bugcheck! __get_free_pages calls __init function w/ CONFIG_NUMA

[Alex Williamson]

   I'm not sure why we haven't seen this before, but I started seeing a
stack trace on reboot on my rx2600 ia64 box running latest 2.6.  Here it
is:

 [<a0000001006ba020>] get_boot_pages+0x0/0x2c0
                                sp=e00000010267fc80 bsp=e000000102679130
 [<a0000001004a8fb0>] sba_alloc_coherent+0x70/0x1a0
                                sp=e00000010267fc80 bsp=e0000001026790f8
 [<a00000010047e3b0>] mptscsih_synchronize_cache+0x1d0/0x640
                                sp=e00000010267fc80 bsp=e000000102679010
 [<a0000001004665e0>] mptbase_shutdown+0xc0/0xe0
                                sp=e00000010267fd30 bsp=e000000102678fd8
 [<a0000001003a9140>] device_shutdown+0x260/0x280
                                sp=e00000010267fd30 bsp=e000000102678fa8
 [<a0000001000b10a0>] sys_reboot+0x2e0/0x720
                                sp=e00000010267fd30 bsp=e000000102678f50


sba_alloc_coherent is simply calling __get_free_pages() to setup a DMA
mapping for the mpt driver to sync a disk.  However, we've already
cleared system_running in sys_reboot, so w/ CONFIG_NUMA, we blowup
trying to call get_boot_pages, which was already freed.  I'm not sure if
the proper fix is to make get_boot_pages not an __init function or if
the CONFIG_NUMA code really intends to be keying off something else. 
Thoughts on the right fix?  Thanks,

	Alex

-- 
Alex Williamson                             HP Linux & Open Source Lab

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/