Messages in this thread | ![/](/images/icornerl.gif) | | Date | Thu, 1 Apr 2010 23:07:17 +0200 (CEST) | From | Thomas Gleixner <> | Subject | Re: [ANNOUNCE] 2.6.33.1-rt11 - BUG? |
| |
On Thu, 1 Apr 2010, Xianghua Xiao wrote:
Can you please stop top posting ?
> Here is the new dmesg output: > BUG: sleeping function called from invalid context at kernel/rtmutex.c:684 > pcnt: 1 0 in_atomic(): 1, irqs_disabled(): 1, pid: 5770, name: insmod > Call Trace: > [ce935dc0] [c00096cc] show_stack+0x6c/0x1a4 (unreliable) > [ce935df0] [c001f928] __might_sleep+0x104/0x108 > [ce935e00] [c03cb414] rt_spin_lock+0xa0/0xa4 > [ce935e10] [c00a4098] kmem_cache_alloc+0x50/0x17c > [ce935e40] [c0073570] irq_to_desc_alloc_node+0x104/0x5ec > [ce935e60] [c00064e0] irq_setup_virq+0x30/0xa8 > [ce935e80] [c000665c] irq_create_mapping+0x104/0x168 > [ce935ea0] [d1f69bc4] dma_init+0x118/0x1f0 [ipc] > [ce935ee0] [d1f75018] ipc_init+0x18/0x140 [ipc] > [ce935ef0] [c00038e0] do_one_initcall+0x54/0x210 > [ce935f20] [c005e424] sys_init_module+0x120/0x240 > [ce935f40] [c00139d4] ret_from_syscall+0x0/0x38 > > I chased from ipc_init to irq_to_desc_alloc_node and found no > interrupt-disabling. > > By looking at irq_to_desc_alloc_node (kernel/irq/handler.c) it has > raw_spin_lock_irqsave(), with this raw spinlock irqsave I'm not sure > if it causes trouble at kmem_cache_alloc after rt11 is applied, still > checking on that.
Can you please disabled CONFIG_SPARSE_IRQ ?
Thanks,
tglx
| ![\](/images/icornerr.gif) |