As a follow-on to the recent patches I submitted that allowed for a sysfs
memory block to span multiple memory sections, we should also update the
probe routine to online all of the memory sections in a memory block. Without
this patch the current code will only add a single memory section. I think
the probe routine should add all of the memory sections in the specified memory
block so that its behavior is in line with memory hotplug actions through
the sysfs interfaces.
This patch applies on top of the previous sysfs memory updates to allow
a sysfs directory o span multiple memory sections.
https://lkml.org/lkml/2011/1/20/245
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
Define a version of memory_block_size_bytes for x86_64 when CONFIG_X86_UV is
set.
Signed-off-by: Robin Holt <[email protected]>
Signed-off-by: Jack Steiner <[email protected]>
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
Define a version of memory_block_size_bytes() for powerpc/pseries such that
a memory block spans an entire lmb.
Signed-off-by: Nathan Fontenot <[email protected]>
Reviewed-by: Robin Holt <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
Update the 'phys_index' property of a the memory_block struct to be
called start_section_nr, and add a end_section_nr property. The
data tracked here is the same but the updated naming is more in line
with what is stored here, namely the first and last section number
that the memory block spans.
The names presented to userspace remain the same, phys_index for
start_section_nr and end_phys_index for end_section_nr, to avoid breaking
anything in userspace.
This also updates the node sysfs code to be aware of the new capability for
a memory block to contain multiple memory sections and be aware of the memory
block structure name changes (start_section_nr). This requires an additional
parameter to unregister_mem_sect_under_nodes so that we know which memory
section of the memory block to unregister.
Signed-off-by: Nathan Fontenot <[email protected]>
Reviewed-by: Robin Holt <[email protected]>
Reviewed-by: KAMEZAWA Hiroyuki <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
Update the memory sysfs code such that each sysfs memory directory is now
considered a memory block that can span multiple memory sections per
memory block. The default size of each memory block is SECTION_SIZE_BITS
to maintain the current behavior of having a single memory section per
memory block (i.e. one sysfs directory per memory section).
For architectures that want to have memory blocks span multiple
memory sections they need only define their own memory_block_size_bytes()
routine.
Update the memory hotplug documentation to reflect the new behaviors of
memory blocks reflected in sysfs.
Signed-off-by: Nathan Fontenot <[email protected]>
Reviewed-by: Robin Holt <[email protected]>
Reviewed-by: KAMEZAWA Hiroyuki <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
Add a section count property to the memory_block struct to track the number
of memory sections that have been added/removed from a memory block. This
allows us to know when the last memory section of a memory block has been
removed so we can remove the memory block.
Signed-off-by: Nathan Fontenot <[email protected]>
Reviewed-by: Robin Holt <[email protected]>
Reviewed-by: KAMEZAWA Hiroyuki <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
Add a new mutex for use in adding and removing of memory blocks. This
is needed to avoid any race conditions in which the same memory block could
be added and removed at the same time.
Signed-off-by: Nathan Fontenot <[email protected]>
Reviewed-by: Robin Holt <[email protected]>
Reviewed-By: KAMEZAWA Hiroyuki <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
Enable partition migration in the kernel. To do this a new sysfs file,
/sys/kernel/mobility/migration, is created. In order to initiate a migration
the stream id (generated by the HMC managing the system) is written to this
file.
After a migration occurs, and what is the majority of this code, the device
tree needs to be updated for the new system the partition is running on. This
is done via the ibm,update-nodes and ibm,update-properties rtas calls which
return information regarding which nodes and properties of the device tree
are to be added/removed/updated.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
Export the rtas_ibm_suspend_me() routine. This is needed to perform
partition migration in the kernel.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
Export routines associated with adding and removing device tree nodes on
pseries needed for device tree updating.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
The dlpar code can cause a deadlock to occur when making the RTAS
configure-connector call. This occurs because we make kmalloc calls,
which can block, while parsing the rtas_data_buf and holding the
rtas_data_buf_lock. This an cause issues if someone else attempts
to grab the rtas_data_bug_lock.
This patch alleviates this issue by copying the contents of the rtas_data_buf
to a local buffer before parsing. This allows us to only hold the
rtas_data_buf_lock around the RTAS configure-connector calls.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
The 'smt_enabled=X' boot option does not handle values of X > 2.
For Power 7 processors with smt modes of 0,1,2,3, and 4 this does
not work. This patch allows the smt_enabled option to be set to
any value limited to a max equal to the number of threads per
core.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
Move the defintion and lock helper routines for the cpu hotplug driver
lock from pseries to powerpc code to avoid build breaks for platforms
other than pseries that use cpu hotplug.
Signed-off-by: Nathan Fontenot <[email protected]>
Acked-by: Michael Ellerman <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
It looks like the previous patch sent out to move RTAS and
other items from /proc/ppc64 to /proc/powerpc missed a few
files needed for RAS and DLPAR functionality.
Original Patch here:
http://lists.ozlabs.org/pipermail/linuxppc-dev/2009-September/076096.html
This patch updates the remaining files to be created under /proc/powerpc.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
The recent patch to add cpu offline/online as part of the DLPAR
process for pseries causes a build break if CONFIG_SMP is not
defined. Original patch here;
http://lists.ozlabs.org/pipermail/linuxppc-dev/2009-November/078299.html
This corrects the build break by moving the online_node_cpus
and offline_node_cpus under the #ifdef CONFIG_ARCH_CPU_PROBE_RELEASE
portions of dlpar.c.
This patch also slightly modifies the online_node_cpus and offline_node_cpus
routines to prepend dlpar_ to the them and make them static. These two
routine are only used in the dlpar add/remove of cpus and these changes
should help clarify that.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
This patch adds the specific routines to probe and release (add and remove)
cpu resource for the powerpc pseries platform and registers these handlers
with the ppc_md callout structure.
Signed-off-by: Nathan Fontenot <[email protected]>
Acked-by: Paul Mackerras <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
Version 3 of this patch is updated with documentation added to
Documentation/ABI. There are no changes to any of the C code from v2
of the patch.
In order to support kernel DLPAR of CPU resources we need to provide an
interface to add (probe) and remove (release) the resource from the system.
This patch Creates new generic probe and release sysfs files to facilitate
cpu probe/release. The probe/release interface provides for allowing each
arch to supply their own routines for implementing the backend of adding
and removing cpus to/from the system.
This also creates the powerpc specific stubs to handle the arch callouts
from writes to the sysfs files.
The creation and use of these files is regulated by the
CONFIG_ARCH_CPU_PROBE_RELEASE option so that only architectures that need the
capability will have the files created.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
The Dynamic Logical Partitioning capabilities of the powerpc pseries platform
allows for the addition and removal of resources (i.e. CPU's, memory, and PCI
devices) from a partition. The removal of a resource involves
removing the resource's node from the device tree and then returning the
resource to firmware via the rtas set-indicator call. To add a resource, it
is first obtained from firmware via the rtas set-indicator call and then a
new device tree node is created using the ibm,configure-coinnector rtas call
and added to the device tree.
This patch provides the kernel DLPAR infrastructure in a new filed named
dlpar.c. The functionality provided is for acquiring and releasing a resource
from firmware and the parsing of information returned from the
ibm,configure-connector rtas call. Additionally this exports the pSeries
reconfiguration notifier chain so that it can be invoked when device tree
updates are made.
Signed-off-by: Nathan Fontenot <[email protected]>
Acked-by: Paul Mackerras <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
This patch updates the output from /proc/ppc64/lparcfg to display the
processor virtualization resource allocations for a shared processor
partition.
This information is already gathered via the h_get_ppp call, we just
have to make sure that the ibm,partition-performance-parameters-level
property is >= 1 to ensure that the information is valid.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
The return code from invoking the notifier chain when updating the
ibm,dynamic-memory property is not handled properly. In failure
cases (rc == NOTIFY_BAD) we should be restoring the original value
of the property. In success (rc == NOTIFY_OK) we should be returning
zero from the calling routine.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
This patch reworks the hot_add_scn_to_nid and its supporting functions
to make them easier to understand. There are no functional changes in
this patch and has been tested on machine with memory represented in the
device tree as memory nodes and in the ibm,dynamic-memory property.
My previous patch that introduced support for hotplug memory add on
systems whose memory was represented by the ibm,dynamic-memory property
of the device tree only left the code more unintelligible. This
will hopefully makes things easier to understand.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
Resources for PHB's that are dynamically added to a system are not
properly allocated in the resource tree.
Not having these resources allocated causes an oops when removing
the PHB when we try to release them.
The diff appears a bit messy, this is mainly due to moving everything
one tab to the left in the pcibios_allocate_bus_resources routine.
The functionality change in this routine is only that the
list_for_each_entry() loop is pulled out and moved to the necessary
calling routine.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
The pfn of the memory to be removed should be validated prior to
attempting to remove the memory. In cases where the probe of a
memory section fails during hotplug add, the pfn for the lmb may
not be valid.
Signed-off-by: Badari Pulavarty <[email protected]>
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
During hotplug memory remove, memory regions should be released on a
PAGES_PER_SECTION size chunks. This mirrors the code in add_memory where
resources are requested on a PAGES_PER_SECTION size.
Attempting to release the entire memory region fails because there is not
a single resource for the total number of pages being removed. Instead
the resources for the pages are split in PAGES_PER_SECTION size chunks as
requested during memory add.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Badari Pulavarty <[email protected]>
Acked-by: Yasunori Goto <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Testing hotplug memory remove has revealed that we can oops in
pseries_lmb_remove(). The incorrect shift causes a NULL pointer
dereference in the page_zone() inline routine.
I have only been able to reproduce the oops on kernels with large pages
enabled.
Tested on Power5 and Power6 with and without large pages enabled.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
The return code from invocation of the notifier for
pSeries_reconfig_chain during update of the device tree is not
checked. This causes writes to /proc/ppc64/ofdt to update memory
properties (i.e. ibm,dyamic-reconfiguration-memory) to always
return success, instead of the result of the notifier chain.
This happens specifically when we remove/add memory from the
device tree on machines using memory specified in the
ibm,dynamic-reconfiguration-memory property of the device tree.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
The kernel copy of the rtas args struct contains the return
value(s) for the specified rtas call. These are copied back
to user space with the assumption that every value has been
set by the rtas call, which turns out to be not always true.
Thus userspace can see random values and think the call failed
when in fact it succeeded, but for some reason didn't set one
of the return values.
This fixes the problem by zeroing out the return value fields
of the rtas args struct before processing the rtas call.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
The code to handle writes to /proc/ppc64/lparcfg incorrectly
assumes that the return code from the helper routines to update
processor or memory entitlement return a hcall return value. It
then assumes any non-hcall return value is bad and sets the return
code for the write to be -EIO.
The update_[mp]pp routines can return values other than a hcall
return value. This patch removes the automatic setting of any
return code that is not an hcall return value from these routines
to -EIO.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
There are only 4 valid name=value pairs for writes to
/proc/ppc64/lparcfg. Current code allocates a buffer to copy
this information in from the user. Since the longest name=value
pair will easily fit into a buffer of 64 characters, simply
put the buffer on the stack instead of allocating the buffer.
Signed-off-by: Nathan Fotenot <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
Update the architecture vector to indicate that Cooperative Memory
Overcommitment is supported if CONFIG_PPC_SMLPAR is set.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Robert Jennings <[email protected]>
Acked-by: Paul Mackerras <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
Update /proc/ppc64/lparcfg to display Cooperative Memory
Overcommitment statistics as reported by the H_GET_MPP hcall. This
also updates the lparcfg interface to allow setting memory entitlement
and weight.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Robert Jennings <[email protected]>
Acked-by: Paul Mackerras <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
Split the retrieval and setting of processor entitlement and weight into
helper routines. This also removes the printing of the raw values
returned from h_get_ppp, the values are already parsed and printed.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Robert Jennings <[email protected]>
Acked-by: Paul Mackerras <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
Remove the extraneous error reporting used when a hcall made from lparcfg fails.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Robert Jennings <[email protected]>
Acked-by: Paul Mackerras <[email protected]>
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
Update the association of a memory section with a numa node that
occurs during hotplug add of a memory section. This adds a check in
the hot_add_scn_to_nid() routine for the
ibm,dynamic-reconfiguration-memory node in the device tree. If
present the new hot_add_drconf_scn_to_nid() routine is invoked, which
can properly parse the ibm,dynamic-reconfiguration-memory node of the
device tree and make the proper numa node associations.
This also introduces the valid_hot_add_scn() routine as a helper
function for code that is common to the hot_add_scn_to_nid() and
hot_add_drconf_scn_to_nid() routines.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This splits off several pieces of code that parse the
ibm,dynamic-reconfiguration-memory node of the device tree into separate
helper routines. This is in preparation for the next commit that will
use these helper routines. There are no functional changes in this patch.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This updates the device tree manipulation routines so that memory
add/remove of lmbs represented under the
ibm,dynamic-reconfiguration-memory node of the device tree invokes the
hotplug notifier chain.
This change is needed because of the change in the way memory is
represented under the ibm,dynamic-reconfiguration-memory node. All lmbs
are described in the ibm,dynamic-memory property instead of having a
separate node for each lmb as in previous device tree layouts. This
requires the update_node() routine to check for updates to the
ibm,dynamic-memory property and invoke the hotplug notifier chain.
This also updates the pseries hotplug notifier to be able to gather information
for lmbs represented under the ibm,dynamic-reconfiguration-memory node and
have the lmbs added/removed.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Use the base address of the lmb to derive the starting page frame number
instead of trying to extract it from the drc index of the lmb. The drc
index should not be used for this as it will, and did, break.
Until this point, systems that have had memory represented in the device
tree with a node for each lmb the drc index would (luckily) closely
track the base address of the lmb. For example a lmb with a drc index
of 8000000a would have a base address of a0000000. This correlation
allowed the current code to derive the starting page frame number from
the drc inddex
Device tree layouts where lmbs are represented under the
ibm,dynamic-reconfiguration-memory node in the ibm,dynamic-memory
property do not have this correlation between the drc index and base
address of the lmb.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Allow the phandle passed to the /proc/ppc64/ofdt file to be specified
in formats other than decimal. This allows us to easily specify phandle
values in hex that would otherwise appear as negative integers.
This is an issue on systems where the value of
/proc/device-tree/ibm,dynamic-reconfiguration-memory.ibm,phandle is
fffffff9. Having to pass this to the ofdt file as a string results in
a large negative number, and simple_strtoul() does not handle negative
numbers.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
The xics code does update the default server information when the boot
cpu is removed. This patch recognizes when the boot cpu is being
removed and updates the appropriate information based on the new 'boot
cpu'.
Failure to update this information can causes us to leave irqs pinned
to cpus that are being removed, especially when removing the boot cpu.
The cpu is removed from the kernel, but cpu dlpar remove operations
fail since we cannot return the cpu to the hypervisor.
Signed-off-by: Nathan Fonteno <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
It appears that xics.c has its own of_get_cpu_node(). Remove this and
use the common one from prom.c.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This splits off the kexec path bits of the xics_teardown_cpu() routine
into its own xics_kexec_teardown_cpu() routine. With the previous
combined routine the CPPR for a cpu that is being removed may have its
CPPR reset in the plpar_eoi() call (which explicitly sets the CPPR to
a non-zero value). Splitting of the kexec bits of the code prevents
this from happening in the cpu remove path.
Once again, this does not cause the cpu remove from the kernel to
fail, but it does cause cpu dlpar operations to not be able to return
the cpu to the hypervisor.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
The affinity mask in the virq descriptor needs to be set before we
reset the affinity for the virq. Without doing this the call to get
the new irq server fails and we end up leaving the virq pinned to the
cpu we are removing.
This does not fail the cpu remove from the kernel, but it does prevent
cpu dlpar remove operations from returning the cpu to the hypervisor.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
This patch removes unnecessary exports, marks functions as static when
possible, and simplifies some list-related code.
Signed-off-by: Nathan Fontenot <[email protected]>
Signed-off-by: Linas Vepstas <[email protected]>
Signed-off-by: Paul Mackerras <[email protected]>
Nathan Fontenot