what's up with "make -j N" output?
Script does effectively:
make -k -j5 &>1.log
What's inside:
C LD drivers/input/built-in.o
^^^^^
CC fs/sysfs/symlink.o CC [M] net/bluetooth/l2cap.o CC [M] drivers/ide/ide-eh.o CC [M] drivers/ide/ide-proc.o CC net/core/skbuff.o CC fs/sysfs/mount.o CC fs/sysfs/bin.o o <=== CC [M] drivers/ide/ide-gd.o CC fs/sysfs/group.o LD fs/sysfs/built-in.o CC [M] drivers/ide/ide-disk.o CC fs/anon_inodes.o
...
CC drivers/mmc/core/sdio_cis.o
CC lib/div64.o54/built-in.o <===
CC net/ipv4/route.o
IIRC, I never saw this before.
Maybe something wasn't guaranteed before, but, hey, logs are crap now.
Kernel is 2.6.33-rc4-00204-g7dc9c48, x86_64
CONFIG_PREEMPT=yCONFIG_HZ=100
# CONFIG_DEBUG_PERF_USE_VMALLOC is not set
# CONFIG_DEBUG_DRIVER is not set
# CONFIG_DEBUG_DEVRES is not set
# CONFIG_DEBUG_FS is not setCONFIG_DEBUG_KERNEL=yCONFIG_DEBUG_SHIRQ=yCONFIG_DEBUG_OBJECTS=y
# CONFIG_DEBUG_OBJECTS_SELFTEST is not setCONFIG_DEBUG_OBJECTS_FREE=yCONFIG_DEBUG_OBJECTS_TIMERS=yCONFIG_DEBUG_OBJECTS_WORK=yCONFIG_DEBUG_OBJECTS_ENABLE_DEFAULT=1
# CONFIG_DEBUG_KMEMLEAK is not setCONFIG_DEBUG_PREEMPT=yCONFIG_DEBUG_RT_MUTEXES=yCONFIG_DEBUG_PI_LIST=yCONFIG_DEBUG_SPINLOCK=yCONFIG_DEBUG_MUTEXES=yCONFIG_DEBUG_LOCK_ALLOC=y
# CONFIG_DEBUG_LOCKDEP is not setCONFIG_DEBUG_SPINLOCK_SLEEP=y
# CONFIG_DEBUG_LOCKING_API_SELFTESTS is not set
# CONFIG_DEBUG_KOBJECT is not setCONFIG_DEBUG_BUGVERBOSE=yCONFIG_DEBUG_INFO=yCONFIG_DEBUG_VM=yCONFIG_DEBUG_VIRTUAL=yCONFIG_DEBUG_WRITECOUNT=yCONFIG_DEBUG_MEMORY_INIT=yCONFIG_DEBUG_LIST=yCONFIG_DEBUG_SG=yCONFIG_DEBUG_NOTIFIERS=y
# CONFIG_DEBUG_CREDENTIALS is not set
# CONFIG_DEBUG_BLOCK_EXT_DEVT is not set
# CONFIG_DEBUG_FORCE_WEAK_PER_CPU is not setCONFIG_DEBUG_PAGEALLOC=yCONFIG_DEBUG_STACKOVERFLOW=yCONFIG_DEBUG_STACK_USAGE=yCONFIG_DEBUG_PER_CPU_MAPS=yCONFIG_DEBUG_RODATA=yCONFIG_DEBUG_RODATA_TEST=yCONFIG_DEBUG_STRICT_USER_COPY_CHECKS=y
Quoted 3 lines · Alexey Dobriyan
Script does effectively:
make -k -j5 &>1.log
My automated builds effectively run a number of "make -j 1" builds in parallel so I get sane parsable and readable logs. I don't think we give any guarantees for -j 2 or higher even though that would be highly desirable when things go wrong.
Ralf
Quoted 8 lines · Ralf Baechle
Quoted 3 lines · Alexey Dobriyan
Script does effectively:
make -k -j5 &>1.log
My automated builds effectively run a number of "make -j 1" builds in parallel so I get sane parsable and readable logs.
I do this too, still -j1 is waste of time if only one compile is running for whatever reason.
Quoted 2 lines
I don't think we give any guarantees for -j 2 or higher even though that would be highly desirable when things go wrong.
Meanwhile, bug dissapeared somewhere. :-\