Re: [PATCH 1/2] livepatch: fix non-static warnings

From: Joe Lawrence
Date: Sat Dec 15 2018 - 11:23:47 EST


On Sat, Dec 15, 2018 at 09:50:52AM +0100, Nicholas Mc Guire wrote:
> On Fri, Dec 14, 2018 at 04:34:23PM -0500, Joe Lawrence wrote:
> > On 12/14/2018 11:56 AM, Nicholas Mc Guire wrote:
> > > Sparse reported warnings about non-static symbols. For the variables a
> > > simple static attribute is fine - for those symbols referenced by
> > > livepatch via klp_func the symbol-names must be unmodified in the
> > > relocation table - to resolve this the __noclone attribute (as
> > ^^^^^^^^^^
> > nit: symbol table
>
> that should have been relocation section as described in
> Documentation/livepatch/module-elf-format.txt - atleast that is how
> I currently undderstand the livepatch mechanism and its seperate
> relocation section.
>
Hi Nicholas,

Jessica can explain module-elf-format.txt in more detail, but the
highlight is that it outlines the format for _livepatch_ modules, in
this case livepatch-shadow-fix{1,2}.ko. The special relocations
detailed in that file are needed when livepatch modules reference
symbols defined elsewhere (vmlinux, other modules) that may not
ordinarily be visible (non-exported, local, etc.) to the module loader.

When livepatch modules register themselves on load, the livepatching
core needs to find the address of the _target_ to-be-patched function
using a kallsyms lookup. If that can't be found, the kernel will emit
an error, "livepatch: symbol 'dummy_free' not found in symbol table".
The call path is:

klp_register_patch
klp_init_object_loaded
klp_find_object_symbol

if you want to trace the execution path.

>
> thanks for your patience - so I did not yet understand how this really
> works together - will give it a rerun and repost a hopefully proper
> solution.
>
> thx!
> hofrat

Thanks for sticking with it and learning some livepatching internals
along the way.

-- Joe