Default RAM backend depends on numa_uses_legacy_mem(), which is
infulenced by -numa options on CLI or set-numa-node QMP command
at preconfig time. If QEMU is started with '-preconfig'
without -numa, it will lead to creating default RAM backend
even if later set-numa-node is used to assing RAM to NUMA nodes
using 'memdev' NUMA option.
That at best will waste RAM object created by default and with
next patch adding a check to prevent usage of conflicting
'-M memory-backend' and '-numa memdev'
options, it will make QEMU error out if user tries to configure
NUMA at preconfig time with memdev option, making set-numa-node
unusable.
To fix issue, move preconfig loop before default RAM backend is
created, so that numa_uses_legacy_mem() would take into account
effects of set-numa-node commands executed at preconfig time.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20200511141103.43768-2-imammedo@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This allows us to keep pc-bios in executable_dir/pc-bios, rather than
executable_dir/../pc-bios, which is incompatible with oss-fuzz' file
structure.
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-id: 20200512030133.29896-2-alxndr@bu.edu
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
The only way object_property_add() can fail is when a property with
the same name already exists. Since our property names are all
hardcoded, failure is a programming error, and the appropriate way to
handle it is passing &error_abort.
Same for its variants, except for object_property_add_child(), which
additionally fails when the child already has a parent. Parentage is
also under program control, so this is a programming error, too.
We have a bit over 500 callers. Almost half of them pass
&error_abort, slightly fewer ignore errors, one test case handles
errors, and the remaining few callers pass them to their own callers.
The previous few commits demonstrated once again that ignoring
programming errors is a bad idea.
Of the few ones that pass on errors, several violate the Error API.
The Error ** argument must be NULL, &error_abort, &error_fatal, or a
pointer to a variable containing NULL. Passing an argument of the
latter kind twice without clearing it in between is wrong: if the
first call sets an error, it no longer points to NULL for the second
call. ich9_pm_add_properties(), sparc32_ledma_realize(),
sparc32_dma_realize(), xilinx_axidma_realize(), xilinx_enet_realize()
are wrong that way.
When the one appropriate choice of argument is &error_abort, letting
users pick the argument is a bad idea.
Drop parameter @errp and assert the preconditions instead.
There's one exception to "duplicate property name is a programming
error": the way object_property_add() implements the magic (and
undocumented) "automatic arrayification". Don't drop @errp there.
Instead, rename object_property_add() to object_property_try_add(),
and add the obvious wrapper object_property_add().
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200505152926.18877-15-armbru@redhat.com>
[Two semantic rebase conflicts resolved]
Specifically explain what users should do in case they don't use
-display yet and depend on the qemu picking the ui for them.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-id: 20200407093617.10058-1-kraxel@redhat.com
The second argument 'id' is a pointer. Pass NULL rather than 0.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Message-Id: <20200427005704.2475782-1-masahiroy@kernel.org>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
the former is not actually used by explicit backend, so instead of
silently ignoring the option in non valid context, exit with error.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20200409134133.11339-1-imammedo@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Commit a1b18df9a4, broke virt_kvm_type() logic, which depends on
maxram_size, ram_size, ram_slots being parsed/set on machine instance
at the time accelerator (KVM) is initialized.
set_memory_options() part was already reverted by commit 2a7b18a320,
so revert remaining initialization of above machine fields to make
virt_kvm_type() work as it used to.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reported-by: Auger Eric <eric.auger@redhat.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Tested-by: Eric Auger <eric.auger@redhat.com>
Message-Id: <20200326112829.19989-1-imammedo@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Older QEMU versions did fixup the ram size to match what can be reported
via sclp. We need to mimic this behaviour for machine types 4.2 and
older to not fail on inbound migration for memory sizes that do not fit.
Old machines with proper aligned memory sizes are not affected.
Alignment table:
VM size (<=) | Alignment
--------------------------
1020M | 1M
2040M | 2M
4080M | 4M
8160M | 8M
16320M | 16M
32640M | 32M
65280M | 64M
130560M | 128M
261120M | 256M
522240M | 512M
1044480M | 1G
2088960M | 2G
4177920M | 4G
8355840M | 8G
Suggested action is to replace unaligned -m value with a suitable
aligned one or if a change to a newer machine type is possible, use a
machine version >= 5.0.
A future version might remove the compatibility handling.
For machine types >= 5.0 we can simply use an increment size of 1M and
use the full range of increment number which allows for all possible
memory sizes. The old limitation of having a maximum of 1020 increments
was added for standby memory, which we no longer support. With that we
can now support even weird memory sizes like 10001234 MB.
As we no longer fixup maxram_size as well, make other users use ram_size
instead. Keep using maxram_size when setting the maximum ram size in KVM,
as that will come in handy in the future when supporting memory hotplug
(in contrast, storage keys and storage attributes for hotplugged memory
will have to be migrated per RAM block in the future).
Fixes: 3a12fc61af ("390x/s390-virtio-ccw: use memdev for RAM")
Reported-by: Lukáš Doktor <ldoktor@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Igor Mammedov <imammedo@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20200401123754.109602-1-borntraeger@de.ibm.com>
[CH: fixed up message on memory size fixup]
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Remove the global acpi_enabled bool and replace it with an
acpi OnOffAuto machine property.
qemu throws an error now if you use -no-acpi while the machine
type you are using doesn't support acpi in the first place.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-Id: <20200320100136.11717-1-kraxel@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Migration from QEMU(v4.0) fails when using 3.1 or older machine
type. For example if one attempts to migrate
QEMU-2.12 started as
qemu-system-ppc64 -nodefaults -M pseries-2.12 -m 4096 -mem-path /tmp/
to current master, it will fail with
qemu-system-ppc64: Unknown ramblock "ppc_spapr.ram", cannot accept migration
qemu-system-ppc64: error while loading state for instance 0x0 of device 'ram'
qemu-system-ppc64: load of migration failed: Invalid argument
Caused by 900c0ba373 commit which switches main RAM allocation to
memory backends and the fact in 3.1 and older QEMU, backends used
full[***] QOM path as memory region name instead of backend's name.
That was changed after 3.1 to use prefix-less names by default
(fa0cb34d22) for new machine types.
*** effectively makes main RAM memory region names defined by
MachineClass::default_ram_id being altered with '/objects/' prefix
and therefore migration fails as old QEMU sends prefix-less
name while new QEMU expects name with prefix when using 3.1 and
older machine types.
Fix it by forcing implicit[1] memory backend to always use
prefix-less names for its memory region by setting
'x-use-canonical-path-for-ramblock-id'
property to false.
1) i.e. memory backend created by compat glue which maps
-m/-mem-path/-mem-prealloc/default RAM size into
appropriate backend type/options to match old CLI format.
Fixes: 900c0ba373
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reported-by: Lukáš Doktor <ldoktor@redhat.com>
Message-Id: <20200304172748.15338-1-imammedo@redhat.com>
Tested-by: Lukáš Doktor <ldoktor@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Store the smp sockets in CpuTopology. The socket information required to
build the apic id in EPYC mode. Right now socket information is not passed
to down when decoding the apic id. Add the socket information here.
Signed-off-by: Babu Moger <babu.moger@amd.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Message-Id: <158396718647.58170.2278448323151215741.stgit@naples-babu.amd.com>
Currently if you try to ask for the list of CPUs for a target
architecture which does not specify a default machine type
you just get an error:
$ qemu-system-arm -cpu help
qemu-system-arm: No machine specified, and there is no default
Use -machine help to list supported machines
Since the list of CPUs doesn't depend on the machine, this is
unnecessarily unhelpful. "-device help" has a similar problem.
Move the checks for "did the user ask for -cpu help or -device help"
up so they precede the select_machine() call which checks that the
user specified a valid machine type.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
vl.c calls machine_usb() declared in "hw/boards.h". Include it.
This fixes (when modifying unrelated headers):
vl.c:1283:10: error: implicit declaration of function 'machine_usb' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
if (!machine_usb(current_machine)) {
^
vl.c:1283:10: error: this function declaration is not a prototype [-Werror,-Wstrict-prototypes]
vl.c:1283:22: error: use of undeclared identifier 'current_machine'
if (!machine_usb(current_machine)) {
^
Acked-by: John Snow <jsnow@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20200228114649.12818-2-philmd@redhat.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
It would be confusing to have multiple default machines.
Abort if this ever occurs.
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20200207161948.15972-4-philmd@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Tested-by: Laurent Vivier <laurent@vivier.eu>
So we don't have to compile everything in, or have ifdefs
Signed-off-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Commit a1b18df9a4 moved -m option parsing after configure_accelerators()
that broke TCG accelerator initialization which accesses global ram_size
from size_code_gen_buffer() which is equal to 0 at that moment.
Partially revert a1b18df9a4, by returning set_memory_options() to its
original location and only keep 32-bit host VA check and 'memory-backend'
size check introduced by fe64d06afc at current place.
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This series removes ad hoc RAM allocation API (memory_region_allocate_system_memory)
and consolidates it around hostmem backend. It allows to
* resolve conflicts between global -mem-prealloc and hostmem's "policy" option,
fixing premature allocation before binding policy is applied
* simplify complicated memory allocation routines which had to deal with 2 ways
to allocate RAM.
* reuse hostmem backends of a choice for main RAM without adding extra CLI
options to duplicate hostmem features. A recent case was -mem-shared, to
enable vhost-user on targets that don't support hostmem backends [1] (ex: s390)
* move RAM allocation from individual boards into generic machine code and
provide them with prepared MemoryRegion.
* clean up deprecated NUMA features which were tied to the old API (see patches)
- "numa: remove deprecated -mem-path fallback to anonymous RAM"
- (POSTPONED, waiting on libvirt side) "forbid '-numa node,mem' for 5.0 and newer machine types"
- (POSTPONED) "numa: remove deprecated implicit RAM distribution between nodes"
Introduce a new machine.memory-backend property and wrapper code that aliases
global -mem-path and -mem-alloc into automatically created hostmem backend
properties (provided memory-backend was not set explicitly given by user).
A bulk of trivial patches then follow to incrementally convert individual
boards to using machine.memory-backend provided MemoryRegion.
Board conversion typically involves:
* providing MachineClass::default_ram_size and MachineClass::default_ram_id
so generic code could create default backend if user didn't explicitly provide
memory-backend or -m options
* dropping memory_region_allocate_system_memory() call
* using convenience MachineState::ram MemoryRegion, which points to MemoryRegion
allocated by ram-memdev
On top of that for some boards:
* missing ram_size checks are added (typically it were boards with fixed ram size)
* ram_size fixups are replaced by checks and hard errors, forcing user to
provide correct "-m" values instead of ignoring it and continuing running.
After all boards are converted, the old API is removed and memory allocation
routines are cleaned up.
The qtest-based fuzzer makes use of forking to reset-state between
tests. Keep the callback enabled, so the call_rcu thread gets created
within the child process.
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200220041118.23264-15-alxndr@bu.edu
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
A program might rely on functions implemented in vl.c, but implement its
own main(). By placing main into a separate source file, there are no
complaints about duplicate main()s when linking against vl.o. For
example, the virtual-device fuzzer uses a main() provided by libfuzzer,
and needs to perform some initialization before running the softmmu
initialization. Now, main simply calls three vl.c functions which
handle the guest initialization, main loop and cleanup.
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-id: 20200220041118.23264-3-alxndr@bu.edu
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Move vl.c to a separate directory, similar to linux-user/
Update the chechpatch and get_maintainer scripts, since they relied on
/vl.c for top_of_tree checks.
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-id: 20200220041118.23264-2-alxndr@bu.edu
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
