Move cpu_exec_init() call from instance_init to realize. This allows
any failures from cpu_exec_init() to be handled appropriately.
Also add corresponding cpu_exec_exit() call from unrealize.
cpu_dt_id assignment from instance_init is no longer needed since
correct assignment for cpu_dt_id is already present in realizefn.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
[AF: Keep calling cpu_exec_init() for CONFIG_USER_ONLY]
Signed-off-by: Andreas Färber <afaerber@suse.de>
Currently CPUState::cpu_index is monotonically increasing and a newly
created CPU always gets the next higher index. The next available
index is calculated by counting the existing number of CPUs. This is
fine as long as we only add CPUs, but there are architectures which
are starting to support CPU removal, too. For an architecture like PowerPC
which derives its CPU identifier (device tree ID) from cpu_index, the
existing logic of generating cpu_index values causes problems.
With the currently proposed method of handling vCPU removal by parking
the vCPU fd in QEMU
(Ref: http://lists.gnu.org/archive/html/qemu-devel/2015-02/msg02604.html),
generating cpu_index this way will not work for PowerPC.
This patch changes the way cpu_index is handed out by maintaining
a bit map of the CPUs that tracks both addition and removal of CPUs.
The CPU bitmap allocation logic is part of cpu_exec_init(), which is
called by instance_init routines of various CPU targets. Newly added
cpu_exec_exit() API handles the deallocation part and this routine is
called from generic CPU instance_finalize.
Note: This new CPU enumeration is for !CONFIG_USER_ONLY only.
CONFIG_USER_ONLY continues to have the old enumeration logic.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
[AF: max_cpus -> MAX_CPUMASK_BITS]
Signed-off-by: Andreas Färber <afaerber@suse.de>
Add an Error argument to cpu_exec_init() to let users collect the
error. This is in preparation to change the CPU enumeration logic
in cpu_exec_init(). With the new enumeration logic, cpu_exec_init()
can fail if cpu_index values corresponding to max_cpus have already
been handed out.
Since all current callers of cpu_exec_init() are from instance_init,
use error_abort Error argument to abort in case of an error.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@xilinx.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Peter Crosthwaite <crosthwaite.peter@gmail.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
When supporting CPU hot removal by parking the vCPU fd and reusing
it during hotplug again, there can be cases where we try to reenable
KVM_CAP_IRQ_XICS CAP for the vCPU for which it was already enabled.
Introduce a boolean member in ICPState to track this and don't
reenable the CAP if it was already enabled earlier.
Re-enabling this CAP should ideally work, but currently it results in
kernel trying to create and associate ICP with this vCPU and that
fails since there is already an ICP associated with it. Hence this
patch is needed to work around this problem in the kernel.
This change allows CPU hot removal to work for sPAPR.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Alexander Graf <agraf@suse.de>
Factor out bits of sPAPR specific CPU initialization code into
a separate routine so that it can be called from CPU hotplug
path too.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Alexander Graf <agraf@suse.de>
Reorganize CPU device tree generation code so that it be reused from
hotplug path. CPU dt entries are now generated from spapr_finalize_fdt()
instead of spapr_create_fdt_skel().
Note: This is how the split-up looks like now:
Boot path
---------
spapr_finalize_fdt
spapr_populate_cpus_dt_node
spapr_populate_cpu_dt
spapr_fixup_cpu_numa_dt
spapr_fixup_cpu_smt_dt
ibm,cas path
------------
spapr_h_cas_compose_response
spapr_fixup_cpu_dt
spapr_fixup_cpu_numa_dt
spapr_fixup_cpu_smt_dt
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Alexander Graf <agraf@suse.de>
Add support for ibm,lrdr-capacity since this is needed by the guest
kernel to know about the possible hot-pluggable CPUs and Memory. With
this, pseries kernels will start reporting correct maxcpus in
/sys/devices/system/cpu/possible.
Also define the minimum hotpluggable memory size as 256MB.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
[agraf: Fix compile error on 32bit hosts]
Signed-off-by: Alexander Graf <agraf@suse.de>
Use max_cpus instead of smp_cpus when intializating xics system. Also
report max_cpus in ibm,interrupt-server-ranges device tree property of
interrupt controller node.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Alexander Graf <agraf@suse.de>
Introduce an API numa_get_node(ram_addr_t addr, Error **errp) that
returns the NUMA node to which the given address belongs to. This
API works uniformly for both boot time as well as hotplugged memory.
This API is needed by sPAPR PowerPC to support
ibm,dynamic-reconfiguration-memory device tree node which is needed for
memory hotplug.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Tested-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Store memory address range information of boot memory in address
range list of numa_info.
This helps to have a common NUMA node lookup by address function that
works for both boot-time memory and hotplugged memory.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Tested-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
pc_dimm_plug() has code that will be needed for memory plug handlers
in other archs too. Extract code from pc_dimm_plug() into a generic
routine pc_dimm_memory_plug() that resides in pc-dimm.c. Also
correspondingly refactor re-usable unplug code into pc_dimm_memory_unplug().
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Tested-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Move hotplug_memory_base and hotplug_memory fields of PCMachineState
into a separate structure so that the same can be made use of from
other architectures supporing memory hotplug.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Tested-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Now that pc_existing_dimms_capacity() is an API, include Error pointer
as an argument and modify the caller appropriately.
Suggested-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Move pc_existing_dimms_capacity() to pc-dimm.c since it would be needed
by PowerPC memory hotplug code too.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
pc_existing_dimms_capacity() is returning DIMMs count rather than capacity.
Fix this to return the capacity. Also consider only realized devices for
capacity calculation.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Opcode table has direct, indirect and double indirect handlers, but
ppc_cpu_unrealizefn() frees direct handlers which are never allocated
and never frees double indirect handlers.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
Define and use macros instead of direct numbers wherever
possible in ppc opcodes table handling code.
This doesn't change any code functionality.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
Fix ppc64 arch specific dump code to support all combinations of little/big
endian hosts/guests. FWIW the current code is broken for altivec registers
when guest and host have a different endianness: these 128-bit registers
are written to guest memory as a two 64-bit entities and we should also swap
them.
Unit testing was done with the following program provided by Tom Musta:
#include <stdio.h>
#include <stdint.h>
#include <stdlib.h>
int main(int argc, char** argv)
{
__uint128_t v = ((__uint128_t)0x0001020304050607ull << 64) |
0x08090a0b0c0d0e0full;
register void * vptr asm ("r11");
vptr = &v;
for(;;)
asm volatile ("lvx 30,0,11" );
}
When sending SIGABRT to this program and examining the core file, we get:
- ppc64 : 00 01 02 03 04 05 06 07 08 09 0a 0b 0c 0d 0e 0f
- ppc64le: 0f 0e 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
We expect to find the very same layout in the QEMU dump since they are
real core files. This is what we get:
- ppc64 host, ppc64 guest : 00 01 02 03 04 05 06 07 08 09 0a 0b 0c 0d 0e 0f
- ppc64 host, ppc64le guest : 0f 0e 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
- x86_64 host, ppc64 guest : 00 01 02 03 04 05 06 07 08 09 0a 0b 0c 0d 0e 0f
- x86_64 host, ppc64le guest: 0f 0e 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00
We introduce a NoteFuncArg type to avoid adding extra arguments to all note
functions.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
[ rebased on top of current master branch,
introduced NoteFuncArg,
use new cpu_to_dump{16,32,64} endian helpers,
fix altivec support,
Greg Kurz <gkurz@linux.vnet.ibm.com> ]
Reviewed-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
Make DumpState and endian conversion routines available for arch-specific dump
code by moving into dump.h. DumpState will be needed by arch-specific dump
code to access target endian information from DumpState->ArchDumpInfo. Also
break the dependency of dump.h from stubs/dump.c by creating a separate
dump-arch.h.
This patch doesn't change any functionality.
Signed-off-by: Bharata B Rao <bharata@linux.ibm.com>
[ rebased on top of current master branch,
renamed endian helpers to cpu_to_dump{16,32,64},
pass a DumpState * argument to endian helpers,
Greg Kurz <gkurz@linux.vnet.ibm.com> ]
Signed-off-by: Greg Kurz <gkurz@linux.vnet.ibm.com>
[agraf: fix to apply]
Signed-off-by: Alexander Graf <agraf@suse.de>
PowerPC kernel expects the number of SMT threads in a core to be a power
of 2. Since QEMU doesn't enforce this, it leads to an early guest kernel
crash if invalid threads count is specified.
Prevent this crash and make it a graceful exit from QEMU itself by
validating the user-supplied threads count.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stewart Smith <stewart@linux.vnet.ibm.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
Remove the definitions of GLUSTER_FD_WRITE and GLUSTER_FD_READ which are
no longer used. Also sockets.h isn't needed any more.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Pipe handling mechanism in gluster driver was based on similar implementation
in RBD driver and hence had GPLv2 and associated copyright information.
After changing gluster driver to coroutine based implementation, the pipe
handling code no longer exists and hence change gluster driver's licence to
GPLv2+ and remove RBD copyrights.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
GlusterFS supports creation of zero-filled file on GlusterFS volume
by means of an API called glfs_zerofill(). Use this API from QEMU to
create an image that is filled with zeroes by using the preallocation
option of qemu-img.
qemu-img create gluster://server/volume/image -o preallocation=full 10G
The allowed values for preallocation are 'full' and 'off'. By default
preallocation is off and image is not zero-filled.
glfs_zerofill() offloads the writing of zeroes to the server and if
the storage supports SCSI WRITESAME, GlusterFS server can issue
BLKZEROOUT ioctl to achieve the zeroing.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Support .bdrv_co_write_zeroes() from gluster driver by using GlusterFS API
glfs_zerofill() that off-loads the writing of zeroes to GlusterFS server.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Convert the read, write, flush and discard implementations from aio-based
ones to coroutine based ones.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
qemu-iotests number 007 doesn't do test image cleanup. This will affect
those protocols that expect a clean state before every test. Hence
ensure that test image is cleaned up in this test.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Currently if gluster AIO callback thread fails to notify the QEMU thread about
AIO completion, we try graceful recovery by marking the disk drive as
inaccessible. This error recovery code is race-prone as found by Asias and
Stefan. However as found out by Paolo, this kind of error is impossible and
hence simplify the code that handles this error recovery.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Use pkg-config to determine the version and library dependency
for GlusterFS block driver.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This patch adds gluster as the new block backend in QEMU. This gives
QEMU the ability to boot VM images from gluster volumes. Its already
possible to boot from VM images on gluster volumes using FUSE mount, but
this patchset provides the ability to boot VM images from gluster volumes
by by-passing the FUSE layer in gluster. This is made possible by
using libgfapi routines to perform IO on gluster volumes directly.
VM Image on gluster volume is specified like this:
file=gluster[+transport]://[server[:port]]/volname/image[?socket=...]
'gluster' is the protocol.
'transport' specifies the transport type used to connect to gluster
management daemon (glusterd). Valid transport types are
tcp, unix and rdma. If a transport type isn't specified, then tcp
type is assumed.
'server' specifies the server where the volume file specification for
the given volume resides. This can be either hostname, ipv4 address
or ipv6 address. ipv6 address needs to be within square brackets [ ].
If transport type is 'unix', then 'server' field should not be specifed.
The 'socket' field needs to be populated with the path to unix domain
socket.
'port' is the port number on which glusterd is listening. This is optional
and if not specified, QEMU will send 0 which will make gluster to use the
default port. If the transport type is unix, then 'port' should not be
specified.
'volname' is the name of the gluster volume which contains the VM image.
'image' is the path to the actual VM image that resides on gluster volume.
Examples:
file=gluster://1.2.3.4/testvol/a.img
file=gluster+tcp://1.2.3.4/testvol/a.img
file=gluster+tcp://1.2.3.4:24007/testvol/dir/a.img
file=gluster+tcp://[1:2:3:4:5:6:7:8]/testvol/dir/a.img
file=gluster+tcp://[1:2:3:4:5:6:7:8]:24007/testvol/dir/a.img
file=gluster+tcp://server.domain.com:24007/testvol/dir/a.img
file=gluster+unix:///testvol/dir/a.img?socket=/tmp/glusterd.socket
file=gluster+rdma://1.2.3.4:24007/testvol/a.img
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
GlusterFS support in QEMU depends on libgfapi, libgfrpc and
libgfxdr provided by GlusterFS.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Add NUMA specific properties to guest's device tree to boot a multi-node
guests. This patch adds the following properties:
ibm,associativity
ibm,architecture-vec-5
ibm,associativity-reference-points
With this, it becomes possible to use -numa option on pseries targets.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Alexander Graf <agraf@suse.de>
Bharata B Rao