Skip to contentSkip to conversation

PPC: Fix NUMA node numbering on IBM POWER8 LE machine

PATCH New Chao Zhu 3 participants 6 messages Aug 14, 2015 – Sep 17, 2015

lore
Share & tools
lore

Chao Zhu

author
#0 lore raw

PATCH 1/1 PPC: Fix NUMA node numbering on IBM POWER8 LE machine

When Linux is running on bare metal, it gets the raw hardware information. On POWER8 little endian bare metal machine, the node number is not continuous. It will jump from 0 to other values, for example, it can be 0, 1, 16, 17. This patch modified the CONFIG_RTE_MAX_NUMA_NODES value to make dpdk work on POWER8 bare metal little endian machine.

---
 config/common_linuxapp |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
config/common_linuxapp +1 -1
index 0de43d5..82a027e 100644
@@ -98,7 +98,7 @@ CONFIG_RTE_NEXT_ABI=y
 #
 CONFIG_RTE_LIBRTE_EAL=y
 CONFIG_RTE_MAX_LCORE=128
-CONFIG_RTE_MAX_NUMA_NODES=8
+CONFIG_RTE_MAX_NUMA_NODES=32
 CONFIG_RTE_MAX_MEMSEG=256
 CONFIG_RTE_MAX_MEMZONE=2560
 CONFIG_RTE_MAX_TAILQ=32
Signature · 1 line
1.7.1
author
#1 lore raw

PATCH 1/1 PPC: Fix NUMA node numbering on IBM POWER8 LE machine

Any response of this patch?

-------- Forwarded Message --------
Subject: [dpdk-dev] [PATCH] PPC: Fix NUMA node numbering on IBM POWER8 LE machine
Date: Fri, 14 Aug 2015 20:19:48 +0800
From: Chao Zhu <[email protected]>
To: [email protected]

When Linux is running on bare metal, it gets the raw hardware information. On POWER8 little endian bare metal machine, the node number is not continuous. It will jump from 0 to other values, for example, it can be 0, 1, 16, 17. This patch modified the CONFIG_RTE_MAX_NUMA_NODES value to make dpdk work on POWER8 bare metal little endian machine.

---
 config/common_linuxapp |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
config/common_linuxapp +1 -1
index 0de43d5..82a027e 100644
@@ -98,7 +98,7 @@ CONFIG_RTE_NEXT_ABI=y
 #
 CONFIG_RTE_LIBRTE_EAL=y
 CONFIG_RTE_MAX_LCORE=128
-CONFIG_RTE_MAX_NUMA_NODES=8
+CONFIG_RTE_MAX_NUMA_NODES=32
 CONFIG_RTE_MAX_MEMSEG=256
 CONFIG_RTE_MAX_MEMZONE=2560
 CONFIG_RTE_MAX_TAILQ=32
Signature · 1 line
1.7.1
#2 lore raw
Quoted 2 lines · Chao Zhu

Any response of this patch?

Looks ok to me - pretty trivial change.

/Bruce

Quoted 39 lines

-------- Forwarded Message --------
Subject: [dpdk-dev] [PATCH] PPC: Fix NUMA node numbering on IBM POWER8 LE
machine
Date: Fri, 14 Aug 2015 20:19:48 +0800
From: Chao Zhu <[email protected]>
To: [email protected]

When Linux is running on bare metal, it gets the raw hardware information. On POWER8 little endian bare metal machine, the node number is not continuous. It will jump from 0 to other values, for example, it can be 0, 1, 16, 17. This patch modified the CONFIG_RTE_MAX_NUMA_NODES value to make dpdk work on POWER8 bare metal little endian machine.

---
 config/common_linuxapp |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
config/common_linuxapp +1 -1
index 0de43d5..82a027e 100644
@@ -98,7 +98,7 @@ CONFIG_RTE_NEXT_ABI=y
 #
 CONFIG_RTE_LIBRTE_EAL=y
 CONFIG_RTE_MAX_LCORE=128
-CONFIG_RTE_MAX_NUMA_NODES=8
+CONFIG_RTE_MAX_NUMA_NODES=32
 CONFIG_RTE_MAX_MEMSEG=256
 CONFIG_RTE_MAX_MEMZONE=2560
 CONFIG_RTE_MAX_TAILQ=32
Signature · 1 line
1.7.1
author
#3 lore raw

Actually, without this change, DPDK can't work properly on PPC64 little endian platform. It'll report "EAL: Not enough memory available! Requested: xxxMB, available: xxxMB" such kind of error. But for users, they don't know that changing the value of CONFIG_RTE_MAX_NUMA_NODES can fix this. That why I invoke this patch.

Quoted 43 lines · Bruce Richardson
Quoted 1 line · Chao Zhu

Any response of this patch?

Looks ok to me - pretty trivial change.

/Bruce

Quoted 38 lines

-------- Forwarded Message --------
Subject: [dpdk-dev] [PATCH] PPC: Fix NUMA node numbering on IBM POWER8 LE
machine
Date: Fri, 14 Aug 2015 20:19:48 +0800
From: Chao Zhu <[email protected]>
To: [email protected]

When Linux is running on bare metal, it gets the raw hardware information. On POWER8 little endian bare metal machine, the node number is not continuous. It will jump from 0 to other values, for example, it can be 0, 1, 16, 17. This patch modified the CONFIG_RTE_MAX_NUMA_NODES value to make dpdk work on POWER8 bare metal little endian machine.

---
  config/common_linuxapp |    2 +-
  1 files changed, 1 insertions(+), 1 deletions(-)
config/common_linuxapp +1 -1
index 0de43d5..82a027e 100644
@@ -98,7 +98,7 @@ CONFIG_RTE_NEXT_ABI=y
  #
  CONFIG_RTE_LIBRTE_EAL=y
  CONFIG_RTE_MAX_LCORE=128
-CONFIG_RTE_MAX_NUMA_NODES=8
+CONFIG_RTE_MAX_NUMA_NODES=32
  CONFIG_RTE_MAX_MEMSEG=256
  CONFIG_RTE_MAX_MEMZONE=2560
  CONFIG_RTE_MAX_TAILQ=32
Signature · 1 line
1.7.1
#4 lore raw

Hello Chao,

Quoted 5 lines · Chao Zhu

Actually, without this change, DPDK can't work properly on PPC64 little endian platform. It'll report "EAL: Not enough memory available! Requested:
xxxMB, available: xxxMB" such kind of error. But for users, they don't know that changing the value of CONFIG_RTE_MAX_NUMA_NODES can fix this. That why I invoke this patch.

Sorry, I forgot to reply in this thread.

Well, to me, this is a workaround.
Yes it will work, but what happens if tomorrow we have some hardware that tells us that it has some numa node which 60000 index ?

I think we need a rework in eal to proerly handle this, like I said in this mail :
http://dpdk.org/ml/archives/dev/2015-September/023630.html

Do you think you can look into this ?

Thanks.

Signature · 1 line
David Marchand
author
#5 lore raw

David,

Let me take a look.

Quoted 26 lines · David Marchand

Hello Chao,

On Wed, Sep 16, 2015 at 4:02 AM, Chao Zhu <[email protected] <mailto:[email protected]>> wrote:

Actually, without this change, DPDK can't work properly on PPC64 little endian platform. It'll report "EAL: Not enough memory available! Requested: xxxMB, available: xxxMB" such kind of error. But for users, they don't know that changing the value of CONFIG_RTE_MAX_NUMA_NODES can fix this. That why I invoke this patch.

Sorry, I forgot to reply in this thread.

Well, to me, this is a workaround.
Yes it will work, but what happens if tomorrow we have some hardware that tells us that it has some numa node which 60000 index ?

I think we need a rework in eal to proerly handle this, like I said in this mail :
http://dpdk.org/ml/archives/dev/2015-September/023630.html

Do you think you can look into this ?

Thanks.

Signature · 1 line
David Marchand