Thanks, but I'm going to bring the -rt tree back to making all local_bh_disable into nops, and then investigate all the areas that might need a true bh disable, and do preempt_disable for rt. This is the major cause of these scheduling while atomic bugs. Any time you see a '| preempt count: 00000101 ]' where you have the 0xff00 bits set, is a local_bh_disable bug. Right now I'm investigating a bug that I get with the local_bh_disable as nops, and this is why I have not released it yet. The old -rt trees had local_bh_disable as a nop. -- Steve On Sun, 1 Feb 2009, Luis Henriques wrote: > Hi, > > I am using an x86_64 and getting the following log while executing the master in > the -rt git tree: > > [ 28.182621] BUG: scheduling while atomic: sirq-net-rx/0/8/0x00000101, CPU#0 > [ 28.182744] Modules linked in: parport_pc ppdev parport ipv6 powernow_k8 > cpufreq_stats cpufreq_conservative cpufreq_ondemand freq_table cpufreq_powersave > cpufreq_userspace af_packet sbp2 loop joydev arc4 snd_hda_intel ecb snd_pcm > snd_page_alloc uvcvideo snd_hwdep ath5k snd_seq compat_ioctl32 mac80211 psmouse > snd_timer videodev snd_seq_device k8temp v4l1_compat led_class snd soundcore > cfg80211 rfkill input_polldev battery video output ac button evdev ext3 jbd > mbcache sg sr_mod cdrom ide_pci_generic ide_core ata_generic sd_mod pata_acpi > ohci1394 ahci ehci_hcd pata_atiixp ohci_hcd ieee1394 libata usbcore scsi_mod > r8169 mii thermal processor fan > [ 28.186632] Pid: 8, comm: sirq-net-rx/0 Tainted: G W > 2.6.28-rt-00209-g82e0a07 #1 > [ 28.186747] Call Trace: > [ 28.186831] [<ffffffff80455cb4>] ? netif_receive_skb+0x2c4/0x590 > [ 28.186938] [<ffffffff80242795>] __schedule_bug+0x85/0x90 > [ 28.187033] [<ffffffff804ead38>] thread_return+0x142/0x70a > [ 28.187127] [<ffffffff8020c3c8>] ? ftrace_call+0x5/0x2b > [ 28.187218] [<ffffffff8024f42f>] ksoftirqd+0x1df/0x270 > [ 28.187306] [<ffffffff8024f250>] ? ksoftirqd+0x0/0x270 > [ 28.187395] [<ffffffff802621de>] kthread+0x4e/0x90 > [ 28.187481] [<ffffffff8020d639>] child_rip+0xa/0x11 > [ 28.187569] [<ffffffff802408d9>] ? finish_task_switch+0x59/0x130 > [ 28.187663] [<ffffffff804ed1ab>] ? _spin_unlock_irq+0x1b/0x40 > [ 28.187753] [<ffffffff8020cb25>] ? restore_args+0x0/0x30 > [ 28.187842] [<ffffffff80262190>] ? kthread+0x0/0x90 > [ 28.187929] [<ffffffff8020d62f>] ? child_rip+0x0/0x11 > [ 28.188034] --------------------------- > [ 28.188111] | preempt count: 00000101 ] > [ 28.188189] | 2-level deep critical section nesting: > [ 28.188272] ---------------------------------------- > [ 28.188356] .. [<ffffffff804a9c16>] .... icmp_send+0xc6/0x6a0 > [ 28.188497] .....[<ffffffff804a9c16>] .. ( <= icmp_send+0xc6/0x6a0) > [ 28.188639] .. [<ffffffff804ea7ac>] .... schedule+0x2c/0x476 > [ 28.188777] .....[<ffffffff8024f42f>] .. ( <= ksoftirqd+0x1df/0x270) > [ 28.188918] > > Can also provide my config file if needed. > > -- > Luis Henriques > > -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
