livepatch/core: Restrict minimum size of function that can be patched

hulk inclusion
category: bugfix
bugzilla: https://gitee.com/openeuler/kernel/issues/I60L10


CVE: NA

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

If a function is patched, instructions at the beginning are modified to be
'jump codes' which jump to new function. This requires the function be big
enough, otherwise the modification may be out of function range.

Currently each architecture needs to implement arch_klp_func_can_patch()
to check function size. However, there exists following problems:
  1. arch 'x86' didn't implement arch_klp_func_can_patch();
  2. implementations in arm64 & ppc32, function size is checked only if
     there's a long jump. There is a scenario where a very short function
     is successfully patched, but as kernel module increases, someday long
     jump is required, then the function become unable to be patched.
  3. implementaions look like duplicate.

In this patch, introduce macro KLP_MAX_REPLACE_SIZE to denote the maximum
size that will be replaced on patching, then move the check ahead into
klp_init_object_loaded().

Fixes: c33e4283 ("livepatch/core: Allow implementation without ftrace")
Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com>
Reviewed-by: Kuohai Xu <xukuohai@huawei.com>
Signed-off-by: Zheng Zengkai <zhengzengkai@huawei.com>