JDK-8146119 : java/lang/Math/PowTests.java fails on solaris-x64 using -Xcomp
  • Type: Bug
  • Component: hotspot
  • Sub-Component: compiler
  • Affected Version: 9
  • Priority: P2
  • Status: Closed
  • Resolution: Cannot Reproduce
  • OS: solaris
  • CPU: x86_64
  • Submitted: 2015-12-23
  • Updated: 2016-01-21
  • Resolved: 2016-01-20
The Version table provides details related to the release that this issue/RFE will be addressed.

Unresolved : Release in which this issue/RFE will be addressed.
Resolved: Release in which this issue/RFE has been resolved.
Fixed : Release in which this issue/RFE has been fixed. The release containing this fix may be available for download as an Early Access Release or a General Availability Release.

To download the current JDK release, click here.
JDK 9
9Resolved
Related Reports
Relates :  
JDK-8145688 - update for x86 pow in the math lib
Description
----------System.err:(1607/75091)----------
Failure for Math.pow(double, double):
	For inputs -Infinity	(-Infinity) and 4.9E-324	(0x0.0000000000001p-1022)
	expected  Infinity	(Infinity)
	got       0.0	(0x0.0p0).
Failure for Math.pow(double, double):
	For inputs -Infinity	(-Infinity) and 0.3333333333333333	(0x1.5555555555555p-2)
	expected  Infinity	(Infinity)
	got       0.0	(0x0.0p0).
...
Comments
Closing this bug because it is not reproduced after JDK-8145688 changes.
20-01-2016
Official webrev: http://cr.openjdk.java.net/~kvn/8146119/webrev/ Note, we are not fixing this problem because Hotspot SQE does not approved switch to SS12.4 - they did not do testing. We should rollback changes in JPRT instead.
06-01-2016
Does anyone know if this failure occurred during the testing of JDK9-B98? In theory, JDK9-B97 and JDK9-B98 are the same code built with the two different compilers. However, the JDK9-B98 promotion e-mail lists 58 issues resolved in that build: ------------------ Issues Summary ------------------ Current build: jdk-98 Integration summary |- Issues integrated (hg log ) : 58 Test summary |- Integrations without a test or keyword : 0 |- Integrations with a noreg keyward : 44 |- Integrations with a test : 14 while the JDK9-B97 promotion e-mail lists 457 issues resolved: ------------------ Issues Summary ------------------ Current build: jdk-97 Integration summary |- Issues integrated (hg log ) : 457 Test summary |- Integrations without a test or keyword : 0 |- Integrations with a noreg keyward : 385 |- Integrations with a test : 72 I am less than convinced that JDK9-B98 was a properly done compiler switch only build.
06-01-2016
Product build is also affected.
23-12-2015
make/solaris/makefiles/fastdebug.make ifeq ("${Platform_compiler}", "sparcWorks") OPT_CFLAGS/SLOWER = -xO2 +ifeq ($(shell expr $(COMPILER_REV_NUMERIC) \>= 513), 1) +# trans intrinsics failure +OPT_CFLAGS/library_call.o = -c +endif # COMPILER_NUMERIC_REV >= 513 + ifeq ($(COMPILER_REV_NUMERIC), 510) # Avoid apparent crash because of corrupted methodHandle in a tail call OPT_CFLAGS/simpleThresholdPolicy.o = $(OPT_CFLAGS/DEFAULT) $(OPT_CCFLAGS/NO_TAIL_CALL_OPT)
23-12-2015