From mboxrd@z Thu Jan  1 00:00:00 1970
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
        id S1750879AbeAQISu (ORCPT <rfc822;ralf@linux-mips.org> + 1 other);
        Wed, 17 Jan 2018 03:18:50 -0500
Received: from merlin.infradead.org ([205.233.59.134]:58274 "EHLO
        merlin.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
        with ESMTP id S1750794AbeAQISt (ORCPT
        <rfc822;linux-kernel@vger.kernel.org>);
        Wed, 17 Jan 2018 03:18:49 -0500
Date: Wed, 17 Jan 2018 09:18:33 +0100
From: Peter Zijlstra <peterz@infradead.org>
To: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: David Woodhouse <dwmw2@infradead.org>,
        linux-kernel@vger.kernel.org, Dave Hansen <dave.hansen@intel.com>,
        Ashok Raj <ashok.raj@intel.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Tim Chen <tim.c.chen@linux.intel.com>,
        Andy Lutomirski <luto@kernel.org>,
        Linus Torvalds <torvalds@linux-foundation.org>,
        Greg KH <gregkh@linuxfoundation.org>,
        Andrea Arcangeli <aarcange@redhat.com>,
        Andi Kleen <ak@linux.intel.com>,
        Arjan Van De Ven <arjan.van.de.ven@intel.com>,
        Dan Williams <dan.j.williams@intel.com>,
        Paolo Bonzini <pbonzini@redhat.com>,
        Jun Nakajima <jun.nakajima@intel.com>,
        Asit Mallick <asit.k.mallick@intel.com>
Subject: Re: [PATCH v2 10/10] objtool: More complex static jump implementation
Message-ID: <20180117081833.GK2228@hirez.programming.kicks-ass.net>
References: <20180116142825.376986833@infradead.org>
 <20180116143241.478561014@infradead.org>
 <20180117030531.y3dggogw5hmnr3n4@treble>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20180117030531.y3dggogw5hmnr3n4@treble>
User-Agent: Mutt/1.9.2 (2017-12-15)
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org
Return-Path: <linux-kernel-owner@vger.kernel.org>

On Tue, Jan 16, 2018 at 09:05:31PM -0600, Josh Poimboeuf wrote:
> On Tue, Jan 16, 2018 at 03:28:35PM +0100, Peter Zijlstra wrote:
> > When using something like:
> > 
> >   -#define sched_feat(x) (static_branch_##x(&sched_feat_keys[__SCHED_FEAT_##x]))
> >   +#define sched_feat(x) (static_branch_##x(&sched_feat_keys[__SCHED_FEAT_##x]) && \
> >   +			(arch_static_assert(), true))
> > 
> > we get an objtool assertion fail like:
> > 
> > kernel/sched/fair.o: warning: objtool: hrtick_update()+0xd: static assert FAIL
> > 
> > where:
> > 
> > 0000000000001140 <hrtick_update>:
> >     1140:       0f 1f 44 00 00          nopl   0x0(%rax,%rax,1)
> >     1145:       c3                      retq
> >     1146:       48 8b b7 30 09 00 00    mov    0x930(%rdi),%rsi
> >     114d:       8b 87 d8 09 00 00       mov    0x9d8(%rdi),%eax
> >     1153:       48 0f a3 05 00 00 00    bt     %rax,0x0(%rip)        # 115b <hrtick_update+0x1b>
> >     115a:       00
> >                         1157: R_X86_64_PC32     __cpu_active_mask-0x4
> > 
> > and:
> > 
> > RELOCATION RECORDS FOR [__jump_table]:
> > 0000000000000150 R_X86_64_64       .text+0x0000000000001140
> > 0000000000000158 R_X86_64_64       .text+0x0000000000001146
> > 
> > RELOCATION RECORDS FOR [.discard.jump_assert]:
> > 0000000000000028 R_X86_64_64       .text+0x000000000000114d
> > 
> > IOW, GCC managed to place the assertion 1 instruction _after_ the
> > static jump target.
> > 
> > So while the code generation is fine, the assertion gets placed wrong.
> > We can 'fix' this by not only considering the immediate static jump
> > locations but also all the unconditional code after it, terminating
> > the basic block on any unconditional instruction or branch entry
> > point.
> > 
> > Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> 
> This is pretty similar to something I've been wanting to do, which is to
> track all basic blocks.  But this is fine enough for now.

Right, if we'd have that, we could just mark the entire block as
'static' and be done with it.

> One nit, can you rename "branch_target" to "jump_dest" for consistency
> with the existing naming?

Can't, insn->jump_dest already exists.