Alan,
Can I suggest you do a full build of JDK on JPRT with
SKIP_BUILD_CYCLE=false to test this fix?
-- Jon
Joshua Bloch wrote:
My review of this patch: looks good to me.
Josh
On Thu, Oct 29, 2009 at 4:02 PM, Alan Bateman <[email protected]
<mailto:[email protected]>> wrote:
Jonathan Gibbons wrote:
:
Alan,
My hudson falls over with a stack overflow at
DualPivotQuicksort.java:477 when doing a full build
(SKIP_BOOT_CYCLE=false)
at
java.util.DualPivotQuicksort.dualPivotQuicksort(DualPivotQuicksort.java:477)
at
java.util.DualPivotQuicksort.dualPivotQuicksort(DualPivotQuicksort.java:477)
at
java.util.DualPivotQuicksort.dualPivotQuicksort(DualPivotQuicksort.java:477)
at
java.util.DualPivotQuicksort.dualPivotQuicksort(DualPivotQuicksort.java:477)
at
java.util.DualPivotQuicksort.dualPivotQuicksort(DualPivotQuicksort.java:477)
at
java.util.DualPivotQuicksort.dualPivotQuicksort(DualPivotQuicksort.java:477)
at
java.util.DualPivotQuicksort.dualPivotQuicksort(DualPivotQuicksort.java:477)
gnumake[4]: ***
[/x/hudson/jobs/jdk7.tl.langtools-jdk/workspace/tl/build/solaris-sparc/classes/sun/text/resources/CharacterBreakIteratorData]
Error 1
gnumake[4]: Leaving directory
`/x/hudson/jobs/jdk7.tl.langtools-jdk/workspace/tl/jdk/make/java/text'
gnumake[3]: *** [all] Error 1
gnumake[3]: Leaving directory
`/x/hudson/jobs/jdk7.tl.langtools-jdk/workspace/tl/jdk/make/java'
gnumake[2]: *** [all] Error 1
gnumake[2]: Leaving directory
`/x/hudson/jobs/jdk7.tl.langtools-jdk/workspace/tl/jdk/make'
gnumake[1]: *** [jdk-build] Error 2
gnumake[1]: Leaving directory
`/x/hudson/jobs/jdk7.tl.langtools-jdk/workspace/tl'
gnumake: *** [build_product_image] Error 2
Sending e-mails to: [email protected]
<mailto:[email protected]>
finished: FAILURE
Full log for Sun folk at:
http://javac.sfbay:8080/hudson/job/jdk7.tl.langtools-jdk/80/console
-- Jon
Sorry about that - Vladimir also found this today and I've just
created 6896573 to track it. The fix is attached so we'll get that
in before TL integrates. The lesson here is that we should have
included new tests with this (Vladimir has been developing tests
to integrate - we just need to dial them down so that they run in
a reasonable time).
-Alan.
diff -r b05abb410c52
src/share/classes/java/util/DualPivotQuicksort.java
--- a/src/share/classes/java/util/DualPivotQuicksort.java
Thu Oct 29 11:18:37 2009 +0000
+++ b/src/share/classes/java/util/DualPivotQuicksort.java
Thu Oct 29 22:49:11 2009 +0000
@@ -36,7 +36,7 @@ package java.util;
* @author Jon Bentley
* @author Josh Bloch
*
- * @version 2009.10.22 m765.827.v4
+ * @version 2009.10.29 m765.827.v5
*/
final class DualPivotQuicksort {
@@ -473,9 +473,10 @@ final class DualPivotQuicksort {
a[great--] = pivot2;
}
}
- } else { // Use Dual-Pivot Quicksort on large arrays
- dualPivotQuicksort(a, left, right);
- }
+ }
+
+ // Sort center part recursively, excluding known pivot values
+ sort(a, less, great);
}
/** The number of distinct short values */
@@ -507,7 +508,7 @@ final class DualPivotQuicksort {
for (int i = left; i <= right; i++) {
count[a[i] - Short.MIN_VALUE]++;
}
- for (int i = 0, k = left; i < count.length && k <
right; i++) {
+ for (int i = 0, k = left; i < count.length && k <=
right; i++) {
short value = (short) (i + Short.MIN_VALUE);
for (int s = count[i]; s > 0; s--) {
@@ -723,13 +724,13 @@ final class DualPivotQuicksort {
a[j + 1] = ak;
}
} else if (right - left + 1 >
COUNTING_SORT_THRESHOLD_FOR_BYTE) {
- // Use counting sort on large arrays
+ // Use counting sort on huge arrays
int[] count = new int[NUM_BYTE_VALUES];
for (int i = left; i <= right; i++) {
count[a[i] - Byte.MIN_VALUE]++;
}
- for (int i = 0, k = left; i < count.length && k <
right; i++) {
+ for (int i = 0, k = left; i < count.length && k <=
right; i++) {
byte value = (byte) (i + Byte.MIN_VALUE);
for (int s = count[i]; s > 0; s--) {
@@ -951,7 +952,7 @@ final class DualPivotQuicksort {
for (int i = left; i <= right; i++) {
count[a[i]]++;
}
- for (int i = 0, k = left; i < count.length && k <
right; i++) {
+ for (int i = 0, k = left; i < count.length && k <=
right; i++) {
for (int s = count[i]; s > 0; s--) {
a[k++] = (char) i;
}
