> BTW, this may be out of scope for this patchset, but is it a large amount of > work to support clearing klp-relocations on target module unload? ie, this > test case: > > - (target module and livepatch loaded) > - rmmod target_mod > - modprobe target_mod << fails as reloc target is non-zero > > IIRC, Miroslav had taken a stab at this last year, but I don't remember what > the technical problems were then. Yes, see https://lore.kernel.org/live-patching/alpine.LSU.2.21.1910031110440.9011@xxxxxxxxxxxxx/ I think the "conclusion" there still applies... but yes, it makes testing difficult. Miroslav
