linux-kernel - Re: [PATCH v3 1/2] x86/AMD: Fix cpu_llc

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: <20161108083506.rvqb5h4chrcptj7d@pd.tnic>
Date:   Tue, 8 Nov 2016 09:35:06 +0100
From:   Borislav Petkov <bp@...e.de>
To:     Ingo Molnar <mingo@...nel.org>
Cc:     x86@...nel.org, Yazen Ghannam <Yazen.Ghannam@....com>,
        linux-kernel@...r.kernel.org,
        Peter Zijlstra <a.p.zijlstra@...llo.nl>
Subject: Re: [PATCH v3 1/2] x86/AMD: Fix cpu_llc_id for AMD Fam17h systems

On Tue, Nov 08, 2016 at 07:31:45AM +0100, Ingo Molnar wrote:
> So the point I tried to make is that to people doing -stable
> backporting decisions this description you just gave is much more
> valuable than the previous changelog.

Ok, how's that below? I've integrated the gist of it in the commit message:

---
From: Yazen Ghannam <Yazen.Ghannam@....com>
Date: Tue, 1 Nov 2016 11:51:02 -0500
Subject: [PATCH] x86/AMD: Fix cpu_llc_id for AMD Fam17h systems

cpu_llc_id (Last Level Cache ID) derivation on AMD Fam17h has an
underflow bug when extracting the socket_id value. It starts from 0
so subtracting 1 from it will result in an invalid value. This breaks
scheduling topology later on since the cpu_llc_id will be incorrect.

For example, the the cpu_llc_id of the *other* CPU in the loops in
set_cpu_sibling_map() underflows and we're generating the funniest
thread_siblings masks and then when I run 8 threads of nbench, they get
spread around the LLC domains in a very strange pattern which doesn't
give you the normal scheduling spread one would expect for performance.

Other things like EDAC use cpu_llc_id so they will be b0rked too.

So, the APIC ID is preset in APICx020 for bits 3 and above: they contain
the core complex, node and socket IDs.

The LLC is at the core complex level so we can find a unique cpu_llc_id
by right shifting the APICID by 3 because then the least significant bit
will be the Core Complex ID.

Signed-off-by: Yazen Ghannam <Yazen.Ghannam@....com>
Tested-by: Borislav Petkov <bp@...e.de>
Cc: Aravind Gopalakrishnan  <aravindksg.lkml@...il.com>
Cc: <stable@...r.kernel.org> # v4.4..
Cc: x86-ml <x86@...nel.org>
Link: http://lkml.kernel.org/r/1478019063-2632-1-git-send-email-Yazen.Ghannam@amd.com
Fixes: 3849e91f571d ("x86/AMD: Fix last level cache topology for AMD Fam17h systems")
[ Boris: cleanup and extend commit message. ]
Signed-off-by: Borislav Petkov <bp@...e.de>
---
 arch/x86/kernel/cpu/amd.c | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index b81fe2d63e15..1e81a37c034e 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -347,7 +347,6 @@ static void amd_detect_cmp(struct cpuinfo_x86 *c)
 #ifdef CONFIG_SMP
 	unsigned bits;
 	int cpu = smp_processor_id();
-	unsigned int socket_id, core_complex_id;

 	bits = c->x86_coreid_bits;
 	/* Low order bits define the core id (index of core in socket) */
@@ -365,10 +364,7 @@ static void amd_detect_cmp(struct cpuinfo_x86 *c)
 	 if (c->x86 != 0x17 || !cpuid_edx(0x80000006))
 		return;

-	socket_id	= (c->apicid >> bits) - 1;
-	core_complex_id	= (c->apicid & ((1 << bits) - 1)) >> 3;
-
-	per_cpu(cpu_llc_id, cpu) = (socket_id << 3) | core_complex_id;
+	per_cpu(cpu_llc_id, cpu) = c->apicid >> 3;
 #endif
 }

-- 
2.10.0

SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton, HRB 21284 (AG Nürnberg)
--