[PATCH v2 0/4] Fixing SMP config for DeviceTree platforms

[PATCH v2 0/4] Fixing SMP config for DeviceTree platforms

[Saurabh Sengar]

Original motivation for these changes was to fix an SMT warning observed
in Hyper-V VTL platform. Check V1 for more details:
https://lore.kernel.org/lkml/1710265404-23146-1-git-send-email-ssengar@linux.microsoft.com/

But on further review it was observed that x86 DeviceTree config has
modified recently, and there is opportunity to improve the existing
smp parsing for DeviceTree.

https://lore.kernel.org/lkml/20230807135027.064321764@linutronix.de/

Changes are small but I have logically separated them into four patches.
I am willing to merge them if there are too many 

[V2]
- Modified the approch how to fix the SMT issue, this is patch 3
  in this series. V1 for it is mentioned in above link.
- Added 2 more patches which are related to new SMP config parsing.
  These are patches 2 and 4 in this series.
- Fixed an issue with hv_vtl platform how it should use the smp
  parsing for DeviceTree with a fixes tag.

Saurabh Sengar (4):
  x86/hyperv/vtl: Correct parse_smp_cfg assignment
  x86/of: Set the parse_smp_cfg for all the DeviceTree platforms by
    default
  x86/of: Map NUMA node to CPUs as per DeviceTree
  x86/of: Change x86_dtb_parse_smp_config to static

 arch/x86/hyperv/hv_vtl.c          |  1 -
 arch/x86/include/asm/prom.h       |  9 ++-------
 arch/x86/kernel/devicetree.c      | 24 ++++++++++++++----------
 arch/x86/platform/ce4100/ce4100.c |  1 -
 4 files changed, 16 insertions(+), 19 deletions(-)

-- 
2.34.1

[PATCH v2 1/4] x86/hyperv/vtl: Correct parse_smp_cfg assignment

[Saurabh Sengar]

VTL platform uses DeviceTree for fetching SMP configuration, assign
the correct parsing function 'x86_dtb_parse_smp_config' for it to
parse_smp_cfg.

Fixes: c22e19cd2c8a ("x86/hyperv/vtl: Prepare for separate mpparse callbacks")
Signed-off-by: Saurabh Sengar <ssengar@linux.microsoft.com>
---
 arch/x86/hyperv/hv_vtl.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/hyperv/hv_vtl.c b/arch/x86/hyperv/hv_vtl.c
index 5c7de79423b8..3efd0e03b916 100644
--- a/arch/x86/hyperv/hv_vtl.c
+++ b/arch/x86/hyperv/hv_vtl.c
@@ -34,7 +34,7 @@ void __init hv_vtl_init_platform(void)
 	/* Avoid searching for BIOS MP tables */
 	x86_init.mpparse.find_mptable = x86_init_noop;
 	x86_init.mpparse.early_parse_smp_cfg = x86_init_noop;
-	x86_init.mpparse.parse_smp_cfg = x86_init_noop;
+	x86_init.mpparse.parse_smp_cfg = x86_dtb_parse_smp_config;
 
 	x86_platform.get_wallclock = get_rtc_noop;
 	x86_platform.set_wallclock = set_rtc_noop;
-- 
2.34.1

[PATCH v2 2/4] x86/of: Set the parse_smp_cfg for all the DeviceTree platforms by default

[Saurabh Sengar]

x86_dtb_parse_smp_config must be set by DeviceTree platform for
parsing SMP configuration. Set the parse_smp_cfg pointer to
x86_dtb_parse_smp_config by default so that all the dtb platforms
need not to assign it explicitly. Today there are only two platforms
using DeviceTree in x86, ce4100 and hv_vtl. Remove the explicit
assignment of x86_dtb_parse_smp_config function from these.

Signed-off-by: Saurabh Sengar <ssengar@linux.microsoft.com>
---
 arch/x86/hyperv/hv_vtl.c          | 1 -
 arch/x86/include/asm/prom.h       | 7 ++-----
 arch/x86/kernel/devicetree.c      | 6 ++++--
 arch/x86/platform/ce4100/ce4100.c | 1 -
 4 files changed, 6 insertions(+), 9 deletions(-)

diff --git a/arch/x86/hyperv/hv_vtl.c b/arch/x86/hyperv/hv_vtl.c
index 3efd0e03b916..04775346369c 100644
--- a/arch/x86/hyperv/hv_vtl.c
+++ b/arch/x86/hyperv/hv_vtl.c
@@ -34,7 +34,6 @@ void __init hv_vtl_init_platform(void)
 	/* Avoid searching for BIOS MP tables */
 	x86_init.mpparse.find_mptable = x86_init_noop;
 	x86_init.mpparse.early_parse_smp_cfg = x86_init_noop;
-	x86_init.mpparse.parse_smp_cfg = x86_dtb_parse_smp_config;
 
 	x86_platform.get_wallclock = get_rtc_noop;
 	x86_platform.set_wallclock = set_rtc_noop;
diff --git a/arch/x86/include/asm/prom.h b/arch/x86/include/asm/prom.h
index 043758a2e627..02644e010514 100644
--- a/arch/x86/include/asm/prom.h
+++ b/arch/x86/include/asm/prom.h
@@ -24,18 +24,15 @@ extern u64 initial_dtb;
 extern void add_dtb(u64 data);
 void x86_of_pci_init(void);
 void x86_dtb_parse_smp_config(void);
+void x86_flattree_get_config(void);
 #else
 static inline void add_dtb(u64 data) { }
 static inline void x86_of_pci_init(void) { }
 static inline void x86_dtb_parse_smp_config(void) { }
+static inline void x86_flattree_get_config(void) { }
 #define of_ioapic 0
 #endif
 
-#ifdef CONFIG_OF_EARLY_FLATTREE
-void x86_flattree_get_config(void);
-#else
-static inline void x86_flattree_get_config(void) { }
-#endif
 extern char cmd_line[COMMAND_LINE_SIZE];
 
 #endif /* __ASSEMBLY__ */
diff --git a/arch/x86/kernel/devicetree.c b/arch/x86/kernel/devicetree.c
index 003e0298f46a..0d3a50e8395d 100644
--- a/arch/x86/kernel/devicetree.c
+++ b/arch/x86/kernel/devicetree.c
@@ -277,9 +277,9 @@ static void __init dtb_apic_setup(void)
 	dtb_ioapic_setup();
 }
 
-#ifdef CONFIG_OF_EARLY_FLATTREE
 void __init x86_flattree_get_config(void)
 {
+#ifdef CONFIG_OF_EARLY_FLATTREE
 	u32 size, map_len;
 	void *dt;
 
@@ -301,8 +301,10 @@ void __init x86_flattree_get_config(void)
 
 	if (initial_dtb)
 		early_memunmap(dt, map_len);
-}
 #endif
+	if (of_have_populated_dt())
+		x86_init.mpparse.parse_smp_cfg = x86_dtb_parse_smp_config;
+}
 
 void __init x86_dtb_parse_smp_config(void)
 {
diff --git a/arch/x86/platform/ce4100/ce4100.c b/arch/x86/platform/ce4100/ce4100.c
index f32451bdcfdd..f8126821a94d 100644
--- a/arch/x86/platform/ce4100/ce4100.c
+++ b/arch/x86/platform/ce4100/ce4100.c
@@ -139,7 +139,6 @@ void __init x86_ce4100_early_setup(void)
 	x86_init.resources.probe_roms		= x86_init_noop;
 	x86_init.mpparse.find_mptable		= x86_init_noop;
 	x86_init.mpparse.early_parse_smp_cfg	= x86_init_noop;
-	x86_init.mpparse.parse_smp_cfg		= x86_dtb_parse_smp_config;
 	x86_init.pci.init			= ce4100_pci_init;
 	x86_init.pci.init_irq			= sdv_pci_init;
 
-- 
2.34.1

[PATCH v2 3/4] x86/of: Map NUMA node to CPUs as per DeviceTree

[Saurabh Sengar]

Currently for DeviceTree bootup, x86 code does the default mapping of
CPUs to NUMA, which is wrong. This can cause incorrect mapping and WARN
on a SMT enabled systems like below:

CPU #1's smt-sibling CPU #0 is not on the same node! [node: 1 != 0]. Ignoring dependency.
WARNING: CPU: 1 PID: 0 at topology_sane.isra.0+0x5c/0x6d
match_smt+0xf6/0xfc
set_cpu_sibling_map.cold+0x24f/0x512
start_secondary+0x5c/0x110

Add the set_apicid_to_node() function in dtb_cpu_setup() for allowing
the NUMA to CPU mapping for DeviceTree platforms.

Signed-off-by: Saurabh Sengar <ssengar@linux.microsoft.com>
---
 arch/x86/kernel/devicetree.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/arch/x86/kernel/devicetree.c b/arch/x86/kernel/devicetree.c
index 0d3a50e8395d..b93ce8a39ff7 100644
--- a/arch/x86/kernel/devicetree.c
+++ b/arch/x86/kernel/devicetree.c
@@ -24,6 +24,7 @@
 #include <asm/pci_x86.h>
 #include <asm/setup.h>
 #include <asm/i8259.h>
+#include <asm/numa.h>
 #include <asm/prom.h>
 
 __initdata u64 initial_dtb;
@@ -137,6 +138,7 @@ static void __init dtb_cpu_setup(void)
 			continue;
 		}
 		topology_register_apic(apic_id, CPU_ACPIID_INVALID, true);
+		set_apicid_to_node(apic_id, of_node_to_nid(dn));
 	}
 }
 
-- 
2.34.1

[PATCH v2 4/4] x86/of: Change x86_dtb_parse_smp_config to static

[Saurabh Sengar]

x86_dtb_parse_smp_config is called locally only, change it to static.

Signed-off-by: Saurabh Sengar <ssengar@linux.microsoft.com>
---
 arch/x86/include/asm/prom.h  |  2 --
 arch/x86/kernel/devicetree.c | 18 +++++++++---------
 2 files changed, 9 insertions(+), 11 deletions(-)

diff --git a/arch/x86/include/asm/prom.h b/arch/x86/include/asm/prom.h
index 02644e010514..365798cb4408 100644
--- a/arch/x86/include/asm/prom.h
+++ b/arch/x86/include/asm/prom.h
@@ -23,12 +23,10 @@ extern int of_ioapic;
 extern u64 initial_dtb;
 extern void add_dtb(u64 data);
 void x86_of_pci_init(void);
-void x86_dtb_parse_smp_config(void);
 void x86_flattree_get_config(void);
 #else
 static inline void add_dtb(u64 data) { }
 static inline void x86_of_pci_init(void) { }
-static inline void x86_dtb_parse_smp_config(void) { }
 static inline void x86_flattree_get_config(void) { }
 #define of_ioapic 0
 #endif
diff --git a/arch/x86/kernel/devicetree.c b/arch/x86/kernel/devicetree.c
index b93ce8a39ff7..8e3c53b4d070 100644
--- a/arch/x86/kernel/devicetree.c
+++ b/arch/x86/kernel/devicetree.c
@@ -279,6 +279,15 @@ static void __init dtb_apic_setup(void)
 	dtb_ioapic_setup();
 }
 
+static void __init x86_dtb_parse_smp_config(void)
+{
+	if (!of_have_populated_dt())
+		return;
+
+	dtb_setup_hpet();
+	dtb_apic_setup();
+}
+
 void __init x86_flattree_get_config(void)
 {
 #ifdef CONFIG_OF_EARLY_FLATTREE
@@ -307,12 +316,3 @@ void __init x86_flattree_get_config(void)
 	if (of_have_populated_dt())
 		x86_init.mpparse.parse_smp_cfg = x86_dtb_parse_smp_config;
 }
-
-void __init x86_dtb_parse_smp_config(void)
-{
-	if (!of_have_populated_dt())
-		return;
-
-	dtb_setup_hpet();
-	dtb_apic_setup();
-}
-- 
2.34.1

Re: [PATCH v2 1/4] x86/hyperv/vtl: Correct parse_smp_cfg assignment

[Wei Liu]

On Tue, Apr 02, 2024 at 07:40:27AM -0700, Saurabh Sengar wrote:
> VTL platform uses DeviceTree for fetching SMP configuration, assign
> the correct parsing function 'x86_dtb_parse_smp_config' for it to
> parse_smp_cfg.
> 
> Fixes: c22e19cd2c8a ("x86/hyperv/vtl: Prepare for separate mpparse callbacks")
> Signed-off-by: Saurabh Sengar <ssengar@linux.microsoft.com>

Acked-by: Wei Liu <wei.liu@kernel.org>