tests/docker: flatten debian-powerpc-test-cross

Flatten into a single dockerfile. We really don't need the rest of the
stuff from the QEMU base image just to compile test images.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20220914155950.804707-15-alex.bennee@linaro.org>

tests/docker: update and flatten debian-sparc64-cross

Update to the latest stable Debian. While we are at it flatten into a
single dockerfile. We really don't need the rest of the stuff from
the QEMU base image just to compile test images.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20220914155950.804707-14-alex.bennee@linaro.org>

tests/docker: update and flatten debian-sh4-cross

Update to the latest stable Debian. While we are at it flatten into a
single dockerfile. We really don't need the rest of the stuff from
the QEMU base image just to compile test images.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20220914155950.804707-13-alex.bennee@linaro.org>

tests/docker: update and flatten debian-mips64-cross

Update to the latest stable Debian. While we are at it flatten into a
single dockerfile. We really don't need the rest of the stuff from
the QEMU base image just to compile test images.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20220914155950.804707-12-alex.bennee@linaro.org>

tests/docker: update and flatten debian-m68k-cross

Update to the latest stable Debian. While we are at it flatten into a
single dockerfile. We really don't need the rest of the stuff from
the QEMU base image just to compile test images.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20220914155950.804707-11-alex.bennee@linaro.org>

tests/docker: update and flatten debian-hppa-cross

Update to the latest stable Debian. While we are at it flatten into a
single dockerfile. We really don't need the rest of the stuff from
the QEMU base image just to compile test images.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20220914155950.804707-10-alex.bennee@linaro.org>

tests/docker: update and flatten debian-alpha-cross

Update to the latest stable Debian. While we are at it flatten into a
single dockerfile. We really don't need the rest of the stuff from
the QEMU base image just to compile test images.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20220914155950.804707-9-alex.bennee@linaro.org>

tests/avocado: reduce the default timeout to 120s

We should be aiming to keep our tests under 2 minutes so lets reduce
the default timeout to that. Tests that we know take longer should
explicitly set a longer timeout.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220914155950.804707-8-alex.bennee@linaro.org>

tests/avocado: split the AST2x00Machine classes

The SDK tests take a lot longer to run and hence need a longer
timeout. As they run well over the 60 second maximum for CI lets also
disable them for CI as well.

I suspect they also suffer from the inability to detect the login
prompt due to no newlines being processed.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220914155950.804707-7-alex.bennee@linaro.org>

tests/avocado: add explicit timeout for ppc64le TCG tests

We don't want to rely on the soon to be reduced default time. These
tests are still slow for something we want to run in CI though.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220914155950.804707-6-alex.bennee@linaro.org>

tests/avocado: add explicit timeout for s390 TCG tests

We don't want to rely on the soon to be reduced default time. These
tests are still slow for something we want to run in CI though.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20220914155950.804707-5-alex.bennee@linaro.org>

tests/avocado: add explicit timeout for Aarch64 TCG tests

We don't want to rely on the soon to be reduced default time. These
tests are still slow for something we want to run in CI though.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220914155950.804707-4-alex.bennee@linaro.org>

tests/avocado/boot_linux_console: Fix the test_aarch64_xlnx_versal_virt test

The assets that this test tries to download have been removed from the
server. Update to a newer version to get it working again.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20220829080940.110831-1-thuth@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20220914155950.804707-3-alex.bennee@linaro.org>

gitlab: reduce targets in cross_user_build_job

We already limit the scope of the cross system build to reduce the
cross build times. With the recent addition of more targets we are
also running into timeout issues for some of the cross user builds.

I've selected a few of those linux-user targets which are less likely
to be in common use as distros don't have pre-built rootfs for them.
I've also added the same CROSS_SKIP_TARGETS variable as is
occasionally used to further limit cross system builds.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220914155950.804707-2-alex.bennee@linaro.org>

Merge tag 'pull-hmp-20220915a' of https://gitlab.com/dagrh/qemu into staging

HMP pull 2022-09-15

A set of 3 small additions/fixes.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEERfXHG0oMt/uXep+pBRYzHrxb/ecFAmMjNG4ACgkQBRYzHrxb
# /edRhhAAibHXFdK1EHtcABInxMVg50wZ7oD2QiYqKb1m0t9rQDSmZ/cX22qdYF7D
# hBvjsMiniy6isGYyiPhpi+7ZMnid12E6EpZ49tFKbDz+DPfpT3mk/MOKvNQu150I
# K8NuqD40HImpUzo7/OVVIzo26TsKMFUo7WzuzqP4PEguSCqVAlzuoVxId8oZ3DbT
# c/VQTZjwLbi+i7DuuFc1pqfzC3euNEApM7DRAYNios+oGxd2kd+DC7JLqMKK4Uk1
# 5t5YmCz3Q+aFZ1kOvpxUW+3xT0LI4wzR0XA0ImA3jADaqt5G3dNOzosY2E7eaQen
# AXpqu6dH5wbSl2y5LWSscgN8ObAs5N6n0+ncXtIYwMENBBkZpWtaDbTYp6ezcrww
# 2st2qQ0MlUj8oiH3jMr5TtkZxcx9wXdEfrCDZ0MQt3275Bp6JuSW6DjqXjjpKimY
# 2HsAPDKYFNVJl/9SX5PxW6cGPZyb/YMa+14YdIdMbWc8+q+4yMSQnTJbGvxLFakt
# S/agxGu3hpEGrYUjCdl6JzIyOxdUVFp+Lp4NwhU+DI/1UaL1/QbiK+NxDXvkk+Wp
# wcI7VbIXb6YMWf9Gmv4NQ1b4uTf8qoA7J0SNONrw9ywU4lhsm9UKoIoakKaSK9F8
# kI+vBt/mmZTCqNw8h4w/KNvrRFF5nP/vw117xdOGGbtCS7G8438=
# =Isyw
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 15 Sep 2022 10:19:26 EDT
# gpg:                using RSA key 45F5C71B4A0CB7FB977A9FA90516331EBC5BFDE7
# gpg: Good signature from "Dr. David Alan Gilbert (RH2) <dgilbert@redhat.com>" [full]
# Primary key fingerprint: 45F5 C71B 4A0C B7FB 977A  9FA9 0516 331E BC5B FDE7

* tag 'pull-hmp-20220915a' of https://gitlab.com/dagrh/qemu:
  hmp: Fix ordering of text
  monitor/hmp: print trace as option in help for log command
  monitor: Support specified vCPU registers

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'pull-semi-20220914' of https://gitlab.com/rth7680/qemu into staging

Convert m68k to semihosting/syscalls.h.
Convert nios2 to semihosting/syscalls.h.
Allow optional use of semihosting from userspace.

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmMh1W8dHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV8ptggAimuNN6IiD19Huu5F
# PMjzDqFPvWFOf82O16WTBM1xN0lwVH8+02PYRL3AhOIw9ZTgxezOo9/KXZpr8a8Z
# gocr4Ge/J7zHzHahYuqcyOqqkur2dM4lFiK9rfDD6vdNBMbi0kQZVuaNlQK6rV6Z
# 2LHEwKKh64MXJVfwGzK7OLMv4pu0wpWcuCTH2/6U4E1325SOKmEos1VzIePxY1bw
# +AMNnairGEdBX1b3JlzZfrLSaOapJcgl0HZdrg6Mflm6ttTuuykGGtjkWBfcu3Nw
# utNI1zmUYfD/iJbnbsCNpZSLv6LVOQ2l5S6dOWV+JJ1HukVTZu3DoyfTr8t95kwK
# UuUoqA==
# =W7Yh
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 14 Sep 2022 09:21:51 EDT
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [full]
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A  05C0 64DF 38E8 AF7E 215F

* tag 'pull-semi-20220914' of https://gitlab.com/rth7680/qemu:
  target/riscv: Honour -semihosting-config userspace=on and enable=on
  target/xtensa: Honour -semihosting-config userspace=on
  target/nios2: Honour -semihosting-config userspace=on
  target/mips: Honour -semihosting-config userspace=on
  target/m68k: Honour -semihosting-config userspace=on
  target/arm: Honour -semihosting-config userspace=on
  semihosting: Allow optional use of semihosting from userspace
  target/m68k: Convert semihosting errno to gdb remote errno
  target/m68k: Use semihosting/syscalls.h
  target/nios2: Convert semihosting errno to gdb remote errno
  target/nios2: Use semihosting/syscalls.h

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'pull-arm-20220914' of https://gitlab.com/rth7680/qemu into staging

Add cortex-a35.
Fix bcm2835 framebuffer for rpi firmware.
Add FEAT_ETS.
Add FEAT_PMUv3p5.
Cleanups to armv7m_load_kernel.

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmMhwAsdHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV/7Xgf9Ezg+etwsEzY0EWtH
# yoQ7ymJHM1VKqtLkbk9U+Ju18YHEi3fwbYbfLgzkAMFHKDUteKJivKm2w41tsw5g
# lE+5PojQT3k8PffaqeFzKG/JsDPoqKtTa23WjkafbIQAbjRT8JtmF/JmSLaaxyqW
# DozPXJ9jXMS+Q2BxHnoCBuy/Kb3zsqekCr8DOgUIXY5gcAB6q1I8SMX2BrhsIyvV
# vZxNm+hFDCOJtXvkAekMrwGkiIBYoBNpswnv40ldeSvRtD8tnty73JUt+AnWoCwi
# zH4k+CvWICIuZT7oxVecRhlOojtaP5cUTXs0+zhk6GxEx/X6wmXd42heO9ZJL0y1
# 6FH0xw==
# =ohPU
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 14 Sep 2022 07:50:35 EDT
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [full]
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A  05C0 64DF 38E8 AF7E 215F

* tag 'pull-arm-20220914' of https://gitlab.com/rth7680/qemu:
  target/arm: Make boards pass base address to armv7m_load_kernel()
  target/arm: Remove useless TARGET_BIG_ENDIAN check in armv7m_load_kernel()
  target/arm: Report FEAT_PMUv3p5 for TCG '-cpu max'
  target/arm: Support 64-bit event counters for FEAT_PMUv3p5
  target/arm: Implement FEAT_PMUv3p5 cycle counter disable bits
  target/arm: Rename pmu_8_n feature test functions
  target/arm: Detect overflow when calculating next PMU interrupt
  target/arm: Honour MDCR_EL2.HPMD in Secure EL2
  target/arm: Ignore PMCR.D when PMCR.LC is set
  target/arm: Don't mishandle count when enabling or disabling PMU counters
  target/arm: Correct value returned by pmu_counter_mask()
  target/arm: Don't corrupt high half of PMOVSR when cycle counter overflows
  target/arm: Add missing space in comment
  target/arm: Advertise FEAT_ETS for '-cpu max'
  target/arm: Implement ID_DFR1
  target/arm: Implement ID_MMFR5
  target/arm: Sort KVM reads of AArch32 ID registers into encoding order
  target/arm: Make cpregs 0, c0, c{3-15}, {0-7} correctly RAZ in v8
  hw/arm/bcm2835_property: Add support for RPI_FIRMWARE_FRAMEBUFFER_GET_NUM_DISPLAYS
  target/arm: Add cortex-a35

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'tpm-pull-2022-09-13-1' of https://github.com/stefanberger/qemu-tpm into staging

Merge tpm 2022/09/13 v1

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCAAdFiEEuBi5yt+QicLVzsZrda1lgCoLQhEFAmMgtKIACgkQda1lgCoL
# QhG/Zgf9Gs35w+hPwGQdsrwcDmCHiH6s4Eb7i4SgzPP4/EVR9kwYriKja4HoNvK2
# GHQSXgYX5hazwgkRlNKNJSf2zckbZlr3OhPReZMab6YTVSi79xZRl4rWqKbMxk4K
# 82ueaUkLKm/RrCw69sM6ToSUQjbitseMVKorZ9NXVt9SVj+hwQv28o5U/+h8Q76T
# P3t1VraFV2vaiLhAyp4BY52djZ0AMrUox/27EdAYIPPi7om+fGeWcTQP4GsyWUv1
# h8i+ZSU1QMJ5hF1szzP7bENwSzG7mSIiqMbrqtpysu/ET6r9WblLTSkRtojvms1S
# qb7NzQ3S4NwdCWGz0owEbF5kLmMniw==
# =XMPF
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 13 Sep 2022 12:49:38 EDT
# gpg:                using RSA key B818B9CADF9089C2D5CEC66B75AD65802A0B4211
# gpg: Good signature from "Stefan Berger <stefanb@linux.vnet.ibm.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: B818 B9CA DF90 89C2 D5CE  C66B 75AD 6580 2A0B 4211

* tag 'tpm-pull-2022-09-13-1' of https://github.com/stefanberger/qemu-tpm:
  tpm_emulator: Have swtpm relock storage upon migration fall-back
  tpm_emulator: Use latest tpm_ioctl.h from swtpm project
  tpm_crb: Avoid backend startup just before shutdown under Xen
  tpm_emulator: Avoid double initialization during migration

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

hmp: Fix ordering of text

Fix the ordering of the help text so it's always after the commands
being defined.  A few had got out of order.  Keep 'info' at the end.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>

monitor/hmp: print trace as option in help for log command

The below is printed when printing help information in qemu-system-x86_64
command line, and when CONFIG_TRACE_LOG is enabled:

----------------------------
$ qemu-system-x86_64 -d help
... ...
trace:PATTERN   enable trace events

Use "-d trace:help" to get a list of trace events.
----------------------------

However, the options of "trace:PATTERN" are only printed by
"qemu-system-x86_64 -d help", but missing in hmp "help log" command.

Fixes: c84ea00dc2 ("log: add "-d trace:PATTERN"")
Cc: Joe Jin <joe.jin@oracle.com>
Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com>
Message-Id: <20220831213943.8155-1-dongli.zhang@oracle.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>

monitor: Support specified vCPU registers

Originally we have to get all the vCPU registers and parse the
specified one. To improve the performance of this usage, allow user
specified vCPU id to query registers.

Run a VM with 16 vCPU, use bcc tool to track the latency of
'hmp_info_registers':
'info registers -a' uses about 3ms;
'info registers 12' uses about 150us.

Cc: Darren Kenny <darren.kenny@oracle.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20220802073720.1236988-2-pizhenwei@bytedance.com>
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>

target/arm: Make boards pass base address to armv7m_load_kernel()

Currently armv7m_load_kernel() takes the size of the block of memory
where it should load the initial guest image, but assumes that it
should always load it at address 0.  This happens to be true of all
our M-profile boards at the moment, but it isn't guaranteed to always
be so: M-profile CPUs can be configured (via init-svtor and
init-nsvtor, which match equivalent hardware configuration signals)
to have the initial vector table at any address, not just zero.  (For
instance the Teeny board has the boot ROM at address 0x0200_0000.)

Add a base address argument to armv7m_load_kernel(), so that
callers now pass in both base address and size. All the current
callers pass 0, so this is not a behaviour change.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220823160417.3858216-3-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Remove useless TARGET_BIG_ENDIAN check in armv7m_load_kernel()

Arm system emulation targets always have TARGET_BIG_ENDIAN clear, so
there is no need to have handling in armv7m_load_kernel() for the
case when it is defined.  Remove the unnecessary code.

Side notes:
 * our M-profile implementation is always little-endian (that is, it
   makes the IMPDEF choice that the read-only AIRCR.ENDIANNESS is 0)
 * if we did want to handle big-endian ELF files here we should do it
   the way that hw/arm/boot.c:arm_load_elf() does, by looking at the
   ELF header to see what endianness the file itself is

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20220823160417.3858216-2-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Report FEAT_PMUv3p5 for TCG '-cpu max'

Update the ID registers for TCG's '-cpu max' to report a FEAT_PMUv3p5
compliant PMU.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822132358.3524971-11-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Support 64-bit event counters for FEAT_PMUv3p5

With FEAT_PMUv3p5, the event counters are now 64 bit, rather than 32
bit.  (Previously, only the cycle counter could be 64 bit, and other
event counters were always 32 bits).  For any given event counter,
whether the overflow event is noted for overflow from bit 31 or from
bit 63 is controlled by a combination of PMCR.LP, MDCR_EL2.HLP and
MDCR_EL2.HPMN.

Implement the 64-bit event counter handling.  We choose to make our
counters always 64 bits, and mask out the top 32 bits on read or
write of PMXEVCNTR for CPUs which don't have FEAT_PMUv3p5.

(Note that the changes to pmenvcntr_op_start() and
pmenvcntr_op_finish() bring their logic closer into line with that of
pmccntr_op_start() and pmccntr_op_finish(), which already had to cope
with the overflow being either at 32 or 64 bits.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822132358.3524971-10-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Implement FEAT_PMUv3p5 cycle counter disable bits

FEAT_PMUv3p5 introduces new bits which disable the cycle
counter from counting:
 * MDCR_EL2.HCCD disables the counter when in EL2
 * MDCR_EL3.SCCD disables the counter when Secure

Add the code to support these bits.

(Note that there is a third documented counter-disable
bit, MDCR_EL3.MCCD, which disables the counter when in
EL3. This is not present until FEAT_PMUv3p7, so is
out of scope for now.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822132358.3524971-9-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Rename pmu_8_n feature test functions

Our feature test functions that check the PMU version are named
isar_feature_{aa32,aa64,any}_pmu_8_{1,4}.  This doesn't match the
current Arm ARM official feature names, which are FEAT_PMUv3p1 and
FEAT_PMUv3p4.  Rename these functions to _pmuv3p1 and _pmuv3p4.

This commit was created with:
  sed -i -e 's/pmu_8_/pmuv3p/g' target/arm/*.[ch]

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822132358.3524971-8-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Detect overflow when calculating next PMU interrupt

In pmccntr_op_finish() and pmevcntr_op_finish() we calculate the next
point at which we will get an overflow and need to fire the PMU
interrupt or set the overflow flag.  We do this by calculating the
number of nanoseconds to the overflow event and then adding it to
qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL).  However, we don't check
whether that signed addition overflows, which can happen if the next
PMU interrupt would happen massively far in the future (250 years or
more).

Since QEMU assumes that "when the QEMU_CLOCK_VIRTUAL rolls over" is
"never", the sensible behaviour in this situation is simply to not
try to set the timer if it would be beyond that point.  Detect the
overflow, and skip setting the timer in that case.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822132358.3524971-7-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Honour MDCR_EL2.HPMD in Secure EL2

The logic in pmu_counter_enabled() for handling the 'prohibit event
counting' bits MDCR_EL2.HPMD and MDCR_EL3.SPME is written in a way
that assumes that EL2 is never Secure.  This used to be true, but the
architecture now permits Secure EL2, and QEMU can emulate this.

Refactor the prohibit logic so that we effectively OR together
the various prohibit bits when they apply, rather than trying to
construct an if-else ladder where any particular state of the CPU
ends up in exactly one branch of the ladder.

This fixes the Secure EL2 case and also is a better structure for
adding the PMUv8.5 bits MDCR_EL2.HCCD and MDCR_EL3.SCCD.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822132358.3524971-6-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Ignore PMCR.D when PMCR.LC is set

The architecture requires that if PMCR.LC is set (for a 64-bit cycle
counter) then PMCR.D (which enables the clock divider so the counter
ticks every 64 cycles rather than every cycle) should be ignored.  We
were always honouring PMCR.D; fix the bug so we correctly ignore it
in this situation.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822132358.3524971-5-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Don't mishandle count when enabling or disabling PMU counters

The PMU cycle and event counter infrastructure design requires that
operations on the PMU register fields are wrapped in pmu_op_start()
and pmu_op_finish() calls (or their more specific pmmcntr and
pmevcntr equivalents).  This includes any changes to registers which
affect whether the counter should be enabled or disabled, but we
forgot to do this.

The effect of this bug is that in sequences like:
 * disable the cycle counter (PMCCNTR) using the PMCNTEN register
 * write a value such as 0xfffff000 to the PMCCNTR
 * restart the counter by writing to PMCNTEN
the value written to the cycle counter is corrupted, and it starts
counting from the wrong place. (Essentially, we fail to record that
the QEMU_CLOCK_VIRTUAL timestamp when the counter should be considered
to have started counting is the point when PMCNTEN is written to enable
the counter.)

Add the necessary bracketing calls, so that updates to the various
registers which affect whether the PMU is counting are handled
correctly.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822132358.3524971-4-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Correct value returned by pmu_counter_mask()

pmu_counter_mask() accidentally returns a value with bits [63:32]
set, because the expression it returns is evaluated as a signed value
that gets sign-extended to 64 bits.  Force the whole expression to be
evaluated with 64-bit arithmetic with ULL suffixes.

The main effect of this bug was that a guest could write to the bits
in the high half of registers like PMCNTENSET_EL0 that are supposed
to be RES0.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822132358.3524971-3-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Don't corrupt high half of PMOVSR when cycle counter overflows

When the cycle counter overflows, we are intended to set bit 31 in PMOVSR
to indicate this. However a missing ULL suffix means that we end up
setting all of bits 63-31. Fix the bug.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822132358.3524971-2-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Add missing space in comment

Fix a missing space before a comment terminator.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220819110052.2942289-7-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Advertise FEAT_ETS for '-cpu max'

The architectural feature FEAT_ETS (Enhanced Translation
Synchronization) is a set of tightened guarantees about memory
ordering involving translation table walks:

 * if memory access RW1 is ordered-before memory access RW2 then it
   is also ordered-before any translation table walk generated by RW2
   that generates a translation fault, address size fault or access
   fault

 * TLB maintenance on non-exec-permission translations is guaranteed
   complete after a DSB (ie it does not need the context
   synchronization event that you have to have if you don’t have
   FEAT_ETS)

For QEMU’s implementation we don’t reorder translation table walk
accesses, and we guarantee to finish the TLB maintenance as soon as
the TLB op is done (the tlb_flush functions will complete at the end
of the TLB, and TLB ops always end the TB because they’re sysreg
writes).

So we’re already compliant and all we need to do is say so in the ID
registers for the 'max' CPU.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220819110052.2942289-6-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Implement ID_DFR1

In Armv8.6, a new AArch32 ID register ID_DFR1 is defined; implement
it. We don't have any CPUs with features that they need to advertise
here yet, but plumbing in the ID register gives it the right name
when debugging and will help in future when we do add a CPU that
has non-zero ID_DFR1 fields.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220819110052.2942289-5-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Implement ID_MMFR5

In Armv8.6 a new AArch32 ID register ID_MMFR5 is defined.
Implement this; we want to be able to use it to report to
the guest that we implement FEAT_ETS.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220819110052.2942289-4-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Sort KVM reads of AArch32 ID registers into encoding order

The code that reads the AArch32 ID registers from KVM in
kvm_arm_get_host_cpu_features() does so almost but not quite in
encoding order.  Move the read of ID_PFR2 down so it's really in
encoding order.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220819110052.2942289-3-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Make cpregs 0, c0, c{3-15}, {0-7} correctly RAZ in v8

In the AArch32 ID register scheme, coprocessor registers with
encoding cp15, 0, c0, c{0-7}, {0-7} are all in the space covered by
what in v6 and v7 was called the "CPUID scheme", and are supposed to
RAZ if they're not allocated to a specific ID register.  For our
pre-v8 CPUs we get this right, because the regdefs in
id_pre_v8_midr_cp_reginfo[] cover these RAZ requirements.  However
for v8 we failed to put in the necessary patterns to cover this, so
we end up UNDEFing on everything we didn't have an ID register for.
This is a problem because in Armv8 some encodings in 0, c0, c3, {0-7}
are now being used for new ID registers, and guests might thus start
trying to read them.  (We already have one of these: ID_PFR2.)

For v8 CPUs, we already have regdefs for 0, c0, c{0-2}, {0-7} (that
is, the space is completely allocated with no reserved spaces).  Add
entries to v8_idregs[] covering 0, c0, c3, {0-7}:
 * c3, {0-2} is the reserved AArch32 space corresponding to the
   AArch64 MVFR[012]_EL1
 * c3, {3,5,6,7} are reserved RAZ for both AArch32 and AArch64
   (in fact some of these are given defined meanings in Armv8.6,
   but we don't implement them yet)
 * c3, 4 is ID_PFR2 (already defined)

We then programmatically add RAZ patterns for AArch32 for
0, c0, c{4..15}, {0-7}:
 * c4-c7 are unused, and not shared with AArch64 (these
   are the encodings corresponding to where the AArch64
   specific ID registers live in the system register space)
 * c8-c15 weren't required to RAZ in v6/v7, but v8 extends
   the AArch32 reserved-should-RAZ space to cover these;
   the equivalent area of the AArch64 sysreg space is not
   defined as must-RAZ

Note that the architecture allows some registers in this space
to return an UNKNOWN value; we always return 0.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220819110052.2942289-2-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

hw/arm/bcm2835_property: Add support for RPI_FIRMWARE_FRAMEBUFFER_GET_NUM_DISPLAYS

In more recent Raspbian OS Linux kernels, the fb driver gives up
immediately if RPI_FIRMWARE_FRAMEBUFFER_GET_NUM_DISPLAYS fails or no
displays are reported.

This change simply always reports one display. It makes bcm2835_fb work
again with these more recent kernels.

Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Enrik Berkhan <Enrik.Berkhan@inka.de>
Message-Id: <20220812143519.59134-1-Enrik.Berkhan@inka.de>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Add cortex-a35

Add cortex A35 core and enable it for virt board.

Signed-off-by: Hao Wu <wuhaotsh@google.com>
Reviewed-by: Joe Komlodi <komlodi@google.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20220819002015.1663247-1-wuhaotsh@google.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/riscv: Honour -semihosting-config userspace=on and enable=on

The riscv target incorrectly enabled semihosting always, whether the
user asked for it or not.  Call semihosting_enabled() passing the
correct value to the is_userspace argument, which fixes this and also
handles the userspace=on argument.  Because we do this at translate
time, we no longer need to check the privilege level in
riscv_cpu_do_interrupt().

Note that this is a behaviour change: we used to default to
semihosting being enabled, and now the user must pass
"-semihosting-config enable=on" if they want it.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20220822141230.3658237-8-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/xtensa: Honour -semihosting-config userspace=on

Honour the commandline -semihosting-config userspace=on option,
instead of always permitting userspace semihosting calls in system
emulation mode, by passing the correct value to the is_userspace
argument of semihosting_enabled().

Note that this is a behaviour change: if the user wants to
do semihosting calls from userspace they must now specifically
enable them on the command line.

xtensa semihosting is not implemented for linux-user builds.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Acked-by: Max Filippov <jcmvbkbc@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822141230.3658237-7-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/nios2: Honour -semihosting-config userspace=on

Honour the commandline -semihosting-config userspace=on option,
instead of always permitting userspace semihosting calls in system
emulation mode, by passing the correct value to the is_userspace
argument of semihosting_enabled().

Note that this is a behaviour change: if the user wants to
do semihosting calls from userspace they must now specifically
enable them on the command line.

nios2 semihosting is not implemented for linux-user builds.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822141230.3658237-6-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/mips: Honour -semihosting-config userspace=on

Honour the commandline -semihosting-config userspace=on option,
instead of always permitting userspace semihosting calls in system
emulation mode, by passing the correct value to the is_userspace
argument of semihosting_enabled().

Note that this is a behaviour change: if the user wants to
do semihosting calls from userspace they must now specifically
enable them on the command line.

MIPS semihosting is not implemented for linux-user builds.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822141230.3658237-5-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/m68k: Honour -semihosting-config userspace=on

Honour the commandline -semihosting-config userspace=on option,
instead of never permitting userspace semihosting calls in system
emulation mode, by passing the correct value to the is_userspace
argument of semihosting_enabled(), instead of manually checking and
always forbidding semihosting if the guest is in userspace.

(Note that target/m68k doesn't support semihosting at all
in the linux-user build.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822141230.3658237-4-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Honour -semihosting-config userspace=on

Honour the commandline -semihosting-config userspace=on option,
instead of never permitting userspace semihosting calls in system
emulation mode, by passing the correct value to the is_userspace
argument of semihosting_enabled(), instead of manually checking and
always forbidding semihosting if the guest is in userspace and this
isn't the linux-user build.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822141230.3658237-3-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

semihosting: Allow optional use of semihosting from userspace

Currently our semihosting implementations generally prohibit use of
semihosting calls in system emulation from the guest userspace.  This
is a very long standing behaviour justified originally "to provide
some semblance of security" (since code with access to the
semihosting ABI can do things like read and write arbitrary files on
the host system).  However, it is sometimes useful to be able to run
trusted guest code which performs semihosting calls from guest
userspace, notably for test code.  Add a command line suboption to
the existing semihosting-config option group so that you can
explicitly opt in to semihosting from guest userspace with
 -semihosting-config userspace=on

(There is no equivalent option for the user-mode emulator, because
there by definition all code runs in userspace and has access to
semihosting already.)

This commit adds the infrastructure for the command line option and
adds a bool 'is_user' parameter to the function
semihosting_userspace_enabled() that target code can use to check
whether it should be permitting the semihosting call for userspace.
It mechanically makes all the callsites pass 'false', so they
continue checking "is semihosting enabled in general".  Subsequent
commits will make each target that implements semihosting honour the
userspace=on option by passing the correct value and removing
whatever "don't do this for userspace" checking they were doing by
hand.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Acked-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822141230.3658237-2-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/m68k: Convert semihosting errno to gdb remote errno

The semihosting abi used by m68k uses the gdb remote
protocol filesys errnos.

Acked-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/m68k: Use semihosting/syscalls.h

This separates guest file descriptors from host file descriptors,
and utilizes shared infrastructure for integration with gdbstub.

Acked-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/nios2: Convert semihosting errno to gdb remote errno

The semihosting abi used by nios2 uses the gdb remote
protocol filesys errnos.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/nios2: Use semihosting/syscalls.h

This separates guest file descriptors from host file descriptors,
and utilizes shared infrastructure for integration with gdbstub.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tpm_emulator: Have swtpm relock storage upon migration fall-back

Swtpm may release the lock once the last one of its state blobs has been
migrated out. In case of VM migration failure QEMU now needs to notify
swtpm that it should again take the lock, which it can otherwise only do
once it has received the first TPM command from the VM.

Only try to send the lock command if swtpm supports it. It will not have
released the lock (and support shared storage setups) if it doesn't
support the locking command since the functionality of releasing the lock
upon state blob reception and the lock command were added to swtpm
'together'.

If QEMU sends the lock command and the storage has already been locked
no error is reported.

If swtpm does not receive the lock command (from older version of QEMU),
it will lock the storage once the first TPM command has been received. So
sending the lock command is an optimization.

Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 20220912174741.1542330-3-stefanb@linux.ibm.com

tpm_emulator: Use latest tpm_ioctl.h from swtpm project

Use the latest tpm_ioctl.h from upstream swtpm project.

Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-id: 20220912174741.1542330-2-stefanb@linux.ibm.com

tpm_crb: Avoid backend startup just before shutdown under Xen

When running under Xen and the guest reboots, it boots into a new domain
with a new QEMU process (and a new swtpm process if using the emulator
backend). The existing reset function is triggered just before the old
QEMU process exists which causes QEMU to startup the TPM backend and
then immediately shut it down. This is probably harmless but when using
the emulated backend, it wastes CPU and IO time reloading state, etc.

Fix this by calling the reset function directly from realize() when
running under Xen. During a reboot, this will be called by the QEMU
process for the new domain.

Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Reviewed-by: Stefan Berger <stefanb@linux.ibm.com>
Message-id: 20220826143841.1515326-1-ross.lagerwall@citrix.com
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>

tpm_emulator: Avoid double initialization during migration

When resuming after a migration, the backend sends CMD_INIT to the
emulator from the startup callback, then it sends the migration state
from the vmstate to the emulator, then it sends CMD_INIT again. Skip the
first CMD_INIT during a migration to avoid initializing the TPM twice.

Signed-off-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Stefan Berger <stefanb@linux.ibm.com>
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>

Merge tag 'pull-qapi-2022-09-07' of git://repo.or.cz/qemu/armbru into staging

QAPI patches patches for 2022-09-07

# -----BEGIN PGP SIGNATURE-----
#
# iQJGBAABCAAwFiEENUvIs9frKmtoZ05fOHC0AOuRhlMFAmMYp0sSHGFybWJydUBy
# ZWRoYXQuY29tAAoJEDhwtADrkYZTtbUP/3bHyjF4xamMwlXHi/nY/zlNKkZyhMul
# SdS+KVLBxufuYmlP2mPGqG/wnK4geWnduwprl/SCSd8NdymWDVxu6tUjdPAdW6S4
# 27WZUADZtXow2tOczcKFKMve4qBZXNKtv4W/eBTDMMpfje3dWBuEx70xN1HgpCWw
# i/h3J5dKwwWmHnyqroFJAEUPTx0jEk9GhzZCQ4hey2CFXDc5sgywliJA+pKv+1sF
# z1g2S0rq35jd7YZnPF0M3J2MdqQo6AaOsoPqRSJMpuma0Cfvwj0UmjzQAIeaahrD
# 1uEP3YEGXeOQlocDv3zqA6VFympAyFv+Ljj4XW3dsu9waBvtg2FYXQOrizxDuj1Y
# nAGJ6/xYrvZ+8Eb0tbdJYnxWv7MESbswhabJA+YfnjKOTD6TV5hmuHEzoR5c5Mly
# RYT+68WANXveOZ+f4oiqHQ4Gxuvp9+rU2uHQ+MvMhScDmJE7E+Gg8zELwKiQWu2q
# lU5vOv55fZTHhLt/gyOAEuNvFCnbfdYluV845GBaaqLRjBPMfSoNKBtswzQbCgKH
# vy5R+omAxQjjXzEL9GO2Wm50xYk5Er14ad5vAQ1256/sx3zuzQ/O0jzvGsKNPIXG
# Y/dnFg3e9eT18gQ+uDyZGpYf0XV5PKlUUcomIPikWogTiZvTXMdYNWz+gt2taatE
# jbxTea7AU3ty
# =L82X
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 07 Sep 2022 10:14:35 EDT
# gpg:                using RSA key 354BC8B3D7EB2A6B68674E5F3870B400EB918653
# gpg:                issuer "armbru@redhat.com"
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>" [full]
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>" [full]
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867  4E5F 3870 B400 EB91 8653

* tag 'pull-qapi-2022-09-07' of git://repo.or.cz/qemu/armbru:
  qapi: fix examples of events missing timestamp
  qapi: fix example of query-hotpluggable-cpus command
  qapi: fix examples of blockdev-add with qcow2
  qapi: fix example of MEM_UNPLUG_ERROR event
  qapi: fix example of DEVICE_UNPLUG_GUEST_ERROR event
  qapi: fix example of NIC_RX_FILTER_CHANGED event
  qapi: fix example of BLOCK_JOB_READY event
  qapi: fix example of query-dump-guest-memory-capability command
  qapi: fix example of query-vnc command
  qapi: fix example of query-ballon command

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

qapi: fix examples of events missing timestamp

I've used real timestamp and changing them one by one so they would
not be all equal.

Problem was noticed when using the example as a test case for Go
bindings.

Signed-off-by: Victor Toso <victortoso@redhat.com>
Message-Id: <20220901085840.22520-11-victortoso@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

qapi: fix example of query-hotpluggable-cpus command

The example return type has the wrong member name. Fix it.

Problem was noticed when using the example as a test case for Go
bindings.

Signed-off-by: Victor Toso <victortoso@redhat.com>
Message-Id: <20220901085840.22520-10-victortoso@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

qapi: fix examples of blockdev-add with qcow2

The examples use "qcow2" driver with the wrong member name for
BlockdevRef alternate type. This patch changes all wrong member names
from "file" to "data-file" which is the correct member name in
BlockdevOptionsQcow2 for the BlockdevRef field.

Problem was noticed when using the example as a test case for Go
bindings.

Signed-off-by: Victor Toso <victortoso@redhat.com>
Message-Id: <20220901085840.22520-9-victortoso@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

qapi: fix example of MEM_UNPLUG_ERROR event

Example output was missing ',' delimiter. Fix it.

Problem was noticed when trying to load the example into python's json
library.

Signed-off-by: Victor Toso <victortoso@redhat.com>
Message-Id: <20220901085840.22520-8-victortoso@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

qapi: fix example of DEVICE_UNPLUG_GUEST_ERROR event

Example output is missing a ',' delimiter and it has an extra ending
curly bracket. Fix it.

Problem was noticed when trying to load the example into python's json
library.

Signed-off-by: Victor Toso <victortoso@redhat.com>
Message-Id: <20220901085840.22520-7-victortoso@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

qapi: fix example of NIC_RX_FILTER_CHANGED event

Example output has an extra ending curly bracket. Fix it.

Problem was noticed when trying to load the example into python's json
library.

Signed-off-by: Victor Toso <victortoso@redhat.com>
Message-Id: <20220901085840.22520-6-victortoso@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

qapi: fix example of BLOCK_JOB_READY event

Example output is missing ',' delimiter. Fix it.

Problem was noticed when trying to load the example into python's json
library.

Signed-off-by: Victor Toso <victortoso@redhat.com>
Message-Id: <20220901085840.22520-5-victortoso@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

qapi: fix example of query-dump-guest-memory-capability command

Example output is missing closing curly brackets. Fix it.

Problem was noticed when trying to load the example into python's json
library.

Signed-off-by: Victor Toso <victortoso@redhat.com>
Message-Id: <20220901085840.22520-4-victortoso@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

qapi: fix example of query-vnc command

Example output has an extra ',' delimiter in member "websocket" and it
lacks it in "family" member. Fix it.

Problem was noticed when trying to load the example into python's json
library.

Signed-off-by: Victor Toso <victortoso@redhat.com>
Message-Id: <20220901085840.22520-3-victortoso@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

qapi: fix example of query-ballon command

Example output has an extra ',' delimiter. Fix it.

Problem was noticed when trying to load the example into python's json
library.

Signed-off-by: Victor Toso <victortoso@redhat.com>
Message-Id: <20220901085840.22520-2-victortoso@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

Merge tag 'pull-riscv-to-apply-20220907' of https://github.com/alistair23/qemu into staging

First RISC-V PR for QEMU 7.2

* Update [m|h]tinst CSR in interrupt handling
* Force disable extensions if priv spec version does not match
* fix shifts shamt value for rv128c
* move zmmul out of the experimental
* virt: pass random seed to fdt
* Add checks for supported extension combinations
* Upgrade OpenSBI to v1.1
* Fix typo and restore Pointer Masking functionality for RISC-V
* Add mask agnostic behaviour (rvv_ma_all_1s) for vector extension
* Add Zihintpause support
* opentitan: bump opentitan version
* microchip_pfsoc: fix kernel panics due to missing peripherals
* Remove additional priv version check for mcountinhibit
* virt machine device tree improvements
* Add xicondops in ISA entry
* Use official extension names for AIA CSRs

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCAAdFiEE9sSsRtSTSGjTuM6PIeENKd+XcFQFAmMYUCUACgkQIeENKd+X
# cFRpEQf/T1FFcGq3TZrEPmqMdFPUSb+SEJNgwYFfloqkNjB2HIFbd2tKWAE1Tgjr
# esV00p7YPyox1Ct+fKdwSxDxRSN9OI56v+nI8ZFwluVu7vpChuTFmOHur8rNxl1T
# 8MZgP2kMxMOJSnyHCS2iV9AUFdTExS65DbmlAKzi5fpBtt9jYTPSXsI49MP8+Ku/
# 1gdv5ZF5BXDJsGs7xHvE92dRzQEVN+As64IjlknFHHpmCM1b+Ah3GekXUbKmBuDG
# /NaZyZNPCYxdRmPm/D7k0SOMZSJ9sLyhXTetZ0ZpBxG1ioClX37yS5wn4NLsCz/2
# fXrnML+MQFUKZ03AZ9lWvxcu7kXfWA==
# =7mGD
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 07 Sep 2022 04:02:45 EDT
# gpg:                using RSA key F6C4AC46D4934868D3B8CE8F21E10D29DF977054
# gpg: Good signature from "Alistair Francis <alistair@alistair23.me>" [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: F6C4 AC46 D493 4868 D3B8  CE8F 21E1 0D29 DF97 7054

* tag 'pull-riscv-to-apply-20220907' of https://github.com/alistair23/qemu: (44 commits)
  target/riscv: Update the privilege field for sscofpmf CSRs
  hw/riscv: virt: Add PMU DT node to the device tree
  target/riscv: Add few cache related PMU events
  target/riscv: Simplify counter predicate function
  target/riscv: Add sscofpmf extension support
  target/riscv: Add vstimecmp support
  target/riscv: Add stimecmp support
  hw/intc: Move mtimer/mtimecmp to aclint
  target/riscv: Use official extension names for AIA CSRs
  target/riscv: Add xicondops in ISA entry
  hw/core: fix platform bus node name
  hw/riscv: virt: fix syscon subnode paths
  hw/riscv: virt: fix the plic's address cells
  hw/riscv: virt: fix uart node name
  target/riscv: Remove additional priv version check for mcountinhibit
  hw/riscv: microchip_pfsoc: fix kernel panics due to missing peripherals
  hw/riscv: opentitan: bump opentitan version
  target/riscv: Fix priority of csr related check in riscv_csrrw_check
  hw/riscv: remove 'fdt' param from riscv_setup_rom_reset_vec()
  target/riscv: Add Zihintpause support
  ...

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

target/riscv: Update the privilege field for sscofpmf CSRs

The sscofpmf extension was ratified as a part of priv spec v1.12.
Mark the csr_ops accordingly.

Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Atish Patra <atishp@rivosinc.com>
Message-Id: <20220824221701.41932-6-atishp@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv: virt: Add PMU DT node to the device tree

Qemu virt machine can support few cache events and cycle/instret counters.
It also supports counter overflow for these events.

Add a DT node so that OpenSBI/Linux kernel is aware of the virt machine
capabilities. There are some dummy nodes added for testing as well.

Acked-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Signed-off-by: Atish Patra <atishp@rivosinc.com>
Message-Id: <20220824221701.41932-5-atishp@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Add few cache related PMU events

Qemu can monitor the following cache related PMU events through
tlb_fill functions.

1. DTLB load/store miss
3. ITLB prefetch miss

Increment the PMU counter in tlb_fill function.

Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Tested-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Signed-off-by: Atish Patra <atishp@rivosinc.com>
Message-Id: <20220824221701.41932-4-atishp@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Simplify counter predicate function

All the hpmcounters and the fixed counters (CY, IR, TM) can be represented
as a unified counter. Thus, the predicate function doesn't need handle each
case separately.

Simplify the predicate function so that we just handle things differently
between RV32/RV64 and S/HS mode.

Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Atish Patra <atishp@rivosinc.com>
Message-Id: <20220824221701.41932-3-atishp@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Add sscofpmf extension support

The Sscofpmf ('Ss' for Privileged arch and Supervisor-level extensions,
and 'cofpmf' for Count OverFlow and Privilege Mode Filtering)
extension allows the perf to handle overflow interrupts and filtering
support. This patch provides a framework for programmable
counters to leverage the extension. As the extension doesn't have any
provision for the overflow bit for fixed counters, the fixed events
can also be monitoring using programmable counters. The underlying
counters for cycle and instruction counters are always running. Thus,
a separate timer device is programmed to handle the overflow.

Tested-by: Heiko Stuebner <heiko@sntech.de>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Signed-off-by: Atish Patra <atishp@rivosinc.com>
Message-Id: <20220824221701.41932-2-atishp@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Add vstimecmp support

vstimecmp CSR allows the guest OS or to program the next guest timer
interrupt directly. Thus, hypervisor no longer need to inject the
timer interrupt to the guest if vstimecmp is used. This was ratified
as a part of the Sstc extension.

Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Atish Patra <atishp@rivosinc.com>
Message-Id: <20220824221357.41070-4-atishp@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Add stimecmp support

stimecmp allows the supervisor mode to update stimecmp CSR directly
to program the next timer interrupt. This CSR is part of the Sstc
extension which was ratified recently.

Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Atish Patra <atishp@rivosinc.com>
Message-Id: <20220824221357.41070-3-atishp@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/intc: Move mtimer/mtimecmp to aclint

Historically, The mtime/mtimecmp has been part of the CPU because
they are per hart entities. However, they actually belong to aclint
which is a MMIO device.

Move them to the ACLINT device. This also emulates the real hardware
more closely.

Reviewed-by: Anup Patel <anup@brainfault.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Signed-off-by: Atish Patra <atishp@rivosinc.com>
Message-Id: <20220824221357.41070-2-atishp@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Use official extension names for AIA CSRs

The arch review of AIA spec is completed and we now have official
extension names for AIA: Smaia (M-mode AIA CSRs) and Ssaia (S-mode
AIA CSRs).

Refer, section 1.6 of the latest AIA v0.3.1 stable specification at
https://github.com/riscv/riscv-aia/releases/download/0.3.1-draft.32/riscv-interrupts-032.pdf)

Based on above, we update QEMU RISC-V to:
1) Have separate config options for Smaia and Ssaia extensions
   which replace RISCV_FEATURE_AIA in CPU features
2) Not generate AIA INTC compatible string in virt machine

Signed-off-by: Anup Patel <apatel@ventanamicro.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220820042958.377018-1-apatel@ventanamicro.com
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Add xicondops in ISA entry

XVentanaCondOps is Ventana custom extension. Add
its extension entry in the ISA Ext array

Signed-off-by: Rahul Pathak <rpathak@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220816045408.1231135-1-rpathak@ventanamicro.com
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/core: fix platform bus node name

"platform" is not a valid name for a bus node in dt-schema, so warnings
can be see in dt-validate on a dump of the riscv virt dtb:

/stuff/qemu/qemu.dtb: platform@4000000: $nodename:0: 'platform@4000000' does not match '^([a-z][a-z0-9\\-]+-bus|bus|soc|axi|ahb|apb)(@[0-9a-f]+)?$'
        From schema: /home/conor/.local/lib/python3.9/site-packages/dtschema/schemas/simple-bus.yaml
"platform-bus" is a valid name, so use that instead.

CC: Rob Herring <robh@kernel.org>
Fixes: 11d306b9df ("hw/arm/sysbus-fdt: helpers for platform bus nodes addition")
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Message-id: 20220810184612.157317-5-mail@conchuod.ie
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv: virt: fix syscon subnode paths

The reset and poweroff features of the syscon were originally added to
top level, which is a valid path for a syscon subnode. Subsequently a
reorganisation was carried out while implementing NUMA in which the
subnodes were moved into the /soc node. As /soc is a "simple-bus", this
path is invalid, and so dt-validate produces the following warnings:

/stuff/qemu/qemu.dtb: soc: poweroff: {'value': [[21845]], 'offset': [[0]], 'regmap': [[4]], 'compatible': ['syscon-poweroff']} should not be valid under {'type': 'object'}
        From schema: /home/conor/.local/lib/python3.9/site-packages/dtschema/schemas/simple-bus.yaml
/stuff/qemu/qemu.dtb: soc: reboot: {'value': [[30583]], 'offset': [[0]], 'regmap': [[4]], 'compatible': ['syscon-reboot']} should not be valid under {'type': 'object'}
        From schema: /home/conor/.local/lib/python3.9/site-packages/dtschema/schemas/simple-bus.yaml

Move the syscon subnodes back to the top level and silence the warnings.

Reported-by: Rob Herring <robh@kernel.org>
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20220810184612.157317-4-mail@conchuod.ie
Link: https://lore.kernel.org/linux-riscv/20220803170552.GA2250266-robh@kernel.org/
Fixes: 18df0b4695 ("hw/riscv: virt: Allow creating multiple NUMA sockets")
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv: virt: fix the plic's address cells

When optional AIA PLIC support was added the to the virt machine, the
address cells property was removed leading the issues with dt-validate
on a dump from the virt machine:
/stuff/qemu/qemu.dtb: plic@c000000: '#address-cells' is a required property
        From schema: /stuff/linux/Documentation/devicetree/bindings/interrupt-controller/sifive,plic-1.0.0.yaml
Add back the property to suppress the warning.

Reported-by: Rob Herring <robh@kernel.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Message-id: 20220810184612.157317-3-mail@conchuod.ie
Link: https://lore.kernel.org/linux-riscv/20220803170552.GA2250266-robh@kernel.org/
Fixes: e6faee6585 ("hw/riscv: virt: Add optional AIA APLIC support to virt machine")
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv: virt: fix uart node name

"uart" is not a node name that complies with the dt-schema.
Change the node name to "serial" to ix warnings seen during
dt-validate on a dtbdump of the virt machine such as:
/stuff/qemu/qemu.dtb: uart@10000000: $nodename:0: 'uart@10000000' does not match '^serial(@.*)?$'
        From schema: /stuff/linux/Documentation/devicetree/bindings/serial/8250.yaml

Reported-by: Rob Herring <robh@kernel.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Message-id: 20220810184612.157317-2-mail@conchuod.ie
Link: https://lore.kernel.org/linux-riscv/20220803170552.GA2250266-robh@kernel.org/
Fixes: 04331d0b56 ("RISC-V VirtIO Machine")
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Remove additional priv version check for mcountinhibit

With .min_priv_version, additiona priv version check is uncessary
for mcountinhibit read/write functions.

Reviewed-by: Heiko Stuebner <heiko@sntech.de>
Tested-by: Heiko Stuebner <heiko@sntech.de>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Atish Patra <atishp@rivosinc.com>
Message-Id: <20220816232321.558250-7-atishp@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv: microchip_pfsoc: fix kernel panics due to missing peripherals

Booting using "Direct Kernel Boot" for PolarFire SoC & skipping u-boot
entirely is probably not advisable, but it does at least show signs of
life. Recent Linux kernel versions make use of peripherals that are
missing definitions in QEMU and lead to kernel panics. These issues
almost certain rear their head for other methods of booting, but I was
unable to figure out a suitable HSS version that is recent enough to
support these peripherals & works with QEMU.

With these peripherals added, booting a kernel with the following hangs
hangs waiting for the system controller's hwrng, but the kernel no
longer panics. With the Linux driver for hwrng disabled, it boots to
console.

qemu-system-riscv64 -M microchip-icicle-kit \
-m 2G -smp 5 \
-kernel $(vmlinux_bin) \
-dtb  $(dtb)\
-initrd $(initramfs) \
-display none -serial null \
-serial stdio

More peripherals are added than strictly required to fix the panics in
the hopes of avoiding a replication of this problem in the future.
Some of the peripherals which are in the device tree for recent kernels
are implemented in the FPGA fabric. The eMMC/SD mux, which exists as
an unimplemented device is replaced by a wider entry. This updated
entry covers both the mux & the remainder of the FPGA fabric connected
to the MSS using Fabric Interrconnect (FIC) 3.

Link: https://github.com/polarfire-soc/icicle-kit-reference-design#fabric-memory-map
Link: https://ww1.microchip.com/downloads/aemDocuments/documents/FPGA/ProductDocuments/SupportingCollateral/V1_4_Register_Map.zip
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20220813135127.2971754-1-mail@conchuod.ie>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv: opentitan: bump opentitan version

The following patch updates opentitan to match the new configuration,
as per, lowRISC/opentitan@217a0168ba118503c166a9587819e3811eeb0c0c

Note: with this patch we now skip the usage of the opentitan
`boot_rom`. The Opentitan boot rom contains hw verification
for devies which we are currently not supporting in qemu. As of now,
the `boot_rom` has no major significance, however, would be good to
support in the future.

Tested by running utests from the latest tock [1]
(that supports this version of OT).

[1] https://github.com/tock/tock/pull/3056

Signed-off-by: Wilfred Mallawa <wilfred.mallawa@wdc.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20220812005229.358850-1-wilfred.mallawa@opensource.wdc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Fix priority of csr related check in riscv_csrrw_check

Normally, riscv_csrrw_check is called when executing Zicsr instructions.
And we can only do access control for existed CSRs. So the priority of
CSR related check, from highest to lowest, should be as follows:
1) check whether Zicsr is supported: raise RISCV_EXCP_ILLEGAL_INST if not
2) check whether csr is existed: raise RISCV_EXCP_ILLEGAL_INST if not
3) do access control: raise RISCV_EXCP_ILLEGAL_INST or RISCV_EXCP_VIRT_
INSTRUCTION_FAULT if not allowed

The predicates contain parts of function of both 2) and 3), So they need
to be placed in the middle of riscv_csrrw_check

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20220803123652.3700-1-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv: remove 'fdt' param from riscv_setup_rom_reset_vec()

The 'fdt' param is not being used in riscv_setup_rom_reset_vec().
Simplify the API by removing it. While we're at it, remove the redundant
'return' statement at the end of function.

Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Alistair Francis <alistair.francis@wdc.com>
Cc: Bin Meng <bin.meng@windriver.com>
Cc: Vijai Kumar K <vijai@behindbytes.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20220728181926.2123771-1-danielhb413@gmail.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Add Zihintpause support

Added support for RISC-V PAUSE instruction from Zihintpause extension,
enabled by default.

Tested-by: Heiko Stuebner <heiko@sntech.de>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Dao Lu <daolu@rivosinc.com>
Message-Id: <20220725034728.2620750-2-daolu@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: rvv: Add option 'rvv_ma_all_1s' to enable optional mask agnostic behavior

According to v-spec, mask agnostic behavior can be either kept as
undisturbed or set elements' bits to all 1s. To distinguish the
difference of mask policies, QEMU should be able to simulate the mask
agnostic behavior as "set mask elements' bits to all 1s".

There are multiple possibility for agnostic elements according to
v-spec. The main intent of this patch-set tries to add option that
can distinguish between mask policies. Setting agnostic elements to
all 1s allows QEMU to express this.

This commit adds option 'rvv_ma_all_1s' is added to enable the
behavior, it is default as disabled.

Signed-off-by: eop Chen <eop.chen@sifive.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <165570784143.17634.35095816584573691-10@git.sr.ht>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: rvv: Add mask agnostic for vector permutation instructions

Signed-off-by: eop Chen <eop.chen@sifive.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <165570784143.17634.35095816584573691-9@git.sr.ht>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: rvv: Add mask agnostic for vector mask instructions

Signed-off-by: eop Chen <eop.chen@sifive.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <165570784143.17634.35095816584573691-8@git.sr.ht>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: rvv: Add mask agnostic for vector floating-point instructions

Signed-off-by: eop Chen <eop.chen@sifive.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <165570784143.17634.35095816584573691-7@git.sr.ht>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: rvv: Add mask agnostic for vector fix-point arithmetic instructions

Signed-off-by: eop Chen <eop.chen@sifive.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <165570784143.17634.35095816584573691-6@git.sr.ht>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: rvv: Add mask agnostic for vector integer comparison instructions

Signed-off-by: eop Chen <eop.chen@sifive.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <165570784143.17634.35095816584573691-5@git.sr.ht>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: rvv: Add mask agnostic for vector integer shift instructions

Signed-off-by: eop Chen <eop.chen@sifive.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <165570784143.17634.35095816584573691-4@git.sr.ht>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: rvv: Add mask agnostic for vx instructions

Signed-off-by: eop Chen <eop.chen@sifive.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <165570784143.17634.35095816584573691-3@git.sr.ht>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: rvv: Add mask agnostic for vector load / store instructions

Signed-off-by: eop Chen <eop.chen@sifive.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <165570784143.17634.35095816584573691-2@git.sr.ht>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: rvv: Add mask agnostic for vv instructions

According to v-spec, mask agnostic behavior can be either kept as
undisturbed or set elements' bits to all 1s. To distinguish the
difference of mask policies, QEMU should be able to simulate the mask
agnostic behavior as "set mask elements' bits to all 1s".

There are multiple possibility for agnostic elements according to
v-spec. The main intent of this patch-set tries to add option that
can distinguish between mask policies. Setting agnostic elements to
all 1s allows QEMU to express this.

This is the first commit regarding the optional mask agnostic
behavior. Follow-up commits will add this optional behavior
for all rvv instructions.

Signed-off-by: eop Chen <eop.chen@sifive.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <165570784143.17634.35095816584573691-1@git.sr.ht>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

docs: List kvm as a supported accelerator on RISC-V

Since commit fbf43c7dbf18 ("target/riscv: enable riscv kvm accel"),
KVM accelerator is supported on RISC-V. Let's document it.

Signed-off-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20220719082635.3741878-1-bin.meng@windriver.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Fix typo and restore Pointer Masking functionality for RISC-V

Fixes: 4302bef9e178 ("target/riscv: Calculate address according to XLEN")
Signed-off-by: Alexey Baturo <baturo.alexey@gmail.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20220717101543.478533-2-space.monkey.delivers@gmail.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>