| From: |
| Peter Zijlstra <peterz-AT-infradead.org> |
| To: |
| David Woodhouse <dwmw2-AT-infradead.org>, Josh Poimboeuf <jpoimboe-AT-redhat.com> |
| Subject: |
| [PATCH v2 00/10] objtool validation of static branches and retpoline |
| Date: |
| Tue, 16 Jan 2018 15:28:25 +0100 |
| Message-ID: |
| <20180116142825.376986833@infradead.org> |
| Cc: |
| linux-kernel-AT-vger.kernel.org, Dave Hansen <dave.hansen-AT-intel.com>, Ashok Raj <ashok.raj-AT-intel.com>, Thomas Gleixner <tglx-AT-linutronix.de>, Tim Chen <tim.c.chen-AT-linux.intel.com>, Andy Lutomirski <luto-AT-kernel.org>, Linus Torvalds <torvalds-AT-linux-foundation.org>, Greg KH <gregkh-AT-linuxfoundation.org>, Andrea Arcangeli <aarcange-AT-redhat.com>, Andi Kleen <ak-AT-linux.intel.com>, Arjan Van De Ven <arjan.van.de.ven-AT-intel.com>, Dan Williams <dan.j.williams-AT-intel.com>, Paolo Bonzini <pbonzini-AT-redhat.com>, Jun Nakajima <jun.nakajima-AT-intel.com>, Asit Mallick <asit.k.mallick-AT-intel.com>, Peter Zijlstra <peterz-AT-infradead.org> |
Because dwmw2 is a wee bit paranoid and GCC has a history of being slightly
retarded at times; here are a few patches that validate it will in fact
generate sensible code for static branches.
Also, per dwmw2's request, code to validate retpoline code does not include
dynamic jump or calls.
---
arch/x86/include/asm/alternative-asm.h | 3 +-
arch/x86/include/asm/alternative.h | 7 +-
arch/x86/include/asm/cpufeature.h | 82 ++++++------
arch/x86/include/asm/jump_label.h | 23 ++++
arch/x86/kernel/head_64.S | 5 +-
scripts/Makefile.build | 4 +
tools/objtool/builtin-check.c | 5 +-
tools/objtool/builtin-orc.c | 4 +-
tools/objtool/check.c | 223 +++++++++++++++++++++++++++++++--
tools/objtool/check.h | 3 +-
tools/objtool/special.c | 13 +-
tools/objtool/special.h | 1 +
12 files changed, 318 insertions(+), 55 deletions(-)
