Max Filippov [Mon, 8 May 2017 18:00:56 +0000 (11:00 -0700)]
hw/xtensa: sim: use g_string/g_new
Replace malloc/free/sprintf with g_string/g_string_printf/g_string_free.
Replace g_malloc with g_new when allocating the MemoryRegion to get more
type safety.
NetBSD ships with traditional BSD curses with compatibility with ncurses.
qemu works nicely with the basesystem version of curses(3) from NetBSD.
The only mismatch between curses(3) and ncurses is the lack of
curses_version() in the NetBSD version. This function is used solely in
the configure script, therefore eliminate it from the curses(3) detection.
With this change applied, configure detects correctly curses frontend.
Thomas Huth [Mon, 22 May 2017 15:40:54 +0000 (17:40 +0200)]
tests/libqtest: Print error instead of aborting when env variable is missing
When you currently try to run a test directly from the command line
without setting the QTEST_QEMU_BINARY environment variable first,
you are presented with an unhelpful assertion message like this:
Thomas Huth [Mon, 22 May 2017 20:53:29 +0000 (22:53 +0200)]
qemu-doc: Move the qemu-ga description into a separate chapter
The qemu-ga description is currently a subsection of the Disk Images
chapter - which does not make much sense since the qemu-ga is not
directly related to disk images. So let's move this information
into a separate chapter instead.
Stefan Hajnoczi [Fri, 2 Jun 2017 09:54:24 +0000 (10:54 +0100)]
virtio-serial: fix segfault on disconnect
Since commit d4c19cdeeb2f1e474bc426a6da261f1d7346eb5b ("virtio-serial:
add missing virtio_detach_element() call") the following commands may
cause QEMU to segfault:
The segfault is non-deterministic: if the event loop notices the socket
has been closed then there is no crash. The disconnect has to happen
right before QEMU attempts to write data to the socket.
The backtrace is as follows:
Thread 1 "qemu-system-x86" received signal SIGSEGV, Segmentation fault.
0x00005555557e0698 in do_flush_queued_data (port=0x5555582cedf0, vq=0x7fffcc854290, vdev=0x55555807b1d0) at hw/char/virtio-serial-bus.c:180
180 for (i = port->iov_idx; i < port->elem->out_num; i++) {
#1 0x000055555580d363 in virtio_queue_notify_vq (vq=0x7fffcc854290) at hw/virtio/virtio.c:1524
#2 0x000055555580d363 in virtio_queue_host_notifier_read (n=0x7fffcc8542f8) at hw/virtio/virtio.c:2430
#3 0x0000555555b3482c in aio_dispatch_handlers (ctx=ctx@entry=0x5555566b8c80) at util/aio-posix.c:399
#4 0x0000555555b350d8 in aio_dispatch (ctx=0x5555566b8c80) at util/aio-posix.c:430
#5 0x0000555555b3212e in aio_ctx_dispatch (source=<optimized out>, callback=<optimized out>, user_data=<optimized out>) at util/async.c:261
#6 0x00007fffde71de52 in g_main_context_dispatch () at /lib64/libglib-2.0.so.0
#7 0x0000555555b34353 in glib_pollfds_poll () at util/main-loop.c:213
#8 0x0000555555b34353 in os_host_main_loop_wait (timeout=<optimized out>) at util/main-loop.c:261
#9 0x0000555555b34353 in main_loop_wait (nonblocking=<optimized out>) at util/main-loop.c:517
#10 0x0000555555773207 in main_loop () at vl.c:1917
#11 0x0000555555773207 in main (argc=<optimized out>, argv=<optimized out>, envp=<optimized out>) at vl.c:4751
The do_flush_queued_data() function does not anticipate chardev close
events during vsc->have_data(). It expects port->elem to remain
non-NULL for the duration its for loop.
The fix is simply to return from do_flush_queued_data() if the port
closes because the close event already frees port->elem and drains the
virtqueue - there is nothing left for do_flush_queued_data() to do.
Ladi Prosek [Tue, 30 May 2017 08:59:43 +0000 (10:59 +0200)]
virtio-serial-bus: Unset hotplug handler when unrealize
Virtio serial device controls the lifetime of virtio-serial-bus and
virtio-serial-bus links back to the device via its hotplug-handler
property. This extra ref-count prevents the device from getting
finalized, leaving the VirtIODevice memory listener registered and
leading to use-after-free later on.
This patch addresses the same issue as Fam Zheng's
"virtio-scsi: Unset hotplug handler when unrealize"
only for a different virtio device.
Niels de Vos [Sun, 28 May 2017 06:31:14 +0000 (12:01 +0530)]
gluster: add support for PREALLOC_MODE_FALLOC
Add missing support for "preallocation=falloc" to the Gluster block
driver. This change bases its logic on that of block/file-posix.c and
removed the gluster_supports_zerofill() and qemu_gluster_zerofill()
functions in favour of #ifdef checks in an easy to read
switch-statement.
Both glfs_zerofill() and glfs_fallocate() have been introduced with
GlusterFS 3.5.0 (pkg-config glusterfs-api = 6). A #define for the
availability of glfs_fallocate() has been added to ./configure.
# gpg: Signature made Thu 01 Jun 2017 17:51:04 BST
# gpg: using RSA key 0xF487EF185872D723
# gpg: Good signature from "Juan Quintela <[email protected]>"
# gpg: aka "Juan Quintela <[email protected]>"
# Primary key fingerprint: 1899 FF8E DEBF 58CC EE03 4B82 F487 EF18 5872 D723
* remotes/juanquintela/tags/migration/20170601:
migration: Move include/migration/block.h into migration/
migration: Export ram.c functions in its own file
migration: Create include for migration snapshots
migration: Export rdma.c functions in its own file
migration: Export tls.c functions in its own file
migration: Export socket.c functions in its own file
migration: Export fd.c functions in its own file
migration: Export exec.c functions in its own file
migration: Split qemu-file.h
migration: Remove unneeded includes of migration/vmstate.h
migration: shut src return path unconditionally
migration: fix leak of src file on dst
migration: Remove section_id parameter from vmstate_load
migration: loadvm handlers are not used
migration: Use savevm_handlers instead of loadvm copy
Peter Maydell [Fri, 2 Jun 2017 12:05:06 +0000 (13:05 +0100)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20170602' into staging
target-arm queue:
* virt: numa: provide ACPI distance info when needed
* aspeed: fix i2c controller bugs
* M profile: support MPU
* gicv3: fix mishandling of BPR1, VBPR1
* load_uboot_image: don't assume a full header read
* libvixl: Correct build failures on NetBSD
* remotes/pmaydell/tags/pull-target-arm-20170602: (25 commits)
hw/arm/virt: fdt: generate distance-map when needed
hw/arm/virt-acpi-build: build SLIT when needed
aspeed: add some I2C devices to the Aspeed machines
aspeed/i2c: introduce a state machine
aspeed/i2c: handle LAST command under the RX command
aspeed/i2c: improve command handling
arm: Implement HFNMIENA support for M profile MPU
arm: add MPU support to M profile CPUs
armv7m: Classify faults as MemManage or BusFault
arm: All M profile cores are PMSA
armv7m: Implement M profile default memory map
armv7m: Improve "-d mmu" tracing for PMSAv7 MPU
arm: Remove unnecessary check on cpu->pmsav7_dregion
arm: Don't let no-MPU PMSA cores write to SCTLR.M
arm: Don't clear ARM_FEATURE_PMSA for no-mpu configs
arm: Clean up handling of no-MPU PMSA CPUs
arm: Use different ARMMMUIdx values for M profile
arm: Add support for M profile CPUs having different MMU index semantics
arm: Use the mmu_idx we're passed in arm_cpu_do_unaligned_access()
target/arm: clear PMUVER field of AA64DFR0 when vPMU=off
...
The Aspeed I2C controller maintains a state machine in the command
register, which is mostly used for debug.
Let's start adding a few states to handle abnormal STOP
commands. Today, the model uses the busy status of the bus as a
condition to do so but it is not precise enough.
Also remove the ABNORMAL bit for failing TX commands. This is
incorrect with respect to the specs.
Peter Maydell [Fri, 2 Jun 2017 10:51:49 +0000 (11:51 +0100)]
arm: Implement HFNMIENA support for M profile MPU
Implement HFNMIENA support for the M profile MPU. This bit controls
whether the MPU is treated as enabled when executing at execution
priorities of less than zero (in NMI, HardFault or with the FAULTMASK
bit set).
Doing this requires us to use a different MMU index for "running
at execution priority < 0", because we will have different
access permissions for that case versus the normal case.
The M series MPU is almost the same as the already implemented R
profile MPU (v7 PMSA). So all we need to implement here is the MPU
register interface in the system register space.
This implementation has the same restriction as the R profile MPU
that it doesn't permit regions to be sized down smaller than 1K.
We also do not yet implement support for MPU_CTRL.HFNMIENA; this
bit should if zero disable use of the MPU when running HardFault,
NMI or with FAULTMASK set to 1 (ie at an execution priority of
less than zero) -- if the MPU is enabled we don't treat these
cases any differently.
Signed-off-by: Michael Davidsaver <[email protected]>
Message-id: 1493122030[email protected]
[PMM: Keep all the bits in mpu_ctrl field, rather than
using SCTLR bits for them; drop broken HFNMIENA support;
various cleanup] Signed-off-by: Peter Maydell <[email protected]>
General logic is that operations stopped by the MPU are MemManage,
and those which go through the MPU and are caught by the unassigned
handle are BusFault. Distinguish these by looking at the
exception.fsr values, and set the CFSR bits and (if appropriate)
fill in the BFAR or MMFAR with the exception address.
Peter Maydell [Fri, 2 Jun 2017 10:51:48 +0000 (11:51 +0100)]
arm: All M profile cores are PMSA
All M profile CPUs are PMSA, so set the feature bit.
(We haven't actually implemented the M profile MPU register
interface yet, but setting this feature bit gives us closer
to correct behaviour for the MPU-disabled case.)
Add support for the M profile default memory map which is used
if the MPU is not present or disabled.
The main differences in behaviour from implementing this
correctly are that we set the PAGE_EXEC attribute on
the right regions of memory, such that device regions
are not executable.
Signed-off-by: Michael Davidsaver <[email protected]>
Message-id: 1493122030[email protected]
[PMM: rephrased comment and commit message; don't mark
the flash memory region as not-writable; list all
the cases in the default map explicitly rather than
using a 'default' case for the non-executable regions] Signed-off-by: Peter Maydell <[email protected]>
Improve the "-d mmu" tracing for the PMSAv7 MPU translation
process as an aid in debugging guest MPU configurations:
* fix a missing newline for a guest-error log
* report the region number with guest-error or unimp
logs of bad region register values
* add a log message for the overall result of the lookup
* print "0x" prefix for hex values
Peter Maydell [Fri, 2 Jun 2017 10:51:48 +0000 (11:51 +0100)]
arm: Remove unnecessary check on cpu->pmsav7_dregion
Now that we enforce both:
* pmsav7_dregion == 0 implies has_mpu == false
* PMSA with has_mpu == false means SCTLR.M cannot be set
we can remove a check on pmsav7_dregion from get_phys_addr_pmsav7(),
because we can only reach this code path if the MPU is enabled
(and so region_translation_disabled() returned false).
Peter Maydell [Fri, 2 Jun 2017 10:51:47 +0000 (11:51 +0100)]
arm: Don't clear ARM_FEATURE_PMSA for no-mpu configs
Fix the handling of QOM properties for PMSA CPUs with no MPU:
Allow no-MPU to be specified by either:
* has-mpu = false
* pmsav7_dregion = 0
and make setting one imply the other. Don't clear the PMSA
feature bit in this situation.
Peter Maydell [Fri, 2 Jun 2017 10:51:47 +0000 (11:51 +0100)]
arm: Clean up handling of no-MPU PMSA CPUs
ARM CPUs come in two flavours:
* proper MMU ("VMSA")
* only an MPU ("PMSA")
For PMSA, the MPU may be implemented, or not (in which case there
is default "always acts the same" behaviour, but it isn't guest
programmable).
QEMU is a bit confused about how we indicate this: we have an
ARM_FEATURE_MPU, but it's not clear whether this indicates
"PMSA, not VMSA" or "PMSA and MPU present" , and sometimes we
use it for one purpose and sometimes the other.
Currently trying to implement a PMSA-without-MPU core won't
work correctly because we turn off the ARM_FEATURE_MPU bit
and then a lot of things which should still exist get
turned off too.
As the first step in cleaning this up, rename the feature
bit to ARM_FEATURE_PMSA, which indicates a PMSA CPU (with
or without MPU).
Peter Maydell [Fri, 2 Jun 2017 10:51:47 +0000 (11:51 +0100)]
arm: Use different ARMMMUIdx values for M profile
Make M profile use completely separate ARMMMUIdx values from
those that A profile CPUs use. This is a prelude to adding
support for the MPU and for v8M, which together will require
6 MMU indexes which don't map cleanly onto the A profile
uses:
non secure User
non secure Privileged
non secure Privileged, execution priority < 0
secure User
secure Privileged
secure Privileged, execution priority < 0
Peter Maydell [Fri, 2 Jun 2017 10:51:47 +0000 (11:51 +0100)]
arm: Add support for M profile CPUs having different MMU index semantics
The M profile CPU's MPU has an awkward corner case which we
would like to implement with a different MMU index.
We can avoid having to bump the number of MMU modes ARM
uses, because some of our existing MMU indexes are only
used by non-M-profile CPUs, so we can borrow one.
To avoid that getting too confusing, clean up the code
to try to keep the two meanings of the index separate.
Instead of ARMMMUIdx enum values being identical to core QEMU
MMU index values, they are now the core index values with some
high bits set. Any particular CPU always uses the same high
bits (so eventually A profile cores and M profile cores will
use different bits). New functions arm_to_core_mmu_idx()
and core_to_arm_mmu_idx() convert between the two.
In general core index values are stored in 'int' types, and
ARM values are stored in ARMMMUIdx types.
Peter Maydell [Fri, 2 Jun 2017 10:51:47 +0000 (11:51 +0100)]
arm: Use the mmu_idx we're passed in arm_cpu_do_unaligned_access()
When identifying the DFSR format for an alignment fault, use
the mmu index that we are passed, rather than calling cpu_mmu_index()
to get the mmu index for the current CPU state. This doesn't actually
make any difference since the only cases where the current MMU index
differs from the index used for the load are the "unprivileged
load/store" instructions, and in that case the mmu index may
differ but the translation regime is the same (apart from the
"use from Hyp mode" case which is UNPREDICTABLE).
However it's the more logical thing to do.
Wei Huang [Fri, 2 Jun 2017 10:51:47 +0000 (11:51 +0100)]
target/arm: clear PMUVER field of AA64DFR0 when vPMU=off
The PMUv3 driver of linux kernel (in arch/arm64/kernel/perf_event.c)
relies on the PMUVER field of id_aa64dfr0_el1 to decide if PMU support
is present or not. This patch clears the PMUVER field under TCG mode
when vPMU=off. Without it, PMUv3 will init insider guest VMs even
with vPMU=off. This patch also removes a redundant line inside the
if-statement.
Peter Maydell [Fri, 2 Jun 2017 10:51:47 +0000 (11:51 +0100)]
hw/intc/arm_gicv3_cpuif: Fix priority masking for NS BPR1
When we calculate the mask to use to get the group priority from
an interrupt priority, the way that NS BPR1 is handled differs
from how BPR0 and S BPR1 work -- a BPR1 value of 1 means
the group priority is in bits [7:1], whereas for BPR0 and S BPR1
this is indicated by a 0 BPR value.
Subtract 1 from the BPR value before creating the mask if
we're using the NS BPR value, for both hardware and virtual
interrupts, as the GICv3 pseudocode does, and fix the comments
accordingly.
Peter Maydell [Fri, 2 Jun 2017 10:51:47 +0000 (11:51 +0100)]
hw/intc/arm_gicv3_cpuif: Don't let BPR be set below its minimum
icc_bpr_write() was not enforcing that writing a value below the
minimum for the BPR should behave as if the BPR was set to the
minimum value. This doesn't make a difference for the secure
BPRs (since we define the minimum for the QEMU implementation
as zero) but did mean we were allowing the NS BPR1 to be set to
0 when 1 should be the lowest value.
Peter Maydell [Fri, 2 Jun 2017 10:51:46 +0000 (11:51 +0100)]
hw/intc/arm_gicv3_cpuif: Fix reset value for VMCR_EL2.VBPR1
We were setting the VBPR1 field of VMCR_EL2 to icv_min_vbpr()
on reset, but this is not correct. The field should reset to
the minimum value of ICV_BPR0_EL1 plus one.
Ensure that C99 macros are defined regardless of the inclusion order of
headers in vixl. This is required at least on NetBSD.
The vixl/globals.h headers defines __STDC_CONSTANT_MACROS and must be
included before other system headers.
This file defines unconditionally the following macros, without altering
the original sources:
- __STDC_CONSTANT_MACROS
- __STDC_LIMIT_MACROS
- __STDC_FORMAT_MACROS
qemu_chr_fe_write() is similar to qemu_chr_write_all(): the later write
all with a chardev backend.
Make qemu_chr_write() and qemu_chr_fe_write_buffer() take an 'all'
argument. If false, handle 'partial' write the way qemu_chr_fe_write()
use to, and call qemu_chr_write() from qemu_chr_fe_write().
char: cast ARRAY_SIZE() as signed to silent warning on empty array
chardev/char.c: In function 'chardev_name_foreach':
chardev/char.c:546:19: error: comparison of unsigned expression < 0 is always false [-Werror=type-limits]
for (i = 0; i < ARRAY_SIZE(chardev_alias_table); i++) {
^ Signed-off-by: Philippe Mathieu-Daudé <[email protected]>
Message-Id: <20170530120919[email protected]> Reviewed-by: Marc-André Lureau <[email protected]>
xiaoqiang zhao [Thu, 25 May 2017 13:34:48 +0000 (21:34 +0800)]
hw/timer: QOM'ify m48txx_sysbus
* split the old SysBus init function into an instance_init
and a Device realize function
* use DeviceClass::realize instead of SysBusDeviceClass::init
* assign DeviceClass::vmsd instead of using vmstate_register function
Juan Quintela [Wed, 5 Apr 2017 15:45:16 +0000 (17:45 +0200)]
migration: Export tls.c functions in its own file
Just for the functions exported from tls.c. Notice that we can't
remove the migration/migration.h include from tls.c because it access
directly MigrationState for the tls params.
Juan Quintela [Thu, 20 Apr 2017 16:52:18 +0000 (18:52 +0200)]
migration: Split qemu-file.h
Split the file into public and internal interfaces. I have to rename
the external one because we can't have two include files with the same
name in the same directory. Build system gets confused. The only
exported functions are the ones that handle basic types.
Peter Maydell [Thu, 1 Jun 2017 15:39:16 +0000 (16:39 +0100)]
Merge remote-tracking branch 'remotes/armbru/tags/pull-qapi-2017-05-31' into staging
QAPI patches for 2017-05-31
# gpg: Signature made Wed 31 May 2017 18:06:39 BST
# gpg: using RSA key 0x3870B400EB918653
# gpg: Good signature from "Markus Armbruster <[email protected]>"
# gpg: aka "Markus Armbruster <[email protected]>"
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867 4E5F 3870 B400 EB91 8653
* remotes/armbru/tags/pull-qapi-2017-05-31:
qapi: Reject alternates that can't work with keyval_parse()
tests/qapi-schema: Avoid 'str' in alternate test cases
qapi: Document visit_type_any() issues with keyval input
qobject-input-visitor: Reject non-finite numbers with keyval
Peter Maydell [Thu, 1 Jun 2017 14:50:40 +0000 (15:50 +0100)]
Merge remote-tracking branch 'remotes/riku/tags/pull-linux-user-20170531' into staging
Misc linux-user updates
# gpg: Signature made Wed 31 May 2017 12:33:17 BST
# gpg: using RSA key 0xB44890DEDE3C9BC0
# gpg: Good signature from "Riku Voipio <[email protected]>"
# gpg: aka "Riku Voipio <[email protected]>"
# Primary key fingerprint: FF82 03C8 C391 98AE 0581 41EF B448 90DE DE3C 9BC0
* remotes/riku/tags/pull-linux-user-20170531:
linux-user: add strace support for uinfo structure of rt_sigqueueinfo() and rt_tgsigqueueinfo()
linux-user: fix inconsistent spaces in print_siginfo() output
linux-user: add rt_tgsigqueueinfo() strace
linux-user: add support for rt_tgsigqueueinfo() system call
linux-user: fix argument type declaration of rt_sigqueinfo() syscall
linux-user: fix mismatch of lock/unlock_user() invocations in rt_sigqueinfo() syscall
linux-user: fix ssetmask() system call
linux-user: add tkill(), tgkill() and rt_sigqueueinfo() strace
linux-user: add strace for getuid(), gettid(), getppid(), geteuid()
linux-user: remove all traces of qemu from /proc/self/cmdline
linux-user: allocate heap memory for execve arguments
linux-user: fix inotify
linux-user: fix fadvise64_64() on ppc
linux-user: fix eventfd
linux-user: call fd_trans_target_to_host_data() for write()
# gpg: Signature made Wed 31 May 2017 08:53:06 BST
# gpg: using RSA key 0xF487EF185872D723
# gpg: Good signature from "Juan Quintela <[email protected]>"
# gpg: aka "Juan Quintela <[email protected]>"
# Primary key fingerprint: 1899 FF8E DEBF 58CC EE03 4B82 F487 EF18 5872 D723
* remotes/juanquintela/tags/migration/20170531:
migration: use dirty_rate_high_cnt more aggressively
migration: set bytes_xfer_* outside of autoconverge logic
migration: set dirty_pages_rate before autoconverge logic
migration: keep bytes_xfer_prev init'd to zero
migration: Create savevm.h for functions exported from savevm.c
Peter Maydell [Thu, 1 Jun 2017 12:12:20 +0000 (13:12 +0100)]
Merge remote-tracking branch 'remotes/aurel/tags/pull-target-sh4-20170530' into staging
Queued target/sh4 patches
# gpg: Signature made Tue 30 May 2017 20:12:10 BST
# gpg: using RSA key 0xBA9C78061DDD8C9B
# gpg: Good signature from "Aurelien Jarno <[email protected]>"
# gpg: aka "Aurelien Jarno <[email protected]>"
# gpg: aka "Aurelien Jarno <[email protected]>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg: It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 7746 2642 A9EF 94FD 0F77 196D BA9C 7806 1DDD 8C9B
* remotes/aurel/tags/pull-target-sh4-20170530:
target/sh4: fix RTE instruction delay slot
target/sh4: ignore interrupts in a delay slot
target/sh4: introduce DELAY_SLOT_MASK
target/sh4: fix reset when using a kernel and an initrd
target/sh4: log unauthorized accesses using qemu_log_mask
Peter Maydell [Thu, 1 Jun 2017 11:06:58 +0000 (12:06 +0100)]
Merge remote-tracking branch 'remotes/gkurz/tags/for-upstream' into staging
Various bugfixes and code cleanups. Most notably, it fixes metadata handling in
mapped-file security mode (especially for the virtfs root).
# gpg: Signature made Tue 30 May 2017 14:36:22 BST
# gpg: using DSA key 0x02FC3AEB0101DBC2
# gpg: Good signature from "Greg Kurz <[email protected]>"
# gpg: aka "Greg Kurz <[email protected]>"
# gpg: aka "Greg Kurz <[email protected]>"
# gpg: aka "Gregory Kurz (Groug) <[email protected]>"
# gpg: aka "[jpeg image of size 3330]"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg: There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 2BD4 3B44 535E C0A7 9894 DBA2 02FC 3AEB 0101 DBC2
* remotes/gkurz/tags/for-upstream:
9pfs: local: metadata file for the VirtFS root
9pfs: local: simplify file opening
9pfs: local: resolve special directories in paths
9pfs: check return value of v9fs_co_name_to_path()
util: drop old utimensat() compat code
9pfs: assume utimensat() and futimens() are present
fsdev: fix virtfs-proxy-helper cwd
9pfs: local: fix unlink of alien files in mapped-file mode
9pfs: drop pdu_push_and_notify()
fsdev: don't allow unknown format in marshal/unmarshal
virtio-9p/xen-9p: move 9p specific bits to core 9p code
qapi: Reject alternates that can't work with keyval_parse()
Alternates are sum types like unions, but use the JSON type on the
wire / QType in QObject instead of an explicit tag. That's why we
require alternate members to have distinct QTypes.
The recently introduced keyval_parse() (commit d454dbe) can only
produce string scalars. The qobject_input_visitor_new_keyval() input
visitor mostly hides the difference, so code using a QObject input
visitor doesn't have to care whether its input was parsed from JSON or
KEY=VALUE,... The difference leaks for alternates, as noted in commit 0ee9ae7: a non-string, non-enum scalar alternate value can't currently
be expressed.
In part, this is just our insufficiently sophisticated implementation.
Consider alternate type 'GuestFileWhence'. It has an integer member
and a 'QGASeek' member. The latter is an enumeration with values
'set', 'cur', 'end'. The meaning of b=set, b=cur, b=end, b=0, b=1 and
so forth is perfectly obvious. However, our current implementation
falls apart at run time for b=0, b=1, and so forth. Fixable, but not
today; add a test case and a TODO comment.
Now consider an alternate type with a string and an integer member.
What's the meaning of a=42? Is it the string "42" or the integer 42?
Whichever meaning you pick makes the other inexpressible. This isn't
just an implementation problem, it's fundamental. Our current
implementation will pick string.
So far, we haven't needed such alternates. To make sure we stop and
think before we add one that cannot sanely work with keyval_parse(),
let's require alternate members to have sufficiently distinct
representation in KEY=VALUE,... syntax:
* A string member clashes with any other scalar member
* An enumeration member clashes with bool members when it has value
'on' or 'off'.
* An enumeration member clashes with numeric members when it has a
value that starts with '-', '+', or a decimal digit. This is a
rather lazy approximation of the actual number syntax accepted by
the visitor.
Note that enumeration values starting with '-' and '+' are rejected
elsewhere already, but better safe than sorry.
tests/qapi-schema: Avoid 'str' in alternate test cases
The next commit is going to make alternate members of type 'str'
conflict with other scalar types. Would break a few test cases that
don't actually require 'str'. Flip them from 'str' to 'bool' or
'EnumOne'.
qobject-input-visitor: Reject non-finite numbers with keyval
The QObject input visitor can produce only finite numbers when its
input comes out of the JSON parser, because the the JSON parser
implements RFC 7159, which provides no syntax for infinity and NaN.
However, it can produce infinity and NaN when its input comes out of
keyval_parse(), because we parse with strtod() then.
The keyval variant should not be able to express things the JSON
variant can't. Rejecting non-finite numbers there is the conservative
fix. It's also minimally invasive.
We could instead extend our JSON dialect to provide for infinity and
NaN. Not today.
Note that the JSON formatter can emit non-finite numbers (marked FIXME
in commit 6e8e5cb).
Felipe Franciosi [Wed, 24 May 2017 16:10:03 +0000 (17:10 +0100)]
migration: use dirty_rate_high_cnt more aggressively
The commit message from 070afca25 suggests that dirty_rate_high_cnt
should be used more aggressively to start throttling after two
iterations instead of four. The code, however, only changes the auto
convergence behaviour to throttle after three iterations. This makes the
behaviour more aggressive by kicking off throttling after two iterations
as originally intended.
Felipe Franciosi [Wed, 24 May 2017 16:10:02 +0000 (17:10 +0100)]
migration: set bytes_xfer_* outside of autoconverge logic
The bytes_xfer_now/prev counters are only used by the auto convergence
logic. However, they are used alongside the dirty_pages_rate counter,
which is calculated (and required) outside of this logic. The problem
with this approach is that if the auto convergence capability is changed
while a migration is ongoing, the relationship of the counters will be
broken.
This moves the management of bytes_xfer_now/prev counters outside of the
auto convergence logic to address this issue.
projects / qemu.git / log