Kees,

On Wed, Jan 15, 2014 at 04:19:34PM -0800, Kees Cook wrote:
> Is it possible this is the bug that was fixed with "x86, relocs: add
> percpu fixup for GNU ld 2.23"?
> 
> 82fa9637a2ba285bcc7c5050c73010b2c1b3d803 doesn't include the full
> x86,kalsr (and x86,relocs) series, so I'm not sure it's a good place
> to test.

I just tested the new tip/x86/kaslr head da2b6fb99 ("x86, kaslr:
Clarify RANDOMIZE_BASE_MAX_OFFSET") and find 4 boot failures out of
100 total boots for kconfig x86_64-randconfig-s1-01101738.

        boot_successes: 96
        boot_failures: 4
        general_protection_fault:SMP_SMP: 2
        RIP:get_pageblock_flags_group: 1
        RIP:memcpy: 1
        Kernel_panic-not_syncing:Fatal_exception: 2
        page_allocation_failure:order:,mode: 1
        BUG:kernel_boot_hang: 1
        RIP:__lock_acquire: 1
        RIP:free_object: 1

Attached are the dmesg files. One of them is OOM triggered by trinity,
which is not a problem.

Thanks,
Fengguang