[PATCH 0/4] x86/platform/UV: SGI UV4 Kernel Fixes

[PATCH 0/4] x86/platform/UV: SGI UV4 Kernel Fixes

[Mike Travis]

A few problems have been found as the UV4 BIOS and hardware is being
developed and tested.  These patches fix those problems.

Patch1: uv4_fix_gam_range_lookup_table
  - Fixes a problem with building the UV4 GAM range table

Patch2: uv4_fix_bios_init_error
  - Fixes a panic associated with the UV4 system table

Patch3: uv4_remove_uvsystab_pxm
  - Fixes a problem with incorrect data from UV4 BIOS

Patch4: uv4_disable_is_uv_system
  - Fixes a panic using the RHEL7 kdump kernel

-- 

[PATCH 1/4] x86/platform/UV: Fix problem with UV4 Socket IDs not being contiguous

[Mike Travis]

[-- Attachment #1: uv4_fix_gam_range_lookup_table --]
[-- Type: text/plain, Size: 1680 bytes --]

The UV4 Socket IDs are not guaranteed to equate to Node values which
can cause the GAM (Global Addressable Memory) table lookups to fail.
Fix this by using an independent index into the GAM table instead of
the Socket ID to reference the base address.

Reviewed-by: Dimitri Sivanich <sivanich@sgi.com>
Reviewed-by: Nathan Zimmer <nzimmer@sgi.com>
Tested-by: Frank Ramsay <framsay@sgi.com>
Tested-by: John Estabrook <estabrook@sgi.com>
Signed-off-by: Mike Travis <travis@sgi.com>
---
 arch/x86/kernel/apic/x2apic_uv_x.c |    9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

--- linux-3.12.orig/arch/x86/kernel/apic/x2apic_uv_x.c
+++ linux-3.12/arch/x86/kernel/apic/x2apic_uv_x.c
@@ -325,7 +325,7 @@ static __init void build_uv_gr_table(voi
 	struct uv_gam_range_entry *gre = uv_gre_table;
 	struct uv_gam_range_s *grt;
 	unsigned long last_limit = 0, ram_limit = 0;
-	int bytes, i, sid, lsid = -1;
+	int bytes, i, sid, lsid = -1, indx = 0, lindx = -1;
 
 	if (!gre)
 		return;
@@ -356,11 +356,12 @@ static __init void build_uv_gr_table(voi
 		}
 		sid = gre->sockid - _min_socket;
 		if (lsid < sid) {		/* new range */
-			grt = &_gr_table[sid];
-			grt->base = lsid;
+			grt = &_gr_table[indx];
+			grt->base = lindx;
 			grt->nasid = gre->nasid;
 			grt->limit = last_limit = gre->limit;
 			lsid = sid;
+			lindx = indx++;
 			continue;
 		}
 		if (lsid == sid && !ram_limit) {	/* update range */
@@ -371,7 +372,7 @@ static __init void build_uv_gr_table(voi
 		}
 		if (!ram_limit) {		/* non-contiguous ram range */
 			grt++;
-			grt->base = sid - 1;
+			grt->base = lindx;
 			grt->nasid = gre->nasid;
 			grt->limit = last_limit = gre->limit;
 			continue;

-- 

[PATCH 2/4] x86/platform/UV: Fix problem with bad UV4 EFI System Table causing panic

[Mike Travis]

[-- Attachment #1: uv4_fix_bios_init_error --]
[-- Type: text/plain, Size: 996 bytes --]

Fix a problem that occurs if for some reason the UV4 EFI System Table
is not available, the check inadvertantly can cause a panic.

Reviewed-by: Dimitri Sivanich <sivanich@sgi.com>
Reviewed-by: Nathan Zimmer <nzimmer@sgi.com>
Tested-by: Frank Ramsay <framsay@sgi.com>
Tested-by: John Estabrook <estabrook@sgi.com>
Signed-off-by: Mike Travis <travis@sgi.com>
---
 arch/x86/platform/uv/bios_uv.c |    7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

--- linux-3.12.orig/arch/x86/platform/uv/bios_uv.c
+++ linux-3.12/arch/x86/platform/uv/bios_uv.c
@@ -201,11 +201,12 @@ void uv_bios_init(void)
 	}
 
 	if (uv_systab->revision >= UV_SYSTAB_VERSION_UV4) {
+		int size = uv_systab->size;
+
 		iounmap(uv_systab);
-		uv_systab = ioremap(efi.uv_systab, uv_systab->size);
+		uv_systab = ioremap(efi.uv_systab, size);
 		if (!uv_systab) {
-			pr_err("UV: UVsystab: ioremap(%d) failed!\n",
-				uv_systab->size);
+			pr_err("UV: UVsystab: ioremap(%d) failed!\n", size);
 			return;
 		}
 	}

-- 

Re: [PATCH 2/4] x86/platform/UV: Fix problem with bad UV4 EFI System Table causing panic

[Thomas Gleixner]

On Wed, 27 Jul 2016, Mike Travis wrote:

> Fix a problem that occurs if for some reason the UV4 EFI System Table
> is not available, the check inadvertantly can cause a panic.
> 
> Reviewed-by: Dimitri Sivanich <sivanich@sgi.com>
> Reviewed-by: Nathan Zimmer <nzimmer@sgi.com>
> Tested-by: Frank Ramsay <framsay@sgi.com>
> Tested-by: John Estabrook <estabrook@sgi.com>
> Signed-off-by: Mike Travis <travis@sgi.com>
> ---
>  arch/x86/platform/uv/bios_uv.c |    7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> --- linux-3.12.orig/arch/x86/platform/uv/bios_uv.c
> +++ linux-3.12/arch/x86/platform/uv/bios_uv.c
> @@ -201,11 +201,12 @@ void uv_bios_init(void)
>  	}
>  
>  	if (uv_systab->revision >= UV_SYSTAB_VERSION_UV4) {
> +		int size = uv_systab->size;
> +
>  		iounmap(uv_systab);
> -		uv_systab = ioremap(efi.uv_systab, uv_systab->size);

I think the changelog is bogus. What's happening here is a classic use after
unmap, which you avoid by this change.

> +		uv_systab = ioremap(efi.uv_systab, size);

Hmm?

	tglx

Re: [PATCH 2/4] x86/platform/UV: Fix problem with bad UV4 EFI System Table causing panic

[Mike Travis]

On 7/29/2016 4:17 AM, Thomas Gleixner wrote:
> On Wed, 27 Jul 2016, Mike Travis wrote:
> 
>> Fix a problem that occurs if for some reason the UV4 EFI System Table
>> is not available, the check inadvertently can cause a panic.
>>
>> Reviewed-by: Dimitri Sivanich <sivanich@sgi.com>
>> Reviewed-by: Nathan Zimmer <nzimmer@sgi.com>
>> Tested-by: Frank Ramsay <framsay@sgi.com>
>> Tested-by: John Estabrook <estabrook@sgi.com>
>> Signed-off-by: Mike Travis <travis@sgi.com>
>> ---
>>  arch/x86/platform/uv/bios_uv.c |    7 ++++---
>>  1 file changed, 4 insertions(+), 3 deletions(-)
>>
>> --- linux-3.12.orig/arch/x86/platform/uv/bios_uv.c
>> +++ linux-3.12/arch/x86/platform/uv/bios_uv.c
>> @@ -201,11 +201,12 @@ void uv_bios_init(void)
>>  	}
>>  
>>  	if (uv_systab->revision >= UV_SYSTAB_VERSION_UV4) {
>> +		int size = uv_systab->size;
>> +
>>  		iounmap(uv_systab);
>> -		uv_systab = ioremap(efi.uv_systab, uv_systab->size);
> 
> I think the changelog is bogus. What's happening here is a classic use after
> unmap, which you avoid by this change.
> 
>> +		uv_systab = ioremap(efi.uv_systab, size);
> 
> Hmm?
> 
> 	tglx
> 

You might be right, I'm not very familiar with the EFI transitions
between physical and virtual addresses.  I do know that it worked
until the later kernels and the EFI changes that have occurred.
Another engineer here (athorlton) is working that problem and he
is the one that discovered this snafu.

But yes, the original code was incorrect in any case.

Thanks,
Mike

[PATCH 3/4] x86/platform/UV: Fix problem with UV4 BIOS providing incorrect PXM values.

[Mike Travis]

[-- Attachment #1: uv4_remove_uvsystab_pxm --]
[-- Type: text/plain, Size: 4373 bytes --]

There are some circumstances where the UV4 BIOS cannot provide the
correct Proximity Node values to associate with specific Sockets and
Physical Nodes.  The decision was made to remove these values from BIOS
and for the kernel to get these values from the standard ACPI tables.

Reviewed-by: Dimitri Sivanich <sivanich@sgi.com>
Reviewed-by: Nathan Zimmer <nzimmer@sgi.com>
Tested-by: Frank Ramsay <framsay@sgi.com>
Tested-by: John Estabrook <estabrook@sgi.com>
Signed-off-by: Mike Travis <travis@sgi.com>
---
 arch/x86/include/asm/uv/bios.h     |    5 +++--
 arch/x86/kernel/apic/x2apic_uv_x.c |   28 ++++++++++------------------
 2 files changed, 13 insertions(+), 20 deletions(-)

--- linux.orig/arch/x86/include/asm/uv/bios.h
+++ linux/arch/x86/include/asm/uv/bios.h
@@ -79,7 +79,7 @@ struct uv_gam_range_entry {
 	u16	nasid;		/* HNasid */
 	u16	sockid;		/* Socket ID, high bits of APIC ID */
 	u16	pnode;		/* Index to MMR and GRU spaces */
-	u32	pxm;		/* ACPI proximity domain number */
+	u32	unused2;
 	u32	limit;		/* PA bits 56:26 (UV_GAM_RANGE_SHFT) */
 };
 
@@ -88,7 +88,8 @@ struct uv_gam_range_entry {
 #define	UV_SYSTAB_VERSION_UV4		0x400	/* UV4 BIOS base version */
 #define	UV_SYSTAB_VERSION_UV4_1		0x401	/* + gpa_shift */
 #define	UV_SYSTAB_VERSION_UV4_2		0x402	/* + TYPE_NVRAM/WINDOW/MBOX */
-#define	UV_SYSTAB_VERSION_UV4_LATEST	UV_SYSTAB_VERSION_UV4_2
+#define	UV_SYSTAB_VERSION_UV4_3		0x403	/* - GAM Range PXM Value */
+#define	UV_SYSTAB_VERSION_UV4_LATEST	UV_SYSTAB_VERSION_UV4_3
 
 #define	UV_SYSTAB_TYPE_UNUSED		0	/* End of table (offset == 0) */
 #define	UV_SYSTAB_TYPE_GAM_PARAMS	1	/* GAM PARAM conversions */
--- linux.orig/arch/x86/kernel/apic/x2apic_uv_x.c
+++ linux/arch/x86/kernel/apic/x2apic_uv_x.c
@@ -1157,19 +1157,18 @@ static void __init decode_gam_rng_tbl(un
 	for (; gre->type != UV_GAM_RANGE_TYPE_UNUSED; gre++) {
 		if (!index) {
 			pr_info("UV: GAM Range Table...\n");
-			pr_info("UV:  # %20s %14s %5s %4s %5s %3s %2s %3s\n",
+			pr_info("UV:  # %20s %14s %5s %4s %5s %3s %2s\n",
 				"Range", "", "Size", "Type", "NASID",
-				"SID", "PN", "PXM");
+				"SID", "PN");
 		}
 		pr_info(
-		"UV: %2d: 0x%014lx-0x%014lx %5luG %3d   %04x  %02x %02x %3d\n",
+		"UV: %2d: 0x%014lx-0x%014lx %5luG %3d   %04x  %02x %02x\n",
 			index++,
 			(unsigned long)lgre << UV_GAM_RANGE_SHFT,
 			(unsigned long)gre->limit << UV_GAM_RANGE_SHFT,
 			((unsigned long)(gre->limit - lgre)) >>
 				(30 - UV_GAM_RANGE_SHFT), /* 64M -> 1G */
-			gre->type, gre->nasid, gre->sockid,
-			gre->pnode, gre->pxm);
+			gre->type, gre->nasid, gre->sockid, gre->pnode);
 
 		lgre = gre->limit;
 		if (sock_min > gre->sockid)
@@ -1288,7 +1287,7 @@ static void __init build_socket_tables(v
 		_pnode_to_socket[i] = SOCK_EMPTY;
 
 	/* fill in pnode/node/addr conversion list values */
-	pr_info("UV: GAM Building socket/pnode/pxm conversion tables\n");
+	pr_info("UV: GAM Building socket/pnode conversion tables\n");
 	for (; gre->type != UV_GAM_RANGE_TYPE_UNUSED; gre++) {
 		if (gre->type == UV_GAM_RANGE_TYPE_HOLE)
 			continue;
@@ -1296,20 +1295,18 @@ static void __init build_socket_tables(v
 		if (_socket_to_pnode[i] != SOCK_EMPTY)
 			continue;	/* duplicate */
 		_socket_to_pnode[i] = gre->pnode;
-		_socket_to_node[i] = gre->pxm;
 
 		i = gre->pnode - minpnode;
 		_pnode_to_socket[i] = gre->sockid;
 
 		pr_info(
-		"UV: sid:%02x type:%d nasid:%04x pn:%02x pxm:%2d pn2s:%2x\n",
+		"UV: sid:%02x type:%d nasid:%04x pn:%02x pn2s:%2x\n",
 			gre->sockid, gre->type, gre->nasid,
 			_socket_to_pnode[gre->sockid - minsock],
-			_socket_to_node[gre->sockid - minsock],
 			_pnode_to_socket[gre->pnode - minpnode]);
 	}
 
-	/* check socket -> node values */
+	/* Set socket -> node values */
 	lnid = -1;
 	for_each_present_cpu(cpu) {
 		int nid = cpu_to_node(cpu);
@@ -1320,14 +1317,9 @@ static void __init build_socket_tables(v
 		lnid = nid;
 		apicid = per_cpu(x86_cpu_to_apicid, cpu);
 		sockid = apicid >> uv_cpuid.socketid_shift;
-		i = sockid - minsock;
-
-		if (nid != _socket_to_node[i]) {
-			pr_warn(
-			"UV: %02x: type:%d socket:%02x PXM:%02x != node:%2d\n",
-				i, sockid, gre->type, _socket_to_node[i], nid);
-			_socket_to_node[i] = nid;
-		}
+		_socket_to_node[sockid - minsock] = nid;
+		pr_info("UV: sid:%02x: apicid:%04x node:%2d\n",
+			sockid, apicid, nid);
 	}
 
 	/* Setup physical blade to pnode translation from GAM Range Table */

-- 

[PATCH 4/4] x86/platform/UV: Fix kernel panic running RHEL kdump kernel on UV

[Mike Travis]

[-- Attachment #1: uv4_disable_is_uv_system --]
[-- Type: text/plain, Size: 912 bytes --]

The latest UV kernel support panics when RHEL7 kexec's the kdump kernel
to make a dumpfile.  This patch fixes the problem by turning off all UV
support if NUMA is off.

Reviewed-by: Dimitri Sivanich <sivanich@sgi.com>
Reviewed-by: Nathan Zimmer <nzimmer@sgi.com>
Tested-by: Frank Ramsay <framsay@sgi.com>
Tested-by: John Estabrook <estabrook@sgi.com>
Signed-off-by: Mike Travis <travis@sgi.com>
---
 arch/x86/kernel/apic/x2apic_uv_x.c |    5 +++++
 1 file changed, 5 insertions(+)

--- linux-3.12.orig/arch/x86/kernel/apic/x2apic_uv_x.c
+++ linux-3.12/arch/x86/kernel/apic/x2apic_uv_x.c
@@ -223,6 +223,11 @@ static int __init uv_acpi_madt_oem_check
 	if (strncmp(oem_id, "SGI", 3) != 0)
 		return 0;
 
+	if (numa_off) {
+		pr_err("UV: NUMA is off, disabling UV support\n");
+		return 0;
+	}
+
 	/* Setup early hub type field in uv_hub_info for Node 0 */
 	uv_cpu_info->p_uv_hub_info = &uv_hub_info_node0;
 

-- 

Re: [PATCH 0/4] x86/platform/UV: SGI UV4 Kernel Fixes

[Greg KH]

On Wed, Jul 27, 2016 at 12:33:31PM -0500, Mike Travis wrote:
> 
> A few problems have been found as the UV4 BIOS and hardware is being
> developed and tested.  These patches fix those problems.
> 
> Patch1: uv4_fix_gam_range_lookup_table
>   - Fixes a problem with building the UV4 GAM range table
> 
> Patch2: uv4_fix_bios_init_error
>   - Fixes a panic associated with the UV4 system table
> 
> Patch3: uv4_remove_uvsystab_pxm
>   - Fixes a problem with incorrect data from UV4 BIOS
> 
> Patch4: uv4_disable_is_uv_system
>   - Fixes a panic using the RHEL7 kdump kernel


<formletter>

This is not the correct way to submit patches for inclusion in the
stable kernel tree.  Please read Documentation/stable_kernel_rules.txt
for how to do this properly.

</formletter>