Greg Kurz [Fri, 26 Jul 2019 14:44:38 +0000 (16:44 +0200)]
spapr/pci: Consolidate de-allocation of MSIs
When freeing MSIs, we need to:
- remove them from the machine's MSI bitmap
- remove them from the IC backend
- remove them from the PHB's MSI cache
This is currently open coded in two places in rtas_ibm_change_msi(),
and we're about to need this in spapr_phb_reset() as well. Instead of
duplicating this code again, make it a destroy function for the PHB's
MSI cache. Removing an MSI device from the cache will call the destroy
function internally.
The logic is broken for multiple vcpu guests, also causing memory leak.
The logic is in place to handle kvm not having KVM_CAP_PPC_IRQ_LEVEL,
which is part of the kernel now since 2.6.37. Instead of fixing the
leak, drop the redundant logic which is not excercised on new kernels
anymore. Exit with error on older kernels.
Nicholas Piggin [Mon, 22 Jul 2019 06:17:52 +0000 (16:17 +1000)]
spapr: Implement ibm,suspend-me
This has been useful to modify and test the Linux pseries suspend
code but it requires modification to the guest to call it (due to
being gated by other unimplemented features). It is not otherwise
used by Linux yet, but work is slowly progressing there.
This allows a (lightly modified) guest kernel to suspend with
`echo mem > /sys/power/state` and be resumed with system_wakeup
monitor command.
Nicholas Piggin [Mon, 22 Jul 2019 05:32:13 +0000 (15:32 +1000)]
machine: Add wakeup method to MachineClass
Waking from suspend is not logically a machine reset on all machines,
particularly in the paravirtualized case rather than hardware
emulated. The ppc spapr machine for example just invokes hypervisor
to suspend, and expects that call to return with the machine in the
same state (modulo some possible migration and reconfiguration
details).
Implement a machine ->wakeup method and use that if it exists.
Provide a better output of the XIVE END structures including the
escalation information and extend the PowerNV machine 'info pic'
command with a dump of the END EAS table used for escalations.
When the 's' bit is set the escalation is said to be 'silent' or
'silent/gather'. In such configuration, the notification sequence is
skipped and only the escalation sequence is performed. This is used to
configure all the EQs of a vCPU to escalate on a single EQ which will
then target the hypervisor.
ppc/xive: Provide unconditional escalation support
When the 'u' bit is set the escalation is said to be 'unconditional'
which means that the ESe PQ bits are not used. Introduce a
xive_router_end_es_notify() routine to share code with the ESn
notification.
If the XIVE presenter can not find the NVT dispatched on any of the HW
threads, it can not deliver the interrupt. XIVE offers an escalation
mechanism to handle such scenarios and inform the hypervisor that an
action should be taken.
Escalation is configured by setting the 'e' bit and the EAS in word 4
& 5 to let the HW look for the escalation END on which to trigger a
new event.
ppc/xive: Implement TM_PULL_OS_CTX special command
When a vCPU is not dispatched anymore on a HW thread, the Hypervisor
(KVM on Linux) invalidates the OS interrupt context of a vCPU with
this special command. It returns the OS CAM line value and resets the
VO bit.
The only change that SLOF does not rely on QEMU providing an RTAS blob
and provides one itself:
https://git.qemu.org/?p=SLOF.git;a=commitdiff;h=5e4ed1fd0f39e
Michael Roth [Wed, 17 Jul 2019 20:58:42 +0000 (15:58 -0500)]
spapr: initial implementation for H_TPM_COMM/spapr-tpm-proxy
This implements the H_TPM_COMM hypercall, which is used by an
Ultravisor to pass TPM commands directly to the host's TPM device, or
a TPM Resource Manager associated with the device.
This also introduces a new virtual device, spapr-tpm-proxy, which
is used to configure the host TPM path to be used to service
requests sent by H_TPM_COMM hcalls, for example:
Michael Roth [Wed, 17 Jul 2019 20:58:41 +0000 (15:58 -0500)]
docs/specs: initial spec summary for Ultravisor-related hcalls
For now this only covers hcalls relating to TPM communication since
it's the only one particularly important from a QEMU perspective atm,
but others can be added here where it makes sense.
The full specification for all hcalls/ucalls will eventually be made
available in the public/OpenPower version of the PAPR specification.
Nicholas Piggin [Thu, 18 Jul 2019 03:42:14 +0000 (13:42 +1000)]
spapr: Implement H_JOIN
This has been useful to modify and test the Linux pseries suspend
code but it requires modification to the guest to call it (due to
being gated by other unimplemented features). It is not otherwise
used by Linux yet, but work is slowly progressing there.
Nicholas Piggin [Thu, 18 Jul 2019 03:42:13 +0000 (13:42 +1000)]
spapr: Implement H_CONFER
This does not do directed yielding and is not quite as strict as PAPR
specifies in terms of precise dispatch behaviour. This generally will
mean suboptimal performance, rather than guest misbehaviour. Linux
does not rely on exact dispatch behaviour.
Nicholas Piggin [Thu, 18 Jul 2019 03:42:11 +0000 (13:42 +1000)]
spapr: Implement dispatch tracking for tcg
Implement cpu_exec_enter/exit on ppc which calls into new methods of
the same name in PPCVirtualHypervisorClass. These are used by spapr
to implement the splpar VPA dispatch counter initially.
Stefan Brankovic [Mon, 15 Jul 2019 14:22:52 +0000 (16:22 +0200)]
target/ppc: Optimize emulation of vclzw instruction
Optimize Altivec instruction vclzw (Vector Count Leading Zeros Word).
This instruction counts the number of leading zeros of each word element
in source register and places result in the appropriate word element of
destination register.
Counting is to be performed in four iterations of for loop(one for each
word elemnt of source register vB). Every iteration consists of loading
appropriate word element from source register, counting leading zeros
with tcg_gen_clzi_i32, and saving the result in appropriate word element
of destination register.
Stefan Brankovic [Mon, 15 Jul 2019 14:22:51 +0000 (16:22 +0200)]
target/ppc: Optimize emulation of vclzd instruction
Optimize Altivec instruction vclzd (Vector Count Leading Zeros Doubleword).
This instruction counts the number of leading zeros of each doubleword element
in source register and places result in the appropriate doubleword element of
destination register.
Using tcg-s count leading zeros instruction two times(once for each
doubleword element of source register vB) and placing result in
appropriate doubleword element of destination register vD.
Stefan Brankovic [Mon, 15 Jul 2019 14:22:50 +0000 (16:22 +0200)]
target/ppc: Optimize emulation of vgbbd instruction
Optimize altivec instruction vgbbd (Vector Gather Bits by Bytes by Doubleword)
All ith bits (i in range 1 to 8) of each byte of doubleword element in
source register are concatenated and placed into ith byte of appropriate
doubleword element in destination register.
Following solution is done for both doubleword elements of source register
in parallel, in order to reduce the number of instructions needed(that's why
arrays are used):
First, both doubleword elements of source register vB are placed in
appropriate element of array avr. Bits are gathered in 2x8 iterations(2 for
loops). In first iteration bit 1 of byte 1, bit 2 of byte 2,... bit 8 of
byte 8 are in their final spots so avr[i], i={0,1} can be and-ed with
tcg_mask. For every following iteration, both avr[i] and tcg_mask variables
have to be shifted right for 7 and 8 places, respectively, in order to get
bit 1 of byte 2, bit 2 of byte 3.. bit 7 of byte 8 in their final spots so
shifted avr values(saved in tmp) can be and-ed with new value of tcg_mask...
After first 8 iteration(first loop), all the first bits are in their final
places, all second bits but second bit from eight byte are in their places...
only 1 eight bit from eight byte is in it's place). In second loop we do all
operations symmetrically, in order to get other half of bits in their final
spots. Results for first and second doubleword elements are saved in
result[0] and result[1] respectively. In the end those results are saved in
appropriate doubleword element of destination register vD.
Alex Bennée [Tue, 16 Jul 2019 12:13:52 +0000 (13:13 +0100)]
target/ppc: move opcode decode tables to PowerPCCPU
The opcode decode tables aren't really part of the CPUPPCState but an
internal implementation detail for the translator. This can cause
problems with memcpy in cpu_copy as any table created during
ppc_cpu_realize get written over causing a memory leak. To avoid this
move the tables into PowerPCCPU which is better suited to hold
internal implementation details.
Stefan Brankovic [Mon, 15 Jul 2019 14:22:48 +0000 (16:22 +0200)]
target/ppc: Optimize emulation of vsl and vsr instructions
Optimization of altivec instructions vsl and vsr(Vector Shift Left/Rigt).
Perform shift operation (left and right respectively) on 128 bit value of
register vA by value specified in bits 125-127 of register vB. Lowest 3
bits in each byte element of register vB must be identical or result is
undefined.
For vsl instruction, the first step is bits 125-127 of register vB have
to be saved in variable sh. Then, the highest sh bits of the lower
doubleword element of register vA are saved in variable shifted,
in order not to lose those bits when shift operation is performed on
the lower doubleword element of register vA, which is the next
step. After shifting the lower doubleword element shift operation
is performed on higher doubleword element of vA, with replacement of
the lowest sh bits(that are now 0) with bits saved in shifted.
For vsr instruction, firstly, the bits 125-127 of register vB have
to be saved in variable sh. Then, the lowest sh bits of the higher
doubleword element of register vA are saved in variable shifted,
in odred not to lose those bits when the shift operation is
performed on the higher doubleword element of register vA, which is
the next step. After shifting higher doubleword element, shift operation
is performed on lower doubleword element of vA, with replacement of
highest sh bits(that are now 0) with bits saved in shifted.
Stefan Brankovic [Mon, 15 Jul 2019 14:22:47 +0000 (16:22 +0200)]
target/ppc: Optimize emulation of lvsl and lvsr instructions
Adding simple macro that is calling tcg implementation of appropriate
instruction if altivec support is active.
Optimization of altivec instruction lvsl (Load Vector for Shift Left).
Place bytes sh:sh+15 of value 0x00 || 0x01 || 0x02 || ... || 0x1E || 0x1F
in destination register. Sh is calculated by adding 2 source registers and
getting bits 60-63 of result.
First, the bits [28-31] are placed from EA to variable sh. After that,
the bytes are created in the following way:
sh:(sh+7) of X(from description) by multiplying sh with 0x0101010101010101
followed by addition of the result with 0x0001020304050607. Value obtained
is placed in higher doubleword element of vD.
(sh+8):(sh+15) by adding the result of previous multiplication with
0x08090a0b0c0d0e0f. Value obtained is placed in lower doubleword element
of vD.
Optimization of altivec instruction lvsr (Load Vector for Shift Right).
Place bytes 16-sh:31-sh of value 0x00 || 0x01 || 0x02 || ... || 0x1E ||
0x1F in destination register. Sh is calculated by adding 2 source
registers and getting bits 60-63 of result.
First, the bits [28-31] are placed from EA to variable sh. After that,
the bytes are created in the following way:
sh:(sh+7) of X(from description) by multiplying sh with 0x0101010101010101
followed by substraction of the result from 0x1011121314151617. Value
obtained is placed in higher doubleword element of vD.
(sh+8):(sh+15) by substracting the result of previous multiplication from
0x18191a1b1c1d1e1f. Value obtained is placed in lower doubleword element
of vD.
migration: Do not re-read the clock on pre_save in case of paused guest
Re-read the timebase before migrate was ported from x86 commit: 6053a86fe7bd: kvmclock: reduce kvmclock difference on migration
The clock move makes the guest knows about the paused time between
the stop and migrate commands. This is an issue in an already-paused
VM because some side effects, like process stalls, could happen
after migration.
So, this patch checks the runstate of guest in the pre_save handler and
do not re-reads the timebase in case of paused state (cold migration).
David Gibson [Fri, 5 Jul 2019 05:03:05 +0000 (15:03 +1000)]
spapr_pci: Allow 2MiB and 16MiB IOMMU pagesizes by default
We've had the qemu and kernel KVM infrastructure to handle larger TCE
page sizes for a while, but forgot to update the defaults to actually
allow them. This turns that change on.
Add 4.2 machine types for arm/i440fx/q35/s390x/spapr.
For i440fx and q35, unversioned cpu models are still translated
to -v1, as 0788a56bd1ae ("i386: Make unversioned CPU models be
aliases") states this should only transition to the latest cpu
model version in 4.3 (or later).
Its not immediately obvious how cap-X=Y setting need to be applied
to the command line so, for spapr capability error messages, this
has been clarified to:
appending -machine cap-X=Y
The wrong value messages have been left as is, as the user has found
the right location.
* remotes/huth-gitlab/tags/pull-request-2019-08-20:
hw/core: Add a config switch for the generic loader device
hw/misc: Add a config switch for the "unimplemented" device
hw/core: Add a config switch for the "split-irq" device
hw/core: Add a config switch for the "or-irq" device
hw/core: Add a config switch for the "register" device
hw/dma: Do not build the xlnx_dpdma device for the MicroBlaze machines
hw/intc: Only build the xlnx-iomod-intc device for the MicroBlaze PMU
hw/Kconfig: Move the generic XLNX_ZYNQMP to the root hw/Kconfig
Peter Maydell [Tue, 20 Aug 2019 12:40:48 +0000 (13:40 +0100)]
Merge remote-tracking branch 'remotes/amarkovic/tags/mips-queue-aug-20-2019' into staging
MIPS queue for August 20th, 2019
# gpg: Signature made Mon 19 Aug 2019 19:07:18 BST
# gpg: using RSA key D4972A8967F75A65
# gpg: Good signature from "Aleksandar Markovic <[email protected]>" [unknown]
# 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: 8526 FBF1 5DA3 811F 4A01 DD75 D497 2A89 67F7 5A65
* remotes/amarkovic/tags/mips-queue-aug-20-2019:
target/mips: tests/tcg: Fix target configurations for MSA tests
target/mips: tests/tcg: Add optional printing of more detailed failure info
target/mips: Style improvements in mips_mipssim.c
target/mips: Style improvements in mips_malta.c
target/mips: Style improvements in mips_int.c
target/mips: Style improvements in mips_fulong2e.c
target/mips: Style improvements in cps.c
target/mips: Style improvements in translate.c
target/mips: Style improvements in machine.c
target/mips: Style improvements in cpu.c
target/mips: Style improvements in cp0_timer.c
* remotes/maxreitz/tags/pull-block-2019-08-19:
doc: Preallocation does not require writing zeroes
iotests: Fix 141 when run with qed
vpc: Do not return RAW from block_status
vmdk: Make block_status recurse for flat extents
vdi: Make block_status recurse for fixed images
iotests: Full mirror to existing non-zero image
iotests: Test convert -n to pre-filled image
iotests: Convert to preallocated encrypted qcow2
vhdx: Fix .bdrv_has_zero_init()
vdi: Fix .bdrv_has_zero_init()
qcow2: Fix .bdrv_has_zero_init()
block: Use bdrv_has_zero_init_truncate()
block: Implement .bdrv_has_zero_init_truncate()
block: Add bdrv_has_zero_init_truncate()
mirror: Fix bdrv_has_zero_init() use
qemu-img: Fix bdrv_has_zero_init() use in convert
LUKS: support preallocation
Thomas Huth [Tue, 30 Jul 2019 13:40:50 +0000 (15:40 +0200)]
hw/core: Add a config switch for the generic loader device
The generic loader device is completely optional. Let's add a proper
config switch for it so that people can disable it if they don't need
it and want to create a minimalistic QEMU binary.
Thomas Huth [Tue, 14 May 2019 08:24:28 +0000 (10:24 +0200)]
hw/core: Add a config switch for the "split-irq" device
The "split-irq" device is currently only used by machines that use
CONFIG_ARMSSE. Let's add a proper CONFIG_SPLIT_IRQ switch for this
so that it only gets compiled when we really need it.
hw/dma: Do not build the xlnx_dpdma device for the MicroBlaze machines
The xlnx_dpdma device is only used by the ZynqMP AArch64 machine
(not the MicroBlaze PMU). Remove it from the ZynqMP generic objects.
(Note, this entry was duplicated for the AArch64).
hw/intc: Only build the xlnx-iomod-intc device for the MicroBlaze PMU
The Xilinx I/O Module Interrupt Controller is only used by the
MicroBlaze PMU, not by the AArch64 machine.
Move it from the generic ZynqMP object list to the PMU specific.
target/mips: tests/tcg: Fix target configurations for MSA tests
At this moment, the only MIPS CPUs that are emulated in QEMU and
support MSA extension are R5600 (mips32r5), and I6400/I6500 (mips64r6).
Therefore, mips32r5 and mips64r6 are the only ISAs that could support
MSA in QEMU. This means mips32r6 currently do not make much sense, and
mips32r5 support for MSA tests is needed, which is done by this patch.
target/mips: tests/tcg: Add optional printing of more detailed failure info
There is a need for printing input and output data for failure cases,
for debugging purpose. This is achieved by this patch, and only if a
preprocessor constant is manually set to 1. (Assumption is that the
need for such printout is relatively rare.)
Max Reitz [Thu, 11 Jul 2019 13:29:35 +0000 (15:29 +0200)]
doc: Preallocation does not require writing zeroes
When preallocating an encrypted qcow2 image, it just lets the protocol
driver write data and then does not mark the clusters as zero.
Therefore, reading this image will yield effectively random data.
As such, we have not fulfilled the promise of always writing zeroes when
preallocating an image in a while. It seems that nobody has really
cared, so change the documentation to conform to qemu's actual behavior.
Max Reitz [Fri, 9 Aug 2019 18:52:53 +0000 (20:52 +0200)]
iotests: Fix 141 when run with qed
69f47505ee has changed qcow2 in such a way that the commit job run in
test 141 (and 144[1]) returns before it emits the READY event. However,
141 also runs with qed, where the order is still the other way around.
Just filter out the {"return": {}} so the test passes for qed again.
[1] 144 only runs with qcow2, so it is fine as it is.
Max Reitz [Thu, 25 Jul 2019 15:55:12 +0000 (17:55 +0200)]
vpc: Do not return RAW from block_status
vpc is not really a passthrough driver, even when using the fixed
subformat (where host and guest offsets are equal). It should handle
preallocation like all other drivers do, namely by returning
DATA | RECURSE instead of RAW.
There is no tangible difference but the fact that bdrv_is_allocated() no
longer falls through to the protocol layer.
Max Reitz [Wed, 24 Jul 2019 17:12:39 +0000 (19:12 +0200)]
iotests: Full mirror to existing non-zero image
The result of a sync=full mirror should always be the equal to the
input. Therefore, existing images should be treated as potentially
non-zero and thus should be explicitly initialized to be zero
beforehand.
Max Reitz [Wed, 24 Jul 2019 17:12:37 +0000 (19:12 +0200)]
iotests: Convert to preallocated encrypted qcow2
Add a test case for converting an empty image (which only returns zeroes
when read) to a preallocated encrypted qcow2 image.
qcow2_has_zero_init() should return 0 then, thus forcing qemu-img
convert to create zero clusters.
Max Reitz [Wed, 24 Jul 2019 17:12:35 +0000 (19:12 +0200)]
vdi: Fix .bdrv_has_zero_init()
Static VDI images cannot guarantee to be zero-initialized. If the image
has been statically allocated, forward the call to the underlying
storage node.
Max Reitz [Wed, 24 Jul 2019 17:12:34 +0000 (19:12 +0200)]
qcow2: Fix .bdrv_has_zero_init()
If a qcow2 file is preallocated, it can no longer guarantee that it
initially appears as filled with zeroes.
So implement .bdrv_has_zero_init() by checking whether the file is
preallocated; if so, forward the call to the underlying storage node,
except for when it is encrypted: Encrypted preallocated images always
return effectively random data, so .bdrv_has_zero_init() must always
return 0 for them.
.bdrv_has_zero_init_truncate() can remain bdrv_has_zero_init_1(),
because it presupposes PREALLOC_MODE_OFF.
Max Reitz [Wed, 24 Jul 2019 17:12:33 +0000 (19:12 +0200)]
block: Use bdrv_has_zero_init_truncate()
vhdx and parallels call bdrv_has_zero_init() when they do not really
care about an image's post-create state but only about what happens when
you grow an image. That is a bit ugly, and also overly safe when
growing preallocated images without preallocating the new areas.
Let them use bdrv_has_zero_init_truncate() instead.
Max Reitz [Wed, 24 Jul 2019 17:12:31 +0000 (19:12 +0200)]
block: Add bdrv_has_zero_init_truncate()
No .bdrv_has_zero_init() implementation returns 1 if growing the file
would add non-zero areas (at least with PREALLOC_MODE_OFF), so using it
in lieu of this new function was always safe.
But on the other hand, it is possible that growing an image that is not
zero-initialized would still add a zero-initialized area, like when
using nonpreallocating truncation on a preallocated image. For callers
that care only about truncation, not about creation with potential
preallocation, this new function is useful.
Alternatively, we could have added a PreallocMode parameter to
bdrv_has_zero_init(). But the only user would have been qemu-img
convert, which does not have a plain PreallocMode value right now -- it
would have to parse the creation option to obtain it. Therefore, the
simpler solution is to let bdrv_has_zero_init() inquire the
preallocation status and add the new bdrv_has_zero_init_truncate() that
presupposes PREALLOC_MODE_OFF.
Max Reitz [Wed, 24 Jul 2019 17:12:30 +0000 (19:12 +0200)]
mirror: Fix bdrv_has_zero_init() use
bdrv_has_zero_init() only has meaning for newly created images or image
areas. If the mirror job itself did not create the image, it cannot
rely on bdrv_has_zero_init()'s result to carry any meaning.
This is the case for drive-mirror with mode=existing and always for
blockdev-mirror.
Note that we only have to zero-initialize the target with sync=full,
because other modes actually do not promise that the target will contain
the same data as the source after the job -- sync=top only promises to
copy anything allocated in the top layer, and sync=none will only copy
new I/O. (Which is how mirror has always handled it.)
Max Reitz [Wed, 24 Jul 2019 17:12:29 +0000 (19:12 +0200)]
qemu-img: Fix bdrv_has_zero_init() use in convert
bdrv_has_zero_init() only has meaning for newly created images or image
areas. If qemu-img convert did not create the image itself, it cannot
rely on bdrv_has_zero_init()'s result to carry any meaning.
We can tell that a decodetree input file is "secondary" when it
uses an argument set marked "!extern". This indicates that at
least one of the insn translation functions will have already
been declared by the "primary" input file, but given only the
secondary we cannot tell which.
Avoid redundant declaration warnings by suppressing them with pragmas.
Peter Maydell [Mon, 19 Aug 2019 14:58:01 +0000 (15:58 +0100)]
Merge remote-tracking branch 'remotes/stsquad/tags/pull-softfloat-headers-190819-1' into staging
Softfloat updates
- minor refactoring of constants
- drop LIT64 macro
- re-organise header inclusion
# gpg: Signature made Mon 19 Aug 2019 12:08:37 BST
# gpg: using RSA key 6685AE99E75167BCAFC8DF35FBD0DB095A9E2A44
# gpg: Good signature from "Alex Bennée (Master Work Key) <[email protected]>" [full]
# Primary key fingerprint: 6685 AE99 E751 67BC AFC8 DF35 FBD0 DB09 5A9E 2A44
* remotes/stsquad/tags/pull-softfloat-headers-190819-1:
targets (various): use softfloat-helpers.h where we can
target/riscv: rationalise softfloat includes
target/mips: rationalise softfloat includes
fpu: rename softfloat-specialize.h -> .inc.c
fpu: make softfloat-macros "self-contained"
fpu: move inline helpers into a separate header
fpu: remove the LIT64 macro
target/m68k: replace LIT64 with UINT64_C macros
fpu: replace LIT64 with UINT64_C macros
fpu: use min/max values from stdint.h for integral overflow
fpu: convert float[16/32/64]_squash_denormal to new modern style
fpu: replace LIT64 usage with UINT64_C for specialize constants
* remotes/huth-gitlab/tags/pull-request-2019-08-17:
gitlab-ci: Remove qcow2 tests that are handled by "make check" already
tests: Run the iotests during "make check" again
block: fix NetBSD qemu-iotests failure
Peter Maydell [Mon, 19 Aug 2019 11:32:15 +0000 (12:32 +0100)]
Merge remote-tracking branch 'remotes/jnsnow/tags/ide-pull-request' into staging
Pull request
Stable notes: patches one and two can be considered
for the next -stable release.
# gpg: Signature made Sat 17 Aug 2019 00:15:50 BST
# gpg: using RSA key F9B7ABDBBCACDF95BE76CBD07DEF8106AAFC390E
# gpg: Good signature from "John Snow (John Huston) <[email protected]>" [full]
# Primary key fingerprint: FAEB 9711 A12C F475 812F 18F2 88A9 064D 1835 61EB
# Subkey fingerprint: F9B7 ABDB BCAC DF95 BE76 CBD0 7DEF 8106 AAFC 390E
* remotes/jnsnow/tags/ide-pull-request:
hw/ide/atapi: Use the ldst API
Revert "ide/ahci: Check for -ECANCELED in aio callbacks"
dma-helpers: ensure AIO callback is invoked after cancellation
Alex Bennée [Thu, 8 Aug 2019 16:30:35 +0000 (17:30 +0100)]
targets (various): use softfloat-helpers.h where we can
Generally the cpu and non-FP helper files just want to manipulate the
softfloat flags. For this they can just use the -helpers.h include
which brings in a minimal number of inline helpers.
Alex Bennée [Thu, 8 Aug 2019 16:29:41 +0000 (17:29 +0100)]
target/riscv: rationalise softfloat includes
We should avoid including the whole of softfloat headers in cpu.h and
explicitly include it only where we will be calling softfloat
functions. We can use the -types.h and -helpers.h in cpu.h for the few
bits that are global.
Alex Bennée [Thu, 8 Aug 2019 16:27:31 +0000 (17:27 +0100)]
target/mips: rationalise softfloat includes
We should avoid including the whole of softfloat headers in cpu.h and
explicitly include it only where we will be calling softfloat
functions. We can use the -types.h in cpu.h for the few bits that are
global. We also move the restore_snan_bit_mode into internal.h and
include -helpers.h there.
Alex Bennée [Thu, 8 Aug 2019 16:18:21 +0000 (17:18 +0100)]
fpu: rename softfloat-specialize.h -> .inc.c
This is not a normal header and should only be included in the main
softfloat.c file to bring in the various target specific
specialisations. Indeed as it contains non-inlined C functions it is
not even a legal header. Rename it to match our included C convention.
Alex Bennée [Thu, 8 Aug 2019 16:05:15 +0000 (17:05 +0100)]
fpu: make softfloat-macros "self-contained"
The macros use the "flags" type and to be consistent if anyone just
needs the macros we should bring in the header we need. There is an
outstanding TODO to audit the use of "flags" and replace with bool at
which point this include could be dropped.
Alex Bennée [Thu, 8 Aug 2019 14:11:08 +0000 (15:11 +0100)]
fpu: move inline helpers into a separate header
There are a bunch of users of the inline helpers who do not need
access to the entire softfloat API. Move those inline helpers into a
new header file which can be included without bringing in the rest of
the world.
Alex Bennée [Tue, 13 Aug 2019 11:16:23 +0000 (12:16 +0100)]
fpu: replace LIT64 with UINT64_C macros
In our quest to eliminate the home rolled LIT64 macro we fixup usage
inside the softfloat code. While we are at it we remove some of the
extraneous spaces to closer fit the house style.
Alex Bennée [Tue, 13 Aug 2019 10:55:32 +0000 (11:55 +0100)]
fpu: use min/max values from stdint.h for integral overflow
Remove some more use of LIT64 while making the meaning more clear. We
also avoid the need of casts as the results by definition fit into the
return type.
Alex Bennée [Mon, 12 Aug 2019 16:19:33 +0000 (17:19 +0100)]
fpu: convert float[16/32/64]_squash_denormal to new modern style
This also allows us to remove the extractFloat16exp/frac helpers. We
avoid using the floatXX_pack_raw functions as they are slight overkill
for masking out all but the top bit of the number. The generated code
is almost exactly the same as makes no difference to the
pre-conversion code.
Peter Maydell [Mon, 19 Aug 2019 09:55:03 +0000 (10:55 +0100)]
Merge remote-tracking branch 'remotes/jnsnow/tags/bitmaps-pull-request' into staging
Pull request
Rebase notes:
011/36:[0003] [FC] 'block/backup: upgrade copy_bitmap to BdrvDirtyBitmap'
016/36:[----] [-C] 'iotests: Add virtio-scsi device helper'
017/36:[0002] [FC] 'iotests: add test 257 for bitmap-mode backups'
030/36:[0011] [FC] 'block/backup: teach TOP to never copy unallocated regions'
032/36:[0018] [FC] 'iotests/257: test traditional sync modes'
11: A new hbitmap call was added late in 4.1, changed to
bdrv_dirty_bitmap_next_zero.
16: Context-only (self.has_quit is new context in 040)
17: Removed 'auto' to follow upstream trends in iotest fashion
30: Handled explicitly on-list with R-B from Max.
32: Fix capitalization in test, as mentioned on-list.
# gpg: Signature made Sat 17 Aug 2019 00:12:13 BST
# gpg: using RSA key F9B7ABDBBCACDF95BE76CBD07DEF8106AAFC390E
# gpg: Good signature from "John Snow (John Huston) <[email protected]>" [full]
# Primary key fingerprint: FAEB 9711 A12C F475 812F 18F2 88A9 064D 1835 61EB
# Subkey fingerprint: F9B7 ABDB BCAC DF95 BE76 CBD0 7DEF 8106 AAFC 390E
* remotes/jnsnow/tags/bitmaps-pull-request: (36 commits)
tests/test-hbitmap: test next_zero and _next_dirty_area after truncate
block/backup: refactor write_flags
block/backup: deal with zero detection
qapi: add dirty-bitmaps to query-named-block-nodes result
iotests/257: test traditional sync modes
block/backup: support bitmap sync modes for non-bitmap backups
block/backup: teach TOP to never copy unallocated regions
block/backup: add backup_is_cluster_allocated
block/backup: centralize copy_bitmap initialization
block/backup: improve sync=bitmap work estimates
iotests/257: test API failures
block/backup: hoist bitmap check into QMP interface
iotests/257: Refactor backup helpers
iotests/257: add EmulatedBitmap class
iotests/257: add Pattern class
iotests: test bitmap moving inside 254
qapi: implement block-dirty-bitmap-remove transaction action
blockdev: reduce aio_context locked sections in bitmap add/remove
block/backup: loosen restriction on readonly bitmaps
iotests: add test 257 for bitmap-mode backups
...
Thomas Huth [Tue, 16 Jul 2019 09:30:35 +0000 (11:30 +0200)]
gitlab-ci: Remove qcow2 tests that are handled by "make check" already
Since most iotests are now run during "make check" already, we do not
need to test them explicitly from the gitlab-ci.yml script anymore.
And while we're at it, add some of the new non-auto tests >= 246 instead.
Thomas Huth [Fri, 12 Jul 2019 15:39:33 +0000 (17:39 +0200)]
tests: Run the iotests during "make check" again
People often forget to run the iotests before submitting patches or pull
requests - this is likely due to the fact that we do not run the tests
during our mandatory "make check" tests yet. Now that we've got a proper
"auto" group of iotests that should be fine to run in every environment,
we can enable the iotests during "make check" again by running the "auto"
tests by default from the check-block.sh script.
Some cases still need to be checked first, though: iotests need bash and
GNU sed (otherwise they fail), and if gprof is enabled, it spoils the
output of some test cases causing them to fail. So if we detect that one
of the required programs is missing or that gprof is enabled, we still
have to skip the iotests to avoid failures.
And finally, since we are using check-block.sh now again, this patch also
removes the qemu-iotests-quick.sh script since we do not need that anymore
(and having two shell wrapper scripts around the block tests seems rather
confusing than helpful).
Paolo Bonzini [Thu, 25 Jul 2019 09:59:20 +0000 (11:59 +0200)]
block: fix NetBSD qemu-iotests failure
Opening a block device on NetBSD has an additional step compared to other OSes,
corresponding to raw_normalize_devicepath. The error message in that function
is slightly different from that in raw_open_common and this was causing spurious
failures in qemu-iotests. However, in general it is not important to know what
exact step was failing, for example in the qemu-iotests case the error message
contains the fairly unequivocal "No such file or directory" text from strerror.
We can thus fix the failures by standardizing on a single error message for
both raw_open_common and raw_normalize_devicepath; in fact, we can even
use error_setg_file_open to make sure the error message is the same as in
the rest of QEMU.
It's not correct to just ignore an error code in a callback; we need to
handle that error and possible report failure to the guest so that they
don't wait indefinitely for an operation that will now never finish.
This ought to help cases reported by Nutanix where iSCSI returns a
legitimate -ECANCELED for certain operations which should be propagated
normally.
Paolo Bonzini [Mon, 29 Jul 2019 21:34:16 +0000 (23:34 +0200)]
dma-helpers: ensure AIO callback is invoked after cancellation
dma_aio_cancel unschedules the BH if there is one, which corresponds
to the reschedule_dma case of dma_blk_cb. This can stall the DMA
permanently, because dma_complete will never get invoked and therefore
nobody will ever invoke the original AIO callback in dbs->common.cb.
Fix this by invoking the callback (which is ensured to happen after
a bdrv_aio_cancel_async, or done manually in the dbs->bh case), and
add assertions to check that the DMA state machine is indeed waiting
for dma_complete or reschedule_dma, but never both.