x86, vsmp: Remove is_vsmp_box() from apic_is_clustered_box()
PATCH 1/1 x86, vsmp: Remove is_vsmp_box() from apic_is_clustered_box()
Remove invalid code which caused TSC to be declared as "unstable" on vSMP Foundation box even if it was stable and let the kernel decide for itself.
When a vSMP Foundation box is detected, the function apic_cluster_num() counts the number of APIC clusters found. If more than one found, a multi board configuration is assumed, and TSC marked as unstable. This behavior is incorrect as vSMP Foundation may use processors from single node only, attached to memory of other nodes - and such node may have more than one APIC cluster (typically any recent intel box has more than single APIC_CLUSTERID(x)).
To fix this, we simply remove the code which detects a vSMP Foundation box and affects apic_is_clusted_box() return value. This can be done because later the kernel checks by itself if the TSC is stable using the check_tsc_sync_[source|target]() functions and marks TSC as unstable if needed.
--- arch/x86/include/asm/apic.h | 8 ------ arch/x86/kernel/apic/apic.c | 60 +------------------------------------------ 2 files changed, 1 insertion(+), 67 deletions(-)
arch/x86/include/asm/apic.h +0 -8
@@ -85,14 +85,6 @@ static inline bool apic_from_smp_config(void)#include <asm/paravirt.h> #endif -#ifdef CONFIG_X86_64 -extern int is_vsmp_box (void ); -#else -static inline int is_vsmp_box (void ) -{ -return 0 ; -} -#endif extern int setup_profiling_timer (unsigned int );static inline void native_apic_mem_write (u32 reg,u32 v)
arch/x86/kernel/apic/apic.c +1 -59
72 lines not on this page · raw message
Quoted 10 lines · Oren Twaig
Remove invalid code which caused TSC to be declared as "unstable" on vSMP Foundation box even if it was stable and let the kernel decide for itself.
When a vSMP Foundation box is detected, the function
apic_cluster_num()counts the number of APIC clusters found. If more than one found, a multi board configuration is assumed, and TSC marked as unstable. This behavior is incorrect as vSMP Foundation may use processors from single node only, attached to memory of other nodes - and such node may have more than one APIC cluster (typically any recent intel box has more than singleAPIC_CLUSTERID(x)).
Quoted 4 lines
To fix this, we simply remove the code which detects a vSMP Foundation box and affects
apic_is_clusted_box()return value. This can be done because later the kernel checks by itself if the TSC is stable using the check_tsc_sync_[source|target]() functions and marks TSC as unstable if needed.
Looks good to me. Yes the APIC cluster check is obsolete.
-andi
ping
Quoted 121 lines · Oren Twaig
Remove invalid code which caused TSC to be declared as "unstable" on vSMP Foundation box even if it was stable and let the kernel decide for itself.
When a vSMP Foundation box is detected, the function
apic_cluster_num()counts the number of APIC clusters found. If more than one found, a multi board configuration is assumed, and TSC marked as unstable. This behavior is incorrect as vSMP Foundation may use processors from single node only, attached to memory of other nodes - and such node may have more than one APIC cluster (typically any recent intel box has more than singleAPIC_CLUSTERID(x)).To fix this, we simply remove the code which detects a vSMP Foundation box and affects
apic_is_clusted_box()return value. This can be done because later the kernel checks by itself if the TSC is stable using the check_tsc_sync_[source|target]() functions and marks TSC as unstable if needed.Signed-off-by Oren Twaig <[email protected]>Acked-by Shai Fultheim <[email protected]>--- arch/x86/include/asm/apic.h | 8 ------ arch/x86/kernel/apic/apic.c | 60 +------------------------------------------ 2 files changed, 1 insertion(+), 67 deletions(-)arch/x86/include/asm/apic.h +0
-8@@ -85,14 +85,6 @@ static inline bool apic_from_smp_config(void)#include <asm/paravirt.h> #endif -#ifdef CONFIG_X86_64 -extern int is_vsmp_box (void ); -#else -static inline int is_vsmp_box (void ) -{ -return 0 ; -} -#endif extern int setup_profiling_timer (unsigned int );static inline void native_apic_mem_write (u32 reg,u32 v)arch/x86/kernel/apic/apic.c +1
-59@@ -2451,51 +2451,6 @@ static void apic_pm_activate(void) { }#ifdef CONFIG_X86_64 -static int apic_cluster_num (void ) -{ -int i, clusters, zeros; -unsigned id; -u16 *bios_cpu_apicid; -DECLARE_BITMAP (clustermap,NUM_APIC_CLUSTERS ); - - bios_cpu_apicid =early_per_cpu_ptr (x86_bios_cpu_apicid); -bitmap_zero (clustermap,NUM_APIC_CLUSTERS ); - -for (i =0 ; i < nr_cpu_ids; i++) { -/* are we being called early in kernel startup? */ -if (bios_cpu_apicid) { - id = bios_cpu_apicid[i]; - }else if (i < nr_cpu_ids) { -if (cpu_present (i)) - id =per_cpu (x86_bios_cpu_apicid, i); -else -continue ; - }else -break ; - -if (id !=BAD_APICID ) -__set_bit (APIC_CLUSTERID (id), clustermap); - } - -/* Problem: Partially populated chassis may not have CPUs in some of -* the APIC clusters they have been allocated. Only present CPUs have -* x86_bios_cpu_apicid entries, thus causing zeroes in the bitmap. -* Since clusters are allocated sequentially, count zeros only if -* they are bounded by ones. -*/ - clusters =0 ; - zeros =0 ; -for (i =0 ; i <NUM_APIC_CLUSTERS ; i++) { -if (test_bit (i, clustermap)) { - clusters +=1 + zeros; - zeros =0 ; - }else - ++zeros; - } - -return clusters; -} -static int multi_checked;static int multi;@@ -2540,20 +2495,7 @@ static void dmi_check_multi(void)int apic_is_clustered_box (void ) {dmi_check_multi (); -if (multi) -return 1 ; - -if (!is_vsmp_box ()) -return 0 ; - -/* -* ScaleMP vSMPowered boxes have one cluster per board and TSCs are -* not guaranteed to be synced between boards -*/ -if (apic_cluster_num () >1 ) -return 1 ; - -return 0 ; +return multi; }#endif
H. Peter Anvin
reviewerQuoted 1 line · Oren Twaig
ping
This patch conflicts with the changes on the tip:x86/apic branch. Could you please rebase the patch on top of that branch?
-hpa
Hi,
There was actually a problem that the mailer which changed a '\t' to spaces. This is why it conflicted with the tip:x86/apic branch. Any how, I've also tested on the x86/apic branch and everything is ok.
Sent :
"[PATCH v2] x86, vsmp: Remove is_vsmp_box() from apic_is_clustered_box()"
Sorry for any troubles.
Thanks,
Oren.
Quoted 15 lines · H. Peter Anvin
Quoted 1 line · Oren Twaig
ping
This patch conflicts with the changes on the tip:x86/apic branch. Could you please rebase the patch on top of that branch?
-hpa
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/