linux-kernel - Re: [PATCH 6/7] x86: objtool: add support for R_X86_64_REX

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAG_fn=W8GDqYy_JV1F=YypD-6qR6vEqMuCi=DKfhdM-5=N3DdA@mail.gmail.com>
Date: Thu, 17 Apr 2025 17:37:09 +0200
From: Alexander Potapenko <glider@...gle.com>
To: Uros Bizjak <ubizjak@...il.com>, Ard Biesheuvel <ardb@...nel.org>
Cc: quic_jiangenj@...cinc.com, linux-kernel@...r.kernel.org, 
	kasan-dev@...glegroups.com, x86@...nel.org, 
	Aleksandr Nogikh <nogikh@...gle.com>, Andrey Konovalov <andreyknvl@...il.com>, Borislav Petkov <bp@...en8.de>, 
	Dave Hansen <dave.hansen@...ux.intel.com>, Dmitry Vyukov <dvyukov@...gle.com>, 
	Ingo Molnar <mingo@...hat.com>, Josh Poimboeuf <jpoimboe@...nel.org>, Marco Elver <elver@...gle.com>, 
	Peter Zijlstra <peterz@...radead.org>, Thomas Gleixner <tglx@...utronix.de>
Subject: Re: [PATCH 6/7] x86: objtool: add support for R_X86_64_REX_GOTPCRELX

On Wed, Apr 16, 2025 at 4:21 PM Uros Bizjak <ubizjak@...il.com> wrote:
>
>
>
> On 16. 04. 25 10:54, Alexander Potapenko wrote:
> > When compiling modules with -fsanitize-coverage=trace-pc-guard, Clang
> > will emit R_X86_64_REX_GOTPCRELX relocations for the
> > __start___sancov_guards and __stop___sancov_guards symbols. Although
> > these relocations can be resolved within the same binary, they are left
> > over by the linker because of the --emit-relocs flag.
> >
> > This patch makes it possible to resolve the R_X86_64_REX_GOTPCRELX
> > relocations at runtime, as doing so does not require a .got section.
> > In addition, add a missing overflow check to R_X86_64_PC32/R_X86_64_PLT32.
> >
> > Cc: x86@...nel.org
> > Signed-off-by: Alexander Potapenko <glider@...gle.com>
> > ---
> >   arch/x86/include/asm/elf.h      | 1 +
> >   arch/x86/kernel/module.c        | 8 ++++++++
> >   arch/x86/um/asm/elf.h           | 1 +
> >   tools/objtool/arch/x86/decode.c | 1 +
> >   4 files changed, 11 insertions(+)
> >
> > diff --git a/arch/x86/include/asm/elf.h b/arch/x86/include/asm/elf.h
> > index 1fb83d47711f9..15d0438467e94 100644
> > --- a/arch/x86/include/asm/elf.h
> > +++ b/arch/x86/include/asm/elf.h
> > @@ -63,6 +63,7 @@ typedef struct user_i387_struct elf_fpregset_t;
> >   #define R_X86_64_8          14      /* Direct 8 bit sign extended  */
> >   #define R_X86_64_PC8                15      /* 8 bit sign extended pc relative */
> >   #define R_X86_64_PC64               24      /* Place relative 64-bit signed */
> > +#define R_X86_64_REX_GOTPCRELX       42      /* R_X86_64_GOTPCREL with optimizations */
> >
> >   /*
> >    * These are used to set parameters in the core dumps.
> > diff --git a/arch/x86/kernel/module.c b/arch/x86/kernel/module.c
> > index 8984abd91c001..6c8b524bfbe3b 100644
> > --- a/arch/x86/kernel/module.c
> > +++ b/arch/x86/kernel/module.c
> > @@ -133,6 +133,14 @@ static int __write_relocate_add(Elf64_Shdr *sechdrs,
> >               case R_X86_64_PC32:
> >               case R_X86_64_PLT32:
> >                       val -= (u64)loc;
> > +                     if ((s64)val != *(s32 *)&val)
> > +                             goto overflow;
> > +                     size = 4;
> > +                     break;
> > +             case R_X86_64_REX_GOTPCRELX:
> > +                     val -= (u64)loc;
> > +                     if ((s64)val != *(s32 *)&val)
> > +                             goto overflow;
> >                       size = 4;
> >                       break;
>
> These two cases are the same. You probably want:
>
>                 case R_X86_64_PC32:
>                 case R_X86_64_PLT32:
>                 case R_X86_64_REX_GOTPCRELX:
>                         val -= (u64)loc;
>                         if ((s64)val != *(s32 *)&val)
>                                 goto overflow;
>                         size = 4;
>                         break;
>

You are right, I overlooked this, as well as the other
R_X86_64_REX_GOTPCRELX case above.
Ard, do you think we can relax the code handling __stack_chk_guard to
accept every R_X86_64_REX_GOTPCRELX relocation?