euler inclusion
category: bugfix
Bugzilla: 5507/5072
CVE: N/A

----------------------------------------

we use arch__klp_enable_func in atomic context to patched instruction
	arch__klp_enable_func
		-=>	kzalloc(XXX, GFP_KERNEL)
but it might_sleep here, when we enale an livepatch module, cause crash,
use GFP_ATOMIC instead of GFP_KERNEL

the call trace is like:

livepatch: enabling patch 'klp_testEL_HOTPATCH_ADDFUNTOMULTIFILE_FUN_001'
BUG: sleeping function called from invalid context at mm/slub.c:1287
in_atomic(): 1, irqs_disabled(): 128, pid: 13, name: migration/1
Preemption disabled at:[<ffffffc0002397b4>] smpboot_thread_fn+0x27c/0x2a4

CPU: 1 PID: 13 Comm: migration/1 Tainted: G        W  O  K 4.4.159+ #3
Hardware name: hisilicon,hi1213-fpga (DT)
Call trace:
[<ffffffc000207f88>] dump_backtrace+0x0/0x13c
[<ffffffc0002080e8>] show_stack+0x24/0x30
[<ffffffc00041d338>] dump_stack+0x90/0xb0
[<ffffffc00023db1c>] ___might_sleep+0x18c/0x19c
[<ffffffc00023dbac>] __might_sleep+0x80/0x90
[<ffffffc0003251d4>] kmem_cache_alloc_trace+0x60/0x248
[<ffffffc000211f28>] arch__klp_enable_func+0x70/0x144
[<ffffffc0002726a8>] klp_try_enable_patch+0x114/0x1e0
[<ffffffc0002a25c0>] multi_cpu_stop+0xb0/0x104
[<ffffffc0002a2828>] cpu_stopper_thread+0xa0/0x130
[<ffffffc0002397b4>] smpboot_thread_fn+0x27c/0x2a4
[<ffffffc000235e90>] kthread+0x114/0x11c
[<ffffffc000203dd0>] ret_from_fork+0x10/0x40
Signed-off-by: NCheng Jian <cj.chengjian@huawei.com>
Reviewed-by: NLi Bin <huawei.libin@huawei.com>
Signed-off-by: NYang Yingliang <yangyingliang@huawei.com>