2010-07-16 15:11:17

by Cliff Wickman

[permalink] [raw]
Subject: [PATCH] x86, UV: initialize BAU MMRs only on hubs with cpus


Remove the initialization of MMRs UVH_LB_BAU_SB_ACTIVATION_CONTROL and
UVH_BAU_DATA_BROADCAST on UV hubs that have no active cpus.
Such initialization on hubs with no active cpus would result in a kernel
page fault.

This is not of real high priority, because we don't have any such systems
(with UV hubs that have no active cpus). But they will be coming.

Diffed against 2.6.35-rc4
Signed-off-by: Cliff Wickman <[email protected]>
---
arch/x86/kernel/tlb_uv.c | 16 ++++++++++------
1 file changed, 10 insertions(+), 6 deletions(-)

Index: linux/arch/x86/kernel/tlb_uv.c
===================================================================
--- linux.orig/arch/x86/kernel/tlb_uv.c
+++ linux/arch/x86/kernel/tlb_uv.c
@@ -1634,12 +1634,16 @@ static int __init uv_bau_init(void)
alloc_intr_gate(vector, uv_bau_message_intr1);

for_each_possible_blade(uvhub) {
- pnode = uv_blade_to_pnode(uvhub);
- /* INIT the bau */
- uv_write_global_mmr64(pnode, UVH_LB_BAU_SB_ACTIVATION_CONTROL,
- ((unsigned long)1 << 63));
- mmr = 1; /* should be 1 to broadcast to both sockets */
- uv_write_global_mmr64(pnode, UVH_BAU_DATA_BROADCAST, mmr);
+ if (uv_blade_nr_possible_cpus(uvhub)) {
+ pnode = uv_blade_to_pnode(uvhub);
+ /* INIT the bau */
+ uv_write_global_mmr64(pnode,
+ UVH_LB_BAU_SB_ACTIVATION_CONTROL,
+ ((unsigned long)1 << 63));
+ mmr = 1; /* should be 1 to broadcast to both sockets */
+ uv_write_global_mmr64(pnode, UVH_BAU_DATA_BROADCAST,
+ mmr);
+ }
}

return 0;


2010-07-17 10:13:25

by Ingo Molnar

[permalink] [raw]
Subject: Re: [PATCH] x86, UV: initialize BAU MMRs only on hubs with cpus


* Cliff Wickman <[email protected]> wrote:

> Remove the initialization of MMRs UVH_LB_BAU_SB_ACTIVATION_CONTROL and
> UVH_BAU_DATA_BROADCAST on UV hubs that have no active cpus.
> Such initialization on hubs with no active cpus would result in a kernel
> page fault.

Applied to tip:x86/uv, thanks.

Ingo

2010-07-17 11:13:50

by Cliff Wickman

[permalink] [raw]
Subject: [tip:x86/uv] x86, UV: Initialize BAU MMRs only on hubs with cpus

Commit-ID: 93a7ca0c3ebe5d931126f1fb732cb9c4518383d4
Gitweb: http://git.kernel.org/tip/93a7ca0c3ebe5d931126f1fb732cb9c4518383d4
Author: Cliff Wickman <[email protected]>
AuthorDate: Fri, 16 Jul 2010 10:11:21 -0500
Committer: Ingo Molnar <[email protected]>
CommitDate: Sat, 17 Jul 2010 12:11:48 +0200

x86, UV: Initialize BAU MMRs only on hubs with cpus

Remove the initialization of MMRs
UVH_LB_BAU_SB_ACTIVATION_CONTROL and UVH_BAU_DATA_BROADCAST on
UV hubs that have no active cpus. Such initialization on hubs
with no active cpus would result in a kernel page fault.

This is not of real high priority, because we don't have any
such systems (with UV hubs that have no active cpus). But they
will be coming.

Signed-off-by: Cliff Wickman <[email protected]>
LKML-Reference: <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
---
arch/x86/kernel/tlb_uv.c | 16 ++++++++++------
1 files changed, 10 insertions(+), 6 deletions(-)

diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c
index abf3c31..59efb53 100644
--- a/arch/x86/kernel/tlb_uv.c
+++ b/arch/x86/kernel/tlb_uv.c
@@ -1635,12 +1635,16 @@ static int __init uv_bau_init(void)
alloc_intr_gate(vector, uv_bau_message_intr1);

for_each_possible_blade(uvhub) {
- pnode = uv_blade_to_pnode(uvhub);
- /* INIT the bau */
- uv_write_global_mmr64(pnode, UVH_LB_BAU_SB_ACTIVATION_CONTROL,
- ((unsigned long)1 << 63));
- mmr = 1; /* should be 1 to broadcast to both sockets */
- uv_write_global_mmr64(pnode, UVH_BAU_DATA_BROADCAST, mmr);
+ if (uv_blade_nr_possible_cpus(uvhub)) {
+ pnode = uv_blade_to_pnode(uvhub);
+ /* INIT the bau */
+ uv_write_global_mmr64(pnode,
+ UVH_LB_BAU_SB_ACTIVATION_CONTROL,
+ ((unsigned long)1 << 63));
+ mmr = 1; /* should be 1 to broadcast to both sockets */
+ uv_write_global_mmr64(pnode, UVH_BAU_DATA_BROADCAST,
+ mmr);
+ }
}

return 0;