Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755019AbcLTHF4 (ORCPT ); Tue, 20 Dec 2016 02:05:56 -0500 Received: from terminus.zytor.com ([198.137.202.10]:37822 "EHLO mail.zytor.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751245AbcLTHFw (ORCPT ); Tue, 20 Dec 2016 02:05:52 -0500 Subject: Re: [tip:x86/urgent] x86/cpu: Probe CPUID leaf 6 even when cpuid_level == 6 To: brgerst@gmail.com, luto@kernel.org, bp@alien8.de, linux-kernel@vger.kernel.org, mingo@kernel.org, tglx@linutronix.de, linux-tip-commits@vger.kernel.org References: <6ea30c0e9daec21e488b54761881a6dfcf3e04d0.1481825597.git.luto@kernel.org> From: "H. Peter Anvin" Message-ID: Date: Mon, 19 Dec 2016 23:05:07 -0800 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Thunderbird/45.5.1 MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 3122 Lines: 94 On 12/19/16 02:56, tip-bot for Andy Lutomirski wrote: > Commit-ID: 3df8d9208569ef0b2313e516566222d745f3b94b > Gitweb: http://git.kernel.org/tip/3df8d9208569ef0b2313e516566222d745f3b94b > Author: Andy Lutomirski > AuthorDate: Thu, 15 Dec 2016 10:14:42 -0800 > Committer: Thomas Gleixner > CommitDate: Mon, 19 Dec 2016 11:50:24 +0100 > > x86/cpu: Probe CPUID leaf 6 even when cpuid_level == 6 > > A typo (or mis-merge?) resulted in leaf 6 only being probed if > cpuid_level >= 7. > > Fixes: 2ccd71f1b278 ("x86/cpufeature: Move some of the scattered feature bits to x86_capability") > Signed-off-by: Andy Lutomirski > Acked-by: Borislav Petkov > Cc: Brian Gerst > Link: http://lkml.kernel.org/r/6ea30c0e9daec21e488b54761881a6dfcf3e04d0.1481825597.git.luto@kernel.org > Signed-off-by: Thomas Gleixner > > --- > arch/x86/kernel/cpu/common.c | 7 ++++--- > 1 file changed, 4 insertions(+), 3 deletions(-) > > diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c > index 1f6b50a..dc1697c 100644 > --- a/arch/x86/kernel/cpu/common.c > +++ b/arch/x86/kernel/cpu/common.c > @@ -667,13 +667,14 @@ void get_cpu_cap(struct cpuinfo_x86 *c) > c->x86_capability[CPUID_1_EDX] = edx; > } > > + /* Thermal and Power Management Leaf: level 0x00000006 (eax) */ > + if (c->cpuid_level >= 0x00000006) > + c->x86_capability[CPUID_6_EAX] = cpuid_eax(0x00000006); > + > /* Additional Intel-defined flags: level 0x00000007 */ > if (c->cpuid_level >= 0x00000007) { > cpuid_count(0x00000007, 0, &eax, &ebx, &ecx, &edx); > - > c->x86_capability[CPUID_7_0_EBX] = ebx; > - > - c->x86_capability[CPUID_6_EAX] = cpuid_eax(0x00000006); > c->x86_capability[CPUID_7_ECX] = ecx; > } Perhaps we should have something like: void cpuid_cond(u32 leaf, u32 subleaf, u32 *eax, u32 *ebx, u32 *ecx, u32 *edx) { const struct cpuinfo_x86 *c = this_cpu_ptr(&cpu_info); u32 maxleaf; int cpuid_level = c->cpuid_level; switch (leaf >> 16) { case 0x0000: maxleaf = c->cpuid_level; break; case 0x8000: maxleaf = c->extended_cpuid_level; break; case 0x4000 .. 0x40ff: /* Not ideal, can we do better? */ maxleaf = cpu_has(X86_FEATURE_HYPERVISOR) ? 0x40ffffff : 0; break; /* Add Transmeta 0x8086 and VIA/Cyrix 0xc000? */ default: /* Can we do better? */ if (c->x86_vendor_id) == X86_VENDOR_INTEL) maxleaf = 0; else maxleaf = leaf | 0xffff; break; } if (cpuid_level >= 0 && leaf >= maxleaf) { cpuid_count(leaf, subleaf, eax, ebx, ecx, edx); } else { *eax = *ebx = *ecx = *edx = 0; } } -hpa P.S. I would love to (a) move the CPUID bits into a structure instead of passing all those crazy pointers around, and (b) stop passing struct cpuinfo * around when we only use it for the current processor anyway (a lot of these functions are in fact completely invalid if we don't); we could define this_cpu_info as (*this_cpu_ptr(&cpu_info)) -- basically what I have above -- or directly use percpu functions to access these variables.