Greetings! This appears to be due to k-a's unique oom processor. The underlying GCL jobs are receiving SIGKILL when enough memory is written to. GCL attempts to size the available runtime memory by probing brk, and then backing off by some multiple, but here (apparently alone) the multiple needs to be significantly smaller. I can put in a heuristic workaround of course, but this leaves the broader question of whether the kernel should allow memory to be brk'ed which cannot later be written.
Take care, Steven Chamberlain <ste...@pyro.eu.org> writes: > Hi, > > Steven Chamberlain wrote: >> | **CERTIFICATION FAILED** for >> /«PKGBUILDDIR»/books/parsers/earley/earley-parser.lisp >> | | *********************************************** >> | | ************ ABORTING from raw Lisp *********** >> | | *********************************************** >> | **CERTIFICATION FAILED** for >> /«PKGBUILDDIR»/books/parsers/earley/earley-parser.lisp >> | make[2]: *** [parsers/earley/earley-parser.cert] Error 1 >> | **CERTIFICATION FAILED** for >> /«PKGBUILDDIR»/books/projects/security/des-ghosh/des.lisp >> | | *********************************************** >> | | ************ ABORTING from raw Lisp *********** >> | | *********************************************** >> | | *** Note: No checkpoints to print. *** >> | **CERTIFICATION FAILED** for >> /«PKGBUILDDIR»/books/projects/security/des-ghosh/des.lisp >> | make[2]: *** [projects/security/des-ghosh/des.cert] Error 1 > > I couldn't reproduce it locally; Christoph, please could you give back > src:acl2 for another build on kfreebsd-amd64 to see if it is still > reproducible on the buildd? > > Thanks, > Regards, -- Camm Maguire c...@maguirefamily.org ========================================================================== "The earth is but one country, and mankind its citizens." -- Baha'u'llah
