....~v.
+0b95890: e329 4e44 1c52 e849 35ca 84a6 d07e 7683 .)ND.R.I5....~v.
0b958a0: 49e2 8426 76ca 17d4 2a10 65cd 8a77 e2b8 I..&v...*.e..w..
0b958b0: 54df a9a6 fe9b 0153 2a0a 4571 c41a c1aa T......S*.Eq....
0b958c0: 5aea 8e8e c719 e657 8e98 468a 1744 dad0 Z......W..F..D..
----- Case 2 failure output, run 2 (xxd only) -----
@@ -762567,7 +762567,7 @@
0ba2c60: 0623 ccc3 f600 4224 67f8 39b4 bd94 f603 .#....B$g.9.....
0ba2c70: a845 0cc3 0e72 860a 403e d0fc dfa1 5903 .E...r..@>....Y.
0ba2c80: ca2d 581a 584d 2b01 2724 fd01 f901 5e87 .-X.XM+.'$....^.
-0ba2c90: b41c f04c 4940 36a4 1ca6 3d48 a977 6344 ...LI@6...=H.wcD
+0ba2c90: b41c f04c 4940 36a4 1ca6 3d48 a877 6344 ...LI@6...=H.wcD
0ba2ca0: 714a 47cb 903b 37ec 0f63 2fa0 e701 1a39 qJG..;7..c/....9
0ba2cb0: 30f1 8031 f743 6a8c d826 aac2 6041 d0bb 0..1.Cj..&..`A..
0ba2cc0: 5b8c 520a 864c 9e90 0de5 dc71 507a 64bb [.R..L.....qPzd.
----- Case 2 failure output, run 3 (xxd only) -----
@@ -892999,7 +892999,7 @@
0da0460: 557f 1ebe ae7a 8adf 1256 e57d 2406 3ea5 U....z...V.}$.>.
0da0470: 4a5b e417 92c3 6727 3339 6e37 cd93 21ca J[....g'39n7..!.
0da0480: 0ac1 c28e 46f0 4093 f5d2 eb99 9ffa 49b9 ....F.@.......I.
-0da0490: 4c6e ad62 43fd 4c6f 8021 c19a 4da9 60bb Ln.bC.Lo.!..M.`.
+0da0490: 4c6e ad62 43fd 4c6f 8021 c19a 4ca9 60bb Ln.bC.Lo.!..L.`.
0da04a0: a04b 1f24 ebbb 9f26 176e 8529 a2c3 d750 .K.$...&.n.)...P
0da04b0: f977 bbfc 98bb bb4c d101 cb4f 8a9c dd7a .w.....L...O...z
0da04c0: 9c25 9c86 961b 70ea 02db c872 1613 74c5 .%....p....r..t.
----- Case 2 failure output, run 4 (xxd only) -----
@@ -471431,7 +471431,7 @@
0731860: 4d46 6ae3 0252 6864 e634 15eb 7ac1 f0ee MFj..Rhd.4..z...
0731870: 9f2b 8d82 33e3 138b 31a2 8da5 4594 5648 .+..3...1...E.VH
0731880: 74fd 00e0 bc48 fe09 d557 f501 70a8 7dfd t....H...W..p.}.
-0731890: ea8f 5010 b963 e2ec 7b84 8ef7 e851 fdfa ..P..c..{....Q..
+0731890: ea8f 5010 b963 e2ec 7b84 8ef7 e751 fdfa ..P..c..{....Q..
07318a0: 6031 670b cd54 fe01 20d6 f3fb c662 dfc3 `1g..T.. ....b..
07318b0: 7605 acd2 1be6 3fee 54ff e15b bc60 77fa v.....?.T..[.`w.
07318c0: 368e 99f9 60a0 a1a2 fbdf ef0d 4bca a201 6...`.......K...
