makefile: handle -n / -k / -q correctly

Use $(findstring) instead of $(filter) to detect -n/-k
as different versions of MAKE fill in $(MAKEFLAGS) differently.
Do not bother running ninja at all if -nq is passed.

Tested-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20201026155854.3074290-1-pbonzini@redhat.com>

gitlab-ci: Clone from GitLab itself

Let GitLab runners use GitLab repository directly.

Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20201022123302.2884788-1-philmd@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>

tests/acceptance: pick a random gdb port for reverse debugging

Currently the test randomly fails if you are using a shared machine
due to contention on the well known port 1234. We can ameliorate this
a bit by picking a random non-ephemeral port although it doesn't
totally avoid the problem. While we could use a totally unique socket
address for debugging it is fiddly to probe for gdb support. While gdb
socket debugging is not yet ubiquitous this a sub-optimal but workable
option.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Pavel Dovgalyuk <pavel.dovgalyuk@ispras.ru>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20201021163136.27324-7-alex.bennee@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>

scripts: fix error from checkpatch.pl when no commits are found

The error message was supposed to mention the input revision list start
point, not the branch flag.

Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20201019143537.283094-3-berrange@redhat.com>
Message-Id: <20201021163136.27324-6-alex.bennee@linaro.org>

gitlab: skip checkpatch.pl checks if no commit delta on branch

If the current branch is synced to the current upstream git master,
there are no commits that need checking. This causes checkpatch.pl
to print an error that it found no commits. We need to avoid calling
checkpatch.pl in this case.

Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20201019143537.283094-2-berrange@redhat.com>
Message-Id: <20201021163136.27324-5-alex.bennee@linaro.org>

tests/docker/dockerfiles/centos: Use SDL2 instead of SDL1

We do not support SDL1 in QEMU anymore. Use SDL2 instead.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20201021072308.9224-1-thuth@redhat.com>
Message-Id: <20201021163136.27324-4-alex.bennee@linaro.org>

contrib/gitdm: Add more individual contributors

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Acked-by: Subbaraya Sundeep <sundeep.lkml@gmail.com>
Acked-by: Michael Rolnik <mrolnik@gmail.com>
Acked-by: Thomas Huth <huth@tuxfamily.org>
Acked-by: James Hogan <jhogan@kernel.org>
Acked-by: Artyom Tarasenko <atar4qemu@gmail.com>
Message-id: <20201004182506.2038515-1-f4bug@amsat.org>
Message-Id: <20201021163136.27324-3-alex.bennee@linaro.org>

Adding ani's email as an individual contributor

Ani is an individual contributor into qemu project. Adding my email into the
correct file to reflect so.

Signed-off-by: Ani Sinha <ani@anisinha.ca>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20201007161940.1478-1-ani@anisinha.ca>
Message-Id: <20201021163136.27324-2-alex.bennee@linaro.org>

Merge remote-tracking branch 'remotes/philmd-gitlab/tags/sd-next-20201026' into staging

SD/MMC patches

- Fix a bug in CMD6/SWITCH_FUNCTION (Bin Meng)
- Minor housekeeping patches

CI jobs result:
. https://cirrus-ci.com/build/5461987880599552
. https://gitlab.com/philmd/qemu/-/pipelines/207532287
. https://travis-ci.org/github/philmd/qemu/builds/738901111
. https://app.shippable.com/github/philmd/qemu/runs/888/summary/console

# gpg: Signature made Mon 26 Oct 2020 11:02:22 GMT
# gpg:                using RSA key FAABE75E12917221DCFD6BB2E3E32C2CDEADC0DE
# gpg: Good signature from "Philippe Mathieu-Daudé (F4BUG) <f4bug@amsat.org>" [full]
# Primary key fingerprint: FAAB E75E 1291 7221 DCFD  6BB2 E3E3 2C2C DEAD C0DE

* remotes/philmd-gitlab/tags/sd-next-20201026:
  hw/sd/sdcard: Zero out function selection fields before being populated
  hw/sd/sdcard: Make iolen unsigned
  hw/sd/sdcard: Constify sd_crc*()'s message argument
  hw/sd/sdcard: Simplify cmd_valid_while_locked()
  hw/sd/sdcard: Update the SDState documentation

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

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

* fix --disable-tcg builds (Claudio)
* Fixes for macOS --enable-modules build and OpenBSD curses/iconv detection (myself)
* Start preparing for meson 0.56 (myself)
* Move directory configuration to meson (myself)
* Start untangling qemu_init (myself)
* Windows fixes (Sunil)
* Remove -no-kbm (Thomas)

# gpg: Signature made Mon 26 Oct 2020 11:12:17 GMT
# gpg:                using RSA key F13338574B662389866C7682BFFBD25F78C7AE83
# gpg:                issuer "pbonzini@redhat.com"
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>" [full]
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>" [full]
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* remotes/bonzini-gitlab/tags/for-upstream:
  machine: move SMP initialization from vl.c
  machine: move UP defaults to class_base_init
  machine: remove deprecated -machine enforce-config-section option
  win32: boot broken when bind & data dir are the same
  WHPX: Fix WHPX build break
  configure: move install_blobs from configure to meson
  configure: remove unused variable from config-host.mak
  configure: move directory options from config-host.mak to meson
  configure: allow configuring localedir
  Makefile: separate meson rerun from the rest of the ninja invocation
  Remove deprecated -no-kvm option
  replay: do not build if TCG is not available
  qtest: unbreak non-TCG builds in bios-tables-test
  hw/core/qdev-clock: add a reference on aliased clocks
  do not use colons in test names
  meson: rewrite curses/iconv test
  build: fix macOS --enable-modules build

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

Merge remote-tracking branch 'remotes/cohuck/tags/s390x-20201026' into staging

some s390x fixes

# gpg: Signature made Mon 26 Oct 2020 10:46:50 GMT
# gpg:                using RSA key C3D0D66DC3624FF6A8C018CEDECF6B93C6F02FAF
# gpg:                issuer "cohuck@redhat.com"
# gpg: Good signature from "Cornelia Huck <conny@cornelia-huck.de>" [unknown]
# gpg:                 aka "Cornelia Huck <huckc@linux.vnet.ibm.com>" [full]
# gpg:                 aka "Cornelia Huck <cornelia.huck@de.ibm.com>" [full]
# gpg:                 aka "Cornelia Huck <cohuck@kernel.org>" [unknown]
# gpg:                 aka "Cornelia Huck <cohuck@redhat.com>" [unknown]
# Primary key fingerprint: C3D0 D66D C362 4FF6 A8C0  18CE DECF 6B93 C6F0 2FAF

* remotes/cohuck/tags/s390x-20201026:
  s390x: pv: Fix diag318 PV fencing
  s390x: pv: Remove sclp boundary checks
  s390x/s390-virtio-ccw: Reset PCI devices during subsystem reset

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

Merge remote-tracking branch 'remotes/huth-gitlab/tags/pull-request-2020-10-26' into staging

* qtest fixes (e.g. memory leaks)
* Fix for Xen dummy cpu loop (which happened due to qtest accel rework)
* Introduction of the generic device fuzzer
* Run more check-acceptance tests in the gitlab-CI

# gpg: Signature made Mon 26 Oct 2020 09:34:04 GMT
# gpg:                using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5
# gpg:                issuer "thuth@redhat.com"
# gpg: Good signature from "Thomas Huth <th.huth@gmx.de>" [full]
# gpg:                 aka "Thomas Huth <thuth@redhat.com>" [full]
# gpg:                 aka "Thomas Huth <huth@tuxfamily.org>" [full]
# gpg:                 aka "Thomas Huth <th.huth@posteo.de>" [unknown]
# Primary key fingerprint: 27B8 8847 EEE0 2501 18F3  EAB9 2ED9 D774 FE70 2DB5

* remotes/huth-gitlab/tags/pull-request-2020-10-26: (31 commits)
  tests/acceptance: Use .ppm extention for Portable PixMap files
  tests/acceptance: Remove unused import
  test/docker/dockerfiles: Add missing packages for acceptance tests
  tests/acceptance: Enable AVOCADO_ALLOW_UNTRUSTED_CODE in the gitlab-CI
  test/acceptance: Remove the CONTINUOUS_INTEGRATION tags
  tests/acceptance/ppc_prep_40p: Fix the URL to the NetBSD-4.0 archive
  scripts/oss-fuzz: ignore the generic-fuzz target
  scripts/oss-fuzz: use hardlinks instead of copying
  fuzz: register predefined generic-fuzz configs
  fuzz: add generic-fuzz configs for oss-fuzz
  fuzz: add an "opaque" to the FuzzTarget struct
  fuzz: Add instructions for using generic-fuzz
  scripts/oss-fuzz: Add crash trace minimization script
  scripts/oss-fuzz: Add script to reorder a generic-fuzzer trace
  fuzz: add a crossover function to generic-fuzzer
  fuzz: add a DISABLE_PCI op to generic-fuzzer
  fuzz: Add support for custom crossover functions
  fuzz: Add fuzzer callbacks to DMA-read functions
  fuzz: Declare DMA Read callback function
  fuzz: Add DMA support to the generic-fuzzer
  ...

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

Merge remote-tracking branch 'remotes/alistair/tags/pull-riscv-to-apply-20201023' into staging

A collection of RISC-V fixes for the next QEMU release.

This includes:
 - Improvements to logging output
 - Hypervisor instruction fixups
 - The ability to load a noMMU kernel
 - SiFive OTP support

# gpg: Signature made Fri 23 Oct 2020 16:13:57 BST
# gpg:                using RSA key F6C4AC46D4934868D3B8CE8F21E10D29DF977054
# gpg: Good signature from "Alistair Francis <alistair@alistair23.me>" [full]
# Primary key fingerprint: F6C4 AC46 D493 4868 D3B8  CE8F 21E1 0D29 DF97 7054

* remotes/alistair/tags/pull-riscv-to-apply-20201023:
  hw/misc/sifive_u_otp: Add backend drive support
  hw/misc/sifive_u_otp: Add write function and write-once protection
  target/riscv: raise exception to HS-mode at get_physical_address
  hw/riscv: Load the kernel after the firmware
  hw/riscv: Add a riscv_is_32_bit() function
  hw/riscv: Return the end address of the loaded firmware
  hw/riscv: sifive_u: Allow specifying the CPU
  target/riscv: Fix implementation of HLVX.WU instruction
  target/riscv: Fix update of hstatus.GVA in riscv_cpu_do_interrupt
  target/riscv: Fix update of hstatus.SPVP
  hw/intc: Move sifive_plic.h to the include directory
  riscv: Convert interrupt logs to use qemu_log_mask()

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

machine: move SMP initialization from vl.c

Initialize the object's values from the class when the object is
created, no need to have vl.c do it for us.

Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

machine: move UP defaults to class_base_init

Clean up vl.c, default min/max/default_cpus to uniprocessor
directly in the QOM class initialization code.

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

machine: remove deprecated -machine enforce-config-section option

Deprecated since 3.1 and complicates the initialization sequence,
remove it.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

win32: boot broken when bind & data dir are the same

With upstream commit#ea1edcd7da1a "vl: relocate paths to data
directories", the data dir logic was unified between POSIX &
Win32. That patch moved to using 'get_relocated_path()', to
find the data dir. There is a latent bug in get_relocated_path
which can cause it to spin indefinitely, when the bind dir is
the same as the passed in dir (in this case, it was the data
dir).

Signed-off-by: Sunil Muthuswamy <sunilmut@microsoft.com>
Message-Id: <SN4PR2101MB08802BF242C429A15DDB32ACC01B0@SN4PR2101MB0880.namprd21.prod.outlook.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

WHPX: Fix WHPX build break

With upstream commit#8a19980e3fc4, logic was introduced to only
allow WHPX build on x64. But, the logic checks for the cpu family
and not the cpu. On my fedora container build, the cpu family is
x86 and the cpu is x86_64. Fixing the build break by checking for
the cpu, instead of the cpu family.

Signed-off-by: Sunil Muthuswamy <sunilmut@microsoft.com>
Message-Id: <SN4PR2101MB0880D706A85793DDFC411304C01D0@SN4PR2101MB0880.namprd21.prod.outlook.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: move install_blobs from configure to meson

Move the conditions under which edk2 blobs are decompressed
and installed to pc-bios/meson.build.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: remove unused variable from config-host.mak

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: move directory options from config-host.mak to meson

Since installation is not part of Makefiles anymore, Make need not
know the directories anymore.  Meson already knows them through
built-in options, do everything using them instead of the config_host
dictionary.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: allow configuring localedir

Meson has a localedir option, so passing the path through that option
is the cleanest way when we move directories out of config-host.mak.
In preparation for doing that without changing semantics and without
special-casing localedir code, add a configure option.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Makefile: separate meson rerun from the rest of the ninja invocation

The rules to build Makefile.mtest are suffering from the "tunnel vision"
problem that is common with recursive makefiles.  Makefile.mtest depends
on build.ninja, but Make does not know when build.ninja needs to be
rebuilt before creating Makefile.mtest.

To fix this, separate the ninja invocation into the "regenerate build
files" phase and the QEMU build phase.  Sentinel files such as
meson-private/coredata.dat or build.ninja are used to figure out the
phases that haven't run yet; however, because those files' timestamps
are not guaranteed to be touched, the usual makefile stamp-file trick
is used on top.

Reported-by: Havard Skinnemoen <hskinnemoen@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Merge remote-tracking branch 'remotes/stefanha-gitlab/tags/block-pull-request' into staging

Pull request

v3:
 * Fix linker error on FreeBSD and other !CONFIG_LINUX host OSes [Peter]

# gpg: Signature made Fri 23 Oct 2020 16:21:19 BST
# gpg:                using RSA key 8695A8BFD3F97CDAAC35775A9CA4ABB381AB73C8
# gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>" [full]
# gpg:                 aka "Stefan Hajnoczi <stefanha@gmail.com>" [full]
# Primary key fingerprint: 8695 A8BF D3F9 7CDA AC35  775A 9CA4 ABB3 81AB 73C8

* remotes/stefanha-gitlab/tags/block-pull-request: (28 commits)
  iotests: add commit top->base cases to 274
  block/io: fix bdrv_is_allocated_above
  block/io: bdrv_common_block_status_above: support bs == base
  block/io: bdrv_common_block_status_above: support include_base
  block/io: fix bdrv_co_block_status_above
  block/export: add vhost-user-blk multi-queue support
  block/export: add iothread and fixed-iothread options
  block: move block exports to libblockdev
  qemu-storage-daemon: avoid compiling blockdev_ss twice
  util/vhost-user-server: use static library in meson.build
  util/vhost-user-server: move header to include/
  block/export: convert vhost-user-blk server to block export API
  block/export: report flush errors
  util/vhost-user-server: rework vu_client_trip() coroutine lifecycle
  util/vhost-user-server: check EOF when reading payload
  util/vhost-user-server: fix memory leak in vu_message_read()
  util/vhost-user-server: drop unused DevicePanicNotifier
  block/export: consolidate request structs into VuBlockReq
  util/vhost-user-server: drop unnecessary watch deletion
  util/vhost-user-server: drop unnecessary QOM cast
  ...

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

tests/acceptance: Use .ppm extention for Portable PixMap files

The HMP 'screendump' command generates Portable PixMap files.
Make it obvious by using the .ppm file extention.

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20201021105035.2477784-3-f4bug@amsat.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/acceptance: Remove unused import

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20201021105035.2477784-2-f4bug@amsat.org>
Tested-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

test/docker/dockerfiles: Add missing packages for acceptance tests

Some of the "check-acceptance" tests are still skipped in the CI
since the docker images do not provide the necessary packages, e.g.
the netcat binary. Add them to get more test coverage.

Message-Id: <20201023073351.251332-5-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/acceptance: Enable AVOCADO_ALLOW_UNTRUSTED_CODE in the gitlab-CI

The tests are running in containers here, so it should be OK to
run with AVOCADO_ALLOW_UNTRUSTED_CODE enabled in this case.

Message-Id: <20201023073351.251332-4-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

test/acceptance: Remove the CONTINUOUS_INTEGRATION tags

We are not running the acceptance tests on Travis anymore, so these
checks can be removed now.

Message-Id: <20201023073351.251332-3-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/acceptance/ppc_prep_40p: Fix the URL to the NetBSD-4.0 archive

The current URL on cdn.netbsd.org is failing - using archive.netbsd.org
instead seems to be fine.

Message-Id: <20201023073351.251332-2-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

scripts/oss-fuzz: ignore the generic-fuzz target

generic-fuzz is not a standalone fuzzer - it requires some env variables
to be set. On oss-fuzz, we set these with some predefined
generic-fuzz-{...} targets, that are thin wrappers around generic-fuzz.
Do not make a link for the generic-fuzz from the oss-fuzz build, so
oss-fuzz does not treat it as a standalone fuzzer.

Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Message-Id: <20201023150746.107063-18-alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
[thuth: Reformatted one comment to stay within the 80 columns limit]
Signed-off-by: Thomas Huth <thuth@redhat.com>

scripts/oss-fuzz: use hardlinks instead of copying

Prior to this, fuzzers in the output oss-fuzz directory were exactly
the same executable, with a different name to do argv[0]-based
fuzz-target selection. This is a waste of space, especially since these
binaries can weigh many MB.

Instead of copying, use hard links, to cut down on wasted space. We need
to place the primary copy of the executable into DEST_DIR, since this is
a separate file-system on oss-fuzz. We should not place it directly into
$DEST_DIR, since oss-fuzz will treat it as an independent fuzzer and try
to run it for fuzzing. Instead, we create a DEST_DIR/bin directory to
store the primary copy.

Suggested-by: Darren Kenny <darren.kenny@oracle.com>
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Message-Id: <20201023150746.107063-17-alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

fuzz: register predefined generic-fuzz configs

We call get_generic_fuzz_configs, which fills an array with
predefined {name, args, objects} triples. For each of these, we add a
new FuzzTarget, that uses a small wrapper to set
QEMU_FUZZ_{ARGS,OBJECTS} to the corresponding predefined values.

Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Message-Id: <20201023150746.107063-16-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

fuzz: add generic-fuzz configs for oss-fuzz

Predefine some generic-fuzz configs. For each of these, we will create a
separate FuzzTarget that can be selected through argv0 and, therefore,
fuzzed on oss-fuzz.

Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Message-Id: <20201023150746.107063-15-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

fuzz: add an "opaque" to the FuzzTarget struct

It can be useful to register FuzzTargets that have nearly-identical
initialization handlers (e.g. for using the same fuzzing code, with
different configuration options). Add an opaque pointer to the
FuzzTarget struct, so that FuzzTargets can hold some data, useful for
storing target-specific configuration options, that can be read by the
get_init_cmdline function.

Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Message-Id: <20201023150746.107063-14-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

fuzz: Add instructions for using generic-fuzz

Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Message-Id: <20201023150746.107063-13-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

scripts/oss-fuzz: Add crash trace minimization script

Once we find a crash, we can convert it into a QTest trace. Usually this
trace will contain many operations that are unneeded to reproduce the
crash. This script tries to minimize the crashing trace, by removing
operations and trimming QTest bufwrite(write addr len data...) commands.

Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20201023150746.107063-12-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

scripts/oss-fuzz: Add script to reorder a generic-fuzzer trace

The generic-fuzzer uses hooks to fulfill DMA requests just-in-time.
This means that if we try to use QTEST_LOG=1 to build a reproducer, the
DMA writes will be logged _after_ the in/out/read/write that triggered
the DMA read. To work work around this, the generic-fuzzer annotates
these just-in time DMA fulfilments with a tag that we can use to
discern them. This script simply iterates over a raw qtest
trace (including log messages, errors, timestamps etc), filters it and
re-orders it so that DMA fulfillments are placed directly _before_ the
qtest command that will cause the DMA access.

Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20201023150746.107063-11-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

fuzz: add a crossover function to generic-fuzzer

Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Message-Id: <20201023150746.107063-10-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

fuzz: add a DISABLE_PCI op to generic-fuzzer

This new operation is used in the next commit, which concatenates two
fuzzer-generated inputs. With this operation, we can prevent the second
input from clobbering the PCI configuration performed by the first.

Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20201023150746.107063-9-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

fuzz: Add support for custom crossover functions

libfuzzer supports a "custom crossover function". Libfuzzer often tries
to blend two inputs to create a new interesting input. Sometimes, we
have a better idea about how to blend inputs together. This change
allows fuzzers to specify a custom function for blending two inputs
together.

Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20201023150746.107063-8-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

fuzz: Add fuzzer callbacks to DMA-read functions

We should be careful to not call any functions besides fuzz_dma_read_cb.
Without --enable-fuzzing, fuzz_dma_read_cb is an empty inlined function.

Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20201023150746.107063-7-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

fuzz: Declare DMA Read callback function

This patch declares the fuzz_dma_read_cb function and uses the
preprocessor and linker(weak symbols) to handle these cases:

When we build softmmu/all with --enable-fuzzing, there should be no
strong symbol defined for fuzz_dma_read_cb, and we link against a weak
stub function.

When we build softmmu/fuzz with --enable-fuzzing, we link against the
strong symbol in generic_fuzz.c

When we build softmmu/all without --enable-fuzzing, fuzz_dma_read_cb is
an empty, inlined function. As long as we don't call any other functions
when building the arguments, there should be no overhead.

Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20201023150746.107063-6-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

fuzz: Add DMA support to the generic-fuzzer

When a virtual-device tries to access some buffer in memory over DMA, we
add call-backs into the fuzzer(next commit). The fuzzer checks verifies
that the DMA request maps to a physical RAM address and fills the memory
with fuzzer-provided data. The patterns that we use to fill this memory
are specified using add_dma_pattern and clear_dma_patterns operations.

Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20201023150746.107063-5-alxndr@bu.edu>
[thuth: Reformatted one comment according to the QEMU coding style]
Signed-off-by: Thomas Huth <thuth@redhat.com>

hw/sd/sdcard: Zero out function selection fields before being populated

The function selection fields (399:376) should be zeroed out to
prevent leftover from being or'ed into the switch function status
data structure.

This fixes the boot failure as seen in the acceptance testing on
the orangepi target.

Fixes: b638627c723a ("hw/sd: Fix incorrect populated function switch status data structure")
Reported-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20201024014954.21330-1-bmeng.cn@gmail.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

hw/sd/sdcard: Make iolen unsigned

I/O request length can not be negative.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20200630133912.9428-15-f4bug@amsat.org>

hw/sd/sdcard: Constify sd_crc*()'s message argument

CRC functions don't modify the buffer argument,
make it const.

Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20200630133912.9428-14-f4bug@amsat.org>

hw/sd/sdcard: Simplify cmd_valid_while_locked()

cmd_valid_while_locked() only needs to read SDRequest->cmd,
pass it directly and make it const.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20200630133912.9428-13-f4bug@amsat.org>

hw/sd/sdcard: Update the SDState documentation

Add more descriptive comments to keep a clear separation
between static property vs runtime changeable.

Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20200630133912.9428-12-f4bug@amsat.org>

fuzz: Add PCI features to the generic fuzzer

This patch compares TYPE_PCI_DEVICE objects against the user-provided
matching pattern. If there is a match, we use some hacks and leverage
QOS to map each possible BAR for that device. Now fuzzed inputs might be
converted to pci_read/write commands which target specific. This means
that we can fuzz a particular device's PCI configuration space,

Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20201023150746.107063-4-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

fuzz: Add generic virtual-device fuzzer

This is a generic fuzzer designed to fuzz a virtual device's
MemoryRegions, as long as they exist within the Memory or Port IO (if it
exists) AddressSpaces. The fuzzer's input is interpreted into a sequence
of qtest commands (outb, readw, etc). The interpreted commands are
separated by a magic seaparator, which should be easy for the fuzzer to
guess. Without ASan, the separator can be specified as a "dictionary
value" using the -dict argument (see libFuzzer documentation).

Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Message-Id: <20201023150746.107063-3-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

memory: Add FlatView foreach function

Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Message-Id: <20201023150746.107063-2-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

libqtest: fix memory leak in the qtest_qmp_event_ref

The g_list_remove_link doesn't free the link element,
opposed to what I thought.
Switch to g_list_delete_link that does free it.

Also refactor the code a bit.
Thanks for Max Reitz for helping me with this.

Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20201019163702.471239-4-mlevitsk@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

libqtest: fix the order of buffered events

By a mistake I added the pending events in a wrong order.
Fix this by using g_list_append.

Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20201019163702.471239-3-mlevitsk@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest: Make npcm7xx_timer-test conditional on CONFIG_NPCM7XX

This test won't work if qemu was compiled without CONFIG_NPCM7XX, as
pointed out by Thomas Huth on a different patch.

Signed-off-by: Havard Skinnemoen <hskinnemoen@google.com>
Message-Id: <20201023210637.351238-2-hskinnemoen@google.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

accel: Add xen CpusAccel using dummy-cpus

Xen was broken by commit 1583a3898853 ("cpus: extract out qtest-specific
code to accel/qtest").  Xen relied on qemu_init_vcpu() calling
qemu_dummy_start_vcpu() in the default case, but that was replaced by
g_assert_not_reached().

Add a minimal "CpusAccel" for Xen using the dummy-cpus implementation
used by qtest.

Signed-off-by: Jason Andryuk <jandryuk@gmail.com>
Message-Id: <20201013140511.5681-4-jandryuk@gmail.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Claudio Fontana <cfontana@suse.de>
Acked-by: Anthony PERARD <anthony.perard@citrix.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

accel: move qtest CpusAccel functions to a common location

Move and rename accel/qtest/qtest-cpus.c files to accel/dummy-cpus.c so
it can be re-used by Xen.

Signed-off-by: Jason Andryuk <jandryuk@gmail.com>
Message-Id: <20201013140511.5681-3-jandryuk@gmail.com>
Reviewed-by: Claudio Fontana <cfontana@suse.de>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

accel: Remove _WIN32 ifdef from qtest-cpus.c

dummy-cpus.c is only compiled with CONFIG_POSIX, so the _WIN32 condition
will never evaluate true.  Remove it.

Signed-off-by: Jason Andryuk <jandryuk@gmail.com>
Message-Id: <20201013140511.5681-2-jandryuk@gmail.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Claudio Fontana <cfontana@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest/libqtest: Fix detection of architecture for binaries without path

The qtests can be run directly by specifying the QEMU binary with the
QTEST_QEMU_BINARY environment variable, for example:

 $ QTEST_QEMU_BINARY=x86_64-softmmu/qemu-system-x86_64 tests/qtest/test-hmp

However, if you specify a binary without a path, for example with
QTEST_QEMU_BINARY=qemu-system-x86_64 if the QEMU binary is in your
$PATH, then the test currently simply crashes.

Let's try a little bit smarter here by looking for the final '-'
instead of the slash.

Message-Id: <20201012114816.43546-1-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/migration: fix memleak in wait_command/wait_command_fd

Properly free each command resp to avoid memory leak.
ASAN shows memory leak stack:

Indirect leak of 2352520 byte(s) in 571 object(s) allocated from:
    #0 0x7f6ca3308d4e in __interceptor_calloc (/lib64/libasan.so.5+0x112d4e)
    #1 0x7f6ca3127a50 in g_malloc0 (/lib64/libglib-2.0.so.0+0x55a50)
    #2 0x557bf3c71d2b in qdict_new ../qobject/qdict.c:29
    #3 0x557bf3c9caba in parse_object ../qobject/json-parser.c:318
    #4 0x557bf3c9ce75 in json_parser_parse ../qobject/json-parser.c:580
    #5 0x557bf3c8c8cf in json_message_process_token ../qobject/json-streamer.c:92
    #6 0x557bf3c9ea59 in json_lexer_feed_char ../qobject/json-lexer.c:313
    #7 0x557bf3c9eeb5 in json_lexer_feed ../qobject/json-lexer.c:350
    #8 0x557bf3c4793a in qmp_fd_receive ../tests/qtest/libqtest.c:608
    #9 0x557bf3c47b58 in qtest_qmp_receive ../tests/qtest/libqtest.c:618
    #10 0x557bf3c44245 in wait_command ../tests/qtest/migration-helpers.c:59
    #11 0x557bf3c445cb in migrate_query_status ../tests/qtest/migration-helpers.c:108
    #12 0x557bf3c44642 in check_migration_status ../tests/qtest/migration-helpers.c:124
    #13 0x557bf3c447e7 in wait_for_migration_status ../tests/qtest/migration-helpers.c:148
    #14 0x557bf3c43b8f in test_migrate_auto_converge ../tests/qtest/migration-test.c:1243
    ......

Fix: 5e34005571af5

Reported-by: Euler Robot <euler.robot@huawei.com>
Signed-off-by: Chen Qun <kuhn.chenqun@huawei.com>
Message-Id: <20201023061218.2080844-2-kuhn.chenqun@huawei.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

iotests: add commit top->base cases to 274

These cases are fixed by previous patches around block_status and
is_allocated.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Message-id: 20200924194003.22080-6-vsementsov@virtuozzo.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/io: fix bdrv_is_allocated_above

bdrv_is_allocated_above wrongly handles short backing files: it reports
after-EOF space as UNALLOCATED which is wrong, as on read the data is
generated on the level of short backing file (if all overlays have
unallocated areas at that place).

Reusing bdrv_common_block_status_above fixes the issue and unifies code
path.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Message-id: 20200924194003.22080-5-vsementsov@virtuozzo.com
[Fix s/has/have/ as suggested by Eric Blake. Fix s/area/areas/.
--Stefan]
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/io: bdrv_common_block_status_above: support bs == base

We are going to reuse bdrv_common_block_status_above in
bdrv_is_allocated_above. bdrv_is_allocated_above may be called with
include_base == false and still bs == base (for ex. from img_rebase()).

So, support this corner case.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Message-id: 20200924194003.22080-4-vsementsov@virtuozzo.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/io: bdrv_common_block_status_above: support include_base

In order to reuse bdrv_common_block_status_above in
bdrv_is_allocated_above, let's support include_base parameter.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 20200924194003.22080-3-vsementsov@virtuozzo.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/io: fix bdrv_co_block_status_above

bdrv_co_block_status_above has several design problems with handling
short backing files:

1. With want_zeros=true, it may return ret with BDRV_BLOCK_ZERO but
without BDRV_BLOCK_ALLOCATED flag, when actually short backing file
which produces these after-EOF zeros is inside requested backing
sequence.

2. With want_zero=false, it may return pnum=0 prior to actual EOF,
because of EOF of short backing file.

Fix these things, making logic about short backing files clearer.

With fixed bdrv_block_status_above we also have to improve is_zero in
qcow2 code, otherwise iotest 154 will fail, because with this patch we
stop to merge zeros of different types (produced by fully unallocated
in the whole backing chain regions vs produced by short backing files).

Note also, that this patch leaves for another day the general problem
around block-status: misuse of BDRV_BLOCK_ALLOCATED as is-fs-allocated
vs go-to-backing.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 20200924194003.22080-2-vsementsov@virtuozzo.com
[Fix s/comes/come/ as suggested by Eric Blake
--Stefan]
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/export: add vhost-user-blk multi-queue support

Allow the number of queues to be configured using --export
vhost-user-blk,num-queues=N. This setting should match the QEMU --device
vhost-user-blk-pci,num-queues=N setting but QEMU vhost-user-blk.c lowers
its own value if the vhost-user-blk backend offers fewer queues than
QEMU.

The vhost-user-blk-server.c code is already capable of multi-queue. All
virtqueue processing runs in the same AioContext. No new locking is
needed.

Add the num-queues=N option and set the VIRTIO_BLK_F_MQ feature bit.
Note that the feature bit only announces the presence of the num_queues
configuration space field. It does not promise that there is more than 1
virtqueue, so we can set it unconditionally.

I tested multi-queue by running a random read fio test with numjobs=4 on
an -smp 4 guest. After the benchmark finished the guest /proc/interrupts
file showed activity on all 4 virtio-blk MSI-X. The /sys/block/vda/mq/
directory shows that Linux blk-mq has 4 queues configured.

An automated test is included in the next commit.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Message-id: 20201001144604.559733-2-stefanha@redhat.com
[Fixed accidental tab characters as suggested by Markus Armbruster
--Stefan]
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/export: add iothread and fixed-iothread options

Make it possible to specify the iothread where the export will run. By
default the block node can be moved to other AioContexts later and the
export will follow. The fixed-iothread option forces strict behavior
that prevents changing AioContext while the export is active. See the
QAPI docs for details.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200929125516.186715-5-stefanha@redhat.com
[Fix stray '#' character in block-export.json and add missing "(since:
5.2)" as suggested by Eric Blake.
--Stefan]
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block: move block exports to libblockdev

Block exports are used by softmmu, qemu-storage-daemon, and qemu-nbd.
They are not used by other programs and are not otherwise needed in
libblock.

Undo the recent move of blockdev-nbd.c from blockdev_ss into block_ss.
Since bdrv_close_all() (libblock) calls blk_exp_close_all()
(libblockdev) a stub function is required..

Make qemu-nbd.c use signal handling utility functions instead of
duplicating the code. This helps because os-posix.c is in libblockdev
and it depends on a qemu_system_killed() symbol that qemu-nbd.c lacks.
Once we use the signal handling utility functions we also end up
providing the necessary symbol.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 20200929125516.186715-4-stefanha@redhat.com
[Fixed s/ndb/nbd/ typo in commit description as suggested by Eric Blake
--Stefan]
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qemu-storage-daemon: avoid compiling blockdev_ss twice

Introduce libblkdev.fa to avoid recompiling blockdev_ss twice.

Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200929125516.186715-3-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

util/vhost-user-server: use static library in meson.build

Don't compile contrib/libvhost-user/libvhost-user.c again. Instead build
the static library once and then reuse it throughout QEMU.

Also switch from CONFIG_LINUX to CONFIG_VHOST_USER, which is what the
vhost-user tools (vhost-user-gpu, etc) do.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200924151549.913737-14-stefanha@redhat.com
[Added CONFIG_LINUX again because libvhost-user doesn't build on macOS.
--Stefan]
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

util/vhost-user-server: move header to include/

Headers used by other subsystems are located in include/. Also add the
vhost-user-server and vhost-user-blk-server headers to MAINTAINERS.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200924151549.913737-13-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/export: convert vhost-user-blk server to block export API

Use the new QAPI block exports API instead of defining our own QOM
objects.

This is a large change because the lifecycle of VuBlockDev needs to
follow BlockExportDriver. QOM properties are replaced by QAPI options
objects.

VuBlockDev is renamed VuBlkExport and contains a BlockExport field.
Several fields can be dropped since BlockExport already has equivalents.

The file names and meson build integration will be adjusted in a future
patch. libvhost-user should probably be built as a static library that
is linked into QEMU instead of as a .c file that results in duplicate
compilation.

The new command-line syntax is:

  $ qemu-storage-daemon \
      --blockdev file,node-name=drive0,filename=test.img \
      --export vhost-user-blk,node-name=drive0,id=export0,unix-socket=/tmp/vhost-user-blk.sock

Note that unix-socket is optional because we may wish to accept chardevs
too in the future.

Markus noted that supported address families are not explicit in the
QAPI schema. It is unlikely that support for more address families will
be added since file descriptor passing is required and few address
families support it. If a new address family needs to be added, then the
QAPI 'features' syntax can be used to advertize them.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Message-id: 20200924151549.913737-12-stefanha@redhat.com
[Skip test on big-endian host architectures because this device doesn't
support them yet (as already mentioned in a code comment).
--Stefan]
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/export: report flush errors

Propagate the flush return value since errors are possible.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200924151549.913737-11-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

util/vhost-user-server: rework vu_client_trip() coroutine lifecycle

The vu_client_trip() coroutine is leaked during AioContext switching. It
is also unsafe to destroy the vu_dev in panic_cb() since its callers
still access it in some cases.

Rework the lifecycle to solve these safety issues.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200924151549.913737-10-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

util/vhost-user-server: check EOF when reading payload

Unexpected EOF is an error that must be reported.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200924151549.913737-9-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

util/vhost-user-server: fix memory leak in vu_message_read()

fds[] is leaked when qio_channel_readv_full() fails.

Use vmsg->fds[] instead of keeping a local fds[] array. Then we can
reuse goto fail to clean up fds. vmsg->fd_num must be zeroed before the
loop to make this safe.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200924151549.913737-8-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

util/vhost-user-server: drop unused DevicePanicNotifier

The device panic notifier callback is not used. Drop it.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200924151549.913737-7-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/export: consolidate request structs into VuBlockReq

Only one struct is needed per request. Drop req_data and the separate
VuBlockReq instance. Instead let vu_queue_pop() allocate everything at
once.

This fixes the req_data memory leak in vu_block_virtio_process_req().

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200924151549.913737-6-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

util/vhost-user-server: drop unnecessary watch deletion

Explicitly deleting watches is not necessary since libvhost-user calls
remove_watch() during vu_deinit(). Add an assertion to check this
though.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200924151549.913737-5-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

util/vhost-user-server: drop unnecessary QOM cast

We already have access to the value with the correct type (ioc and sioc
are the same QIOChannel).

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200924151549.913737-4-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

util/vhost-user-server: s/fileds/fields/ typo fix

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200924151549.913737-3-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

MAINTAINERS: Add vhost-user block device backend server maintainer

Suggested-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Coiby Xu <coiby.xu@gmail.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 20200918080912.321299-8-coiby.xu@gmail.com
[Removed reference to vhost-user-blk-test.c, it will be sent in a
separate pull request.
--Stefan]
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/export: vhost-user block device backend server

By making use of libvhost-user, block device drive can be shared to
the connected vhost-user client. Only one client can connect to the
server one time.

Since vhost-user-server needs a block drive to be created first, delay
the creation of this object.

Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Coiby Xu <coiby.xu@gmail.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 20200918080912.321299-6-coiby.xu@gmail.com
[Shorten "vhost_user_blk_server" string to "vhost_user_blk" to avoid the
following compiler warning:
../block/export/vhost-user-blk-server.c:178:50: error: ‘%s’ directive output truncated writing 21 bytes into a region of size 20 [-Werror=format-truncation=]
and fix "Invalid size %ld ..." ssize_t format string arguments for
32-bit hosts.
--Stefan]
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block: move logical block size check function to a common utility function

Move the constants from hw/core/qdev-properties.c to
util/block-helpers.h so that knowledge of the min/max values is

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Coiby Xu <coiby.xu@gmail.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Acked-by: Eduardo Habkost <ehabkost@redhat.com>
Message-id: 20200918080912.321299-5-coiby.xu@gmail.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

util/vhost-user-server: generic vhost user server

Sharing QEMU devices via vhost-user protocol.

Only one vhost-user client can connect to the server one time.

Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Coiby Xu <coiby.xu@gmail.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 20200918080912.321299-4-coiby.xu@gmail.com
[Fixed size_t %lu -> %zu format string compiler error.
--Stefan]
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

libvhost-user: remove watch for kick_fd when de-initialize vu-dev

When the client is running in gdb and quit command is run in gdb,
QEMU will still dispatch the event which will cause segment fault in
the callback function.

Signed-off-by: Coiby Xu <coiby.xu@gmail.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 20200918080912.321299-3-coiby.xu@gmail.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

libvhost-user: Allow vu_message_read to be replaced

Allow vu_message_read to be replaced by one which will make use of the
QIOChannel functions. Thus reading vhost-user message won't stall the
guest. For slave channel, we still use the default vu_message_read.

Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Coiby Xu <coiby.xu@gmail.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20200918080912.321299-2-coiby.xu@gmail.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

block/nvme: Add driver statistics for access alignment and hw errors

Keep statistics of some hardware errors, and number of
aligned/unaligned I/O accesses.

QMP example booting a full RHEL 8.3 aarch64 guest:

{ "execute": "query-blockstats" }
{
    "return": [
        {
            "device": "",
            "node-name": "drive0",
            "stats": {
                "flush_total_time_ns": 6026948,
                "wr_highest_offset": 3383991230464,
                "wr_total_time_ns": 807450995,
                "failed_wr_operations": 0,
                "failed_rd_operations": 0,
                "wr_merged": 3,
                "wr_bytes": 50133504,
                "failed_unmap_operations": 0,
                "failed_flush_operations": 0,
                "account_invalid": false,
                "rd_total_time_ns": 1846979900,
                "flush_operations": 130,
                "wr_operations": 659,
                "rd_merged": 1192,
                "rd_bytes": 218244096,
                "account_failed": false,
                "idle_time_ns": 2678641497,
                "rd_operations": 7406,
            },
            "driver-specific": {
                "driver": "nvme",
                "completion-errors": 0,
                "unaligned-accesses": 2959,
                "aligned-accesses": 4477
            },
            "qdev": "/machine/peripheral-anon/device[0]/virtio-backend"
        }
    ]
}

Suggested-by: Stefan Hajnoczi <stefanha@gmail.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Message-id: 20201001162939.1567915-1-philmd@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

hw/misc/sifive_u_otp: Add backend drive support

Add '-drive' support to OTP device. Allow users to assign a raw file
as OTP image.

test commands for 16k otp.img filled with zero:

$ dd if=/dev/zero of=./otp.img bs=1k count=16
$ ./qemu-system-riscv64 -M sifive_u -m 256M -nographic -bios none \
-kernel ../opensbi/build/platform/sifive/fu540/firmware/fw_payload.elf \
-d guest_errors -drive if=none,format=raw,file=otp.img

Signed-off-by: Green Wan <green.wan@sifive.com>
Reviewed-by: Bin Meng <bin.meng@windriver.com>
Tested-by: Bin Meng <bin.meng@windriver.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20201020033732.12921-3-green.wan@sifive.com
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/misc/sifive_u_otp: Add write function and write-once protection

 - Add write operation to update fuse data bit when PWE bit is on.
 - Add array, fuse_wo, to store the 'written' status for all bits
   of OTP to block the write operation.

Signed-off-by: Green Wan <green.wan@sifive.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Bin Meng <bin.meng@windriver.com>
Tested-by: Bin Meng <bin.meng@windriver.com>
Message-id: 20201020033732.12921-2-green.wan@sifive.com
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: raise exception to HS-mode at get_physical_address

VS-stage translation at get_physical_address needs to translate pte
address by G-stage translation. But the G-stage translation error
can not be distinguished from VS-stage translation error in
riscv_cpu_tlb_fill. On migration, destination needs to rebuild pte,
and this G-stage translation error must be handled by HS-mode. So
introduce TRANSLATE_STAGE2_FAIL so that riscv_cpu_tlb_fill could
distinguish and raise it to HS-mode.

Signed-off-by: Yifei Jiang <jiangyifei@huawei.com>
Signed-off-by: Yipeng Yin <yinyipeng1@huawei.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20201014101728.848-1-jiangyifei@huawei.com
[ Change by AF:
 - Clarify the fault_pte_addr shift
]
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv: Load the kernel after the firmware

Instead of loading the kernel at a hardcoded start address, let's load
the kernel at the next aligned address after the end of the firmware.

This should have no impact for current users of OpenSBI, but will
allow loading a noMMU kernel at the start of memory.

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Palmer Dabbelt <palmerdabbelt@google.com>
Reviewed-by: Bin Meng <bin.meng@windriver.com>
Tested-by: Bin Meng <bin.meng@windriver.com>
Message-id: 46c00c4f15b42feb792090e3d74359e180a6d954.1602634524.git.alistair.francis@wdc.com

hw/riscv: Add a riscv_is_32_bit() function

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Palmer Dabbelt <palmerdabbelt@google.com>
Reviewed-by: Bin Meng <bin.meng@windriver.com>
Tested-by: Bin Meng <bin.meng@windriver.com>
Message-id: 4c6a85dfb6dd470aa79356ebc1b02f479c2758e0.1602634524.git.alistair.francis@wdc.com

hw/riscv: Return the end address of the loaded firmware

Instead of returning the unused entry address from riscv_load_firmware()
instead return the end address. Also return the end address from
riscv_find_and_load_firmware().

This tells the caller if a firmware was loaded and how big it is. This
can be used to determine the load address of the next image (usually the
kernel).

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Palmer Dabbelt <palmerdabbelt@google.com>
Reviewed-by: Bin Meng <bin.meng@windriver.com>
Tested-by: Bin Meng <bin.meng@windriver.com>
Message-id: 558cf67162342d65a23262248b040563716628b2.1602634524.git.alistair.francis@wdc.com

hw/riscv: sifive_u: Allow specifying the CPU

Allow the user to specify the main application CPU for the sifive_u
machine.

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Palmer Dabbelt <palmerdabbelt@google.com>
Tested-by: Bin Meng <bin.meng@windriver.com>
Message-id: b8412086c8aea0eff30fb7a17f0acf2943381b6a.1602634524.git.alistair.francis@wdc.com

target/riscv: Fix implementation of HLVX.WU instruction

The HLVX.WU instruction is supposed to read a machine word,
but prior to this change it read a byte instead.

Fixes: 8c5362acb57 ("target/riscv: Allow generating hlv/hlvx/hsv instructions")
Signed-off-by: Georg Kotheimer <georg.kotheimer@kernkonzept.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20201013172223.443645-1-georg.kotheimer@kernkonzept.com
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Fix update of hstatus.GVA in riscv_cpu_do_interrupt

The hstatus.GVA bit was not set if the faulting guest virtual address
was zero.

Signed-off-by: Georg Kotheimer <georg.kotheimer@kernkonzept.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20201013173054.451135-1-georg.kotheimer@kernkonzept.com
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Fix update of hstatus.SPVP

When trapping from virt into HS mode, hstatus.SPVP was set to
the value of sstatus.SPP, as according to the specification both
flags should be set to the same value.
However, the assignment of SPVP takes place before SPP itself is
updated, which results in SPVP having an outdated value.

Signed-off-by: Georg Kotheimer <georg.kotheimer@kernkonzept.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20201013151054.396481-1-georg.kotheimer@kernkonzept.com
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/intc: Move sifive_plic.h to the include directory

Since sifive_plic.h is used by hw/intc/sifive_plic.c,
it has to be in the public include directory. Move it.

Fixes: 84fcf3c15111 ("hw/riscv: Move sifive_plic model to hw/intc")
Signed-off-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 1602578033-68384-1-git-send-email-bmeng.cn@gmail.com
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

riscv: Convert interrupt logs to use qemu_log_mask()

Currently we log interrupts and exceptions using the trace backend in
riscv_cpu_do_interrupt(). We also log exceptions using the interrupt log
mask (-d int) in riscv_raise_exception().

This patch converts riscv_cpu_do_interrupt() to log both interrupts and
exceptions with the interrupt log mask, so that both are printed when a
user runs QEMU with -d int.

Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 29a8c766c7c4748d0f2711c3a0abb81208138c5e.1601652179.git.alistair.francis@wdc.com