[SVN-4283] When the identical suffix starts at chunk boundary, "svn diff" produces incorrect diff

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: trunk
• Fix Version/s: 1.8.0
• Component/s: libsvn_diff
• Assignee: Unassigned
• Reporter: Subversion Importer
• Created: 29/Dec/12 12:06
• Updated: 27/Oct/18 16:30
• Resolved: 29/Dec/12 13:52

Description

The optimization of diff inclued in version 1.7 has a bug that
produces incorrect diff on a certain condition.
I will attach a patch fix it. 

Detail of the bug
-----------------

When the identical suffix begins at the boundary of a chunk,
datasource_get_next_token() defined in subversion/libsvn_diff/diff_file.c
does not stop at head of the identical suffix.
Therefore, when one of the identical suffixes of the original file
and the modified file begin from the boundary of a chunk,
excessive tokens are added to the diff tree.

How to reproduce
----------------

$ for ((i=0;i<8256;i++)); do echo 0123456789abcde; done > test.txt
$ hexdump -C test.txt
00000000  30 31 32 33 34 35 36 37  38 39 61 62 63 64 65 0a  |0123456789abcde.|
*
00020400
$ svn add test.txt; svn ci -m test
A         test.txt
Adding         test.txt
Transmitting file data .
Committed revision 2.
$ echo 0123456789ABCDE |dd of=test.txt bs=16 seek=64 conv=notrunc
1+0 records in
1+0 records out
$ echo 0123456789ABCDE |dd of=test.txt bs=16 seek=8141 conv=notrunc
1+0 records in
1+0 records out
$ echo 0123456789abcde >> test.txt
$ echo 0123456789abcde >> test.txt
$ hexdump -C test.txt
00000000  30 31 32 33 34 35 36 37  38 39 61 62 63 64 65 0a  |0123456789abcde.|
*
00000400  30 31 32 33 34 35 36 37  38 39 41 42 43 44 45 0a  |0123456789ABCDE.|
00000410  30 31 32 33 34 35 36 37  38 39 61 62 63 64 65 0a  |0123456789abcde.|
*
0001fcd0  30 31 32 33 34 35 36 37  38 39 41 42 43 44 45 0a  |0123456789ABCDE.|
0001fce0  30 31 32 33 34 35 36 37  38 39 61 62 63 64 65 0a  |0123456789abcde.|
*
00020420
$ svn cat test.txt | diff -u - test.txt
--- -   2012-12-24 22:30:18.760832000 +0900
+++ test.txt    2012-12-24 22:29:24.000000000 +0900
@@ -62,6 +62,7 @@
 0123456789abcde
 0123456789abcde
 0123456789abcde
+0123456789ABCDE
 0123456789abcde
 0123456789abcde
 0123456789abcde
@@ -8138,6 +8139,7 @@
 0123456789abcde
 0123456789abcde
 0123456789abcde
+0123456789ABCDE
 0123456789abcde
 0123456789abcde
 0123456789abcde
$ svn di test.txt
Index: test.txt
===================================================================
--- test.txt    (revision 2)
+++ test.txt    (working copy)
@@ -62,6 +62,7 @@
 0123456789abcde
 0123456789abcde
 0123456789abcde
+0123456789ABCDE
 0123456789abcde
 0123456789abcde
 0123456789abcde
@@ -8138,6 +8139,7 @@
 0123456789abcde
 0123456789abcde
 0123456789abcde
+0123456789ABCDE
 0123456789abcde
 0123456789abcde
 0123456789abcde
@@ -8188,6 +8190,72 @@
 0123456789abcde
 0123456789abcde
 0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
+0123456789abcde
 0123456789abcde
 0123456789abcde
 0123456789abcde

Original issue reported by hideki_iwamoto