Badness in interruptible_sleep_on in 2.6.2-rc2-mm1
Quad P3 Xeon boots with a series of these:
Badness in interruptible_sleep_on at kernel/sched.c:2239
Call Trace: [<c011cea4>] interruptible_sleep_on+0xf4/0x100 [<c011cab0>] default_wake_function+0x0/0x20 [<c025948b>] pagebuf_daemon+0x13b/0x240 [<c034a22e>] ret_from_fork+0x6/0x14 [<c0259320>] pagebuf_daemon_wakeup+0x0/0x30 [<c0259350>] pagebuf_daemon+0x0/0x240 [<c0108f69>] kernel_thread_helper+0x5/0xc
They repeat over and over. A login prompt never comes. I saw this in 2.6.2-rc2-mm1 and 2.6.2-rc1-mm3. It didn't occur with 2.6.1-rc1-mm2. 2.6.2-rc1-bk2 was okay too.
There are no modules in the configuration.
Here is the .config:
CONFIG_X86=yCONFIG_MMU=yCONFIG_UID16=yCONFIG_GENERIC_ISA_DMA=yCONFIG_EXPERIMENTAL=yCONFIG_CLEAN_COMPILE=yCONFIG_STANDALONE=yCONFIG_SWAP=yCONFIG_SYSVIPC=yCONFIG_SYSCTL=yCONFIG_LOG_BUF_SHIFT=15CONFIG_KALLSYMS=yCONFIG_FUTEX=yCONFIG_EPOLL=yCONFIG_IOSCHED_NOOP=yCONFIG_IOSCHED_AS=yCONFIG_IOSCHED_DEADLINE=yCONFIG_IOSCHED_CFQ=yCONFIG_X86_PC=yCONFIG_CPU_PENTIUMIII=yCONFIG_CPU_INTEL=yCONFIG_X86_CMPXCHG=yCONFIG_X86_XADD=yCONFIG_X86_L1_CACHE_SHIFT=5CONFIG_RWSEM_XCHGADD_ALGORITHM=yCONFIG_X86_WP_WORKS_OK=yCONFIG_X86_INVLPG=yCONFIG_X86_BSWAP=yCONFIG_X86_POPAD_OK=yCONFIG_X86_INTEL_USERCOPY=yCONFIG_X86_USE_PPRO_CHECKSUM=yCONFIG_SMP=yCONFIG_NR_CPUS=4CONFIG_X86_LOCAL_APIC=yCONFIG_X86_IO_APIC=yCONFIG_X86_TSC=yCONFIG_HIGHMEM4G=yCONFIG_HIGHMEM=yCONFIG_MTRR=yCONFIG_HAVE_DEC_LOCK=yCONFIG_ACPI_BOOT=yCONFIG_PCI=yCONFIG_PCI_GOANY=yCONFIG_PCI_BIOS=yCONFIG_PCI_DIRECT=yCONFIG_HOTPLUG=yCONFIG_BINFMT_ELF=yCONFIG_SCSI=yCONFIG_SCSI_PROC_FS=yCONFIG_BLK_DEV_SD=yCONFIG_MAX_SD_DISKS=256CONFIG_SCSI_AIC7XXX=yCONFIG_AIC7XXX_CMDS_PER_DEVICE=4CONFIG_AIC7XXX_RESET_DELAY_MS=15000CONFIG_AIC7XXX_DEBUG_ENABLE=yCONFIG_AIC7XXX_DEBUG_MASK=0CONFIG_AIC7XXX_REG_PRETTY_PRINT=yCONFIG_SCSI_QLA2XXX_CONFIG=yCONFIG_SCSI_QLA2XXX=yCONFIG_SCSI_QLA22XX=yCONFIG_NET=yCONFIG_UNIX=yCONFIG_NET_KEY=yCONFIG_INET=yCONFIG_XFRM=yCONFIG_IPV6_SCTP__=yCONFIG_NETDEVICES=yCONFIG_NET_ETHERNET=yCONFIG_MII=yCONFIG_NET_PCI=yCONFIG_EEPRO100=yCONFIG_INPUT=yCONFIG_INPUT_MOUSEDEV=yCONFIG_INPUT_MOUSEDEV_PSAUX=yCONFIG_INPUT_MOUSEDEV_SCREEN_X=1024CONFIG_INPUT_MOUSEDEV_SCREEN_Y=768CONFIG_SOUND_GAMEPORT=yCONFIG_SERIO=yCONFIG_SERIO_I8042=yCONFIG_SERIO_SERPORT=yCONFIG_INPUT_KEYBOARD=yCONFIG_KEYBOARD_ATKBD=yCONFIG_VT=yCONFIG_VT_CONSOLE=yCONFIG_HW_CONSOLE=yCONFIG_SERIAL_8250=yCONFIG_SERIAL_8250_CONSOLE=yCONFIG_SERIAL_8250_NR_UARTS=2CONFIG_SERIAL_CORE=yCONFIG_SERIAL_CORE_CONSOLE=yCONFIG_UNIX98_PTYS=yCONFIG_UNIX98_PTY_COUNT=256CONFIG_RTC=yCONFIG_VGA_CONSOLE=yCONFIG_DUMMY_CONSOLE=yCONFIG_EXT2_FS=yCONFIG_EXT3_FS=yCONFIG_JBD=yCONFIG_REISERFS_FS=yCONFIG_JFS_FS=yCONFIG_XFS_FS=yCONFIG_PROC_FS=yCONFIG_PROC_KCORE=yCONFIG_SYSFS=yCONFIG_DEVPTS_FS=yCONFIG_RAMFS=yCONFIG_MSDOS_PARTITION=yCONFIG_NLS=yCONFIG_NLS_DEFAULT="iso8859-1"CONFIG_DEBUG_KERNEL=yCONFIG_MAGIC_SYSRQ=yCONFIG_X86_FIND_SMP_CONFIG=yCONFIG_X86_MPPARSE=yCONFIG_X86_SMP=yCONFIG_X86_HT=yCONFIG_X86_BIOS_REBOOT=yCONFIG_X86_TRAMPOLINE=yCONFIG_PC=y
Signature · 2 lines
Randy Hron http://home.earthlink.net/~rwhron/kernel/bigbox.html
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
"Randy" == rwhron <[email protected]> writes:
Randy> Quad P3 Xeon boots with a series of these: Badness in Randy> interruptible_sleep_on at kernel/sched.c:2239 Call Trace:
It's because XFS's pagebuf_daemon calls interruptible_sleep_on without holding the big kernel lock. While what it's doing is pretty ugly, it's in fact not dangerous in thise case. I am seeing the same on an ia64, but it only shows up a couple of times and then quiets down and it boots fine.
If you want, you can remove the SLEEP_ON_BKLCHECK in kernel/sched.c:interruptible_sleep_on() until a proper fix comes out. Or even better, rewrite the code in xfs/pagebuf/pagebuf.c to do the right thing instead.
Cheers,
Jes