linux-kernel - Re: [PATCH V2 7/9] x86, amd: Factor out number of nodes calculation

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]

Date:	Wed, 3 Jun 2015 10:58:05 +0200
From:	Borislav Petkov <bp@...en8.de>
To:	Aravind Gopalakrishnan <Aravind.Gopalakrishnan@....com>
Cc:	dougthompson@...ssion.com, mchehab@....samsung.com,
	linux-edac@...r.kernel.org, linux-kernel@...r.kernel.org,
	x86@...nel.org, Thomas Gleixner <tglx@...utronix.de>,
	Ingo Molnar <mingo@...hat.com>,
	"H. Peter Anvin" <hpa@...or.com>, Borislav Petkov <bp@...e.de>,
	Jacob Shin <jacob.w.shin@...il.com>,
	Dave Hansen <dave.hansen@...ux.intel.com>,
	Andy Lutomirski <luto@...nel.org>,
	Paolo Bonzini <pbonzini@...hat.com>
Subject: Re: [PATCH V2 7/9] x86, amd: Factor out number of nodes calculation

On Tue, Jun 02, 2015 at 03:36:00PM -0500, Aravind Gopalakrishnan wrote:
> Factoring out number of nodes calculation out of amd_get_topology()
> and saving the value in a static variable.
> 
> A later patch will introduce a accessor for this value so we
> can use the information elsewhere in EDAC. The usage will be
> included in a future patch too.
> 
> While at it, remove X86_HT #ifdefs around the code block for
> amd_get_topology() and it's caller amd_detect_cmp(). Since
> CONFIG_X86_HT defaults to Y, this code is always built-in.
> Besides, amd_get_topology() extracts necessary info from
> cpuid_[eax|ebx](0x8000001e) for platforms that are not 'HT'

...

> +static void amd_set_num_nodes(void)
> +{
> +	if (cpu_has_topoext) {
> +		u32 ecx;
> +
> +		ecx = cpuid_ecx(0x8000001e);
> +		nodes_per_processor = ((ecx >> 8) & 7) + 1;
> +	} else if (static_cpu_has(X86_FEATURE_NODEID_MSR)) {
> +		u64 value;
> +
> +		rdmsrl(MSR_FAM10H_NODE_ID, value);
> +		nodes_per_processor = ((value >> 3) & 7) + 1;
> +	}
> +}
> +
>  /*
>   * Fixup core topology information for
>   * (1) AMD multi-node processors
>   *     Assumption: Number of cores in each internal node is the same.
>   * (2) AMD processors supporting compute units
>   */
> -#ifdef CONFIG_X86_HT
>  static void amd_get_topology(struct cpuinfo_x86 *c)
>  {
> -	u32 nodes, cores_per_cu = 1;
> +	u32 cores_per_cu = 1;
>  	u8 node_id;
>  	int cpu = smp_processor_id();
>  
> +	amd_set_num_nodes();

When I said, you like to complicate stuff, I wasn't joking. :-)

See below for what I actually meant. Diff is ontop of the X86_HT removal
patch from today:

---
diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index 5bd3a99dc20b..bdca795743e8 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -19,6 +19,8 @@
 
 #include "cpu.h"
 
+static u32 nodes_per_processor = 1;
+
 static inline int rdmsrl_amd_safe(unsigned msr, unsigned long long *p)
 {
 	u32 gprs[8] = { 0 };
@@ -291,7 +293,7 @@ static int nearby_node(int apicid)
 #ifdef CONFIG_SMP
 static void amd_get_topology(struct cpuinfo_x86 *c)
 {
-	u32 nodes, cores_per_cu = 1;
+	u32 cores_per_cu = 1;
 	u8 node_id;
 	int cpu = smp_processor_id();
 
@@ -300,7 +302,7 @@ static void amd_get_topology(struct cpuinfo_x86 *c)
 		u32 eax, ebx, ecx, edx;
 
 		cpuid(0x8000001e, &eax, &ebx, &ecx, &edx);
-		nodes = ((ecx >> 8) & 7) + 1;
+		nodes_per_processor = ((ecx >> 8) & 7) + 1;
 		node_id = ecx & 7;
 
 		/* get compute unit information */
@@ -311,18 +313,18 @@ static void amd_get_topology(struct cpuinfo_x86 *c)
 		u64 value;
 
 		rdmsrl(MSR_FAM10H_NODE_ID, value);
-		nodes = ((value >> 3) & 7) + 1;
+		nodes_per_processor = ((value >> 3) & 7) + 1;
 		node_id = value & 7;
 	} else
 		return;
 
 	/* fixup multi-node processor information */
-	if (nodes > 1) {
+	if (nodes_per_processor > 1) {
 		u32 cores_per_node;
 		u32 cus_per_node;
 
 		set_cpu_cap(c, X86_FEATURE_AMD_DCM);
-		cores_per_node = c->x86_max_cores / nodes;
+		cores_per_node = c->x86_max_cores / nodes_per_processor;
 		cus_per_node = cores_per_node / cores_per_cu;
 
 		/* store NodeID, use llc_shared_map to store sibling info */
-- 
Regards/Gruss,
    Boris.

ECO tip #101: Trim your mails when you reply.
--
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/