Thomas, I was seeing a bunch of scheduling while atomic backtraces from rc4-rt5 on my laptop: Dec 2 17:50:56 localhost kernel: [ 1397.730931] Call Trace: Dec 2 17:50:56 localhost kernel: [ 1397.730938] [] __schedule_bug+0x75/0x7a Dec 2 17:50:56 localhost kernel: [ 1397.730943] [] __schedule+0x8a/0x505 Dec 2 17:50:56 localhost kernel: [ 1397.730947] [] schedule+0x75/0x91 Dec 2 17:50:56 localhost kernel: [ 1397.730951] [] rt_spin_lock_slowlock+0x182/0x21e Dec 2 17:50:56 localhost kernel: [ 1397.730954] [] ? rt_spin_lock_slowlock+0x10d/0x21e Dec 2 17:50:56 localhost kernel: [ 1397.730961] [] rt_spin_lock_fastlock.constprop.12+0x24/0x26 Dec 2 17:50:56 localhost kernel: [ 1397.730965] [] rt_spin_lock+0x16/0x3d Dec 2 17:50:56 localhost kernel: [ 1397.730969] [] advance_transaction+0x31/0x141 Dec 2 17:50:56 localhost kernel: [ 1397.730972] [] acpi_ec_gpe_handler+0x3f/0xc0 Dec 2 17:50:56 localhost kernel: [ 1397.730977] [] acpi_ev_gpe_dispatch+0xc5/0x13e Dec 2 17:50:56 localhost kernel: [ 1397.730983] [] acpi_ev_gpe_detect+0xb5/0x10d Dec 2 17:50:56 localhost kernel: [ 1397.730987] [] ? irq_thread_fn+0x3a/0x3a Dec 2 17:50:56 localhost kernel: [ 1397.730990] [] acpi_ev_sci_xrupt_handler+0x22/0x2b Dec 2 17:50:56 localhost kernel: [ 1397.730994] [] acpi_irq+0x16/0x31 Dec 2 17:50:56 localhost kernel: [ 1397.730996] [] irq_forced_thread_fn+0x25/0x46 Dec 2 17:50:56 localhost kernel: [ 1397.730998] [] ? _raw_spin_unlock_irq+0x53/0x60 Dec 2 17:50:56 localhost kernel: [ 1397.731001] [] irq_thread+0xa5/0x18d Dec 2 17:50:56 localhost kernel: [ 1397.731003] [] ? irq_finalize_oneshot+0x19/0x19 Dec 2 17:50:56 localhost kernel: [ 1397.731006] [] kthread+0x99/0xa1 Dec 2 17:50:56 localhost kernel: [ 1397.731009] [] ? retint_restore_args+0x13/0x13 Dec 2 17:50:56 localhost kernel: [ 1397.731014] [] kernel_thread_helper+0x4/0x10 Dec 2 17:50:56 localhost kernel: [ 1397.731016] [] ? retint_restore_args+0x13/0x13 Dec 2 17:50:56 localhost kernel: [ 1397.731019] [] ? __init_kthread_worker+0x89/0x89 Dec 2 17:50:56 localhost kernel: [ 1397.731022] [] ? gs_change+0x13/0x13 The attached patch converts the ACPI Embedded Controller lock to a raw spinlock. Boots and suspends/resumes with no backtraces (from this anyway...let's not talk about nouveau right now...) Clark