The second attempt to resolve the issue reported by Josh last year [1] and also reported earlier this year again [2]. The first attempt [3] tried to deny the patched modules to be removed. It did not solve the issue completely. It would be possible, but we decided to try the arch-specific approach first, which I am sending now. Sending early as RFC, because I am leaving on holiday tomorrow and it would be great if you took a look meanwhile. - I decided not to CC the arch maintainers yet. If we decide that the approach is feasible first on our livepatch side, I will split the second patch and resend properly. - the first patch could go in regardless the rest, I guess. - I am not sure about the placement in klp_cleanup_module_patches_limited(). I think it is the best possible, but I would really appreciate double-checking. - I am also not sure about the naming, so ideas also welcome Lightly tested on both x86_64 and ppc64le and it looked ok. [1] 20180602161151.apuhs2dygsexmcg2@treble [2] 1561019068-132672-1-git-send-email-cj.chengjian@xxxxxxxxxx [3] 20180607092949.1706-1-mbenes@xxxxxxx Miroslav Benes (2): livepatch: Nullify obj->mod in klp_module_coming()'s error path livepatch: Clear relocation targets on a module removal arch/powerpc/kernel/Makefile | 1 + arch/powerpc/kernel/livepatch.c | 75 +++++++++++++++++++++++++++++++++ arch/powerpc/kernel/module.h | 15 +++++++ arch/powerpc/kernel/module_64.c | 7 +-- arch/x86/kernel/livepatch.c | 67 +++++++++++++++++++++++++++++ include/linux/livepatch.h | 5 +++ kernel/livepatch/core.c | 18 +++++--- 7 files changed, 177 insertions(+), 11 deletions(-) create mode 100644 arch/powerpc/kernel/livepatch.c create mode 100644 arch/powerpc/kernel/module.h -- 2.22.0
