qemu/hw/ppc
Laurent Vivier 3bfe57165b numa: equally distribute memory on nodes
When there are more nodes than available memory to put the minimum
allowed memory by node, all the memory is put on the last node.

This is because we put (ram_size / nb_numa_nodes) &
~((1 << mc->numa_mem_align_shift) - 1); on each node, and in this
case the value is 0. This is particularly true with pseries,
as the memory must be aligned to 256MB.

To avoid this problem, this patch uses an error diffusion algorithm [1]
to distribute equally the memory on nodes.

We introduce numa_auto_assign_ram() function in MachineClass
to keep compatibility between machine type versions.
The legacy function is used with pseries-2.9, pc-q35-2.9 and
pc-i440fx-2.9 (and previous), the new one with all others.

Example:

qemu-system-ppc64 -S -nographic  -nodefaults -monitor stdio -m 1G -smp 8 \
                  -numa node -numa node -numa node \
                  -numa node -numa node -numa node

Before:

(qemu) info numa
6 nodes
node 0 cpus: 0 6
node 0 size: 0 MB
node 1 cpus: 1 7
node 1 size: 0 MB
node 2 cpus: 2
node 2 size: 0 MB
node 3 cpus: 3
node 3 size: 0 MB
node 4 cpus: 4
node 4 size: 0 MB
node 5 cpus: 5
node 5 size: 1024 MB

After:
(qemu) info numa
6 nodes
node 0 cpus: 0 6
node 0 size: 0 MB
node 1 cpus: 1 7
node 1 size: 256 MB
node 2 cpus: 2
node 2 size: 0 MB
node 3 cpus: 3
node 3 size: 256 MB
node 4 cpus: 4
node 4 size: 256 MB
node 5 cpus: 5
node 5 size: 256 MB

[1] https://en.wikipedia.org/wiki/Error_diffusion

Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Message-Id: <20170502162955.1610-2-lvivier@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
[ehabkost: s/ram_size/size/ at numa_default_auto_assign_ram()]
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
2017-05-11 16:08:47 -03:00
..
e500-ccsr.h ppc: do not use ../ in include files 2013-03-01 13:57:33 +01:00
e500.c target-ppc: Add MMU model check for booke machines 2017-02-02 09:30:06 +11:00
e500.h target-ppc: Eliminate redundant and incorrect function booke206_page_size_to_tlb 2016-07-01 09:57:01 +10:00
e500plat.c dma: do not depend on kvm_enabled() 2016-05-19 16:42:28 +02:00
fdt.c Move target-* CPU file into a target/ folder 2016-12-20 21:52:12 +01:00
mac_newworld.c mac99: replace debug printf with trace points 2017-02-22 11:28:28 +11:00
mac_oldworld.c hw: Default -drive to if=ide explicitly where it works 2017-02-21 13:10:53 +01:00
mac.h Clean up ill-advised or unusual header guards 2016-07-12 16:20:46 +02:00
Makefile.objs ppc/pnv: add initial IPMI sensors for the BMC simulator 2017-04-26 12:41:56 +10:00
mpc8544_guts.c hw: explicitly include qemu-common.h and cpu.h 2016-03-22 22:20:17 +01:00
mpc8544ds.c ppc: Clean up includes 2016-01-29 15:07:22 +00:00
pnv_bmc.c ppc/pnv: generate an OEM SEL event on shutdown 2017-04-26 12:41:56 +10:00
pnv_core.c ppc/pnv: create the ICP object under PnvCore 2017-04-26 12:00:42 +10:00
pnv_lpc.c ppc/pnv: enable only one LPC bus 2017-04-26 12:41:55 +10:00
pnv_occ.c ppc/pnv: Add OCC model stub with interrupt support 2017-04-26 12:00:42 +10:00
pnv_psi.c ppc/pnv: Add cut down PSI bridge model and hookup external interrupt 2017-04-26 12:00:42 +10:00
pnv_xscom.c kvm: move cpu synchronization code 2017-01-19 22:07:46 +01:00
pnv.c ppc/pnv: generate an OEM SEL event on shutdown 2017-04-26 12:41:56 +10:00
ppc4xx_devs.c qemu-common: push cpu.h inclusion out of qemu-common.h 2016-05-19 16:42:29 +02:00
ppc4xx_pci.c ppc4xx: replace debug printf with trace points 2017-02-22 11:28:28 +11:00
ppc405_boards.c clean-up: removed duplicate #includes 2016-10-28 18:17:24 +03:00
ppc405_uc.c hw/ppc/ppc405_uc.c: Avoid integer overflows 2017-02-22 14:28:53 +11:00
ppc405.h Remove unused function declarations 2016-09-15 15:32:22 +03:00
ppc440_bamboo.c target-ppc: Add MMU model check for booke machines 2017-02-02 09:30:06 +11:00
ppc_booke.c ppc: Prevent inifnite loop in decrementer auto-reload. 2017-01-31 10:10:14 +11:00
ppc.c tcg: drop global lock during TCG code execution 2017-02-24 10:32:45 +00:00
ppce500_spin.c hw/ppc: QOM'ify ppce500_spin.c 2017-01-31 10:10:13 +11:00
prep_systemio.c prep: add PReP System I/O 2017-01-31 10:10:13 +11:00
prep.c hw: Default -drive to if=ide explicitly where it works 2017-02-21 13:10:53 +01:00
rs6000_mc.c prep: add IBM RS/6000 7020 (40p) memory controller 2017-01-31 10:10:13 +11:00
spapr_cpu_core.c spapr-cpu-core: Release ICPState object during CPU unrealization 2017-04-26 12:41:56 +10:00
spapr_drc.c spapr: fix memory hot-unplugging 2017-03-29 11:35:16 +11:00
spapr_events.c ppc/spapr: QOM'ify sPAPRRTCState 2017-04-26 12:00:41 +10:00
spapr_hcall.c spapr: Workaround for broken radix guests 2017-04-26 12:00:41 +10:00
spapr_iommu.c target-ppc: kvm: make use of KVM_CREATE_SPAPR_TCE_64 2017-04-26 12:00:41 +10:00
spapr_ovec.c spapr: replace debug printf with trace points 2017-02-22 11:28:28 +11:00
spapr_pci_vfio.c Use #include "..." for our own headers, <...> for others 2016-07-12 16:19:16 +02:00
spapr_pci.c spapr_pci: Removed unused include 2017-04-26 12:00:41 +10:00
spapr_rng.c hw: explicitly include qemu-common.h and cpu.h 2016-03-22 22:20:17 +01:00
spapr_rtas_ddw.c spapr_pci/spapr_pci_vfio: Support Dynamic DMA Windows (DDW) 2016-07-05 14:31:08 +10:00
spapr_rtas.c report guest crash information in GUEST_PANICKED event 2017-02-16 15:30:49 +01:00
spapr_rtc.c ppc/spapr: QOM'ify sPAPRRTCState 2017-04-26 12:00:41 +10:00
spapr_vio.c ppc/xics: store the ICS object under the sPAPR machine 2017-03-01 11:23:39 +11:00
spapr.c numa: equally distribute memory on nodes 2017-05-11 16:08:47 -03:00
trace-events spapr: replace debug printf with trace points 2017-02-22 11:28:28 +11:00
virtex_ml507.c target-ppc: Add MMU model check for booke machines 2017-02-02 09:30:06 +11:00