ACPI ERST: PCI device_id for ERST

This change reserves the PCI device_id for the new ACPI ERST
device.

Signed-off-by: Eric DeVolder <eric.devolder@oracle.com>
Acked-by: Igor Mammedov <imammedo@redhat.com>
Acked-by: Ani Sinha <ani@anisinha.ca>
Message-Id: <1643402289-22216-4-git-send-email-eric.devolder@oracle.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

ACPI ERST: bios-tables-test.c steps 1 and 2

Following the guidelines in tests/qtest/bios-tables-test.c, this
change adds empty placeholder files per step 1 for the new ERST
table, and excludes resulting changed files in bios-tables-test-allowed-diff.h
per step 2.

Signed-off-by: Eric DeVolder <eric.devolder@oracle.com>
Acked-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <1643402289-22216-2-git-send-email-eric.devolder@oracle.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

libvhost-user: Map shared RAM with MAP_NORESERVE to support virtio-mem with hugetlb

For fd-based shared memory, MAP_NORESERVE is only effective for hugetlb,
otherwise it's ignored. Older Linux versions that didn't support
reservation of huge pages ignored MAP_NORESERVE completely.

The first client to mmap a hugetlb fd without MAP_NORESERVE will
trigger reservation of huge pages for the whole mmapped range. There are
two cases to consider:

1) QEMU mapped RAM without MAP_NORESERVE

We're not dealing with a sparse mapping, huge pages for the whole range
have already been reserved by QEMU. An additional mmap() without
MAP_NORESERVE won't have any effect on the reservation.

2) QEMU mapped RAM with MAP_NORESERVE

We're delaing with a sparse mapping, no huge pages should be reserved.
Further mappings without MAP_NORESERVE should be avoided.

For 1), it doesn't matter if we set MAP_NORESERVE or not, so we can
simply set it. For 2), we'd be overriding QEMUs decision and trigger
reservation of huge pages, which might just fail if there are not
sufficient huge pages around. We must map with MAP_NORESERVE.

This change is required to support virtio-mem with hugetlb: a
virtio-mem device mapped into the guest physical memory corresponds to
a sparse memory mapping and QEMU maps this memory with MAP_NORESERVE.
Whenever memory in that sparse region will be accessed by the VM, QEMU
populates huge pages for the affected range by preallocating memory
and handling any preallocation errors gracefully.

So let's map shared RAM with MAP_NORESERVE. As libvhost-user only
supports Linux, there shouldn't be anything to take care of in regard of
other OS support.

Without this change, libvhost-user will fail mapping the region if there
are currently not enough huge pages to perform the reservation:
 fv_panic: libvhost-user: region mmap error: Cannot allocate memory

Cc: "Marc-André Lureau" <marcandre.lureau@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Raphael Norwitz <raphael.norwitz@nutanix.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20220111123939.132659-1-david@redhat.com>
Acked-by: Raphael Norwitz <raphael.norwitz@nutanix.com>

libvhost-user: handle removal of identical regions

Today if QEMU (or any other VMM) has sent multiple copies of the same
region to a libvhost-user based backend and then attempts to remove the
region, only one instance of the region will be removed, leaving stale
copies of the region in dev->regions[].

This change resolves this by having vu_rem_mem_reg() iterate through all
regions in dev->regions[] and delete all matching regions.

Suggested-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
Message-Id: <20220117041050.19718-7-raphael.norwitz@nutanix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>

libvhost-user: prevent over-running max RAM slots

When VHOST_USER_PROTOCOL_F_CONFIGURE_MEM_SLOTS support was added to
libvhost-user, no guardrails were added to protect against QEMU
attempting to hot-add too many RAM slots to a VM with a libvhost-user
based backed attached.

This change adds the missing error handling by introducing a check on
the number of RAM slots the device has available before proceeding to
process the VHOST_USER_ADD_MEM_REG message.

Suggested-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
Message-Id: <20220117041050.19718-6-raphael.norwitz@nutanix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

libvhost-user: fix VHOST_USER_REM_MEM_REG not closing the fd

We end up not closing the file descriptor, resulting in leaking one
file descriptor for each VHOST_USER_REM_MEM_REG message.

Fixes: 875b9fd97b34 ("Support individual region unmap in libvhost-user")
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Raphael Norwitz <raphael.norwitz@nutanix.com>
Cc: "Marc-André Lureau" <marcandre.lureau@redhat.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Coiby Xu <coiby.xu@gmail.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
Message-Id: <20220117041050.19718-5-raphael.norwitz@nutanix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

libvhost-user: Simplify VHOST_USER_REM_MEM_REG

Let's avoid having to manually copy all elements. Copy only the ones
necessary to close the hole and perform the operation in-place without
a second array.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
Message-Id: <20220117041050.19718-4-raphael.norwitz@nutanix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

libvhost-user: Add vu_add_mem_reg input validation

Today if multiple FDs are sent from the VMM to the backend in a
VHOST_USER_ADD_MEM_REG message, one FD will be mapped and the remaining
FDs will be leaked. Therefore if multiple FDs are sent we report an
error and fail the operation, closing all FDs in the message.

Likewise in case the VMM sends a message with a size less than that
of a memory region descriptor, we add a check to gracefully report an
error and fail the operation rather than crashing.

Signed-off-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
Message-Id: <20220117041050.19718-3-raphael.norwitz@nutanix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>

libvhost-user: Add vu_rem_mem_reg input validation

Today if multiple FDs are sent from the VMM to the backend in a
VHOST_USER_REM_MEM_REG message, one FD will be unmapped and the remaining
FDs will be leaked. Therefore if multiple FDs are sent we report an
error and fail the operation, closing all FDs in the message.

Likewise in case the VMM sends a message with a size less than that of a
memory region descriptor, we add a check to gracefully report an error
and fail the operation rather than crashing.

Signed-off-by: Raphael Norwitz <raphael.norwitz@nutanix.com>
Message-Id: <20220117041050.19718-2-raphael.norwitz@nutanix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>

tests: acpi: test short OEM_ID/OEM_TABLE_ID values in test_oem_fields()

Previous patch [1] added explicit whitespace padding to OEM_ID/OEM_TABLE_ID
values used in test_oem_fields() testcase to avoid false positive and
bisection issues when QEMU is switched to \0' padding. As result
testcase ceased to test values that were shorter than max possible
length values.

Update testcase to make sure that it's testing shorter IDs like it
used to before [2].

1) "tests: acpi: manually pad OEM_ID/OEM_TABLE_ID for  test_oem_fields() test"
2) 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20220114142641.1727679-1-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: update expected blobs

Expected changes caused by previous commit:

nvdimm ssdt (q35/pc/virt):
  - *     OEM Table ID     "NVDIMM  "
  + *     OEM Table ID     "NVDIMM"

SLIC test FADT (tests/data/acpi/q35/FACP.slic):
  -[010h 0016   8]                 Oem Table ID : "ME      "
  +[010h 0016   8]                 Oem Table ID : "ME"

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20220112130332.1648664-5-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

acpi: fix OEM ID/OEM Table ID padding

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID
fields in headers of ACPI tables. While it doesn't have impact on
default values since QEMU uses 6 and 8 characters long values
respectively, it broke usecase where IDs are provided on QEMU CLI.
It shouldn't affect guest (but may cause licensing verification
issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs
shorter than max possible length, where [2] mangles IDs with extra
spaces in RSDT and FADT tables whereas guest OS expects those to
mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in
accordance with [1] and expectations of guest OS

1) ACPI spec, v2.0b
       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.

2)
Fixes: 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/707
Reported-by: Dmitry V. Orekhov <dima.orekhov@gmail.com>
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Cc: qemu-stable@nongnu.org
Message-Id: <20220112130332.1648664-4-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Ani Sinha <ani@anisinha.ca>
Tested-by: Dmitry V. Orekhov dima.orekhov@gmail.com

tests: acpi: whitelist nvdimm's SSDT and FACP.slic expected blobs

The next commit will revert OEM fields whitespace padding to
padding with '\0' as it was before [1]. That will change OEM
Table ID for:
  * SSDT.*: where it was padded from 6 characters to 8
  * FACP.slic: where it was padded from 2 characters to 8
after reverting whitespace padding, it will be replaced with
'\0' which effectively will shorten OEM table ID to 6 and 2
characters.

Whitelist affected tables before introducing the change.

1) 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20220112130332.1648664-3-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: manually pad OEM_ID/OEM_TABLE_ID for test_oem_fields() test

The next commit will revert OEM fields padding with whitespace to
padding with '\0' as it was before [1]. As result test_oem_fields() will
fail due to unexpectedly smaller ID sizes read from QEMU ACPI tables.

Pad OEM_ID/OEM_TABLE_ID manually with spaces so that values the test
puts on QEMU CLI and expected values match.

1) 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20220112130332.1648664-2-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

hw/i386: Add the possibility to disable the 'isapc' machine

We already have a CONFIG_ISAPC switch - but we're not using it yet.
Add some "#ifdefs" to make it possible to disable this machine now.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20220107160713.235918-1-thuth@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

cpuid: use unsigned for max cpuid

__get_cpuid_max returns an unsigned value.
For consistency, store the result in an unsigned variable.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Richard Henderson <rth@twiddle.net>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

Merge remote-tracking branch 'remotes/hdeller/tags/hppa-updates-pull-request' into staging

Fixes and updates for hppa target

This patchset fixes some important bugs in the hppa artist graphics driver:
- Fix artist graphics for HP-UX and Linux
- Mouse cursor fixes for HP-UX
- Fix draw_line() function on artist graphic

and it adds new qemu features for hppa:
- Allow up to 16 emulated CPUs (instead of 8)
- Add support for an emulated TOC/NMI button

A new Seabios-hppa firmware is included as well:
- Update SeaBIOS-hppa to VERSION 3
- New opt/hostid fw_cfg option to change hostid
- Add opt/console fw_cfg option to select default console
- Added 16x32 font to STI firmware

Signed-off-by: Helge Deller <deller@gmx.de>
# gpg: Signature made Wed 02 Feb 2022 18:08:34 GMT
# gpg:                using EDDSA key BCE9123E1AD29F07C049BBDEF712B510A23A0F5F
# gpg: Good signature from "Helge Deller <deller@gmx.de>" [unknown]
# gpg:                 aka "Helge Deller <deller@kernel.org>" [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: 4544 8228 2CD9 10DB EF3D  25F8 3E5F 3D04 A7A2 4603
#      Subkey fingerprint: BCE9 123E 1AD2 9F07 C049  BBDE F712 B510 A23A 0F5F

* remotes/hdeller/tags/hppa-updates-pull-request:
  hw/display/artist: Fix draw_line() artefacts
  hw/display/artist: Mouse cursor fixes for HP-UX
  hw/display/artist: rewrite vram access mode handling
  hppa: Add support for an emulated TOC/NMI button.
  hw/hppa: Allow up to 16 emulated CPUs
  seabios-hppa: Update SeaBIOS-hppa to VERSION 3

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/display/artist: Fix draw_line() artefacts

The draw_line() function left artefacts on the screen because it was using the
x/y variables which were incremented in the loop before. Fix it by using the
unmodified x1/x2 variables instead.

Signed-off-by: Sven Schnelle <svens@stackframe.org>
Signed-off-by: Helge Deller <deller@gmx.de>
Cc: qemu-stable@nongnu.org
Signed-off-by: Helge Deller <deller@gmx.de>

hw/display/artist: Mouse cursor fixes for HP-UX

This patch fix the behaviour and positioning of the X11 mouse cursor in HP-UX.

The current code missed to subtract the offset of the CURSOR_CTRL register from
the current mouse cursor position. The HP-UX graphics driver stores in this
register the offset of the mouse graphics compared to the current cursor
position.  Without this adjustment the mouse behaves strange at the screen
borders.

Additionally, depending on the HP-UX version, the mouse cursor position
in the cursor_pos register reports different values. To accommodate this
track the current min and max reported values and auto-adjust at runtime.

With this fix the mouse now behaves as expected on HP-UX 10 and 11.

Signed-off-by: Helge Deller <deller@gmx.de>
Cc: qemu-stable@nongnu.org
Signed-off-by: Helge Deller <deller@gmx.de>

hw/display/artist: rewrite vram access mode handling

When writing this code it was assumed that register 0x118000 is the
buffer access mode for color map accesses. It turned out that this
is wrong. Instead register 0x118000 sets both src and dst buffer
access mode at the same time.

This required a larger rewrite of the code. The good thing is that
both the linear framebuffer and the register based vram access can
now be combined into one function.

This makes the linux 'stifb' framebuffer work, and both HP-UX 10.20
and HP-UX 11.11 are still working.

Signed-off-by: Sven Schnelle <svens@stackframe.org>
Signed-off-by: Helge Deller <deller@gmx.de>
Cc: qemu-stable@nongnu.org
Signed-off-by: Helge Deller <deller@gmx.de>

hppa: Add support for an emulated TOC/NMI button.

Almost all PA-RISC machines have either a button that is labeled with 'TOC' or
a BMC/GSP function to trigger a TOC.  TOC is a non-maskable interrupt that is
sent to the processor.  This can be used for diagnostic purposes like obtaining
a stack trace/register dump or to enter KDB/KGDB in Linux.

This patch adds support for such an emulated TOC button.

It wires up the qemu monitor "nmi" command to trigger a TOC.  For that it
provides the hppa_nmi function which is assigned to the nmi_monitor_handler
function pointer.  When called it raises the EXCP_TOC hardware interrupt in the
hppa_cpu_do_interrupt() function.  The interrupt function then calls the
architecturally defined TOC function in SeaBIOS-hppa firmware (at fixed address
0xf0000000).

According to the PA-RISC PDC specification, the SeaBIOS firmware then writes
the CPU registers into PIM (processor internal memmory) for later analysis.  In
order to write all registers it needs to know the contents of the CPU "shadow
registers" and the IASQ- and IAOQ-back values. The IAOQ/IASQ values are
provided by qemu in shadow registers when entering the SeaBIOS TOC function.
This patch adds a new aritificial opcode "getshadowregs" (0xfffdead2) which
restores the original values of the shadow registers. With this opcode SeaBIOS
can store those registers as well into PIM before calling an OS-provided TOC
handler.

To trigger a TOC, switch to the qemu monitor with Ctrl-A C, and type in the
command "nmi".  After the TOC started the OS-debugger, exit the qemu monitor
with Ctrl-A C.

Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

hw/hppa: Allow up to 16 emulated CPUs

This brings the hppa_hardware.h file in sync with the copy in the
SeaBIOS-hppa sources.

In order to support up to 16 CPUs, it's required to move the HPA for
MEMORY_HPA out of the address space of the new 16th CPU.
The new address of 0xfffff000 worked well for Linux and HP-UX, while
other addresses close to the former 0xfffbf000 area are used by the
architecture for local and global broadcasts.

The PIM_STORAGE_SIZE constant is used in SeaBIOS sources and
is relevant for the TOC/NMI feature.

Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

seabios-hppa: Update SeaBIOS-hppa to VERSION 3

New firmware features and fixes:

* Allow up to 16 CPUs

* Add TOC button support:
    To trigger a TOC, execute "nmi" in the qemu monitor (Ctrl-A C)

* New opt/hostid fw_cfg option to change hostid:
    -fw_cfg opt/hostid,string=334455

* Add opt/console fw_cfg option to select default console:
    -fw_cfg opt/console,string=serial
    -fw_cfg opt/console,string=graphics

* Add Linux TER16x32 font to STI firmware:
    -fw_cfg opt/font,string=2

* Leave IRQs disabled after rendevouz

Signed-off-by: Helge Deller <deller@gmx.de>

Merge remote-tracking branch 'remotes/bsdimp/tags/bsd-user-fix-pull-request' into staging

Pull request

Fix FreeBSD 12 and 13 builds.

# gpg: Signature made Tue 01 Feb 2022 21:58:37 GMT
# gpg:                using RSA key 2035F894B00AA3CF7CCDE1B76C1CD1287DB01100
# gpg: Good signature from "Warner Losh <wlosh@netflix.com>" [unknown]
# gpg:                 aka "Warner Losh <imp@bsdimp.com>" [unknown]
# gpg:                 aka "Warner Losh <imp@freebsd.org>" [unknown]
# gpg:                 aka "Warner Losh <imp@village.org>" [unknown]
# gpg:                 aka "Warner Losh <wlosh@bsdimp.com>" [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: 2035 F894 B00A A3CF 7CCD  E1B7 6C1C D128 7DB0 1100

* remotes/bsdimp/tags/bsd-user-fix-pull-request:
  bsd-user/signal.c: Only copy the _capsicum for FreeBSD_version > 1400026

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

bsd-user/signal.c: Only copy the _capsicum for FreeBSD_version > 1400026

The capsicum signal stuff is new with FreeBSD 14, rev 1400026, so only
define QEMU_SI_CAPSICUM there. Only copy _capsicum when QEMU_SI_CAPSICUM
is defined. Default to no info being passed for signals we make no guess
about.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

Merge remote-tracking branch 'remotes/kwolf-gitlab/tags/for-upstream' into staging

Block layer patches

- rbd: fix handling of holes in .bdrv_co_block_status
- Fix potential crash in bdrv_set_backing_hd()
- vhost-user-blk export: Fix shutdown with requests in flight
- FUSE export: Fix build failure on FreeBSD
- Documentation improvements

# gpg: Signature made Tue 01 Feb 2022 15:14:24 GMT
# gpg:                using RSA key DC3DEB159A9AF95D3D7456FE7F09B272C88F2FD6
# gpg:                issuer "kwolf@redhat.com"
# gpg: Good signature from "Kevin Wolf <kwolf@redhat.com>" [full]
# Primary key fingerprint: DC3D EB15 9A9A F95D 3D74  56FE 7F09 B272 C88F 2FD6

* remotes/kwolf-gitlab/tags/for-upstream:
  block/rbd: workaround for ceph issue #53784
  block/rbd: fix handling of holes in .bdrv_co_block_status
  qemu-img: Unify [-b [-F]] documentation
  qsd: Document fuse's allow-other option
  block.h: remove outdated comment
  block/export/fuse: Fix build failure on FreeBSD
  block/export/fuse: Rearrange if-else-if ladder in fuse_fallocate()
  block/export: Fix vhost-user-blk shutdown with requests in flight
  block: bdrv_set_backing_hd(): use drained section
  qemu-storage-daemon: Fix typo in vhost-user-blk help

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge remote-tracking branch 'remotes/hreitz-gitlab/tags/pull-block-2022-02-01' into staging

Block patches:
- Add support to the iotests to test qcow2's zstd compression mode
- Fix post-migration block node permissions
- iotests fixes (051 and mirror-ready-cancel-error)
- Remove an outdated comment

# gpg: Signature made Tue 01 Feb 2022 13:34:54 GMT
# gpg:                using RSA key CB62D7A0EE3829E45F004D34A1FA40D098019CDF
# gpg:                issuer "hreitz@redhat.com"
# gpg: Good signature from "Hanna Reitz <hreitz@redhat.com>" [marginal]
# 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: CB62 D7A0 EE38 29E4 5F00  4D34 A1FA 40D0 9801 9CDF

* remotes/hreitz-gitlab/tags/pull-block-2022-02-01: (24 commits)
  block.h: remove outdated comment
  iotests/migration-permissions: New test
  block-backend: Retain permissions after migration
  iotests: declare lack of support for compresion_type in IMGOPTS
  iotest 214: explicit compression type
  iotests 60: more accurate set dirty bit in qcow2 header
  iotests: bash tests: filter compression type
  iotest 39: use _qcow2_dump_header
  iotests: massive use _qcow2_dump_header
  iotests/common.rc: introduce _qcow2_dump_header helper
  qcow2: simple case support for downgrading of qcow2 images with zstd
  iotest 302: use img_info_log() helper
  iotests.py: filter compression type out
  iotests.py: filter out successful output of qemu-img create
  iotest 065: explicit compression type
  iotest 303: explicit compression type
  iotests.py: rewrite default luks support in qemu_img
  iotests: drop qemu_img_verbose() helper
  iotests.py: qemu_img*("create"): support IMGOPTS='compression_type=zstd'
  iotests: specify some unsupported_imgopts for python iotests
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

block/rbd: workaround for ceph issue #53784

librbd had a bug until early 2022 that affected all versions of ceph that
supported fast-diff. This bug results in reporting of incorrect offsets
if the offset parameter to rbd_diff_iterate2 is not object aligned.

This patch works around this bug for pre Quincy versions of librbd.

Fixes: 0347a8fd4c3faaedf119be04c197804be40a384b
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Lieven <pl@kamp.de>
Message-Id: <20220113144426.4036493-3-pl@kamp.de>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Tested-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block/rbd: fix handling of holes in .bdrv_co_block_status

the assumption that we can't hit a hole if we do not diff against a snapshot was wrong.

We can see a hole in an image if we diff against base if there exists an older snapshot
of the image and we have discarded blocks in the image where the snapshot has data.

Fix this by simply handling a hole like an unallocated area. There are no callbacks
for unallocated areas so just bail out if we hit a hole.

Fixes: 0347a8fd4c3faaedf119be04c197804be40a384b
Suggested-by: Ilya Dryomov <idryomov@gmail.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Lieven <pl@kamp.de>
Message-Id: <20220113144426.4036493-2-pl@kamp.de>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-img: Unify [-b [-F]] documentation

qemu-img convert documents the backing file and backing format options
as follows:
    [-B backing_file [-F backing_fmt]]
whereas qemu-img create has this:
    [-b backing_file] [-F backing_fmt]

That is, for convert, we document that -F cannot be given without -B,
while for create, way say that they are independent.

Indeed, it is technically possible to give -F without -b, because it is
left to the block driver to decide whether this is an error or not, so
sometimes it is:

$ qemu-img create -f qed -F qed test.qed 64M
Formatting 'test.qed', fmt=qed size=67108864 backing_fmt=qed [...]

And sometimes it is not:

$ qemu-img create -f qcow2 -F qcow2 test.qcow2 64M
Formatting 'test.qcow2', fmt=qcow2 cluster_size=65536 [...]
qemu-img: test.qcow2: Backing format cannot be used without backing file

Generally, it does not make much sense, though, and users should only
give -F with -b, so document it that way, as we have already done for
qemu-img convert (commit 1899bf47375ad40555dcdff12ba49b4b8b82df38).

Reported-by: Tingting Mao <timao@redhat.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220131135908.32393-1-hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qsd: Document fuse's allow-other option

We did not add documentation to the storage daemon's man page for fuse's
allow-other option when it was introduced, so do that now.

Fixes: 8fc54f9428b9763f800 ("export/fuse: Add allow-other option")
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220131103124.20325-1-hreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block.h: remove outdated comment

The comment "disk I/O throttling" doesn't make any sense at all
any more. It was added in commit 0563e191516 to describe
bdrv_io_limits_enable()/disable(), which were removed in commit
97148076, so the comment is just a forgotten leftover.

Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-Id: <20220131125615.74612-1-eesposit@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block/export/fuse: Fix build failure on FreeBSD

When building on FreeBSD we get:

  [816/6851] Compiling C object libblockdev.fa.p/block_export_fuse.c.o
  ../block/export/fuse.c:628:16: error: use of undeclared identifier 'FALLOC_FL_KEEP_SIZE'
      if (mode & FALLOC_FL_KEEP_SIZE) {
                 ^
  ../block/export/fuse.c:651:16: error: use of undeclared identifier 'FALLOC_FL_PUNCH_HOLE'
      if (mode & FALLOC_FL_PUNCH_HOLE) {
                 ^
  ../block/export/fuse.c:652:22: error: use of undeclared identifier 'FALLOC_FL_KEEP_SIZE'
          if (!(mode & FALLOC_FL_KEEP_SIZE)) {
                       ^
  3 errors generated.
  FAILED: libblockdev.fa.p/block_export_fuse.c.o

Meson indeed reported FALLOC_FL_PUNCH_HOLE is not available:

  C compiler for the host machine: cc (clang 10.0.1 "FreeBSD clang version 10.0.1")
  Checking for function "fallocate" : NO
  Checking for function "posix_fallocate" : YES
  Header <linux/falloc.h> has symbol "FALLOC_FL_PUNCH_HOLE" : NO
  Header <linux/falloc.h> has symbol "FALLOC_FL_ZERO_RANGE" : NO
  ...

Similarly to commit 304332039 ("block/export/fuse.c: fix musl build"),
guard the code requiring FALLOC_FL_KEEP_SIZE / FALLOC_FL_PUNCH_HOLE
definitions under CONFIG_FALLOCATE_PUNCH_HOLE #ifdef'ry.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220201112655.344373-3-f4bug@amsat.org>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block/export/fuse: Rearrange if-else-if ladder in fuse_fallocate()

In order to safely maintain a mixture of #ifdef'ry with if-else-if
ladder, rearrange the last statement (!mode) first. Since it is
mutually exclusive with the other conditions, checking it first
doesn't make any logical difference, but allows to add #ifdef'ry
around in a more cleanly way.

Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220201112655.344373-2-f4bug@amsat.org>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block/export: Fix vhost-user-blk shutdown with requests in flight

The vhost-user-blk export runs requests asynchronously in their own
coroutine. When the vhost connection goes away and we want to stop the
vhost-user server, we need to wait for these coroutines to stop before
we can unmap the shared memory. Otherwise, they would still access the
unmapped memory and crash.

This introduces a refcount to VuServer which is increased when spawning
a new request coroutine and decreased before the coroutine exits. The
memory is only unmapped when the refcount reaches zero.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20220125151435.48792-1-kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: bdrv_set_backing_hd(): use drained section

Graph modifications should be done in drained section. stream_prepare()
handler of block stream job call bdrv_set_backing_hd() without using
drained section and it's theoretically possible that some IO request
will interleave with graph modification and will use outdated pointers
to removed block nodes.

Some other callers use bdrv_set_backing_hd() not caring about drained
sections too. So it seems good to make a drained section exactly in
bdrv_set_backing_hd().

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20220124173741.2984056-1-vsementsov@virtuozzo.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-storage-daemon: Fix typo in vhost-user-blk help

The syntax of the fd passing case misses the "addr.type=" key. Add it.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20220125151514.49035-1-kwolf@redhat.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block.h: remove outdated comment

The comment "disk I/O throttling" doesn't make any sense at all
any more. It was added in commit 0563e191516 to describe
bdrv_io_limits_enable()/disable(), which were removed in commit
97148076, so the comment is just a forgotten leftover.

Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Message-Id: <20220131125615.74612-1-eesposit@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests/migration-permissions: New test

This test checks that a raw image in use by a virtio-blk device does not
share the WRITE permission both before and after migration.

Signed-off-by: Hanna Reitz <hreitz@redhat.com>

block-backend: Retain permissions after migration

After migration, the permissions the guest device wants to impose on its
BlockBackend are stored in blk->perm and blk->shared_perm.  In
blk_root_activate(), we take our permissions, but keep all shared
permissions open by calling `blk_set_perm(blk->perm, BLK_PERM_ALL)`.

Only afterwards (immediately or later, depending on the runstate) do we
restrict the shared permissions by calling
`blk_set_perm(blk->perm, blk->shared_perm)`.  Unfortunately, our first
call with shared_perm=BLK_PERM_ALL has overwritten blk->shared_perm to
be BLK_PERM_ALL, so this is a no-op and the set of shared permissions is
not restricted.

Fix this bug by saving the set of shared permissions before invoking
blk_set_perm() with BLK_PERM_ALL and restoring it afterwards.

Fixes: 5f7772c4d0cf32f4e779fcd5a69ae4dae24aeebf
       ("block-backend: Defer shared_perm tightening migration
       completion")
Reported-by: Peng Liang <liangpeng10@huawei.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20211125135317.186576-2-hreitz@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Tested-by: Peng Liang <liangpeng10@huawei.com>

iotests: declare lack of support for compresion_type in IMGOPTS

compression_type can't be used if we want to create image with
compat=0.10. So, skip these tests, not many of them.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20211223160144.1097696-20-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotest 214: explicit compression type

The test-case "Corrupted size field in compressed cluster descriptor"
heavily depends on zlib compression type. So, make it explicit. This
way test passes with IMGOPTS='compression_type=zstd'.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <20211223160144.1097696-19-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests 60: more accurate set dirty bit in qcow2 header

Don't touch other incompatible bits, like compression-type. This makes
the test pass with IMGOPTS='compression_type=zstd'.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <20211223160144.1097696-18-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests: bash tests: filter compression type

We want iotests pass with both the default zlib compression and with
IMGOPTS='compression_type=zstd'.

Actually the only test that is interested in real compression type in
test output is 287 (test for qcow2 compression type), so implement
specific option for it.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20211223160144.1097696-17-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotest 39: use _qcow2_dump_header

_qcow2_dump_header has filter for compression type, so this change
makes test pass with IMGOPTS='compression_type=zstd'.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <20211223160144.1097696-16-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests: massive use _qcow2_dump_header

We are going to add filtering in _qcow2_dump_header and want all tests
use it.

The patch is generated by commands:
  cd tests/qemu-iotests
  sed -ie 's/$PYTHON qcow2.py "$TEST_IMG" dump-header\($\| \)/_qcow2_dump_header\1/' ??? tests/*

(the difficulty is to avoid converting dump-header-exts)

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <20211223160144.1097696-15-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests/common.rc: introduce _qcow2_dump_header helper

We'll use it in tests instead of explicit qcow2.py. Then we are going
to add some filtering in _qcow2_dump_header.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <20211223160144.1097696-14-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

qcow2: simple case support for downgrading of qcow2 images with zstd

If image doesn't have any compressed cluster we can easily switch to
zlib compression, which may allow to downgrade the image.

That's mostly needed to support IMGOPTS='compression_type=zstd' in some
iotests which do qcow2 downgrade.

While being here also fix checkpatch complain against '#' in printf
formatting.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <20211223160144.1097696-13-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotest 302: use img_info_log() helper

Instead of qemu_img_log("info", ..) use generic helper img_info_log().

img_info_log() has smarter logic. For example it use filter_img_info()
to filter output, which in turns filter a compression type. So it will
help us in future when we implement a possibility to use zstd
compression by default (with help of some runtime config file or maybe
build option). For now to test you should recompile qemu with a small
addition into block/qcow2.c before
"if (qcow2_opts->has_compression_type":

    if (!qcow2_opts->has_compression_type && version >= 3) {
        qcow2_opts->has_compression_type = true;
        qcow2_opts->compression_type = QCOW2_COMPRESSION_TYPE_ZSTD;
    }

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <20211223160144.1097696-12-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests.py: filter compression type out

We want iotests pass with both the default zlib compression and with
IMGOPTS='compression_type=zstd'.

Actually the only test that is interested in real compression type in
test output is 287 (test for qcow2 compression type) and it's in bash.
So for now we can safely filter out compression type in all qcow2
tests.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20211223160144.1097696-11-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests.py: filter out successful output of qemu-img create

The only "feature" of this "Formatting ..." line is that we have to
update it every time we add new option. Let's drop it.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20211223160144.1097696-10-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotest 065: explicit compression type

The test checks different options. It of course fails if set
IMGOPTS='compression_type=zstd'. So, let's be explicit in what
compression type we want and independent of IMGOPTS. Test both existing
compression types.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20211223160144.1097696-9-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotest 303: explicit compression type

The test prints qcow2 header fields which depends on chosen compression
type. So, let's be explicit in what compression type we want and
independent of IMGOPTS. Test both existing compression types.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <20211223160144.1097696-8-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests.py: rewrite default luks support in qemu_img

Move the logic to more generic qemu_img_pipe_and_status(). Also behave
better when we have several -o options. And reuse argument parser of
course.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20211223160144.1097696-7-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests: drop qemu_img_verbose() helper

qemu_img_verbose() has a drawback of not going through generic
qemu_img_pipe_and_status(). qemu_img_verbose() is not very popular, so
update the only two users to qemu_img_log() and drop qemu_img_verbose()
at all.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20211223160144.1097696-6-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests.py: qemu_img*("create"): support IMGOPTS='compression_type=zstd'

Adding support of IMGOPTS (like in bash tests) allows user to pass a
lot of different options. Still, some may require additional logic.

Now we want compression_type option, so add some smart logic around it:
ignore compression_type=zstd in IMGOPTS, if test want qcow2 in
compatibility mode. As well, ignore compression_type for non-qcow2
formats.

Note that we may instead add support only to qemu_img_create(), but
that works bad:

1. We'll have to update a lot of tests to use qemu_img_create instead
   of qemu_img('create'). (still, we may want do it anyway, but no
   reason to create a dependancy between task of supporting IMGOPTS and
   updating a lot of tests)

2. Some tests use qemu_img_pipe('create', ..) - even more work on
   updating

3. Even if we update all tests to go through qemu_img_create, we'll
   need a way to avoid creating new tests using qemu_img*('create') -
   add assertions.. That doesn't seem good.

So, let's add support of IMGOPTS to most generic
qemu_img_pipe_and_status().

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20211223160144.1097696-5-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests: specify some unsupported_imgopts for python iotests

We are going to support IMGOPTS for python iotests. Still some iotests
will not work with common IMGOPTS used with bash iotests like
specifying refcount_bits and compat qcow2 options. So we
should define corresponding unsupported_imgopts for now.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20211223160144.1097696-4-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests.py: implement unsupported_imgopts

We are going to support some addition IMGOPTS in python iotests like
in bash iotests. Similarly to bash iotests, we want a way to skip some
tests which can't work with specific IMGOPTS.

Globally for python iotests we will not support things like
'data_file=$TEST_IMG.ext_data_file' in IMGOPTS, so, forbid this
globally in iotests.py.

Suggested-by: Hanna Reitz <hreitz@redhat.com>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20211223160144.1097696-3-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests.py: img_info_log(): rename imgopts argument

We are going to support IMGOPTS environment variable like in bash
tests. Corresponding global variable in iotests.py should be called
imgopts. So to not interfere with function argument, rename it in
advance.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-Id: <20211223160144.1097696-2-vsementsov@virtuozzo.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

iotests/MRCE: Write data to source

This test assumes that mirror flushes the source when entering the READY
state, and that the format level will pass that flush on to the protocol
level (where we intercept it with blkdebug).

However, apparently that does not happen when using a VMDK image with
zeroed_grain=on, which actually is the default set by testenv.py.  Right
now, Python tests ignore IMGOPTS, though, so this has no effect; but
Vladimir has a series that will change this, so we need to fix this test
before that series lands.

We can fix it by writing data to the source before we start the mirror
job; apparently that makes the (VMDK) format layer change its mind and
pass on the pre-READY flush to the protocol level, so the test passes
again.  (I presume, without any data written, mirror just does a 64M
zero write on the target, which VMDK with zeroed_grain=on basically just
ignores.)

Without this, we do not get a flush, and so blkdebug only sees a single
flush at the end of the job instead of two, and therefore does not
inject an error, which makes the block job complete instead of raising
an error.

Signed-off-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20211223165308.103793-1-hreitz@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

tests/qemu-iotests: Fix 051 for binaries without 'lsi53c895a'

The lsi53c895a SCSI adaptor might not be enabled in each and every
x86 QEMU binary, e.g. it's disabled in the RHEL/CentOS build.
Thus let's add a check to the 051 test so that it does not fail if
this device is not available.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20211206143404.247032-1-thuth@redhat.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>

Merge remote-tracking branch 'remotes/bsdimp/tags/bsd-user-arm-2022q1-pull-request' into staging

bsd-user: upstream signal implementation

Upstream the bsd-user fork signal implementation, for the most part.  This
series of commits represents nearly all of the infrastructure that surround
signals, except the actual system call glue (that was also reworked in the
fork and needs its own series). In addition, this adds the sigsegv and sigbus
code to arm. Even in the fork, we don't have good x86 signal implementation,
so there's little to upstream for that at the moment.

bsd-user's signal implementation is similar to linux-user's. The full context
can be found in the bsd-user's fork's 'blitz branch' at
https://github.com/qemu-bsd-user/qemu-bsd-user/tree/blitz which shows how these
are used to implement various system calls. Since this was built from
linux-user's stack stuff, evolved for BSD with the passage of a few years, it
no-doubt missed some bug fixes from linux-user (though nothing obvious stood out
in the quick comparison I made). After the first round of reviews, many of these
improvements have been incorporated.

Patchew history: https://patchew.org/QEMU/20220125012947.14974-1-imp@bsdimp.com/

# gpg: Signature made Mon 31 Jan 2022 19:55:51 GMT
# gpg:                using RSA key 2035F894B00AA3CF7CCDE1B76C1CD1287DB01100
# gpg: Good signature from "Warner Losh <wlosh@netflix.com>" [unknown]
# gpg:                 aka "Warner Losh <imp@bsdimp.com>" [unknown]
# gpg:                 aka "Warner Losh <imp@freebsd.org>" [unknown]
# gpg:                 aka "Warner Losh <imp@village.org>" [unknown]
# gpg:                 aka "Warner Losh <wlosh@bsdimp.com>" [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: 2035 F894 B00A A3CF 7CCD  E1B7 6C1C D128 7DB0 1100

* remotes/bsdimp/tags/bsd-user-arm-2022q1-pull-request: (40 commits)
  bsd-user/freebsd/target_os_ucontext.h: Prefer env as arg name for CPUArchState args
  bsd-user: Rename arg name for target_cpu_reset to env
  MAINTAINERS: Add tests/vm/*bsd to the list to get reviews on
  bsd-user/signal.c: do_sigaltstack
  bsd-user/signal.c: implement do_sigaction
  bsd-user/signal.c: implement do_sigreturn
  bsd-user/signal.c: process_pending_signals
  bsd-user/signal.c: tswap_siginfo
  bsd-user/signal.c: handle_pending_signal
  bsd-user/signal.c: setup_frame
  bsd-user/signal.c: sigset manipulation routines.
  bsd-user/signal.c: Fill in queue_signal
  bsd-user/signal.c: Implement dump_core_and_abort
  bsd-user/strace.c: print_taken_signal
  bsd-user/signal.c: Implement host_signal_handler
  bsd-user/signal.c: Implement rewind_if_in_safe_syscall
  bsd-user/signal.c: host_to_target_siginfo_noswap
  bsd-user: Add trace events for bsd-user
  bsd-user: Add host signals to the build
  bsd-user/host/x86_64/host-signal.h: Implement host_signal_*
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge remote-tracking branch 'remotes/legoater/tags/pull-ppc-20220130' into staging

ppc 7.0 queue:

* Exception and TLB fixes for the 405 CPU (Fabiano and Cedric)
* spapr fixes (Alexey and Daniel)
* PowerNV PHB3/4 fixes (Frederic and Daniel)
* PowerNV XIVE improvements (Cedric)
* 603 CPUs fixes (Christophe)
* Book-E exception fixes (Vitaly)
* Misc compile issues  (Philippe and Fabiano)
* Exception model rework for the BookS CPUs (Fabiano)
* Exception model rework for the 74xx CPUs (Fabiano)
* Removal of 602 CPUs

# gpg: Signature made Sun 30 Jan 2022 17:42:23 GMT
# gpg:                using RSA key A0F66548F04895EBFE6B0B6051A343C7CFFBECA1
# gpg: Good signature from "Cédric Le Goater <clg@kaod.org>" [undefined]
# 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: A0F6 6548 F048 95EB FE6B  0B60 51A3 43C7 CFFB ECA1

* remotes/legoater/tags/pull-ppc-20220130: (41 commits)
  target/ppc: Remove support for the PowerPC 602 CPU
  target/ppc: 74xx: Set SRRs directly in exception code
  target/ppc: 74xx: System Reset interrupt cleanup
  target/ppc: 74xx: System Call exception cleanup
  target/ppc: 74xx: Program exception cleanup
  target/ppc: 74xx: External interrupt cleanup
  target/ppc: 74xx: Machine Check exception cleanup
  target/ppc: Simplify powerpc_excp_74xx
  target/ppc: Introduce powerpc_excp_74xx
  target/ppc: books: Program exception cleanup
  target/ppc: books: External interrupt cleanup
  target/ppc: books: Machine Check exception cleanup
  target/ppc: Simplify powerpc_excp_books
  target/ppc: Introduce powerpc_excp_books
  target/ppc: 405: Watchdog timer exception cleanup
  target/ppc: 405: Program exception cleanup
  target/ppc: 405: Instruction storage interrupt cleanup
  target/ppc: 405: Data Storage exception cleanup
  target/ppc: 405: Debug exception cleanup
  target/ppc: 405: Alignment exception cleanup
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

bsd-user/freebsd/target_os_ucontext.h: Prefer env as arg name for CPUArchState args

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user: Rename arg name for target_cpu_reset to env

Rename the parameter name for target_cpu_reset's CPUArchState * arg from
cpu to env.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

MAINTAINERS: Add tests/vm/*bsd to the list to get reviews on

tests/vm/*bsd (especailly tests/vm/freebsd) are adjacent to the bsd-user
stuff and we're keen on keeping them working as well.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: do_sigaltstack

Implement the meat of the sigaltstack(2) system call with do_sigaltstack.

With that, all the stubbed out routines are complete, so remove
now-incorrect comment.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: implement do_sigaction

Implement the meat of the sigaction(2) system call with do_sigaction and
helper routiner block_signals (which is also used to implemement signal
masking so it's global).

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: implement do_sigreturn

Implements the meat of a sigreturn(2) system call via do_sigreturn, and
helper reset_signal_mask. Fix the prototype of do_sigreturn in qemu.h
and remove do_rt_sigreturn since it's linux only.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

Merge remote-tracking branch 'remotes/ericb/tags/pull-nbd-2022-01-28' into staging

nbd patches for 2022-01-28

- Hanna Reitz: regression fix for block status caching
- Philippe Mathieu-Daude: documentation formatting
- Nir Soffer: dead code removal

# gpg: Signature made Fri 28 Jan 2022 22:58:50 GMT
# gpg:                using RSA key 71C2CC22B1C4602927D2F3AAA7A16B4A2527436A
# gpg: Good signature from "Eric Blake <eblake@redhat.com>" [full]
# gpg:                 aka "Eric Blake (Free Software Programmer) <ebb9@byu.net>" [full]
# gpg:                 aka "[jpeg image of size 6874]" [full]
# Primary key fingerprint: 71C2 CC22 B1C4 6029 27D2  F3AA A7A1 6B4A 2527 436A

* remotes/ericb/tags/pull-nbd-2022-01-28:
  iotests/block-status-cache: New test
  block/io: Update BSC only if want_zero is true
  qapi/block: Cosmetic change in BlockExportType schema
  nbd/server.c: Remove unused field

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge remote-tracking branch 'remotes/quintela-gitlab/tags/migration-20220128-pull-request' into staging

Migration Pull request (Take 2)

Hi

This time I have disabled vmstate canary patches form Dave Gilbert.

Let's see if it works.

Later, Juan.

# gpg: Signature made Fri 28 Jan 2022 18:30:25 GMT
# gpg:                using RSA key 1899FF8EDEBF58CCEE034B82F487EF185872D723
# gpg: Good signature from "Juan Quintela <quintela@redhat.com>" [full]
# gpg:                 aka "Juan Quintela <quintela@trasno.org>" [full]
# Primary key fingerprint: 1899 FF8E DEBF 58CC EE03  4B82 F487 EF18 5872 D723

* remotes/quintela-gitlab/tags/migration-20220128-pull-request: (36 commits)
  migration: Move temp page setup and cleanup into separate functions
  migration: Simplify unqueue_page()
  migration: Add postcopy_has_request()
  migration: Enable UFFD_FEATURE_THREAD_ID even without blocktime feat
  migration: No off-by-one for pss->page update in host page size
  migration: Tally pre-copy, downtime and post-copy bytes independently
  migration: Introduce ram_transferred_add()
  migration: Don't return for postcopy_send_discard_bm_ram()
  migration: Drop return code for disgard ram process
  migration: Do chunk page in postcopy_each_ram_send_discard()
  migration: Drop postcopy_chunk_hostpages()
  migration: Don't return for postcopy_chunk_hostpages()
  migration: Drop dead code of ram_debug_dump_bitmap()
  migration/ram: clean up unused comment.
  migration: Report the error returned when save_live_iterate fails
  migration/migration.c: Remove the MIGRATION_STATUS_ACTIVE when migration finished
  migration/migration.c: Avoid COLO boot in postcopy migration
  migration/migration.c: Add missed default error handler for migration state
  Remove unnecessary minimum_version_id_old fields
  multifd: Rename pages_used to normal_pages
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

iotests/block-status-cache: New test

Add a new test to verify that want_zero=false block-status calls do not
pollute the block-status cache for want_zero=true calls.

We check want_zero=true calls and their results using `qemu-img map`
(over NBD), and want_zero=false calls also using `qemu-img map` over
NBD, but using the qemu:allocation-depth context.

(This test case cannot be integrated into nbd-qemu-allocation, because
that is a qcow2 test, and this is a raw test.)

Signed-off-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220118170000.49423-3-hreitz@redhat.com>
Reviewed-by: Nir Soffer <nsoffer@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Tested-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>

bsd-user/signal.c: process_pending_signals

Process the currently queued signals.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: tswap_siginfo

Convert siginfo from targer to host.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: handle_pending_signal

Handle a queued signal.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: setup_frame

setup_frame sets up a signalled stack frame. Associated routines to
extract the pointer to the stack frame and to support alternate stacks.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: sigset manipulation routines.

target_sigemptyset: resets a set to having no bits set
target_sigaddset:   adds a signal to a set
target_sigismember: returns true when signal is a member
host_to_target_sigset_internal: convert host sigset to target
host_to_target_sigset: convert host sigset to target
target_to_host_sigset_internal: convert target sigset to host
target_to_host_sigset: convert target sigset to host

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: Fill in queue_signal

Fill in queue signal implementation, as well as routines allocate and
delete elements of the signal queue.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: Implement dump_core_and_abort

Force delivering a signal and generating a core file. It's a global
function for the moment...

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/strace.c: print_taken_signal

print_taken_signal() prints signals when we're tracing signals.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: Implement host_signal_handler

Implement host_signal_handler to handle signals generated by the host
and to do safe system calls.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: Implement rewind_if_in_safe_syscall

Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: host_to_target_siginfo_noswap

Implement conversion of host to target siginfo.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user: Add trace events for bsd-user

Add the bsd-user specific events and infrastructure. Only include the
linux-user trace events for linux-user, not bsd-user.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user: Add host signals to the build

Start to add the host signal functionality to the build.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

block/io: Update BSC only if want_zero is true

We update the block-status cache whenever we get new information from a
bdrv_co_block_status() call to the block driver.  However, if we have
passed want_zero=false to that call, it may flag areas containing zeroes
as data, and so we would update the block-status cache with wrong
information.

Therefore, we should not update the cache with want_zero=false.

Reported-by: Nir Soffer <nsoffer@redhat.com>
Fixes: 0bc329fbb00 ("block: block-status cache for data regions")
Reviewed-by: Nir Soffer <nsoffer@redhat.com>
Cc: qemu-stable@nongnu.org
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
Message-Id: <20220118170000.49423-2-hreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>

bsd-user/host/x86_64/host-signal.h: Implement host_signal_*

Implement host_signal_pc, host_signal_set_pc and host_signal_write for
x86_64.

Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/host/i386/host-signal.h: Implement host_signal_*

Implement host_signal_pc, host_signal_set_pc and host_signal_write for
i386.

Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/host/arm/host-signal.h: Implement host_signal_*

Implement host_signal_pc, host_signal_set_pc and host_signal_write for
arm.

Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: Add si_type argument to queue_signal

Mirror the linux-user practice and add a si_type argument to queue
signal. This will be transported as the upper 8 bits in the si_type
element of siginfo so that we know what bits of the structure are valid
and so we can properly implement host_to_target_siginfo_noswap and
tswap_siginfo. Adapt the one caller of queue_signal to the new
interface.  Use all the same names as Linux (except _RT which we don't
treat differently, unlike Linux), though some are unused. Place this
into signal-common.h since that's a better place given bsd-user's
structure. Move prototype of queue_signal to signal-common.h to mirror
linux-user's location.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: Implement signal_init()

Initialize the signal state for the emulator. Setup a set of sane
default signal handlers, mirroring the host's signals. For fatal signals
(those that exit by default), establish our own set of signal
handlers. Stub out the actual signal handler we use for the moment.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org> XXX SIGPROF PENDING

bsd-user/signal.c: implement abstract target / host signal translation

Implement host_to_target_signal and target_to_host_signal.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_cpu.h: Implement data faults

Update for the richer set of data faults that are now possible. Copied
largely from linux-user/arm/cpu_loop.c, with minor typo fixes.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_cpu.h: Use force_sig_fault for EXCP_UDEF

Use force_sig_fault to implement unknown opcode. This just uninlines
that function, so simplify things by using it. Fold in EXCP_NOCP and
EXCP_INVSTATE, as is done in linux-user. Make a note about slight
differences with FreeBSD in case any of them turn out to be important
later.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/target_arch_cpu.h: Correct code pointer

The code has moved in FreeBSD since the emulator was started, update the
comment to reflect that change.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/arm/arget_arch_cpu.h: Move EXCP_DEBUG and EXCP_BKPT together

Implement EXCP_DEBUG and EXCP_BKPT the same, as is done in
linux-user. The prior adjustment of register 15 isn't needed, so remove
that. Remove a redunant comment (that code in FreeBSD never handled
break points). It's unclear why BKPT was an alias for system calls,
but FreeBSD doesn't do that today.

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: implement cpu_loop_exit_sigbus

First attempt at implementing cpu_loop_exit_sigbus, mostly copied from
linux-user version of this function.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: Implement cpu_loop_exit_sigsegv

First attempt at implementing cpu_loop_exit_sigsegv, mostly copied from
linux-user version of this function.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal-common.h: Move signal functions prototypes to here

Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

bsd-user/signal.c: implement force_sig_fault

Start to implement the force_sig_fault code. This currently just calls
queue_signal(). The bsd-user fork version of that will handle this the
synchronous nature of this call. Add signal-common.h to hold signal
helper functions like force_sig_fault.

Signed-off-by: Stacey Son <sson@FreeBSD.org>
Signed-off-by: Kyle Evans <kevans@freebsd.org>
Signed-off-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>