On Wed, Feb 10, 2010 at 07:31:49PM -0600, Dirk Eddelbuettel wrote: > > Hi Dann,
hey Dirk! > > On 10 February 2010 at 11:12, dann frazier wrote: > | Source: quantlib > | Version: 1.0.0~20100112-1 > | Severity: serious > | > | >From a recent build attempt on hppa: > | [...] > | ====================== > | Testing QuantLib 1.0b1 > | ====================== > | *** glibc detected *** > /build/buildd-quantlib_1.0.0~20100112-1-hppa-H12qa6/quantlib-1.0.0~20100112/test-suite/.libs/lt-quantlib-test-suite: > free(): invalid pointer: 0x0093e938 *** > | make[1]: *** wait: No child processes. Stop. > | make[1]: *** Waiting for unfinished jobs.... > | make[1]: *** wait: No child processes. Stop. > | E: Caught signal 'Terminated': terminating immediately > | make[3]: *** wait: No child processes. Stop. > | make[3]: *** Waiting for unfinished jobs.... > | make[3]: *** wait: No child processes. Stop. > | make: *** [test-stamp] Terminated > | make[2]: *** [check-am] Error 2 > | Build killed with signal TERM after 300 minutes of inactivity > > 300 minutes is a long time. What do you suggest we do? It would probably be useful to see what is causing glibc to detect an invalid pointer in a call to free(). This has been consistent in the last 3 build attempts, and it maybe related to this hang. I don't have any reason to believe glibc is detecting a false positive here. > I already skip the > tests here in debian/rules: > > cpu := $(shell dpkg-architecture -qDEB_HOST_GNU_CPU) > > [...] > > test: test-stamp > test-stamp: build-stamp > ifeq "$(findstring $(cpu), m68k arm armeb mipsel mips)" "" > -LD_LIBRARY_PATH=$(CURDIR)/ql/.libs $(MAKE) check > -LD_LIBRARY_PATH=$(CURDIR)/ql/.libs $(MAKE) benchmark > endif > touch test-stamp > > It may make sense to add hppa to the list. What do you think? Could you try > a build without this 'make check; make benchmark' step ? You should be able to do this test in the sid chroot on paer - it should be practically identical to the buildd environment. If you have any arch-specific questions, debian-hppa is a good place to ask. -- dann frazier -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
