On Thu, 2023-06-15 at 17:44 +0200, Florent Revest wrote:
> On Thu, Jun 15, 2023 at 4:56 PM Florent Revest <revest@xxxxxxxxxxxx> wrote:
> >
> > When building a kernel with LLVM=1, LLVM_IAS=0 and CONFIG_KASAN=y, LLVM
> > leaves DWARF tags for the "asan.module_ctor" & co symbols.
>
> To be fair I can't tell if this is an LLVM bug. It's sort of curious
> that with LLVM_IAS=1, these debugging symbols are not kept and they
> are with LLVM_IAS=0 but I don't know what the expected behavior should
> be and how BTF should deal with it. I'll let people with more context
> comment on this! :)
>
> An easy reproducer is:
>
> $ touch pwet.c
>
> $ clang -g -fsanitize=kernel-address -c -o pwet.o pwet.c
> $ llvm-dwarfdump pwet.o | grep module_ctor
>
> $ clang -fno-integrated-as -g -fsanitize=kernel-address -c -o pwet.o pwet.c
> $ llvm-dwarfdump pwet.o | grep module_ctor
> DW_AT_name ("asan.module_ctor")
Interestingly, I am unable to reproduce it using either
clang version 14.0.0-1ubuntu1 or clang main (bd66f4b1da30).
>
> > In a dramatic turn of event, this BTF verification failure can cause
> > the netfilter_bpf initialization to fail, causing netfilter_core to
> > free the netfilter_helper hashmap and netfilter_ftp to trigger a
> > use-after-free. The risk of u-a-f in netfilter will be addressed
> > separately
>
> To be precise, I meant "netfilter conntrack".
>
> I sent the following patch as a more targeted mitigation for the uaf
> https://lore.kernel.org/netfilter-devel/20230615152918.3484699-1-revest@xxxxxxxxxxxx/T/#u
>
