migrate: Update ram_block_discard_range for shared
The choice of call to discard a block is getting more complicated
for other cases. We use fallocate PUNCH_HOLE in any file cases;
it works for both hugepage and for tmpfs.
We use the DONTNEED for non-hugepage cases either where they're
anonymous or where they're private.
Care should be taken when trying other backing files.
Haozhong Zhang [Sun, 11 Mar 2018 03:02:14 +0000 (11:02 +0800)]
tests/bios-tables-test: add test cases for DIMM proximity
QEMU now builds one SRAT memory affinity structure for each PC-DIMM
and NVDIMM device presented at boot time with the proximity domain
specified in the device option 'node', rather than only one SRAT
memory affinity structure covering the entire hotpluggable address
space with the proximity domain of the last node.
Add test cases on PC and Q35 machines with 4 proximity domains, and
one PC-DIMM and one NVDIMM attached to the 2nd and 3rd proximity
domains respectively. Check whether the QEMU-built SRAT tables match
with the expected ones.
The following ACPI tables need to be added for this test:
tests/acpi-test-data/pc/APIC.dimmpxm
tests/acpi-test-data/pc/DSDT.dimmpxm
tests/acpi-test-data/pc/NFIT.dimmpxm
tests/acpi-test-data/pc/SRAT.dimmpxm
tests/acpi-test-data/pc/SSDT.dimmpxm
tests/acpi-test-data/q35/APIC.dimmpxm
tests/acpi-test-data/q35/DSDT.dimmpxm
tests/acpi-test-data/q35/NFIT.dimmpxm
tests/acpi-test-data/q35/SRAT.dimmpxm
tests/acpi-test-data/q35/SSDT.dimmpxm
New APIC and DSDT are needed because of the multiple processors
configuration. New NFIT and SSDT are needed because of NVDIMM.
Haozhong Zhang [Sun, 11 Mar 2018 03:02:13 +0000 (11:02 +0800)]
hw/acpi-build: build SRAT memory affinity structures for DIMM devices
ACPI 6.2A Table 5-129 "SPA Range Structure" requires the proximity
domain of a NVDIMM SPA range must match with corresponding entry in
SRAT table.
The address ranges of vNVDIMM in QEMU are allocated from the
hot-pluggable address space, which is entirely covered by one SRAT
memory affinity structure. However, users can set the vNVDIMM
proximity domain in NFIT SPA range structure by the 'node' property of
'-device nvdimm' to a value different than the one in the above SRAT
memory affinity structure.
In order to solve such proximity domain mismatch, this patch builds
one SRAT memory affinity structure for each DIMM device present at
boot time, including both PC-DIMM and NVDIMM, with the proximity
domain specified in '-device pc-dimm' or '-device nvdimm'.
The remaining hot-pluggable address space is covered by one or multiple
SRAT memory affinity structures with the proximity domain of the last
node as before.
Haozhong Zhang [Sun, 11 Mar 2018 03:02:12 +0000 (11:02 +0800)]
qmp: distinguish PC-DIMM and NVDIMM in MemoryDeviceInfoList
It may need to treat PC-DIMM and NVDIMM differently, e.g., when
deciding the necessity of non-volatile flag bit in SRAT memory
affinity structures.
A new field 'nvdimm' is added to the union type MemoryDeviceInfo for
such purpose. Its type is currently PCDIMMDeviceInfo and will be
updated when necessary in the future.
It also fixes "info memory-devices"/query-memory-devices which
currently show nvdimm devices as dimm devices since
object_dynamic_cast(obj, TYPE_PC_DIMM) happily cast nvdimm to
TYPE_PC_DIMM which it's been inherited from.
Haozhong Zhang [Sun, 11 Mar 2018 03:02:11 +0000 (11:02 +0800)]
pc-dimm: make qmp_pc_dimm_device_list() sort devices by address
Make qmp_pc_dimm_device_list() return sorted by start address
list of devices so that it could be reused in places that
would need sorted list*. Reuse existing pc_dimm_built_list()
to get sorted list.
While at it hide recursive callbacks from callers, so that:
Igor Mammedov [Wed, 28 Feb 2018 14:23:55 +0000 (15:23 +0100)]
tests: acpi: don't read all fields in test_acpi_fadt_table()
there is no point to read fields here but not actually
checking them so drop it and read only header + dsdt/facs
addresses since it's needed later to fetch that tables.
With this cleanup we can get rid of AcpiFadtDescriptorRev3/
ACPI_FADT_COMMON_DEF which have no users left.
Igor Mammedov [Wed, 28 Feb 2018 14:23:54 +0000 (15:23 +0100)]
virt_arm: acpi: reuse common build_fadt()
Extend generic build_fadt() to support rev5.1 FADT
and reuse it for 'virt' board, it would allow to
phase out usage of AcpiFadtDescriptorRev5_1 and
later ACPI_FADT_COMMON_DEF.
Igor Mammedov [Wed, 28 Feb 2018 14:23:53 +0000 (15:23 +0100)]
acpi: move build_fadt() from i386 specific to generic ACPI source
It will be extended and reused by follow up patch for ARM target.
PS:
Since it's generic function now, don't patch FIRMWARE_CTRL, DSDT
fields if they don't point to tables since platform might not
provide them and use X_ variants instead if applicable.
Igor Mammedov [Wed, 28 Feb 2018 14:23:52 +0000 (15:23 +0100)]
pc: acpi: use build_append_foo() API to construct FADT
build_append_foo() API doesn't need explicit endianness
conversions which eliminates a source of errors and
it makes build_fadt() look like declarative definition of
FADT table in ACPI spec, which makes it easy to review.
Also it allows easily extending FADT to support other
revisions which will be used by follow up patches
where build_fadt() will be reused for ARM target.
Igor Mammedov [Wed, 28 Feb 2018 14:23:51 +0000 (15:23 +0100)]
pc: acpi: isolate FADT specific data into AcpiFadtData structure
move FADT data initialization out of fadt_setup() into dedicated
init_fadt_data() that will set common for pc/q35 values in
AcpiFadtData structure and acpi_get_pm_info() will complement
it with pc/q35 specific values initialization.
That will allow to get rid of fadt_setup() and generalize
build_fadt() so it could be easily extended for rev5 and
reused by ARM target.
While at it also move facs/dsdt/xdsdt offsets from build_fadt()
arg list into AcpiFadtData, as they belong to the same dataset.
Igor Mammedov [Wed, 28 Feb 2018 14:23:49 +0000 (15:23 +0100)]
acpi: add build_append_gas() helper for Generic Address Structure
it will help to add Generic Address Structure to ACPI tables
without using packed C structures and avoid endianness
issues as API doesn't need an explicit conversion.
Igor Mammedov [Wed, 28 Feb 2018 14:23:46 +0000 (15:23 +0100)]
acpi: remove unused acpi-dsdt.aml
SeaBIOS blob which is currently shipped with QEMU
doesn't need acpi-dsdt.aml nor is able to use it
and code that loaded it in QEMU was removed by
(commit 9fb7aaaf4c "pc: drop external DSDT loading")
in 2013.
Jason Baron [Thu, 8 Mar 2018 03:25:41 +0000 (22:25 -0500)]
virtio-net: add linkspeed and duplex settings to virtio-net
Although linkspeed and duplex can be set in a linux guest via 'ethtool -s',
this requires custom ethtool commands for virtio-net by default.
Introduce a new feature flag, VIRTIO_NET_F_SPEED_DUPLEX, which allows
the hypervisor to export a linkspeed and duplex setting. The user can
subsequently overwrite it later if desired via: 'ethtool -s'.
Linkspeed and duplex settings can be set as:
'-device virtio-net,speed=10000,duplex=full'
where speed is [0...INT_MAX], and duplex is ["half"|"full"].
Jason Baron [Thu, 8 Mar 2018 03:25:39 +0000 (22:25 -0500)]
scripts/update-linux-headers: add ethtool.h and update to 4.16.0-rc4
A subsequent patch to add support for setting linkspeed/duplex in
virtio-net, requires a few definitions from ethtool.h, which ends up
pulling in kernel.h and sysinfo.h as well.
Peter Maydell [Tue, 13 Mar 2018 16:26:44 +0000 (16:26 +0000)]
Merge remote-tracking branch 'remotes/ehabkost/tags/python-next-pull-request' into staging
Python queue, 2018-03-12
# gpg: Signature made Mon 12 Mar 2018 22:10:36 GMT
# gpg: using RSA key 2807936F984DC5A6
# gpg: Good signature from "Eduardo Habkost <[email protected]>"
# Primary key fingerprint: 5A32 2FD5 ABC4 D3DB ACCF D1AA 2807 936F 984D C5A6
* remotes/ehabkost/tags/python-next-pull-request:
device-crash-test: Use 'python' binary
qmp.py: Encode json data before sending
qemu.py: Use items() instead of iteritems()
device-crash-test: New known crashes
Peter Maydell [Tue, 13 Mar 2018 09:43:43 +0000 (09:43 +0000)]
Merge remote-tracking branch 'remotes/stefanha/tags/tracing-pull-request' into staging
# gpg: Signature made Mon 12 Mar 2018 15:59:54 GMT
# gpg: using RSA key 9CA4ABB381AB73C8
# gpg: Good signature from "Stefan Hajnoczi <[email protected]>"
# gpg: aka "Stefan Hajnoczi <[email protected]>"
# Primary key fingerprint: 8695 A8BF D3F9 7CDA AC35 775A 9CA4 ABB3 81AB 73C8
* remotes/stefanha/tags/tracing-pull-request:
trace: only permit standard C types and fixed size integer types
trace: remove use of QEMU specific types from trace probes
trace: include filename when printing parser error messages
simpletrace: fix timestamp argument type
log-for-trace.h: Split out parts of log.h used by trace.h
Eduardo Habkost [Mon, 12 Mar 2018 18:55:02 +0000 (15:55 -0300)]
qmp.py: Encode json data before sending
On Python 3, json.dumps() return a str object, which can't be
sent directly through a socket and must be encoded into a bytes
object. Use .encode('utf-8'), which will work on both Python 2
and Python 3.
Chao Peng [Sun, 4 Mar 2018 16:48:35 +0000 (00:48 +0800)]
i386: Add Intel Processor Trace feature support
Expose Intel Processor Trace feature to guest.
To make Intel PT live migration safe and get same CPUID information
with same CPU model on diffrent host. CPUID[14] is constant in this
patch. Intel PT use EPT is first supported in IceLake, the CPUID[14]
get on this machine as default value. Intel PT would be disabled
if any machine don't support this minial feature list.
Wanpeng Li [Fri, 9 Feb 2018 14:15:25 +0000 (06:15 -0800)]
target-i386: add KVM_HINTS_DEDICATED performance hint
Add KVM_HINTS_DEDICATED performance hint, guest checks this feature bit
to determine if they run on dedicated vCPUs, allowing optimizations such
as usage of qspinlocks.
Peter Maydell [Mon, 12 Mar 2018 14:06:23 +0000 (14:06 +0000)]
Merge remote-tracking branch 'remotes/kraxel/tags/ui-20180312-pull-request' into staging
gtk,spice: add dmabuf support.
sdl,vnc,gtk: bugfixes.
ui/qapi: add device ID and head parameters to screendump.
build: try improve handling of clang warnings.
* remotes/kraxel/tags/ui-20180312-pull-request:
qapi: Add device ID and head parameters to screendump
spice: add cursor_dmabuf support
spice: add scanout_dmabuf support
spice: drop dprint() debug logging
vnc: deal with surface NULL pointers
ui/gtk-egl: add cursor_dmabuf support
ui/gtk-egl: add scanout_dmabuf support
ui/gtk: use GtkGlArea on wayland only
ui/opengl: Makefile cleanup
ui/gtk: group gtk.mo declarations in Makefile
ui/gtk: make GtkGlArea usage a runtime option
sdl: workaround bug in sdl 2.0.8 headers
make: switch language file build to be gtk module aware
build: try improve handling of clang warnings
Peter Maydell [Mon, 12 Mar 2018 13:21:53 +0000 (13:21 +0000)]
Merge remote-tracking branch 'remotes/dgilbert/tags/pull-migration-20180309a' into staging
Migration pull 2018-03-09
# gpg: Signature made Fri 09 Mar 2018 17:52:46 GMT
# gpg: using RSA key 0516331EBC5BFDE7
# gpg: Good signature from "Dr. David Alan Gilbert (RH2) <[email protected]>"
# Primary key fingerprint: 45F5 C71B 4A0C B7FB 977A 9FA9 0516 331E BC5B FDE7
* remotes/dgilbert/tags/pull-migration-20180309a:
tests: Silence migration-test 'bad' test
migration: fix applying wrong capabilities
migration/block: rename MAX_INFLIGHT_IO to MAX_IO_BUFFERS
migration/block: reset dirty bitmap before read in bulk phase
migration: do not transfer ram during bulk storage migration
migration: fix minor finalize leak
Peter Maydell [Mon, 12 Mar 2018 11:47:52 +0000 (11:47 +0000)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20180309' into staging
target-arm queue:
* i.MX: Add i.MX7 SOC implementation and i.MX7 Sabre board
* Report the correct core count in A53 L2CTLR on the ZynqMP board
* linux-user: preliminary SVE support work (signal handling)
* hw/arm/boot: fix memory leak in case of error loading ELF file
* hw/arm/boot: avoid reading off end of buffer if passed very
small image file
* hw/arm: Use more CONFIG switches for the object files
* target/arm: Add "-cpu max" support
* hw/arm/virt: Support -machine gic-version=max
* hw/sd: improve debug tracing
* hw/sd: sdcard: Add the Tuning Command (CMD 19)
* MAINTAINERS: add Philippe as odd-fixes maintainer for SD
* remotes/pmaydell/tags/pull-target-arm-20180309: (25 commits)
MAINTAINERS: Add entries for SD (SDHCI, SDBus, SDCard)
sdhci: Fix a typo in comment
sdcard: Add the Tuning Command (CMD19)
sdcard: Display which protocol is used when tracing (SD or SPI)
sdcard: Display command name when tracing CMD/ACMD
sdcard: Do not trace CMD55, except when we already expect an ACMD
hw/arm/virt: Support -machine gic-version=max
hw/arm/virt: Add "max" to the list of CPU types "virt" supports
target/arm: Make 'any' CPU just an alias for 'max'
target/arm: Add "-cpu max" support
target/arm: Move definition of 'host' cpu type into cpu.c
target/arm: Query host CPU features on-demand at instance init
arm: avoid heap-buffer-overflow in load_aarch64_image
arm: fix load ELF error leak
hw/arm: Use more CONFIG switches for the object files
aarch64-linux-user: Add support for SVE signal frame records
aarch64-linux-user: Add support for EXTRA signal frame records
aarch64-linux-user: Remove struct target_aux_context
aarch64-linux-user: Split out helpers for guest signal handling
linux-user: Implement aarch64 PR_SVE_SET/GET_VL
...
trace: only permit standard C types and fixed size integer types
Some trace backends will compile code based on the declared trace
events. It should not be assumed that the backends can resolve any QEMU
specific typedefs. So trace events should restrict their argument
types to the standard C types and fixed size integer types. Any complex
pointer types can be declared as "void *" for purposes of trace events,
since nothing will be dereferencing these pointer arguments.
trace: remove use of QEMU specific types from trace probes
Any compound structs / unions / etc, should always be declared as
'void *' pointers, since it cannot be assumed that trace backends
are able to resolve QEMU typedefs.
Stefan Hajnoczi [Thu, 22 Feb 2018 16:39:01 +0000 (16:39 +0000)]
simpletrace: fix timestamp argument type
The timestamp argument to a trace event method is documented as follows:
The method can also take a timestamp argument before the trace event
arguments:
def runstate_set(self, timestamp, new_state):
...
Timestamps have the uint64_t type and are in nanoseconds.
In reality methods with a timestamp argument actually receive a tuple
like (123456789,) as the timestamp argument. This is due to a bug in
simpletrace.py.
This patch unpacks the tuple so that methods receive the correct
timestamp argument type.
Peter Maydell [Tue, 13 Feb 2018 14:00:29 +0000 (14:00 +0000)]
log-for-trace.h: Split out parts of log.h used by trace.h
A persistent build problem we see is where a source file
accidentally omits the #include of log.h. This slips through
local developer testing because if you configure with the
default (log) trace backend trace.h will pull in log.h for you.
Compilation fails only if some other backend is selected.
To make this error cause a compile failure regardless of
the configured trace backend, split out the parts of log.h
that trace.h requires into a new log-for-trace.h header.
Since almost all manual uses of the log.h functions will
use constants or functions which aren't in log-for-trace.h,
this will let us catch missing #include "qemu/log.h" more
consistently.
Stefan Hajnoczi [Wed, 7 Mar 2018 12:46:19 +0000 (12:46 +0000)]
block: make BDRV_POLL_WHILE() re-entrancy safe
Nested BDRV_POLL_WHILE() calls can occur. Currently
assert(!wait_->wakeup) fails in AIO_WAIT_WHILE() when this happens.
This patch converts the bool wait_->need_kick flag to an unsigned
wait_->num_waiters counter.
Nesting works correctly because outer AIO_WAIT_WHILE() callers evaluate
the condition again after the inner caller completes (invoking the inner
caller counts as aio_poll() progress).
Gerd Hoffmann [Fri, 9 Mar 2018 14:37:04 +0000 (15:37 +0100)]
vga: fix region calculation
Typically the scanline length and the line offset are identical. But
in case they are not our calculation for region_end is incorrect. Using
line_offset is fine for all scanlines, except the last one where we have
to use the actual scanline length.
Gerd Hoffmann [Tue, 6 Mar 2018 07:40:48 +0000 (08:40 +0100)]
audio: add module loading support
Make audio_driver_lookup() try load the module in case it doesn't find
the driver in the registry. Also load all modules for -audio-help, so
the help output includes the help text for modular audio drivers.
Gerd Hoffmann [Tue, 6 Mar 2018 07:40:47 +0000 (08:40 +0100)]
audio: add driver registry
Add registry for audio drivers, using the existing audio_driver struct.
Make all drivers register themself. The old list of audio_driver struct
pointers is now a list of audio driver names, specifying the priority
(aka probe order) in case no driver is explicitly asked for.
Peter Maydell [Mon, 12 Mar 2018 10:08:09 +0000 (10:08 +0000)]
Merge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging
Block layer patches
# gpg: Signature made Fri 09 Mar 2018 15:09:20 GMT
# gpg: using RSA key 7F09B272C88F2FD6
# gpg: Good signature from "Kevin Wolf <[email protected]>"
# Primary key fingerprint: DC3D EB15 9A9A F95D 3D74 56FE 7F09 B272 C88F 2FD6
* remotes/kevin/tags/for-upstream: (56 commits)
qemu-iotests: fix 203 migration completion race
iotests: Tweak 030 in order to trigger a race condition with parallel jobs
iotests: Skip test for ENOMEM error
iotests: Mark all tests executable
iotests: Test creating overlay when guest running
qemu-iotests: Test ssh image creation over QMP
qemu-iotests: Test qcow2 over file image creation with QMP
block: Fail bdrv_truncate() with negative size
file-posix: Fix no-op bdrv_truncate() with falloc preallocation
ssh: Support .bdrv_co_create
ssh: Pass BlockdevOptionsSsh to connect_to_ssh()
ssh: QAPIfy host-key-check option
ssh: Use QAPI BlockdevOptionsSsh object
sheepdog: Support .bdrv_co_create
sheepdog: QAPIfy "redundancy" create option
nfs: Support .bdrv_co_create
nfs: Use QAPI options in nfs_client_open()
rbd: Use qemu_rbd_connect() in qemu_rbd_do_create()
rbd: Assign s->snap/image_name in qemu_rbd_open()
rbd: Support .bdrv_co_create
...
Thomas Huth [Mon, 5 Mar 2018 16:37:48 +0000 (17:37 +0100)]
qapi: Add device ID and head parameters to screendump
QEMU's screendump command can only take dumps from the primary display.
When using multiple VGA cards, there is no way to get a dump from a
secondary card or other display heads yet. So let's add a 'device' and
a 'head' parameter to the HMP and QMP commands to be able to specify
alternative devices and heads with the screendump command, too.
Gerd Hoffmann [Thu, 8 Mar 2018 09:06:17 +0000 (10:06 +0100)]
spice: add cursor_dmabuf support
Add support for cursor dmabufs. qemu has to render the cursor for
that, so in case a cursor is present qemu allocates a new dmabuf, blits
the scanout, blends in the pointer and passes on the new dmabuf to
spice-server. Without cursor qemu continues to simply pass on the
scanout dmabuf as-is.
Gerd Hoffmann [Thu, 8 Mar 2018 16:18:03 +0000 (17:18 +0100)]
vnc: deal with surface NULL pointers
Secondary displays in multihead setups are allowed to have a NULL
DisplaySurface. Typically user interfaces handle this by hiding the
window which shows the display in question.
This isn't an option for vnc though because it simply hasn't a concept
of windows or outputs. So handle the situation by showing a placeholder
DisplaySurface instead. Also check in console_select whenever a surface
is preset in the first place before requesting an update.
This fixes a segfault which can be triggered by switching to an unused
display (via vtrl-alt-<nr>) in a multihead setup, for example using
-device virtio-vga,max_outputs=2.
Gerd Hoffmann [Tue, 6 Mar 2018 09:09:49 +0000 (10:09 +0100)]
ui/gtk: use GtkGlArea on wayland only
For dma-buf support we need a egl context. The gtk x11 backend uses glx
contexts though. We can't use the GtkGlArea widget on x11 because of
that, so use our own gtk-egl code instead. wayland continues to use
the GtkGlArea widget.
Gerd Hoffmann [Fri, 9 Mar 2018 13:59:45 +0000 (14:59 +0100)]
build: try improve handling of clang warnings
This patch disables the pragma diagnostic -Wunused-but-set-variable for
clang in util/coroutine-ucontext.c.
This in turn allows us to remove it from the configure check, so the
CONFIG_PRAGMA_DIAGNOSTIC_AVAILABLE will succeed for clang.
With that in place clang builds (linux) will use -Werror by default,
which breaks the build due to warning about unaligned struct members.
Just turning off this warning isn't a good idea as it indicates
portability problems. So make it a warning again, using
-Wno-error=address-of-packed-member. That way it doesn't break the
build but still shows up in the logs.
Now clang builds qemu without errors. Well, almost. There are some
left in the rdma code. Leaving that to the rdma people. All others can
use --disable-rdma to workarounds this.
In 2c9bb29703c I added a migration test that purposely fails;
unfortunately it prints a copy of the failure message to stderr
which makes the output a bit messy.
Peter Lieven [Thu, 8 Mar 2018 11:18:26 +0000 (12:18 +0100)]
migration/block: rename MAX_INFLIGHT_IO to MAX_IO_BUFFERS
this actually limits (as the original commit mesage suggests) the
number of I/O buffers that can be allocated and not the number
of parallel (inflight) I/O requests.
Peter Lieven [Thu, 8 Mar 2018 11:18:24 +0000 (12:18 +0100)]
migration: do not transfer ram during bulk storage migration
this patch makes the bulk phase of a block migration to take
place before we start transferring ram. As the bulk block migration
can take a long time its pointless to transfer ram during that phase.
Direct leak of 48 byte(s) in 1 object(s) allocated from:
#0 0x7f60efba1a38 in __interceptor_calloc (/lib64/libasan.so.4+0xdea38)
#1 0x7f60eef3cf75 in g_malloc0 ../glib/gmem.c:124
#2 0x55ca9094702c in error_copy /home/elmarco/src/qemu/util/error.c:203
#3 0x55ca9037a30f in migrate_set_error /home/elmarco/src/qemu/migration/migration.c:1139
#4 0x55ca9037a462 in migrate_fd_error /home/elmarco/src/qemu/migration/migration.c:1150
#5 0x55ca9038162b in migrate_fd_connect /home/elmarco/src/qemu/migration/migration.c:2411
#6 0x55ca90386e41 in migration_channel_connect /home/elmarco/src/qemu/migration/channel.c:81
#7 0x55ca9038335e in socket_outgoing_migration /home/elmarco/src/qemu/migration/socket.c:85
#8 0x55ca9083dd3a in qio_task_complete /home/elmarco/src/qemu/io/task.c:142
#9 0x55ca9083d6cc in gio_task_thread_result /home/elmarco/src/qemu/io/task.c:88
#10 0x7f60eef37317 in g_idle_dispatch ../glib/gmain.c:5552
#11 0x7f60eef3490b in g_main_dispatch ../glib/gmain.c:3182
#12 0x7f60eef357ac in g_main_context_dispatch ../glib/gmain.c:3847
#13 0x55ca90927231 in glib_pollfds_poll /home/elmarco/src/qemu/util/main-loop.c:214
#14 0x55ca90927420 in os_host_main_loop_wait /home/elmarco/src/qemu/util/main-loop.c:261
#15 0x55ca909275fa in main_loop_wait /home/elmarco/src/qemu/util/main-loop.c:515
#16 0x55ca8fc1c2a4 in main_loop /home/elmarco/src/qemu/vl.c:1942
#17 0x55ca8fc2eb3a in main /home/elmarco/src/qemu/vl.c:4724
#18 0x7f60e4082009 in __libc_start_main (/lib64/libc.so.6+0x21009)
Indirect leak of 45 byte(s) in 1 object(s) allocated from:
#0 0x7f60efba1850 in malloc (/lib64/libasan.so.4+0xde850)
#1 0x7f60eef3cf0c in g_malloc ../glib/gmem.c:94
#2 0x7f60eef3d1cf in g_malloc_n ../glib/gmem.c:331
#3 0x7f60eef596eb in g_strdup ../glib/gstrfuncs.c:363
#4 0x55ca90947085 in error_copy /home/elmarco/src/qemu/util/error.c:204
#5 0x55ca9037a30f in migrate_set_error /home/elmarco/src/qemu/migration/migration.c:1139
#6 0x55ca9037a462 in migrate_fd_error /home/elmarco/src/qemu/migration/migration.c:1150
#7 0x55ca9038162b in migrate_fd_connect /home/elmarco/src/qemu/migration/migration.c:2411
#8 0x55ca90386e41 in migration_channel_connect /home/elmarco/src/qemu/migration/channel.c:81
#9 0x55ca9038335e in socket_outgoing_migration /home/elmarco/src/qemu/migration/socket.c:85
#10 0x55ca9083dd3a in qio_task_complete /home/elmarco/src/qemu/io/task.c:142
#11 0x55ca9083d6cc in gio_task_thread_result /home/elmarco/src/qemu/io/task.c:88
#12 0x7f60eef37317 in g_idle_dispatch ../glib/gmain.c:5552
#13 0x7f60eef3490b in g_main_dispatch ../glib/gmain.c:3182
#14 0x7f60eef357ac in g_main_context_dispatch ../glib/gmain.c:3847
#15 0x55ca90927231 in glib_pollfds_poll /home/elmarco/src/qemu/util/main-loop.c:214
#16 0x55ca90927420 in os_host_main_loop_wait /home/elmarco/src/qemu/util/main-loop.c:261
#17 0x55ca909275fa in main_loop_wait /home/elmarco/src/qemu/util/main-loop.c:515
#18 0x55ca8fc1c2a4 in main_loop /home/elmarco/src/qemu/vl.c:1942
#19 0x55ca8fc2eb3a in main /home/elmarco/src/qemu/vl.c:4724
#20 0x7f60e4082009 in __libc_start_main (/lib64/libc.so.6+0x21009)
From the "Physical Layer Simplified Specification Version 3.01":
A known data block ("Tuning block") can be used to tune sampling
point for tuning required hosts. [...]
This procedure gives the system optimal timing for each specific
host and card combination and compensates for static delays in
the timing budget including process, voltage and different PCB
loads and skews. [...]
Data block, carried by DAT[3:0], contains a pattern for tuning
sampling position to receive data on the CMD and DAT[3:0] line.
Peter Maydell [Fri, 9 Mar 2018 17:09:44 +0000 (17:09 +0000)]
hw/arm/virt: Support -machine gic-version=max
Add support for passing 'max' to -machine gic-version. By analogy
with the -cpu max option, this picks the "best available" GIC version
whether you're using KVM or TCG, so it behaves like 'host' when
using KVM, and gives you GICv3 when using TCG.
Also like '-cpu host', using -machine gic-version=max' means there
is no guarantee of migration compatibility between QEMU versions;
in future 'max' might mean '4'.
Peter Maydell [Fri, 9 Mar 2018 17:09:44 +0000 (17:09 +0000)]
target/arm: Make 'any' CPU just an alias for 'max'
Now we have a working '-cpu max', the linux-user-only
'any' CPU is pretty much the same thing, so implement it
that way.
For the moment we don't add any of the extra feature bits
to the system-emulation "max", because we don't set the
ID register bits we would need to to advertise those
features as present.
Peter Maydell [Fri, 9 Mar 2018 17:09:44 +0000 (17:09 +0000)]
target/arm: Add "-cpu max" support
Add support for "-cpu max" for ARM guests. This CPU type behaves
like "-cpu host" when KVM is enabled, and like a system CPU with
the maximum possible feature set otherwise. (Note that this means
it won't be migratable across versions, as we will likely add
features to it in future.)
Peter Maydell [Fri, 9 Mar 2018 17:09:44 +0000 (17:09 +0000)]
target/arm: Move definition of 'host' cpu type into cpu.c
Move the definition of the 'host' cpu type into cpu.c, where all the
other CPU types are defined. We can do this now we've decoupled it
from the KVM-specific host feature probing. This means we now create
the type unconditionally (assuming we were built with KVM support at
all), but if you try to use it without -enable-kvm this will end
up in the "host cpu probe failed and KVM not enabled" path in
arm_cpu_realizefn(), for an appropriate error message.
Peter Maydell [Fri, 9 Mar 2018 17:09:44 +0000 (17:09 +0000)]
target/arm: Query host CPU features on-demand at instance init
Currently we query the host CPU features in the class init function
for the TYPE_ARM_HOST_CPU class, so that we can later copy them
from the class object into the instance object in the object
instance init function. This is awkward for implementing "-cpu max",
which should work like "-cpu host" for KVM but like "cpu with all
implemented features" for TCG.
Move the place where we store the information about the host CPU from
a class object to static variables in kvm.c, and then in the instance
init function call a new kvm_arm_set_cpu_features_from_host()
function which will query the host kernel if necessary and then
fill in the CPU instance fields.
This allows us to drop the special class struct and class init
function for TYPE_ARM_HOST_CPU entirely.
We can't delay the probe until realize, because the ARM
instance_post_init hook needs to look at the feature bits we
set, so we need to do it in the initfn. This is safe because
the probing doesn't affect the actual VM state (it creates a
separate scratch VM to do its testing), but the probe might fail.
Because we can't report errors in retrieving the host features
in the initfn, we check this belatedly in the realize function
(the intervening code will be able to cope with the relevant
fields in the CPU structure being zero).
Spotted by ASAN:
QTEST_QEMU_BINARY=aarch64-softmmu/qemu-system-aarch64 tests/boot-serial-test
Direct leak of 48 byte(s) in 1 object(s) allocated from:
#0 0x7ff8a9b0ca38 in __interceptor_calloc (/lib64/libasan.so.4+0xdea38)
#1 0x7ff8a8ea7f75 in g_malloc0 ../glib/gmem.c:124
#2 0x55fef3d99129 in error_setv /home/elmarco/src/qemu/util/error.c:59
#3 0x55fef3d99738 in error_setg_internal /home/elmarco/src/qemu/util/error.c:95
#4 0x55fef323acb2 in load_elf_hdr /home/elmarco/src/qemu/hw/core/loader.c:393
#5 0x55fef2d15776 in arm_load_elf /home/elmarco/src/qemu/hw/arm/boot.c:830
#6 0x55fef2d16d39 in arm_load_kernel_notify /home/elmarco/src/qemu/hw/arm/boot.c:1022
#7 0x55fef3dc634d in notifier_list_notify /home/elmarco/src/qemu/util/notify.c:40
#8 0x55fef2fc3182 in qemu_run_machine_init_done_notifiers /home/elmarco/src/qemu/vl.c:2716
#9 0x55fef2fcbbd1 in main /home/elmarco/src/qemu/vl.c:4679
#10 0x7ff89dfed009 in __libc_start_main (/lib64/libc.so.6+0x21009)
Thomas Huth [Fri, 9 Mar 2018 17:09:44 +0000 (17:09 +0000)]
hw/arm: Use more CONFIG switches for the object files
A lot of ARM object files are linked into the executable unconditionally,
even though we have corresponding CONFIG switches like CONFIG_PXA2XX or
CONFIG_OMAP. We should make sure to use these switches in the Makefile so
that the users can disable certain unwanted boards and devices more easily.
While we're at it, also add some new switches for the boards that do not
have a CONFIG option yet.
aarch64-linux-user: Add support for SVE signal frame records
Depending on the currently selected size of the SVE vector registers,
we can either store the data within the "standard" allocation, or we
may beedn to allocate additional space with an EXTRA record.
projects / qemu.git / log