NUMA emulation needs to update node distance information. It did it
by remapping apicid to PXM mapping, even when amdtopology is being
used. There is no reason to go through such convolution. The generic
code has all the information necessary to transform the distance table
to the emulated nid space.
Implement generic distance table transformation in numa_emulation()
and drop private implementations in srat_64 and amdtopology_64. This
makes find_node_by_addr() and fake_physnodes() and related functions
unnecessary, drop them.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
NUMA emulation changes node mappings and thus apicid -> node mapping
needs to be updated accordingly. srat_64 and amdtopology_64 did this
separately; however, all the necessary information is the mapping from
emulated nodes to physical nodes which is available in
emu_nid_to_phys[].
Implement common __apicid_to_node[] transformation in numa_emulation()
and drop duplicate implementations.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
Node distance either used direct node comparison, ACPI PXM comparison
or ACPI SLIT table lookup. This patch implements generic node
distance handling. NUMA init methods can call numa_set_distance() to
set distance between nodes and the common __node_distance()
implementation will report the set distance.
Due to the way NUMA emulation is implemented, the generic node
distance handling is used only when emulation is not used. Later
patches will update NUMA emulation to use the generic distance
mechanism.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
With all memory configuration information now carried in numa_meminfo,
there's no need to keep mem_nodes_parsed separate. Drop it and use
numa_nodes_parsed for CPU / memory-less nodes.
A new helper numa_nodemask_from_meminfo() is added to calculate
memnode mask on the fly which is currently used to set
node_possible_map.
This simplifies NUMA init methods a bit and removes a source of
possible inconsistencies.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
It's no longer necessary to keep both cpu_nodes_parsed and
mem_nodes_parsed. In preparation for merge, rename cpu_nodes_parsed
to numa_nodes_parsed.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
numa_nodes[] doesn't carry any information which isn't present in
numa_meminfo. Each entry is simply min/max range of all the memblks
for the node. This is not only redundant but also inaccurate when
memblks for different nodes interleave - for example,
find_node_by_addr() can return the wrong nodeid.
Kill numa_nodes[] and always use numa_meminfo instead.
* nodes_cover_memory() is renamed to numa_meminfo_cover_memory() and
now operations on numa_meminfo and returns bool.
* setup_node_bootmem() needs min/max range. Compute the range on the
fly. setup_node_bootmem() invocation is restructured to use outer
loop instead of hardcoding the double invocations.
* find_node_by_addr() now operates on numa_meminfo.
* setup_physnodes() builds physnodes[] from memblks. This will go
away when emulation code is updated to use struct numa_meminfo.
This patch also makes the following misc changes.
* Clearing of nodes_add[] clearing is converted to memset().
* numa_add_memblk() in amd_numa_init() is moved down a bit for
consistency.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
srat_64.c and amdtopology_64.c had their own versions of
find_node_by_addr() which were basically the same. Add common one in
numa_64.c and remove the duplicates.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
Move the remaining memblk registration logic from acpi_scan_nodes() to
numa_register_memblks() and initmem_init().
This applies nodes_cover_memory() sanity check, memory node sorting
and node_online() checking, which were only applied to acpi, to all
init methods.
As all memblk registration is moved to common code, active range
clearing is moved to initmem_init() too and removed from bad_srat().
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
Make both amd and dummy use numa_add_memblk() to describe the detected
memory blocks. This allows initmem_init() to call
numa_register_memblk() regardless of init method in use. Drop custom
memory registration codes from amd and dummy.
After this change, memblk merge/cleanup in numa_register_memblks() is
applied to all init methods.
As this makes compute_hash_shift() and numa_register_memblks() used
only inside numa_64.c, make them static.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
Factor out memblk handling from srat_64.c into two functions in
numa_64.c. This patch doesn't introduce any behavior change. The
next patch will make all init methods use these functions.
- v2: Fixed build failure on 32bit due to misplaced NR_NODE_MEMBLKS.
Reported by Ingo.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
With common numa_nodes[], common code in numa_64.c can access it
directly. Copy directly and kill {acpi|amd}_get_nodes().
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
ACPI and amd are using separate nodes[] array. Add numa_nodes[] and
use them in all NUMA init methods. cutoff_node() cleanup is moved
from srat_64.c to numa_64.c and applied in initmem_init() regardless
of init methods.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
ACPI and amd are using separate nodes_parsed masks. Add
{cpu|mem}_nodes_parsed and use them in all NUMA init methods.
Initialization of the masks and building node_possible_map are now
handled commonly by initmem_init().
dummy_numa_init() is updated to set node 0 on both masks. While at
it, move the info messages from scan to init.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
Reorganize initmem_init() such that,
* Different NUMA init methods are iterated in a consistent way.
* Each iteration re-initializes all the parameters and different
method can be tried after a failure.
* Dummy init is handled the same as other methods.
Apart from how retry after failure, this patch doesn't change the
behavior. The call sequences are kept equivalent across the
conversion.
After the change, bad_srat() doesn't need to clear apic to node
mapping or worry about numa_off. Simplified accordingly.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
Because of the way ACPI tables are parsed, the generic
acpi_numa_init() couldn't return failure when error was detected by
arch hooks. Instead, the failure state was recorded and later arch
dependent init hook - acpi_scan_nodes() - would fail.
Wrap acpi_numa_init() with x86_acpi_numa_init() so that failure can be
indicated as return value immediately. This is in preparation for
further NUMA init cleanups.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
The functions used during NUMA initialization - *_numa_init() and
*_scan_nodes() - have different arguments and return values. Unify
them such that they all take no argument and return 0 on success and
-errno on failure. This is in preparation for further NUMA init
cleanups.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
Hotplug node handling in acpi_numa_memory_affinity_init() was
unnecessarily complicated with storing the original nodes[] entry and
restoring it afterwards. Simplify it by not modifying the nodes[]
entry for hotplug nodes from the beginning.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
Conflicts:
arch/x86/mm/numa_64.c
Merge reason: fix the conflict, update to latest -rc and pick up this
dependent fix from Yinghai:
e6d2e2b2b1: memblock: don't adjust size in memblock_find_base()
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The mapping between cpu/apicid and node is done via
apicid_to_node[] on 64bit and apicid_2_node[] +
apic->x86_32_numa_cpu_node() on 32bit. This difference makes it
difficult to further unify 32 and 64bit NUMA handling.
This patch unifies it by replacing both apicid_to_node[] and
apicid_2_node[] with __apicid_to_node[] array, which is accessed
by two accessors - set_apicid_to_node() and numa_cpu_node(). On
64bit, numa_cpu_node() always consults __apicid_to_node[]
directly while 32bit goes through apic->numa_cpu_node() method
to allow apic implementations to override it.
srat_detect_node() for amd cpus contains workaround for broken
NUMA configuration which assumes relationship between APIC ID,
HT node ID and NUMA topology. Leave it to access
__apicid_to_node[] directly as mapping through CPU might result
in undesirable behavior change. The comment is reformatted and
updated to note the ugliness.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Pekka Enberg <penberg@kernel.org>
Cc: eric.dumazet@gmail.com
Cc: yinghai@kernel.org
Cc: brgerst@gmail.com
Cc: gorcunov@gmail.com
Cc: shaohui.zheng@intel.com
Cc: rientjes@google.com
LKML-Reference: <1295789862-25482-14-git-send-email-tj@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Cc: David Rientjes <rientjes@google.com>
NUMA boot code assumes that physical node ids start at 0, but the DIMMs
that the apic id represents may not be reachable. If this is the case,
node 0 is never online and cpus never end up getting appropriately
assigned to a node. This causes the cpumask of all online nodes to be
empty and machines crash with kernel code assuming online nodes have
valid cpus.
The fix is to appropriately map all the address ranges for physical nodes
and ensure the cpu to node mapping function checks all possible nodes (up
to MAX_NUMNODES) instead of simply checking nodes 0-N, where N is the
number of physical nodes, for valid address ranges.
This requires no longer "compressing" the address ranges of nodes in the
physical node map from 0-N, but rather leave indices in physnodes[] to
represent the actual node id of the physical node. Accordingly, the
topology exported by both amd_get_nodes() and acpi_get_nodes() no longer
must return the number of nodes to iterate through; all such iterations
will now be to MAX_NUMNODES.
This change also passes the end address of system RAM (which may be
different from normal operation if mem= is specified on the command line)
before the physnodes[] array is populated. ACPI parsed nodes are
truncated to fit within the address range that respect the mem=
boundaries and even some physical nodes may become unreachable in such
cases.
When NUMA emulation does succeed, any apicid to node mapping that exists
for unreachable nodes are given default values so that proximity domains
can still be assigned. This is important for node_distance() to
function as desired.
Signed-off-by: David Rientjes <rientjes@google.com>
LKML-Reference: <alpine.DEB.2.00.1012221702090.3701@chino.kir.corp.google.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
This patch adds the equivalent of acpi_fake_nodes() for AMD Northbridge
platforms. The goal is to fake the apicid-to-node mappings for NUMA
emulation so the physical topology of the machine is correctly maintained
within the kernel.
This change also fakes proximity domains for both ACPI and k8 code so the
physical distance between emulated nodes is maintained via
node_distance(). This exports the correct distances via
/sys/devices/system/node/.../distance based on the underlying topology.
A new helper function, fake_physnodes(), is introduced to correctly
invoke the correct NUMA code to fake these two mappings based on the
system type. If there is no underlying NUMA configuration, all cpus are
mapped to node 0 for local distance.
Since acpi_fake_nodes() is no longer called with CONFIG_ACPI_NUMA, it's
prototype can be removed from the header file for such a configuration.
Signed-off-by: David Rientjes <rientjes@google.com>
LKML-Reference: <alpine.DEB.2.00.1012221701360.3701@chino.kir.corp.google.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Both acpi_get_nodes() and amd_get_nodes() are only necessary when
CONFIG_NUMA_EMU is enabled, so avoid compiling them when the option is
disabled.
Signed-off-by: David Rientjes <rientjes@google.com>
LKML-Reference: <alpine.DEB.2.00.1012221701210.3701@chino.kir.corp.google.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Recent Intel new system have different order in MADT, aka will list all thread0
at first, then all thread1.
But SRAT table still old order, it will list cpus in one socket all together.
If the user have compiled limited NR_CPUS or boot with nr_cpus=, could have missed
to put some cpus apic id to node mapping into apicid_to_node[].
for example for 4 sockets system with 64 cpus with nr_cpus=32 will get crash...
[ 9.106288] Total of 32 processors activated (136190.88 BogoMIPS).
[ 9.235021] divide error: 0000 [#1] SMP
[ 9.235315] last sysfs file:
[ 9.235481] CPU 1
[ 9.235592] Modules linked in:
[ 9.245398]
[ 9.245478] Pid: 2, comm: kthreadd Not tainted 2.6.37-rc1-tip-yh-01782-ge92ef79-dirty #274 /Sun Fire x4800
[ 9.265415] RIP: 0010:[<ffffffff81075a8f>] [<ffffffff81075a8f>] select_task_rq_fair+0x4f0/0x623
...
[ 9.645938] RIP [<ffffffff81075a8f>] select_task_rq_fair+0x4f0/0x623
[ 9.665356] RSP <ffff88103f8d1c40>
[ 9.665568] ---[ end trace 2296156d35fdfc87 ]---
So let just parse all cpu entries in SRAT.
Also add apicid checking with MAX_LOCAL_APIC, in case We could out of boundaries of
apicid_to_node[].
it fixes following bug too.
https://bugzilla.kernel.org/show_bug.cgi?id=22662
-v2: expand to 32bit according to hpa
need to add MAX_LOCAL_APIC for 32bit
Reported-and-Tested-by: Wu Fengguang <fengguang.wu@intel.com>
Reported-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Tested-by: Myron Stowe <myron.stowe@hp.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
LKML-Reference: <4D0AD486.9020704@kernel.org>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Reason for merge:
Forward-port urgent change to arch/x86/mm/srat_64.c to the memblock tree.
Resolved Conflicts:
arch/x86/mm/srat_64.c
Originally-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Russ reported SGI UV is broken recently. He said:
| The SRAT table shows that memory range is spread over two nodes.
|
| SRAT: Node 0 PXM 0 100000000-800000000
| SRAT: Node 1 PXM 1 800000000-1000000000
| SRAT: Node 0 PXM 0 1000000000-1080000000
|
|Previously, the kernel early_node_map[] would show three entries
|with the proper node.
|
|[ 0.000000] 0: 0x00100000 -> 0x00800000
|[ 0.000000] 1: 0x00800000 -> 0x01000000
|[ 0.000000] 0: 0x01000000 -> 0x01080000
|
|The problem is recent community kernel early_node_map[] shows
|only two entries with the node 0 entry overlapping the node 1
|entry.
|
| 0: 0x00100000 -> 0x01080000
| 1: 0x00800000 -> 0x01000000
After looking at the changelog, Found out that it has been broken for a while by
following commit
|commit 8716273cae
|Author: David Rientjes <rientjes@google.com>
|Date: Fri Sep 25 15:20:04 2009 -0700
|
| x86: Export srat physical topology
Before that commit, register_active_regions() is called for every SRAT memory
entry right away.
Use nodememblk_range[] instead of nodes[] in order to make sure we
capture the actual memory blocks registered with each node. nodes[]
contains an extended range which spans all memory regions associated
with a node, but that does not mean that all the memory in between are
included.
Reported-by: Russ Anderson <rja@sgi.com>
Tested-by: Russ Anderson <rja@sgi.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
LKML-Reference: <4CB27BDF.5000800@kernel.org>
Acked-by: David Rientjes <rientjes@google.com>
Cc: <stable@kernel.org> 2.6.33 .34 .35 .36
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
1.include linux/memblock.h directly. so later could reduce e820.h reference.
2 this patch is done by sed scripts mainly
-v2: use MEMBLOCK_ERROR instead of -1ULL or -1UL
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
With NUMA emulation, it's possible for a single cpu to be bound
to multiple nodes since more than one may have affinity if
allocated on a physical node that is local to the cpu.
APIC ids must therefore be mapped to the lowest node ids to
maintain generic kernel use of functions such as cpu_to_node()
that determine device affinity. For example, if a device has
proximity to physical node 1, for instance, and a cpu happens to
be mapped to a higher emulated node id 8, the proximity may not
be correctly determined by comparison in generic code even
though the cpu may be truly local and allocated on physical node 1.
When this happens, the true topology of the machine isn't
accurately represented in the emulated environment; although
this isn't critical to the system's uptime, any generic code
that is NUMA aware benefits from the physical topology being
accurately represented.
This can affect any system that maps multiple APIC ids to a
single node and is booted with numa=fake=N where N is greater
than the number of physical nodes.
Signed-off-by: David Rientjes <rientjes@google.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Suresh Siddha <suresh.b.siddha@intel.com>
LKML-Reference: <alpine.DEB.2.00.1005060224140.19473@chino.kir.corp.google.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
... i.e. when the hole between two regions isn't occupied by memory on
another node. This reduces the memory->node table size, thus reducing
cache footprint of lookups, which got increased significantly some
time ago, and things go back to how they were before that change on
the systems I looked at.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
LKML-Reference: <4BCF3230020000780003B3CA@vpn.id2.novell.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
nodes_possible_map does not currently include nodes that have SRAT
entries that are all ACPI_SRAT_MEM_HOT_PLUGGABLE since the bit is
cleared in nodes_parsed if it does not have an online address range.
Unequivocally setting the bit in nodes_parsed is insufficient since
existing code, such as acpi_get_nodes(), assumes all nodes in the map
have online address ranges. In fact, all code using nodes_parsed
assumes such nodes represent an address range of online memory.
nodes_possible_map is created by unioning nodes_parsed and
cpu_nodes_parsed; the former represents nodes with online memory and
the latter represents memoryless nodes. We now set the bit for
hotpluggable nodes in cpu_nodes_parsed so that it also gets set in
nodes_possible_map.
[ hpa: Haicheng Li points out that this makes the naming of the
variable cpu_nodes_parsed somewhat counterintuitive. However, leave
it as is in the interest of keeping the pure bug fix patch small. ]
Signed-off-by: David Rientjes <rientjes@google.com>
Tested-by: Haicheng Li <haicheng.li@linux.intel.com>
LKML-Reference: <alpine.DEB.2.00.1001201152040.30528@chino.kir.corp.google.com>
Cc: <stable@kernel.org>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Found one system that boot from socket1 instead of socket0, SRAT get rejected...
[ 0.000000] SRAT: Node 1 PXM 0 0-a0000
[ 0.000000] SRAT: Node 1 PXM 0 100000-80000000
[ 0.000000] SRAT: Node 1 PXM 0 100000000-2080000000
[ 0.000000] SRAT: Node 0 PXM 1 2080000000-4080000000
[ 0.000000] SRAT: Node 2 PXM 2 4080000000-6080000000
[ 0.000000] SRAT: Node 3 PXM 3 6080000000-8080000000
[ 0.000000] SRAT: Node 4 PXM 4 8080000000-a080000000
[ 0.000000] SRAT: Node 5 PXM 5 a080000000-c080000000
[ 0.000000] SRAT: Node 6 PXM 6 c080000000-e080000000
[ 0.000000] SRAT: Node 7 PXM 7 e080000000-10080000000
...
[ 0.000000] NUMA: Allocated memnodemap from 500000 - 701040
[ 0.000000] NUMA: Using 20 for the hash shift.
[ 0.000000] Adding active range (0, 0x2080000, 0x4080000) 0 entries of 3200 used
[ 0.000000] Adding active range (1, 0x0, 0x96) 1 entries of 3200 used
[ 0.000000] Adding active range (1, 0x100, 0x7f750) 2 entries of 3200 used
[ 0.000000] Adding active range (1, 0x100000, 0x2080000) 3 entries of 3200 used
[ 0.000000] Adding active range (2, 0x4080000, 0x6080000) 4 entries of 3200 used
[ 0.000000] Adding active range (3, 0x6080000, 0x8080000) 5 entries of 3200 used
[ 0.000000] Adding active range (4, 0x8080000, 0xa080000) 6 entries of 3200 used
[ 0.000000] Adding active range (5, 0xa080000, 0xc080000) 7 entries of 3200 used
[ 0.000000] Adding active range (6, 0xc080000, 0xe080000) 8 entries of 3200 used
[ 0.000000] Adding active range (7, 0xe080000, 0x10080000) 9 entries of 3200 used
[ 0.000000] SRAT: PXMs only cover 917504MB of your 1048566MB e820 RAM. Not used.
[ 0.000000] SRAT: SRAT not used.
the early_node_map is not sorted because node0 with non zero start come first.
so try to sort it right away after all regions are registered.
also fixs refression by 8716273c (x86: Export srat physical topology)
-v2: make it more solid to handle cross node case like node0 [0,4g), [8,12g) and node1 [4g, 8g), [12g, 16g)
-v3: update comments.
Reported-and-tested-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
LKML-Reference: <4B2579D2.3010201@kernel.org>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
* 'x86-mm-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (36 commits)
x86, mm: Correct the implementation of is_untracked_pat_range()
x86/pat: Trivial: don't create debugfs for memtype if pat is disabled
x86, mtrr: Fix sorting of mtrr after subtracting
x86: Move find_smp_config() earlier and avoid bootmem usage
x86, platform: Change is_untracked_pat_range() to bool; cleanup init
x86: Change is_ISA_range() into an inline function
x86, mm: is_untracked_pat_range() takes a normal semiclosed range
x86, mm: Call is_untracked_pat_range() rather than is_ISA_range()
x86: UV SGI: Don't track GRU space in PAT
x86: SGI UV: Fix BAU initialization
x86, numa: Use near(er) online node instead of roundrobin for NUMA
x86, numa, bootmem: Only free bootmem on NUMA failure path
x86: Change crash kernel to reserve via reserve_early()
x86: Eliminate redundant/contradicting cache line size config options
x86: When cleaning MTRRs, do not fold WP into UC
x86: remove "extern" from function prototypes in <asm/proto.h>
x86, mm: Report state of NX protections during boot
x86, mm: Clean up and simplify NX enablement
x86, pageattr: Make set_memory_(x|nx) aware of NX support
x86, sleep: Always save the value of EFER
...
Fix up conflicts (added both iommu_shutdown and is_untracked_pat_range)
to 'struct x86_platform_ops') in
arch/x86/include/asm/x86_init.h
arch/x86/kernel/x86_init.c
Make it consistent with APIC MADT print out,
for big systems APIC id in hex is more readable.
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
LKML-Reference: <4B07A739.3030104@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Add interleaved NUMA emulation support
This patch interleaves emulated nodes over the system's physical
nodes. This is required for interleave optimizations since
mempolicies, for example, operate by iterating over a nodemask and
act without knowledge of node distances. It can also be used for
testing memory latencies and NUMA bugs in the kernel.
There're a couple of ways to do this:
- divide the number of emulated nodes by the number of physical
nodes and allocate the result on each physical node, or
- allocate each successive emulated node on a different physical
node until all memory is exhausted.
The disadvantage of the first option is, depending on the asymmetry
in node capacities of each physical node, emulated nodes may
substantially differ in size on a particular physical node compared
to another.
The disadvantage of the second option is, also depending on the
asymmetry in node capacities of each physical node, there may be
more emulated nodes allocated on a single physical node as another.
This patch implements the second option; we sacrifice the
possibility that we may have slightly more emulated nodes on a
particular physical node compared to another in lieu of node size
asymmetry.
[ Note that "node capacity" of a physical node is not only a
function of its addressable range, but also is affected by
subtracting out the amount of reserved memory over that range.
NUMA emulation only deals with available, non-reserved memory
quantities. ]
We ensure there is at least a minimal amount of available memory
allocated to each node. We also make sure that at least this
amount of available memory is available in ZONE_DMA32 for any node
that includes both ZONE_DMA32 and ZONE_NORMAL.
This patch also cleans the emulation code up by no longer passing
the statically allocated struct bootnode array among the various
functions. This init.data array is not allocated on the stack since
it may be very large and thus it may be accessed at file scope.
The WARN_ON() for nodes_cover_memory() when faking proximity
domains is removed since it relies on successive nodes always
having greater start addresses than previous nodes; with
interleaving this is no longer always true.
Signed-off-by: David Rientjes <rientjes@google.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andreas Herrmann <andreas.herrmann3@amd.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Ankita Garg <ankita@in.ibm.com>
Cc: Len Brown <len.brown@intel.com>
LKML-Reference: <alpine.DEB.1.00.0909251519150.14754@chino.kir.corp.google.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This is the counterpart to "x86: export k8 physical topology" for
SRAT. It is not as invasive because the acpi code already seperates
node setup into detection and registration steps, with the
exception of registering e820 active regions in
acpi_numa_memory_affinity_init(). This is now moved to
acpi_scan_nodes() if NUMA emulation is disabled or deferred.
acpi_numa_init() now returns a value which specifies whether an
underlying SRAT was located. If so, that topology can be used by
the emulation code to interleave emulated nodes over physical nodes
or to register the nodes for ACPI.
acpi_get_nodes() may now be used to export the srat physical
topology of the machine for NUMA emulation.
Signed-off-by: David Rientjes <rientjes@google.com>
Cc: Andreas Herrmann <andreas.herrmann3@amd.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Ankita Garg <ankita@in.ibm.com>
Cc: Len Brown <len.brown@intel.com>
LKML-Reference: <alpine.DEB.1.00.0909251518580.14754@chino.kir.corp.google.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Need to clear both nodes and nodes_add state for start/end.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
LKML-Reference: <20090718065657.GA2898@basil.fritz.box>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Cc: stable@kernel.org
Recently there were some changes to the meaning of node_possible_map,
and it is quite strange:
- the node without memory would be set in node_possible_map
- but some node with less NODE_MIN_SIZE will be kicked out of node_possible_map.
fix it by adding strict_setup_node_bootmem().
Also, remove unparse_node().
so result will be:
1. cpu_to_node() will return online node only (nearest one)
2. apicid_to_node() still returns the node that could be not online but is set
in node_possible_map.
3. node_possible_map will include nodes that mem on it are less NODE_MIN_SIZE
v2: after move_cpus_to_node change.
[ Impact: get node_possible_map right ]
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Tested-by: Jack Steiner <steiner@sgi.com>
LKML-Reference: <4A0C49BE.6080800@kernel.org>
[ v3: various small cleanups and comment clarifications ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
after:
| commit b263295dbf
| Author: Christoph Lameter <clameter@sgi.com>
| Date: Wed Jan 30 13:30:47 2008 +0100
|
| x86: 64-bit, make sparsemem vmemmap the only memory model
we don't have MEMORY_HOTPLUG_RESERVE anymore.
Historically, x86-64 had an architecture-specific method for memory hotplug
whereby it scanned the SRAT for physical memory ranges that could be
potentially used for memory hot-add later. By reserving those ranges
without physical memory, the memmap would be allocated and left dormant
until needed. This depended on the DISCONTIG memory model which has been
removed so the code implementing HOTPLUG_RESERVE is now dead.
This patch removes the dead code used by MEMORY_HOTPLUG_RESERVE.
(Changelog authored by Mel.)
v2: updated changelog, and remove hotadd= in doc
[ Impact: remove dead code ]
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Reviewed-by: Mel Gorman <mel@csn.ul.ie>
Workflow-found-OK-by: Andrew Morton <akpm@linux-foundation.org>
LKML-Reference: <4A0C4910.7090508@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
It is expected that there might be slight differences between the e820
map and the SRAT table and the intention was that 1MB of slack be allowed.
The calculation comparing e820ram and pxmram assumes the units are bytes,
when they are in fact pages. This means 4GB of slack is being allowed,
not 1MB. This patch makes the correct comparison.
comment is from Mel.
[ Impact: don't accept buggy SRATs that could dump up to 4G of RAM ]
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Cc: Andrew Morton <akpm@linux-foundation.org>
LKML-Reference: <4A03E13E.6050107@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
node_cover_memory() sanity checks the SRAT table by ensuring that all
PXMs cover the memory reported in the e820.
However, when calculating the size of the holes in the e820, it uses
the early_node_map[] which contains information taken from both SRAT
and e820. If the SRAT is missing an entry, then it is not detected
that the SRAT table is incorrect and missing entries.
This patch uses the e820 map to calculate the holes instead of
early_node_map[].
comment is from Mel.
[ Impact: reject incorrect SRAT tables ]
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Cc: Andrew Morton <akpm@linux-foundation.org>
LKML-Reference: <4A03E10C.60906@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The mem= option will truncate the memory map at a specified address so
it's not possible to register nodes with memory beyond the e820 upper
bound.
unparse_node() is only called when then node had memory associated with
it, although with the mem= option it is no longer addressable.
[ Impact: fix boot hang on certain (large) systems ]
Reported-by: "Zhang, Yanmin" <yanmin_zhang@linux.intel.com>
Signed-off-by: David Rientjes <rientjes@google.com>
Acked-by: Jack Steiner <steiner@sgi.com>
LKML-Reference: <alpine.DEB.2.00.0905051248150.20021@chino.kir.corp.google.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Add support for nodes that have cpus but no memory.
The current code was failing to add these nodes
to the nodes_present_map.
v2: Fixes case caught by David Rientjes - missed support
for the x2apic SRAT table.
[ Impact: fix potential boot crash on memory-less UV nodes. ]
Reported-by: David Rientjes <rientjes@google.com>
Signed-off-by: Jack Steiner <steiner@sgi.com>
LKML-Reference: <20090417142242.GA23743@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
All logical processors with APIC ID values of 255 and greater will have their
APIC reported through Processor X2APIC structure (type-9 entry type) and all
logical processors with APIC ID less than 255 will have their APIC reported
through legacy Processor Local APIC (type-0 entry type) only. This is the
same case even for NMI structure reporting.
The Processor X2APIC Affinity structure provides the association between the
X2APIC ID of a logical processor and the proximity domain to which the logical
processor belongs.
For OSPM, Procssor IDs outside the 0-254 range are to be declared as Device()
objects in the ACPI namespace.
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Fix:
arch/x86/mm/srat_64.c: In function ‘acpi_numa_processor_affinity_init’:
arch/x86/mm/srat_64.c:141: error: implicit declaration of function ‘get_uv_system_type’
arch/x86/mm/srat_64.c:141: error: ‘UV_X2APIC’ undeclared (first use in this function)
arch/x86/mm/srat_64.c:141: error: (Each undeclared identifier is reported only once
arch/x86/mm/srat_64.c:141: error: for each function it appears in.)
A couple of UV definitions were moved to asm/uv/uv.h, but srat_64.c did
not include that header. Add it.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
