Updated version with Petr's feedback. It looks a bit different and better now (I would say). Not that it should be considered before we decide what to do with late module patching, but I finished it before the discussion started and someone could be interested. v1: http://lore.kernel.org/r/20190719122840.15353-1-mbenes@xxxxxxx Tested on x86_64, ppc64le and s390x. Cross-compiled on arm64 to verify that nothing is broken. [1] 20180602161151.apuhs2dygsexmcg2@treble [2] 1561019068-132672-1-git-send-email-cj.chengjian@xxxxxxxxxx [3] 20180607092949.1706-1-mbenes@xxxxxxx Miroslav Benes (3): livepatch: Clear relocation targets on a module removal livepatch: Unify functions for writing and clearing object relocations livepatch: Clean up klp_update_object_relocations() return paths arch/powerpc/kernel/module_64.c | 45 +++++++++++++++++++++++++ arch/s390/kernel/module.c | 8 +++++ arch/x86/kernel/module.c | 43 ++++++++++++++++++++++++ include/linux/moduleloader.h | 7 ++++ kernel/livepatch/core.c | 58 ++++++++++++++++++++++++--------- 5 files changed, 146 insertions(+), 15 deletions(-) -- 2.23.0
