Peter Maydell [Fri, 20 Apr 2018 14:52:43 +0000 (15:52 +0100)]
Change references to serial_hds[] to serial_hd()
Change all the uses of serial_hds[] to go via the new
serial_hd() function. Code change produced with:
find hw -name '*.[ch]' | xargs sed -i -e 's/serial_hds\[\([^]]*\)\]/serial_hd(\1)/g'
Peter Maydell [Fri, 20 Apr 2018 14:52:42 +0000 (15:52 +0100)]
vl.c: Provide accessor function serial_hd() for serial_hds[] array
Provide an accessor function serial_hd() to return the Chardev
(if any) associated with the numbered serial port. This will
be used to replace direct accesses to the serial_hds[] array,
so that calling code doesn't need to care about the size of
that array.
Peter Maydell [Fri, 20 Apr 2018 14:52:41 +0000 (15:52 +0100)]
hw/xtensa/xtfpga.c: Don't create "null" chardevs for serial devices
Following commit 12051d82f004024, UART devices should handle
being passed a NULL pointer chardev, so we don't need to
create "null" backends in board code. Remove the code that
does this and updates serial_hds[].
Peter Maydell [Fri, 20 Apr 2018 14:52:40 +0000 (15:52 +0100)]
hw/mips/mips_malta: Don't create "null" chardevs for serial devices
Following commit 12051d82f004024, UART devices should handle
being passed a NULL pointer chardev, so we don't need to
create "null" backends in board code. Remove the code that
does this and updates serial_hds[].
Peter Maydell [Fri, 20 Apr 2018 14:52:39 +0000 (15:52 +0100)]
hw/mips/boston.c: Don't create "null" chardevs for serial devices
Following commit 12051d82f004024, UART devices should handle
being passed a NULL pointer chardev, so we don't need to
create "null" backends in board code. Remove the code that
does this and updates serial_hds[].
Peter Maydell [Fri, 20 Apr 2018 14:52:38 +0000 (15:52 +0100)]
hw/arm/fsl-imx*: Don't create "null" chardevs for serial devices
Following commit 12051d82f004024, UART devices should handle
being passed a NULL pointer chardev, so we don't need to
create "null" backends in board code. Remove the code that
does this and updates serial_hds[].
Peter Maydell [Fri, 20 Apr 2018 14:52:37 +0000 (15:52 +0100)]
hw/char/serial: Allow disconnected chardevs
Currently the serial.c realize code has an explicit check that it is not
connected to a disconnected backend (ie one with a NULL chardev).
This isn't what we want -- you should be able to create a serial device
even if it isn't attached to anything. Remove the check.
Peter Maydell [Thu, 26 Apr 2018 10:56:57 +0000 (11:56 +0100)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20180426' into staging
target-arm queue:
* xilinx_spips: Correct SNOOP_NONE state when flushing the txfifo
* timer/aspeed: fix vmstate version id
* hw/arm/aspeed_soc: don't use vmstate_register_ram_global for SRAM
* hw/arm/aspeed: don't make 'boot_rom' region 'nomigrate'
* hw/arm/highbank: don't make sysram 'nomigrate'
* hw/arm/raspi: Don't bother setting default_cpu_type
* PMU emulation: some minor bugfixes and preparation for
support of other events than just the cycle counter
* target/arm: Use v7m_stack_read() for reading the frame signature
* target/arm: Remove stale TODO comment
* arm: always start from first_cpu when registering loader cpu reset callback
* device_tree: Increase FDT_MAX_SIZE to 1 MiB
* remotes/pmaydell/tags/pull-target-arm-20180426:
xilinx_spips: Correct SNOOP_NONE state when flushing the txfifo
timer/aspeed: fix vmstate version id
hw/arm/aspeed_soc: don't use vmstate_register_ram_global for SRAM
hw/arm/aspeed: don't make 'boot_rom' region 'nomigrate'
hw/arm/highbank: don't make sysram 'nomigrate'
hw/arm/raspi: Don't bother setting default_cpu_type
target/arm: Make PMOVSCLR and PMUSERENR 64 bits wide
target/arm: Fix bitmask for PMCCFILTR writes
target/arm: Allow EL change hooks to do IO
target/arm: Add pre-EL change hooks
target/arm: Support multiple EL change hooks
target/arm: Fetch GICv3 state directly from CPUARMState
target/arm: Mask PMU register writes based on PMCR_EL0.N
target/arm: Treat PMCCNTR as alias of PMCCNTR_EL0
target/arm: Check PMCNTEN for whether PMCCNTR is enabled
target/arm: Use v7m_stack_read() for reading the frame signature
target/arm: Remove stale TODO comment
arm: always start from first_cpu when registering loader cpu reset callback
device_tree: Increase FDT_MAX_SIZE to 1 MiB
Peter Maydell [Thu, 26 Apr 2018 10:48:20 +0000 (11:48 +0100)]
Open 2.13 development tree
Unfortunately I forgot to do this before applying the merge
in commit 8e383d19b44863556, so that commit will incorrectly
claim to be 2.12 even though it isn't in the official 2.12
release. Oops.
commit 1d3e65aa7ac5 ("hw/timer: Add value matching support to
aspeed_timer") increased the vmstate version of aspeed.timer because
the state had changed, but it also bumped the version of the
VMSTATE_STRUCT_ARRAY under the aspeed.timerctrl which did not need to.
Peter Maydell [Thu, 26 Apr 2018 10:04:39 +0000 (11:04 +0100)]
hw/arm/aspeed_soc: don't use vmstate_register_ram_global for SRAM
Currently we use vmstate_register_ram_global() for the SRAM;
this is not a good idea for devices, because it means that
you can only ever create one instance of the device, as
the second instance would get a RAM block name clash.
Instead, use memory_region_init_ram(), which automatically
registers the RAM block with a local-to-the-device name.
Note that this would be a cross-version migration compatibility break
for the "palmetto-bmc", "ast2500-evb" and "romulus-bmc" machines,
but migration is currently broken for them.
Peter Maydell [Thu, 26 Apr 2018 10:04:39 +0000 (11:04 +0100)]
hw/arm/aspeed: don't make 'boot_rom' region 'nomigrate'
Currently we use memory_region_init_ram_nomigrate() to create
the "aspeed.boot_rom" memory region, and we don't manually
register it with vmstate_register_ram(). This currently
means that its contents are migrated but as a ram block
whose name is the empty string; in future it may mean they
are not migrated at all. Use memory_region_init_ram() instead.
Note that would be a cross-version migration compatibility break
for the "palmetto-bmc", "ast2500-evb" and "romulus-bmc" machines,
but migration is currently broken for them.
Peter Maydell [Thu, 26 Apr 2018 10:04:39 +0000 (11:04 +0100)]
hw/arm/highbank: don't make sysram 'nomigrate'
Currently we use memory_region_init_ram_nomigrate() to create
the "highbank.sysram" memory region, and we don't manually
register it with vmstate_register_ram(). This currently
means that its contents are migrated but as a ram block
whose name is the empty string; in future it may mean they
are not migrated at all. Use memory_region_init_ram() instead.
Note that this is a cross-version migration compatibility
break for the "highbank" and "midway" machines.
In commit 210f47840dd62, we changed the bcm2836 SoC object to
always create a CPU of the correct type for that SoC model. This
makes the default_cpu_type settings in the MachineClass structs
for the raspi2 and raspi3 boards redundant. We didn't change
those at the time because it would have meant a temporary
regression in a corner case of error handling if the user
requested a non-existing CPU type. The -cpu parse handling
changes in 2278b93941d42c3 mean that it no longer implicitly
depends on default_cpu_type for this to work, so we can now
delete the redundant default_cpu_type fields.
During code generation, surround CPSR writes and exception returns which
call the EL change hooks with gen_io_start/end. The immediate need is
for the PMU to access the clock and icount during EL change to support
mode filtering.
Because the design of the PMU requires that the counter values be
converted between their delta and guest-visible forms for mode
filtering, an additional hook which occurs before the EL is changed is
necessary.
target/arm: Fetch GICv3 state directly from CPUARMState
This eliminates the need for fetching it from el_change_hook_opaque, and
allows for supporting multiple el_change_hooks without having to hack
something together to find the registered opaque belonging to GICv3.
Peter Maydell [Thu, 26 Apr 2018 10:04:38 +0000 (11:04 +0100)]
target/arm: Use v7m_stack_read() for reading the frame signature
In commit 95695effe8caa552b8f2 we changed the v7M/v8M stack
pop code to use a new v7m_stack_read() function that checks
whether the read should fail due to an MPU or bus abort.
We missed one call though, the one which reads the signature
word for the callee-saved register part of the frame.
Peter Maydell [Thu, 26 Apr 2018 10:04:38 +0000 (11:04 +0100)]
target/arm: Remove stale TODO comment
Remove a stale TODO comment -- we have now made the arm_ldl_ptw()
and arm_ldq_ptw() functions propagate physical memory read errors
out to their callers.
Igor Mammedov [Thu, 26 Apr 2018 10:04:38 +0000 (11:04 +0100)]
arm: always start from first_cpu when registering loader cpu reset callback
if arm_load_kernel() were passed non first_cpu, QEMU would end up
with partially set do_cpu_reset() callback leaving some CPUs without it.
Make sure that do_cpu_reset() is registered for all CPUs by enumerating
CPUs from first_cpu.
(In practice every board that we have was passing us the first CPU
as the boot CPU, either directly or indirectly, so this wasn't
causing incorrect behaviour.)
# gpg: Signature made Wed 25 Apr 2018 20:21:13 BST
# 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-20180425a:
migration: remove ram_save_compressed_page()
migration: introduce save_normal_page()
migration: move calling save_zero_page to the common place
migration: move calling control_save_page to the common place
migration: move some code to ram_save_host_page
migration: introduce control_save_page()
migration: detect compression and decompression errors
migration: stop decompression to allocate and free memory frequently
migration: stop compression to allocate and free memory frequently
migration: stop compressing page in migration thread
migration: add postcopy total blocktime into query-migrate
migration: add blocktime calculation into migration-test
migration: postcopy_blocktime documentation
migration: calculate vCPU blocktime on dst side
migration: add postcopy blocktime ctx into MigrationIncomingState
migration: introduce postcopy-blocktime capability
Xiao Guangrong [Fri, 30 Mar 2018 07:51:28 +0000 (15:51 +0800)]
migration: remove ram_save_compressed_page()
Now, we can reuse the path in ram_save_page() to post the page out
as normal, then the only thing remained in ram_save_compressed_page()
is compression that we can move it out to the caller
Xiao Guangrong [Fri, 30 Mar 2018 07:51:24 +0000 (15:51 +0800)]
migration: move some code to ram_save_host_page
Move some code from ram_save_target_page() to ram_save_host_page()
to make it be more readable for latter patches that dramatically
clean ram_save_target_page() up
Xiao Guangrong [Fri, 30 Mar 2018 07:51:22 +0000 (15:51 +0800)]
migration: detect compression and decompression errors
Currently the page being compressed is allowed to be updated by
the VM on the source QEMU, correspondingly the destination QEMU
just ignores the decompression error. However, we completely miss
the chance to catch real errors, then the VM is corrupted silently
To make the migration more robuster, we copy the page to a buffer
first to avoid it being written by VM, then detect and handle the
errors of both compression and decompression errors properly
Xiao Guangrong [Fri, 30 Mar 2018 07:51:21 +0000 (15:51 +0800)]
migration: stop decompression to allocate and free memory frequently
Current code uses uncompress() to decompress memory which manages
memory internally, that causes huge memory is allocated and freed
very frequently, more worse, frequently returning memory to kernel
will flush TLBs
So, we maintain the memory by ourselves and reuse it for each
decompression
Xiao Guangrong [Fri, 30 Mar 2018 07:51:20 +0000 (15:51 +0800)]
migration: stop compression to allocate and free memory frequently
Current code uses compress2() to compress memory which manages memory
internally, that causes huge memory is allocated and freed very
frequently
More worse, frequently returning memory to kernel will flush TLBs
and trigger invalidation callbacks on mmu-notification which
interacts with KVM MMU, that dramatically reduce the performance
of VM
So, we maintain the memory by ourselves and reuse it for each
compression
Alexey Perevalov [Thu, 22 Mar 2018 18:17:27 +0000 (21:17 +0300)]
migration: add postcopy total blocktime into query-migrate
Postcopy total blocktime is available on destination side only.
But query-migrate was possible only for source. This patch
adds ability to call query-migrate on destination.
To be able to see postcopy blocktime, need to request postcopy-blocktime
capability.
The query-migrate command will show following sample result:
{"return":
"postcopy-vcpu-blocktime": [115, 100],
"status": "completed",
"postcopy-blocktime": 100
}}
postcopy_vcpu_blocktime contains list, where the first item is the first
vCPU in QEMU.
This patch has a drawback, it combines states of incoming and
outgoing migration. Ongoing migration state will overwrite incoming
state. Looks like better to separate query-migrate for incoming and
outgoing migration or add parameter to indicate type of migration.
Alexey Perevalov [Thu, 22 Mar 2018 18:17:24 +0000 (21:17 +0300)]
migration: calculate vCPU blocktime on dst side
This patch provides blocktime calculation per vCPU,
as a summary and as a overlapped value for all vCPUs.
This approach was suggested by Peter Xu, as an improvements of
previous approch where QEMU kept tree with faulted page address and cpus bitmask
in it. Now QEMU is keeping array with faulted page address as value and vCPU
as index. It helps to find proper vCPU at UFFD_COPY time. Also it keeps
list for blocktime per vCPU (could be traced with page_fault_addr)
Blocktime will not calculated if postcopy_blocktime field of
MigrationIncomingState wasn't initialized.
Alexey Perevalov [Thu, 22 Mar 2018 18:17:23 +0000 (21:17 +0300)]
migration: add postcopy blocktime ctx into MigrationIncomingState
This patch adds request to kernel space for UFFD_FEATURE_THREAD_ID, in
case this feature is provided by kernel.
PostcopyBlocktimeContext is encapsulated inside postcopy-ram.c,
due to it being a postcopy-only feature.
Also it defines PostcopyBlocktimeContext's instance live time.
Information from PostcopyBlocktimeContext instance will be provided
much after postcopy migration end, instance of PostcopyBlocktimeContext
will live till QEMU exit, but part of it (vcpu_addr,
page_fault_vcpu_time) used only during calculation, will be released
when postcopy ended or failed.
To enable postcopy blocktime calculation on destination, need to
request proper compatibility (Patch for documentation will be at the
tail of the patch set).
As an example following command enable that capability, assume QEMU was
started with
-chardev socket,id=charmonitor,path=/var/lib/migrate-vm-monitor.sock
option to control it
Right now it could be used on destination side to
enable vCPU blocktime calculation for postcopy live migration.
vCPU blocktime - it's time since vCPU thread was put into
interruptible sleep, till memory page was copied and thread awake.
Unfortunately this fix regresses console handling on MIPS Malta;
since the mux ctrl-a b bug is not a regression since 2.11, we
take the conservative approach and just drop it from 2.12.
Without bounding the increment, we can overflow exp either here
in scalbn_decomposed or when adding the bias in round_canonical.
This can result in e.g. underflowing to 0 instead of overflowing
to infinity.
Peter Maydell [Mon, 16 Apr 2018 15:19:23 +0000 (16:19 +0100)]
linux-user: check that all of AArch64 SVE extended sigframe is writable
In commit 8c5931de0ac7738809 we added support for SVE extended
sigframe records. These mean that the signal frame might now be
larger than the size of the target_rt_sigframe record, so make sure
we call lock_user on the entire frame size when we're creating it.
(The code for restoring the signal frame already correctly handles
the extended records by locking the 'extra' section separately to the
main section.)
In particular, this fixes a bug even for non-SVE signal frames,
because it extends the locked section to cover the
target_rt_frame_record. Previously this was part of 'struct
target_rt_sigframe', but in commit e1eecd1d9d4c1ade3 we pulled
it out into its own struct, and so locking the target_rt_sigframe
alone doesn't cover it. This bug would mean that we would fail
to correctly handle the case where a signal was taken with
SP pointing 16 bytes into an unwritable page, with the page
immediately below it in memory being writable.
Alex Bennée [Mon, 16 Apr 2018 13:54:42 +0000 (14:54 +0100)]
fpu/softfloat: check for Inf / x or 0 / x before /0
The re-factoring of div_floats changed the order of checking meaning
an operation like -inf/0 erroneously raises the divbyzero flag.
IEEE-754 (2008) specifies this should only occur for operations on
finite operands.
We fix this by moving the check on the dividend being Inf/0 to before
the divisor is zero check.
The assumption in the cpu->max_features code is that anything
enabled on GET_SUPPORTED_CPUID should be enabled on "-cpu host".
This shouldn't be the case for FEAT_KVM_HINTS.
This adds a new FeatureWordInfo::no_autoenable_flags field, that
can be used to prevent FEAT_KVM_HINTS bits to be enabled
automatically.
Jason Wang [Fri, 13 Apr 2018 03:01:49 +0000 (11:01 +0800)]
vhost: do not verify ring mappings when IOMMU is enabled
When IOMMU is enabled, we store virtqueue metadata as iova (though it
may has _phys suffix) and access them through dma helpers. Any
translation failures could be reported by IOMMU.
In this case, trying to validate iova against gpa won't work and will
cause a false error reporting. So this patch bypasses the ring
verification if IOMMU is enabled which is similar to the behavior
before 0ca1fd2d6878 that calls vhost_memory_map() which is a nop when
IOMMU is enabled.
Improve and fix 169:
- use MIGRATION events instead of RESUME
- make a TODO: enable dirty-bitmaps capability for offline case
- recreate vm_b without -incoming near test end
This (likely) fixes racy faults at least of the following types:
- timeout on waiting for RESUME event
- sha256 mismatch on line 136 (142 after this patch)
- fail to self.vm_b.launch() on line 135 (141 now after this patch)
And surely fixes cat processes, left after test finish.
Checking reopen by existence of some bitmaps is wrong, as it may be
some other bitmaps, or on the other hand, user may remove bitmaps. This
criteria is bad. To simplify things and make behavior more predictable
let's just add a flag to remember, that we've already tried to load
bitmaps on open and do not want do it again.
Peter Maydell [Thu, 12 Apr 2018 14:02:22 +0000 (15:02 +0100)]
linux-user/signal.c: Put AArch64 frame record in the right place
AArch64 stack frames include a 'frame record' which holds a pointer
to the next frame record in the chain and the LR on entry to the
function. The procedure calling standard doesn't mandate where
exactly this frame record is in the stack frame, but for signal
frames the kernel puts it right at the top. We used to put it
there too, but in commit 7f0f4208b3a96f22 we accidentally put
the "enlarge to the 4K reserved space minimum" check after the
"allow for the frame record" code, rather than before it, with
the effect that the frame record would be inside the reserved
space and immediately after the last used part of it.
Move the frame record back out of the reserved space to where
we used to put it.
This bug shouldn't break any sensible guest code, but test
programs that deliberately look at the internal details
of the signal frame layout will not find what they are
expecting to see.
Peter Maydell [Fri, 13 Apr 2018 14:23:36 +0000 (15:23 +0100)]
tcg/mips: Handle large offsets from target env to tlb_table
The MIPS TCG target makes the assumption that the offset from the
target env pointer to the tlb_table is less than about 64K. This
used to be true, but gradual addition of features to the Arm
target means that it's no longer true there. This results in
the build-time assertion failing:
In file included from /home/pm215/qemu/include/qemu/osdep.h:36:0,
from /home/pm215/qemu/tcg/tcg.c:28:
/home/pm215/qemu/tcg/mips/tcg-target.inc.c: In function ‘tcg_out_tlb_load’:
/home/pm215/qemu/include/qemu/compiler.h:90:36: error: static assertion failed: "not expecting: offsetof(CPUArchState, tlb_table[NB_MMU_MODES - 1][1]) > 0x7ff0 + 0x7fff"
#define QEMU_BUILD_BUG_MSG(x, msg) _Static_assert(!(x), msg)
^
/home/pm215/qemu/include/qemu/compiler.h:98:30: note: in expansion of macro ‘QEMU_BUILD_BUG_MSG’
#define QEMU_BUILD_BUG_ON(x) QEMU_BUILD_BUG_MSG(x, "not expecting: " #x)
^
/home/pm215/qemu/tcg/mips/tcg-target.inc.c:1236:9: note: in expansion of macro ‘QEMU_BUILD_BUG_ON’
QEMU_BUILD_BUG_ON(offsetof(CPUArchState,
^
/home/pm215/qemu/rules.mak:66: recipe for target 'tcg/tcg.o' failed
An ideal long term approach would be to rearrange the CPU state
so that the tlb_table was not so far along it, but this is tricky
because it would move it from the "not cleared on CPU reset" part
of the struct to the "cleared on CPU reset" part. As a simple fix
for the 2.12 release, make the MIPS TCG target handle an arbitrary
offset by emitting more add instructions. This will mean an extra
instruction in the fastpath for TCG loads and stores for the
affected guests (currently just aarch64-softmmu).
Alex Bennée [Fri, 13 Apr 2018 14:03:34 +0000 (15:03 +0100)]
fpu/softfloat: raise float_invalid for NaN/Inf in round_to_int_and_pack
The re-factor broke the raising of INVALID when NaN/Inf is passed to
the float_to_int conversion functions. round_to_uint_and_pack got this
right for NaN but also missed out the Inf handling.
Pavel Dovgalyuk [Fri, 13 Apr 2018 13:30:41 +0000 (16:30 +0300)]
m68k: fix exception stack frame for 68000
68000 CPUs do not save format in the exception stack frame.
This patch adds feature checking to prevent format saving for 68000.
m68k_ret() already includes this modification, this patch fixes
the exception processing function too.
softfloat: fix {min, max}nummag for same-abs-value inputs
Before 8936006 ("fpu/softfloat: re-factor minmax", 2018-02-21),
we used to return +Zero for maxnummag(-Zero,+Zero); after that
commit, we return -Zero.
Fix it by making {min,max}nummag consistent with {min,max}num,
deferring to the latter when the absolute value of the operands
is the same.
Peter Maydell [Tue, 10 Apr 2018 13:42:03 +0000 (14:42 +0100)]
hw/char/cmsdk-apb-uart.c: Correctly clear INTSTATUS bits on writes
The CMSDK APB UART INTSTATUS register bits are all write-one-to-clear.
We were getting this correct for the TXO and RXO bits (which need
special casing because their state lives in the STATE register),
but had forgotten to handle the normal bits for RX and TX which
we do store in our s->intstatus field.
Perform the W1C operation on the bits in s->intstatus too.
Pavel Dovgalyuk [Mon, 9 Apr 2018 09:13:20 +0000 (12:13 +0300)]
icount: fix cpu_restore_state_from_tb for non-tb-exit cases
In icount mode, instructions that access io memory spaces in the middle
of the translation block invoke TB recompilation. After recompilation,
such instructions become last in the TB and are allowed to access io
memory spaces.
When the code includes instruction like i386 'xchg eax, 0xffffd080'
which accesses APIC, QEMU goes into an infinite loop of the recompilation.
This instruction includes two memory accesses - one read and one write.
After the first access, APIC calls cpu_report_tpr_access, which restores
the CPU state to get the current eip. But cpu_restore_state_from_tb
resets the cpu->can_do_io flag which makes the second memory access invalid.
Therefore the second memory access causes a recompilation of the block.
Then these operations repeat again and again.
This patch moves resetting cpu->can_do_io flag from
cpu_restore_state_from_tb to cpu_loop_exit* functions.
It also adds a parameter for cpu_restore_state which controls restoring
icount. There is no need to restore icount when we only query CPU state
without breaking the TB. Restoring it in such cases leads to the
incorrect flow of the virtual time.
In most cases new parameter is true (icount should be recalculated).
But there are two cases in i386 and openrisc when the CPU state is only
queried without the need to break the TB. This patch fixes both of
these cases.
Signed-off-by: Pavel Dovgalyuk <[email protected]>
Message-Id: <20180409091320.12504.35329.stgit@pasha-VirtualBox>
[rth: Make can_do_io setting unconditional; move from cpu_exec;
make cpu_loop_exit_{noexc,restore} call cpu_loop_exit.] Signed-off-by: Richard Henderson <[email protected]>
Peter Maydell [Tue, 10 Apr 2018 15:58:19 +0000 (16:58 +0100)]
Merge remote-tracking branch 'remotes/dgibson/tags/ppc-for-2.12-20180410' into staging
ppc patch queue 2018-04-10
Here's a rather late pull request with a handful of fixes for 2.12.
These have been blocked for some time, because I wasn't able to
complete my usual test set due to the SCSI problem fixed in 37c5174
"scsi-disk: Don't enlarge min_io_size to max_io_size".
Since we're in hard freeze, these are all bugfixes. Most are also
regressions, although in one case it's only a "regression" because a
longstanding bug has been exposed by a new machine type (sam460ex) in
the testcases. There are also a couple of sam460ex fixes that aren't
regressions since the board didn't exist before. On the flipside
though, they're low risk because they only touch board specific code
for a board that doesn't exist in any released version.
* remotes/dgibson/tags/ppc-for-2.12-20180410:
roms/u-boot-sam460ex: Change to qemu git mirror and update
sam460ex: Fix timer frequency and clock multipliers
tests/boot-serial: Test the sam460ex board
spapr: Initialize reserved areas list in FDT in H_CAS handler
target/ppc: Fix backwards migration of msr_mask
hw/misc/macio: Fix crash when listing device properties of macio device
target/ppc: Initialize lazy_tlb_flush correctly
Stefan Hajnoczi [Wed, 4 Apr 2018 15:01:46 +0000 (16:01 +0100)]
qemu-iotests: update 185 output
Commit 4486e89c219c0d1b9bd8dfa0b1dd5b0d51ff2268 ("vl: introduce
vm_shutdown()") added a bdrv_drain_all() call. As a side-effect of the
drain operation the block job iterates one more time than before. The
185 output no longer matches and the test is failing now.
It may be possible to avoid the superfluous block job iteration, but
that type of patch is not suitable late in the QEMU 2.12 release cycle.
This patch simply updates the 185 output file. The new behavior is
correct, just not optimal, so make the test pass again.
Kevin Wolf [Thu, 18 Jan 2018 20:23:52 +0000 (21:23 +0100)]
commit/stream: Reset delay_ns
Streaming and the commit block job only want to apply throttling when
they actually copied data instead of skipping it, so they made the
calculation of delay_ns conditional. However, delay_ns isn't reset when
skipping some sectors, so instead of not waiting, the old delay is
applied again.
Kevin Wolf [Tue, 10 Apr 2018 08:40:04 +0000 (10:40 +0200)]
qemu-iotests: Remove _supported_fmt dmg
qemu-iotests doesn't support dmg, and the dmg block driver doesn't
support image creation. Two test cases declare dmg as supported, but
that's obviously wrong for both reasons. Remove the declaration.
iotests: blacklist bochs and cloop for 205 and 208
Blacklist these formats, as they don't support image creation, as they
say:
> ./qemu-img create -f bochs x 1m
qemu-img: x: Format driver 'bochs' does not support image creation
> ./qemu-img create -f cloop x 1m
qemu-img: x: Format driver 'cloop' does not support image creation
Support "generic" formats like in bash tests with their
_supported_fmt generic
The test, supporting "generic" formats will run if IMGFMT_GENERIC =
true, which is default, except for bochs and cloop. However, you can
use verify_image_format(['generic', 'bochs']), which will run for all
except cloop (for this moment).
Also, add an assert (we don't want set both arguments) and remove
duplication.
Revert "migration: Don't activate block devices if using -S"
This reverts commit 0746a92612276aee69e66dfe6782b0f882d221d5.
Discussion with kwolf suggests this is actually an API change that
we need to gate on a capability. Push to 2.13.
Peter Maydell [Tue, 10 Apr 2018 14:18:58 +0000 (15:18 +0100)]
Merge remote-tracking branch 'remotes/pmaydell/tags/pull-target-arm-20180410' into staging
target-arm queue:
* fpu: Fix rounding mode for floatN_to_uintM_round_to_zero
* tcg: Fix guest state corruption when running 64-bit Arm
guests on a 32-bit host (especially when using icount)
* linux-user/signal.c: Ensure AArch64 signal frame isn't too small
* cpus.c: ensure running CPU recalculates icount deadlines on timer expiry
* target/arm: Report unsupported MPU region sizes more clearly
* hw/arm/fsl-imx: Fix introspection problem with fsl-imx6 and fsl-imx7
* hw/arm/allwinner-a10: Do not use nd_table in instance_init function
* hw/sd/bcm2835_sdhost: Don't raise spurious interrupts
* hw/sd/bcm2835_sdhost: Add tracepoints
* target-arm: Check undefined opcodes for SWP in A32 decoder
* hw/arm/integratorcp: Don't do things that could be fatal in the instance_init
* hw/arm: Allow manually specified /psci node
* remotes/pmaydell/tags/pull-target-arm-20180410:
fpu: Fix rounding mode for floatN_to_uintM_round_to_zero
tcg: Introduce tcg_set_insn_start_param
linux-user/signal.c: Ensure AArch64 signal frame isn't too small
cpus.c: ensure running CPU recalculates icount deadlines on timer expiry
target/arm: Report unsupported MPU region sizes more clearly
hw/arm/fsl-imx: Fix introspection problem with fsl-imx6 and fsl-imx7
hw/arm/allwinner-a10: Do not use nd_table in instance_init function
hw/sd/bcm2835_sdhost: Don't raise spurious interrupts
hw/sd/bcm2835_sdhost: Add tracepoints
target-arm: Check undefined opcodes for SWP in A32 decoder
hw/arm/integratorcp: Don't do things that could be fatal in the instance_init
hw/arm: Allow manually specified /psci node
* remotes/kraxel/tags/ui-20180410-pull-request:
configure: don't warn SDL abi if disabled
configure: don't warn GTK if disabled
gtk: drop pointless code from gd_window_close
ui: fix keymap detection under Xwayland
The problem was that the monitor dispatcher bottom half was bound to
qemu_aio_context now, which could be polled unexpectedly in block code.
We should keep the dispatchers run in iohandler_ctx just like what we
did before the Out-Of-Band series (chardev uses qio, and qio binds
everything with iohandler_ctx).
If without this change, QMP dispatcher might be run even before reaching
main loop in block IO path, for example, in a stack like (the ARM case,
"cont" command handler run even during machine init phase):
#0 qmp_cont ()
#1 0x00000000006bd210 in qmp_marshal_cont ()
#2 0x0000000000ac05c4 in do_qmp_dispatch ()
#3 0x0000000000ac07a0 in qmp_dispatch ()
#4 0x0000000000472d60 in monitor_qmp_dispatch_one ()
#5 0x000000000047302c in monitor_qmp_bh_dispatcher ()
#6 0x0000000000acf374 in aio_bh_call ()
#7 0x0000000000acf428 in aio_bh_poll ()
#8 0x0000000000ad5110 in aio_poll ()
#9 0x0000000000a08ab8 in blk_prw ()
#10 0x0000000000a091c4 in blk_pread ()
#11 0x0000000000734f94 in pflash_cfi01_realize ()
#12 0x000000000075a3a4 in device_set_realized ()
#13 0x00000000009a26cc in property_set_bool ()
#14 0x00000000009a0a40 in object_property_set ()
#15 0x00000000009a3a08 in object_property_set_qobject ()
#16 0x00000000009a0c8c in object_property_set_bool ()
#17 0x0000000000758f94 in qdev_init_nofail ()
#18 0x000000000058e190 in create_one_flash ()
#19 0x000000000058e2f4 in create_flash ()
#20 0x00000000005902f0 in machvirt_init ()
#21 0x00000000007635cc in machine_run_board_init ()
#22 0x00000000006b135c in main ()
Actually the problem is more severe than that. After we switched to the
qemu AIO handler it means the monitor dispatcher code can even be called
with nested aio_poll(), then it can be an explicit aio_poll() inside
another main loop aio_poll() which could be racy too; breaking code
like TPM and 9p that use nested event loops.
Switch to use the iohandler_ctx for monitor dispatchers.
My sincere thanks to Eric Auger who offered great help during both
debugging and verifying the problem. The ARM test was carried out by
applying this patch upon QEMU 2.12.0-rc0 and problem is gone after the
patch.
A quick test of mine shows that after this patch applied we can pass all
raw iotests even with OOB on by default.
Peter Xu [Mon, 9 Apr 2018 08:39:56 +0000 (16:39 +0800)]
iothread: workaround glib bug which hangs qmp-test
Free the AIO context earlier than the GMainContext (if we have) to
workaround a glib2 bug that GSource context pointer is not cleared even
if the context has already been destroyed (while it should).
The patch itself only changed the order to destroy the objects, no
functional change at all. Without this workaround, we can encounter
qmp-test hang with oob (and possibly any other use case when iothread is
used with GMainContexts):
#0 0x00007f35ffe45334 in __lll_lock_wait () from /lib64/libpthread.so.0
#1 0x00007f35ffe405d8 in _L_lock_854 () from /lib64/libpthread.so.0
#2 0x00007f35ffe404a7 in pthread_mutex_lock () from /lib64/libpthread.so.0
#3 0x00007f35fc5b9c9d in g_source_unref_internal (source=0x24f0600, context=0x7f35f0000960, have_lock=0) at gmain.c:1685
#4 0x0000000000aa6672 in aio_context_unref (ctx=0x24f0600) at /root/qemu/util/async.c:497
#5 0x000000000065851c in iothread_instance_finalize (obj=0x24f0380) at /root/qemu/iothread.c:129
#6 0x0000000000962d79 in object_deinit (obj=0x24f0380, type=0x242e960) at /root/qemu/qom/object.c:462
#7 0x0000000000962e0d in object_finalize (data=0x24f0380) at /root/qemu/qom/object.c:476
#8 0x0000000000964146 in object_unref (obj=0x24f0380) at /root/qemu/qom/object.c:924
#9 0x0000000000965880 in object_finalize_child_property (obj=0x24ec640, name=0x24efca0 "mon_iothread", opaque=0x24f0380) at /root/qemu/qom/object.c:1436
#10 0x0000000000962c33 in object_property_del_child (obj=0x24ec640, child=0x24f0380, errp=0x0) at /root/qemu/qom/object.c:436
#11 0x0000000000962d26 in object_unparent (obj=0x24f0380) at /root/qemu/qom/object.c:455
#12 0x0000000000658f00 in iothread_destroy (iothread=0x24f0380) at /root/qemu/iothread.c:365
#13 0x00000000004c67a8 in monitor_cleanup () at /root/qemu/monitor.c:4663
#14 0x0000000000669e27 in main (argc=16, argv=0x7ffc8b1ae2f8, envp=0x7ffc8b1ae380) at /root/qemu/vl.c:4749
The glib2 bug is fixed in commit 26056558b ("gmain: allow
g_source_get_context() on destroyed sources", 2012-07-30), so the first
good version is glib2 2.33.10. But we still support building with
glib as old as 2.28, so we need the workaround.
Let's make sure we destroy the GSources first before its owner context
until we drop support for glib older than 2.33.10.
Peter Xu [Sun, 8 Apr 2018 03:05:42 +0000 (11:05 +0800)]
iotests: fix wait_until_completed()
If there are more than one events, wait_until_completed() might return
the 2nd event even if the 1st event is JOB_COMPLETED, since the for loop
will continue to run even if completed is set to True.
It never happened before, but it can be triggered when OOB is enabled
due to the RESUME startup message. Fix that up.
Peter Maydell [Tue, 10 Apr 2018 12:02:25 +0000 (13:02 +0100)]
linux-user/signal.c: Ensure AArch64 signal frame isn't too small
The AArch64 signal frame design was extended for SVE in commit 8c5931de0ac77388096d79ceb, so that instead of having a fixed setup we
now add various records to the frame, with some of them possibly
overflowing into an extra space outside the original 4K reserved
block in the target_sigcontext. However, we failed to ensure that we
always at least allocate the 4K reserved block. This is ABI, and
some userspace programs rely on it. In particular the dash shell
would segfault if the frame wasn't as big enough.
(Compare the kernel's sigframe_size() function in
arch/arm64/kernel/signal.c.)
Peter Maydell [Tue, 10 Apr 2018 12:02:25 +0000 (13:02 +0100)]
cpus.c: ensure running CPU recalculates icount deadlines on timer expiry
When we run in TCG icount mode, we calculate the number of instructions
to execute using tcg_get_icount_limit(), which ensures that we stop
execution at the next timer deadline. However there is a bug where
currently we do not recalculate that limit if the guest reprograms
a timer so that the next deadline moves closer, and so we will
continue execution until the original limit and fire the timer
later than we should.
Fix this bug in qemu_timer_notify_cb(): if we are currently running
a VCPU in icount mode, we simply need to kick it out of the main
loop and back to tcg_cpu_exec(), where it will recalculate the
icount limit. If we are not currently running a VCPU, then we
retain the existing logic for waking up a halted CPU.
Peter Maydell [Tue, 10 Apr 2018 12:02:25 +0000 (13:02 +0100)]
target/arm: Report unsupported MPU region sizes more clearly
Currently our PMSAv7 and ARMv7M MPU implementation cannot handle
MPU region sizes smaller than our TARGET_PAGE_SIZE. However we
report that in a slightly confusing way:
DRSR[3]: No support for MPU (sub)region alignment of 9 bits. Minimum is 10
The problem is not the alignment of the region, but its size;
tweak the error message to say so:
DRSR[3]: No support for MPU (sub)region size of 512 bytes. Minimum is 1024.
This happens because these devices are doing an exit() from their
instance_init function - which should never be done since instance_init
can be called at any time for device introspection! Fix it by moving
the deadly check into the realize() function instead.
Thomas Huth [Tue, 10 Apr 2018 12:02:25 +0000 (13:02 +0100)]
hw/arm/allwinner-a10: Do not use nd_table in instance_init function
The instance_init function of a device can be called at any time, even
if the device is not going to be used (i.e. not going to be realized).
So a instance_init function must not do things that could cause QEMU
to exit, like calling qemu_check_nic_model(&nd_table[0], ...) for example.
But this is what the instance_init function of the allwinner-a10 device
is currently doing - and this causes QEMU to quit unexpectedly when
you run the 'device-list-properties' QMP command for example:
The Linux bcm2835_sdhost driver doesn't work on QEMU, because our
model raises spurious data interrupts. Our function
bcm2835_sdhost_fifo_run() will flag an interrupt any time it is
called with s->datacnt == 0, even if the host hasn't actually issued
a data read or write command yet. This means that the driver gets a
spurious data interrupt as soon as it enables IRQs and then does
something else that causes us to call the fifo_run routine, like
writing to SDHCFG, and before it does the write to SDCMD to issue the
read. The driver's IRQ handler then spins forever complaining that
there's no data and the SD controller isn't in a state where there's
going to be any data:
Move the interrupt flag setting to more plausible places:
* for BUSY, raise this as soon as a BUSYWAIT command has executed
* for DATA, raise this when the FIFO has any space free (for a write)
or any data in it (for a read)
* for BLOCK, raise this when the data count is 0 and we've
actually done some reading or writing
This is pure guesswork since the documentation for this hardware is
not public, but it is sufficient to get the Linux bcm2835_sdhost
driver to work.
projects / qemu.git / log