[PATCH v5] livepatch: Clear relocation targets on a module removal

[Joe Lawrence]

On Wed, Aug 31, 2022 at 10:05:43AM -0700, Song Liu wrote:
> On Wed, Aug 31, 2022 at 1:01 AM Christophe Leroy
> <christophe.leroy at csgroup.eu> wrote:
> >
> >
> >
> > Le 30/08/2022 à 20:53, Song Liu a écrit :
> > > From: Miroslav Benes <mbenes at suse.cz>
> > >
> > > Josh reported a bug:
> > >
> > >    When the object to be patched is a module, and that module is
> > >    rmmod'ed and reloaded, it fails to load with:
> > >
> > >    module: x86/modules: Skipping invalid relocation target, existing value is nonzero for type 2, loc 00000000ba0302e9, val ffffffffa03e293c
> > >    livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
> > >    livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
> > >
> > >    The livepatch module has a relocation which references a symbol
> > >    in the _previous_ loading of nfsd. When apply_relocate_add()
> > >    tries to replace the old relocation with a new one, it sees that
> > >    the previous one is nonzero and it errors out.
> > >
> > >    On ppc64le, we have a similar issue:
> > >
> > >    module_64: livepatch_nfsd: Expected nop after call, got e8410018 at e_show+0x60/0x548 [livepatch_nfsd]
> > >    livepatch: failed to initialize patch 'livepatch_nfsd' for module 'nfsd' (-8)
> > >    livepatch: patch 'livepatch_nfsd' failed for module 'nfsd', refusing to load module 'nfsd'
> > >
> > > He also proposed three different solutions. We could remove the error
> > > check in apply_relocate_add() introduced by commit eda9cec4c9a1
> > > ("x86/module: Detect and skip invalid relocations"). However the check
> > > is useful for detecting corrupted modules.
> > >
> > > We could also deny the patched modules to be removed. If it proved to be
> > > a major drawback for users, we could still implement a different
> > > approach. The solution would also complicate the existing code a lot.
> > >
> > > We thus decided to reverse the relocation patching (clear all relocation
> > > targets on x86_64). The solution is not
> > > universal and is too much arch-specific, but it may prove to be simpler
> > > in the end.
> > >
> > > Reported-by: Josh Poimboeuf <jpoimboe at redhat.com>
> > > Signed-off-by: Miroslav Benes <mbenes at suse.cz>
> > > Signed-off-by: Song Liu <song at kernel.org>
> > >
> > > ---
> > >
> > > NOTE: powerpc code has not be tested.
> > >
> > > Changes v4 = v5:
> > > 1. Fix compile with powerpc.
> >
> > Not completely it seems.
> >
> >    CC      kernel/livepatch/core.o
> > kernel/livepatch/core.c: In function 'klp_clear_object_relocations':
> > kernel/livepatch/core.c:352:50: error: passing argument 1 of
> > 'clear_relocate_add' from incompatible pointer type
> > [-Werror=incompatible-pointer-types]
> >    352 |                 clear_relocate_add(pmod->klp_info->sechdrs,
> >        |                                    ~~~~~~~~~~~~~~^~~~~~~~~
> >        |                                                  |
> >        |                                                  Elf32_Shdr *
> > {aka struct elf32_shdr *}
> > In file included from kernel/livepatch/core.c:19:
> > ./include/linux/moduleloader.h:76:37: note: expected 'Elf64_Shdr *' {aka
> > 'struct elf64_shdr *'} but argument is of type 'Elf32_Shdr *' {aka
> > 'struct elf32_shdr *'}
> >     76 | void clear_relocate_add(Elf64_Shdr *sechdrs,
> >        |                         ~~~~~~~~~~~~^~~~~~~
> > cc1: some warnings being treated as errors
> >
> >
> > Fixup:
> >
> > diff --git a/include/linux/moduleloader.h b/include/linux/moduleloader.h
> > index d22b36b84b4b..958e6da7f475 100644
> > --- a/include/linux/moduleloader.h
> > +++ b/include/linux/moduleloader.h
> > @@ -73,7 +73,7 @@ int apply_relocate_add(Elf_Shdr *sechdrs,
> >                        unsigned int relsec,
> >                        struct module *mod);
> >   #ifdef CONFIG_LIVEPATCH
> > -void clear_relocate_add(Elf64_Shdr *sechdrs,
> > +void clear_relocate_add(Elf_Shdr *sechdrs,
> >                    const char *strtab,
> >                    unsigned int symindex,
> >                    unsigned int relsec,
> >
> >
> > But then the link fails.
> >
> >    LD      .tmp_vmlinux.kallsyms1
> > powerpc64-linux-ld: kernel/livepatch/core.o: in function
> > `klp_cleanup_module_patches_limited':
> > core.c:(.text+0xdb4): undefined reference to `clear_relocate_add'
> 
> Hmm.. I am not seeing either error. Could you please share your .config file?
> 
> Thanks,
> Song
> 

If it's any help, I see the same build error Christophe reported using
the 'cross-dev' script that's in my klp-convert-tree [1].

  $ BUILD_ARCHES="ppc32" ./cross-dev config
  $ BUILD_ARCHES="ppc32" ./cross-dev build -j$(nproc)

(The kernel will be built in /tmp/klp-convert-ppc32 btw.)

Applying the header file fix results in the same linker error, too.


[1] https://github.com/joe-lawrence/klp-convert-tree/tree/klp-convert-v7-devel+song

-- Joe