target/arm: Implement SME ZERO

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-19-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Implement SME RDSVL, ADDSVL, ADDSPL

These SME instructions are nominally within the SVE decode space,
so we add them to sve.decode and translate-sve.c.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-18-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Handle SME in sve_access_check

The pseudocode for CheckSVEEnabled gains a check for Streaming
SVE mode, and for SME present but SVE absent.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-17-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Add SME enablement checks

These functions will be used to verify that the cpu
is in the correct state for a given instruction.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-16-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Mark LD1RO as non-streaming

Mark these as a non-streaming instructions, which should trap
if full a64 support is not enabled in streaming mode.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-15-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Mark LDFF1 and LDNF1 as non-streaming

Mark these as a non-streaming instructions, which should trap
if full a64 support is not enabled in streaming mode.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-14-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Mark gather prefetch as non-streaming

Mark these as a non-streaming instructions, which should trap if full
a64 support is not enabled in streaming mode.  In this case, introduce
PRF_ns (prefetch non-streaming) to handle the checks.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-13-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Mark gather/scatter load/store as non-streaming

Mark these as a non-streaming instructions, which should trap
if full a64 support is not enabled in streaming mode.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-12-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Mark string/histo/crypto as non-streaming

Mark these as non-streaming instructions, which should trap
if full a64 support is not enabled in streaming mode.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-11-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Mark SMMLA, UMMLA, USMMLA as non-streaming

Mark these as a non-streaming instructions, which should trap
if full a64 support is not enabled in streaming mode.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-10-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Mark FTSMUL, FTMAD, FADDA as non-streaming

Mark these as a non-streaming instructions, which should trap
if full a64 support is not enabled in streaming mode.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-9-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Mark PMULL, FMMLA as non-streaming

Mark these as a non-streaming instructions, which should trap
if full a64 support is not enabled in streaming mode.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-8-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Mark BDEP, BEXT, BGRP, COMPACT, FEXPA, FTSSEL as non-streaming

Mark these as a non-streaming instructions, which should trap
if full a64 support is not enabled in streaming mode.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-7-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Mark RDFFR, WRFFR, SETFFR as non-streaming

Mark these as a non-streaming instructions, which should trap
if full a64 support is not enabled in streaming mode.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-6-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Mark ADR as non-streaming

Mark ADR as a non-streaming instruction, which should trap
if full a64 support is not enabled in streaming mode.

Removing entries from sme-fa64.decode is an easy way to see
what remains to be done.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-5-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Trap non-streaming usage when Streaming SVE is active

This new behaviour is in the ARM pseudocode function
AArch64.CheckFPAdvSIMDEnabled, which applies to AArch32
via AArch32.CheckAdvSIMDOrFPEnabled when the EL to which
the trap would be delivered is in AArch64 mode.

Given that ARMv9 drops support for AArch32 outside EL0, the trap EL
detection ought to be trivially true, but the pseudocode still contains
a number of conditions, and QEMU has not yet committed to dropping A32
support for EL[12] when v9 features are present.

Since the computation of SME_TRAP_NONSTREAMING is necessarily different
for the two modes, we might as well preserve bits within TBFLAG_ANY and
allocate separate bits within TBFLAG_A32 and TBFLAG_A64 instead.

Note that DDI0616A.a has typos for bits [22:21] of LD1RO in the table
of instructions illegal in streaming mode.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-4-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Add infrastructure for disas_sme

This includes the build rules for the decoder, and the
new file for translation, but excludes any instructions.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-3-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Handle SME in aarch64_cpu_dump_state

Dump SVCR, plus use the correct access check for Streaming Mode.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-2-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge tag 'pull-target-arm-20220707' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * hw/arm/virt: dt: add rng-seed property
 * Fix MTE check in sve_ldnfff1_r
 * Record tagged bit for user-only in sve_probe_page
 * Correctly implement OS Lock and OS DoubleLock
 * Implement DBGDEVID, DBGDEVID1, DBGDEVID2 registers
 * Fix qemu-system-arm handling of LPAE block descriptors for highmem

# -----BEGIN PGP SIGNATURE-----
#
# iQJNBAABCAA3FiEE4aXFk81BneKOgxXPPCUl7RQ2DN4FAmLG0O8ZHHBldGVyLm1h
# eWRlbGxAbGluYXJvLm9yZwAKCRA8JSXtFDYM3h1MEACnTd+jfssa/MdtQPP8N+cp
# cvrdrW+V9iho1vLPGG3d6yni6T8IMdulA5OvSkiNz2IaspY3z3u/t01PCpqEF0tA
# nl3HxA65ddehzG8QHpORiP2DIJfuYUcMaruK2zMcC//0EXBOVI215hkR7oNyC0r9
# PrYb+tiDVL1t3xzXWNKMoolZwUjS6M6dloxEu2b/d0tOBvtBLI0E7y9taANHCnmZ
# 8r7ih0WRELo1rzveZVOXqnZGLUjFzbCurHCmshN6xr2V6iilBaLoaGHovZ2c489Z
# Fz/Ui1tyvDUoajr/Ck57GYo0BwDf9dKYkl5RkchdeY+cA88CgJAVK5pT0Rrybpf0
# lMgSAalIFcnIr2kjdnWRUqL02t+HgnOnsBTSUpgiwMNKwjfMN5NDi9294GuMCu2h
# 7UyuAkQvfTwoQyFzJYzuapcnB6i5BvQjq6GxSmogzuR6kNSFptNOGimMhqCO4kN0
# fixOHOOE/aj7QoXq33V9u3ESY1IleDMX6+Zff2RLmVQYZpQKW6LpCfuwgnZrZEN3
# 4hPiL+00bY9pWlXvvEHdu0/XYx8kPXtmdzgzpm80edCLUMFOVMDQMQE1Zy/qzC7S
# 8ryfC2EvSQuigD1+s6I9WxCnHVv5Byz+DXgRCjRsitDPVu2kFcEgAibzz9K1WVUW
# qwQP4WGUzFD/LNUwO/Vryw==
# =TzwD
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 07 Jul 2022 05:56:23 PM +0530
# gpg:                using RSA key E1A5C593CD419DE28E8315CF3C2525ED14360CDE
# gpg:                issuer "peter.maydell@linaro.org"
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>" [full]
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>" [full]
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>" [full]

* tag 'pull-target-arm-20220707' of https://git.linaro.org/people/pmaydell/qemu-arm:
  target/arm: Fix qemu-system-arm handling of LPAE block descriptors for highmem
  target/arm: Correctly implement Feat_DoubleLock
  target/arm: Implement AArch32 DBGDEVID, DBGDEVID1, DBGDEVID2
  target/arm: Suppress debug exceptions when OS Lock set
  target/arm: Move define_debug_regs() to debug_helper.c
  target/arm: Fix code style issues in debug helper functions
  target/arm: Record tagged bit for user-only in sve_probe_page
  target/arm: Fix MTE check in sve_ldnfff1_r
  hw/arm/virt: dt: add rng-seed property

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

Merge tag 'pull-request-2022-07-07' of https://gitlab.com/thuth/qemu into staging

* Check validity of the address in the SET PREFIX instruction
* Fix booting from devices that use 4k sectors, but are not like DASDs
* Re-evaluate pending interrupts after EXECUTE of certain instructions

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmLGhkURHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbU76Q//Y4mEXxWZDpJTg7tL5SZP+UzBVttnCifv
# 6q+2I0keOUs6wFhPq8TzBqkazT9wlv51cNrY2Q3fU9I3dpDeRbAXZR34UD4kw5an
# Q+ZQcebuGSKLjzMrIb1DLAieq8OmZR5FvDUu16BbeJr6GIQIE80lMRfWh9j30UfW
# tlxkXr15BnyPx6m0rSGwzkZD2vgfj5zSUrDtYJcUsfypIA9OOBMA7yNGNlO+d94V
# UZiKgNQtAoBNm4hZh2M86nsUtem+WpMTZQnDnCpMLYvFV/u9jRQBFSR+Ay41hcEN
# WYuLK61rkjc9gPWSjeNNT28x8RMvFJU4YNn1UDiMRSzrigxeui6MOW3SI/h3y6tI
# 94yXmXV2IuDMibvOjK07nkDaEItqPxfj6zuM2xW1Nc+l8Sk12korFBpk/AZiD0Jo
# R3u36efci3zNqDRDJvhGUv8sGcv0mwO7Agq1Bm3h5941gYwzQKILHCShL7DPzvQa
# h+K1MsT7vWfh5++unkGUrN/Zd9CazEylbDuWtywK8lgQcTGDO/9rab8GeXfH/5es
# Tp0RGJwxmalgrAHZPK9lqgpQaGw92ct2G5odvc82EXQhgccnN9mh54BHPfdKs95E
# JZVrXtZH3Gtgl5MGZ+yJevWSc9h1iRnRF4a7QC3UlVBjA/9yAWzQUAnNGZOamE/s
# F+pi89oWLn8=
# =UsTi
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 07 Jul 2022 12:37:49 PM +0530
# gpg:                using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5
# gpg:                issuer "thuth@redhat.com"
# gpg: Good signature from "Thomas Huth <th.huth@gmx.de>" [undefined]
# gpg:                 aka "Thomas Huth <thuth@redhat.com>" [undefined]
# gpg:                 aka "Thomas Huth <th.huth@posteo.de>" [unknown]
# gpg:                 aka "Thomas Huth <huth@tuxfamily.org>" [undefined]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 27B8 8847 EEE0 2501 18F3  EAB9 2ED9 D774 FE70 2DB5

* tag 'pull-request-2022-07-07' of https://gitlab.com/thuth/qemu:
  target/s390x: Exit tb after executing ex_value
  target/s390x: Remove DISAS_PC_STALE_NOCHAIN
  target/s390x: Remove DISAS_PC_STALE
  target/s390x: Remove DISAS_GOTO_TB
  pc-bios/s390-ccw: Update the s390-ccw bios binaries with the virtio-blk fixes
  pc-bios/s390-ccw/netboot.mak: Ignore Clang's warnings about GNU extensions
  pc-bios/s390-ccw/virtio: Remove "extern" keyword from prototypes
  pc-bios/s390-ccw/virtio-blkdev: Request the right feature bits
  pc-bios/s390-ccw: Split virtio-scsi code from virtio_blk_setup_device()
  pc-bios/s390-ccw/virtio: Beautify the code for reading virtqueue configuration
  pc-bios/s390-ccw/virtio: Read device config after feature negotiation
  pc-bios/s390-ccw/virtio: Set missing status bits while initializing
  pc-bios/s390-ccw/virtio-blkdev: Remove virtio_assume_scsi()
  pc-bios/s390-ccw/virtio-blkdev: Simplify/fix virtio_ipl_disk_is_valid()
  pc-bios/s390-ccw/bootmap: Improve the guessing logic in zipl_load_vblk()
  pc-bios/s390-ccw/virtio: Introduce a macro for the DASD block size
  pc-bios/s390-ccw: Add a proper prototype for main()
  target/s390x/tcg: SPX: check validity of new prefix

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

target/arm: Fix qemu-system-arm handling of LPAE block descriptors for highmem

In commit 39a1fd25287f5d we fixed a bug in the handling of LPAE block
descriptors where we weren't correctly zeroing out some RES0 bits.
However this fix has a bug because the calculation of the mask is
done at the wrong width: in
  descaddr &= ~(page_size - 1);
page_size is a target_ulong, so in the 'qemu-system-arm' binary it is
only 32 bits, and the effect is that we always zero out the top 32
bits of the calculated address.  Fix the calculation by forcing the
mask to be calculated with the same type as descaddr.

This only affects 32-bit CPUs which support LPAE (e.g. cortex-a15)
when used on board models which put RAM or devices above the 4GB
mark and when the 'qemu-system-arm' executable is being used.
It was also masked in 7.0 by the main bug reported in
https://gitlab.com/qemu-project/qemu/-/issues/1078 where the
virt board incorrectly does not enable 'highmem' for 32-bit CPUs.

The workaround is to use 'qemu-system-aarch64' with the same
command line.

Reported-by: He Zhe <zhe.he@windriver.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220627134620.3190252-1-peter.maydell@linaro.org
Fixes: 39a1fd25287f5de ("target/arm: Fix handling of LPAE block descriptors")
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Correctly implement Feat_DoubleLock

The architecture defines the OS DoubleLock as a register which
(similarly to the OS Lock) suppresses debug events for use in CPU
powerdown sequences.  This functionality is required in Arm v7 and
v8.0; from v8.2 it becomes optional and in v9 it must not be
implemented.

Currently in QEMU we implement the OSDLR_EL1 register as a NOP.  This
is wrong both for the "feature implemented" and the "feature not
implemented" cases: if the feature is implemented then the DLK bit
should read as written and cause suppression of debug exceptions, and
if it is not implemented then the bit must be RAZ/WI.

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

target/arm: Implement AArch32 DBGDEVID, DBGDEVID1, DBGDEVID2

Starting with v7 of the debug architecture, there are three extra
ID registers that add information on top of that provided in
DBGDIDR. These are DBGDEVID, DBGDEVID1 and DBGDEVID2. In the
v7 debug architecture, DBGDEVID is optional, present only of
DBGDIDR.DEVID_imp is set. In v7.1 all three must be present.

Implement the missing registers.  Note that we only need to set the
values in the ARMISARegisters struct for the CPUs Cortex-A7, A15,
A53, A57 and A72 (plus the 32-bit 'max' which uses the Cortex-A53
values): earlier CPUs didn't implement v7 of the architecture, and
our other 64-bit CPUs (Cortex-A76, Neoverse-N1 and A64fx) don't have
AArch32 support at EL1.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220630194116.3438513-5-peter.maydell@linaro.org

target/arm: Suppress debug exceptions when OS Lock set

The "OS Lock" in the Arm debug architecture is a way for software
to suppress debug exceptions while it is trying to power down
a CPU and save the state of the breakpoint and watchpoint
registers. In QEMU we implemented the support for writing
the OS Lock bit via OSLAR_EL1 and reading it via OSLSR_EL1,
but didn't implement the actual behaviour.

The required behaviour with the OS Lock set is:
 * debug exceptions (apart from BKPT insns) are suppressed
 * some MDSCR_EL1 bits allow write access to the corresponding
   EDSCR external debug status register that they shadow
   (we can ignore this because we don't implement external debug)
 * similarly with the OSECCR_EL1 which shadows the EDECCR
   (but we don't implement OSECCR_EL1 anyway)

Implement the missing behaviour of suppressing debug
exceptions.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220630194116.3438513-4-peter.maydell@linaro.org

target/arm: Move define_debug_regs() to debug_helper.c

The target/arm/helper.c file is very long and is a grabbag of all
kinds of functionality.  We have already a debug_helper.c which has
code for implementing architectural debug.  Move the code which
defines the debug-related system registers out to this file also.
This affects the define_debug_regs() function and the various
functions and arrays which are used only by it.

The functions raw_write() and arm_mdcr_el2_eff() and
define_debug_regs() now need to be global rather than local to
helper.c; everything else is pure code movement.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220630194116.3438513-3-peter.maydell@linaro.org

target/arm: Fix code style issues in debug helper functions

Before moving debug system register helper functions to a
different file, fix the code style issues (mostly block
comment syntax) so checkpatch doesn't complain about the
code-motion patch.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220630194116.3438513-2-peter.maydell@linaro.org

target/arm: Record tagged bit for user-only in sve_probe_page

Fixes a bug in that we were not honoring MTE from user-only
SVE. Copy the user-only MTE logic from allocation_tag_mem
into sve_probe_page.

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

target/arm: Fix MTE check in sve_ldnfff1_r

The comment was correct, but the test was not:
disable mte if tagged is *not* set.

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

hw/arm/virt: dt: add rng-seed property

In 60592cfed2 ("hw/arm/virt: dt: add kaslr-seed property"), the
kaslr-seed property was added, but the equally as important rng-seed
property was forgotten about, which has identical semantics for a
similar purpose. This commit implements it in exactly the same way as
kaslr-seed. It then changes the name of the disabling option to reflect
that this has more to do with randomness vs determinism, rather than
something particular about kaslr.

Cc: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
[PMM: added deprecated.rst section for the deprecation]
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge tag 'block-pull-request' of https://gitlab.com/stefanha/qemu into staging

Pull request

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCAAdFiEEhpWov9P5fNqsNXdanKSrs4Grc8gFAmLGlXYACgkQnKSrs4Gr
# c8jfNwf/fFfx4mPFwdnud9cseb9jFU44/RWJ2rJlYerbHIUUNvW0vAv1nGjuVVWT
# DeHxG2pkHPJQ5u4y/oivq4CCGGQ/Ukx/IEqJsnXEpWyKHosHJHlBrePt+/x+uygA
# RUOfReSq4gsK1FIcluuO/wnBzgREXL6lBsyIHtvG57hYSSESS10fCdRXBPJJMBcF
# pI8LpEgID5KuqPjXoO/Cv4qOXpOpO8540b8mzHP8wY1tYbBP9H4uqO1MJ/OFz8GU
# wDeRJEa66o9kwV7oym8T+KdZHR1f+Yfk0o1WmtfUgECJkM/8aUTvNog5SVeaHKqV
# 0ZQ4L6j8cprKXl9Ui+LXFp2YMMXV9g==
# =IjWZ
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 07 Jul 2022 01:42:38 PM +0530
# gpg:                using RSA key 8695A8BFD3F97CDAAC35775A9CA4ABB381AB73C8
# gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>" [full]
# gpg:                 aka "Stefan Hajnoczi <stefanha@gmail.com>" [full]

* tag 'block-pull-request' of https://gitlab.com/stefanha/qemu:
  block/io_uring: clarify that short reads can happen
  io_uring: fix short read slow path

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

block/io_uring: clarify that short reads can happen

Jens Axboe has confirmed that short reads are rare but can happen:
https://lore.kernel.org/io-uring/YsU%2FCGkl9ZXUI+Tj@stefanha-x1.localdomain/T/#m729963dc577d709b709c191922e98ec79d7eef54

The luring_resubmit_short_read() comment claimed they were only due to a
specific io_uring bug that was fixed in Linux commit 9d93a3f5a0c
("io_uring: punt short reads to async context"), which is wrong.
Dominique Martinet found that a btrfs bug also causes short reads. There
may be more kernel code paths that result in short reads.

Let's consider short reads fair game.

Cc: Dominique Martinet <dominique.martinet@atmark-techno.com>
Based-on: <20220630010137.2518851-1-dominique.martinet@atmark-techno.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Message-id: 20220706080341.1206476-1-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

io_uring: fix short read slow path

sqeq.off here is the offset to read within the disk image, so obviously
not 'nread' (the amount we just read), but as the author meant to write
its current value incremented by the amount we just read.

Normally recent versions of linux will not issue short reads,
but it can happen so we should fix this.

This lead to weird image corruptions when short read happened

Fixes: 6663a0a33764 ("block/io_uring: implements interfaces for io_uring")
Link: https://lkml.kernel.org/r/YrrFGO4A1jS0GI0G@atmark-techno.com
Signed-off-by: Dominique Martinet <dominique.martinet@atmark-techno.com>
Message-Id: <20220630010137.2518851-1-dominique.martinet@atmark-techno.com>
Reviewed-by: Hanna Reitz <hreitz@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'pull-ppc-20220706' of https://gitlab.com/danielhb/qemu into staging

ppc patch queue for 2022-07-06:

This queue consists of improvements and bug fixes in TCG, powernv and
pSeries, with some fixes in other areas as well.

- tcg and target/ppc: BCDA and mffscdrn implementations, Remove CONFIG_INT128
conditional code
- fix '-cpu max' alias
- remove '-cpu default' alias
- spapr: fixes in DDW handling, H_WATCHDOG support
- powernv: cleanups in the pnv-phb3/4 models
- fix core type of MPC8555 and MPC8560 models

# -----BEGIN PGP SIGNATURE-----
#
# iHUEABYKAB0WIQQX6/+ZI9AYAK8oOBk82cqW3gMxZAUCYsXrpgAKCRA82cqW3gMx
# ZBe9AP4oqMTFw7r9EQPJU4QFMUeAVABl4o0xNb2wLyYov9CtKAD+LoVERSmtLTJ1
# kFpgBrRTWKVylaLEdZQoTdFlJeBwzQg=
# =GPG1
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 07 Jul 2022 01:38:06 AM +0530
# gpg:                using EDDSA key 17EBFF9923D01800AF2838193CD9CA96DE033164
# gpg: Good signature from "Daniel Henrique Barboza <danielhb413@gmail.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: 17EB FF99 23D0 1800 AF28  3819 3CD9 CA96 DE03 3164

* tag 'pull-ppc-20220706' of https://gitlab.com/danielhb/qemu: (34 commits)
  target/ppc: Fix MPC8555 and MPC8560 core type to e500v1
  target/ppc/cpu-models: Remove the "default" CPU alias
  target/ppc: Return default CPU for max CPU
  target/ppc: implement cdtbcd
  target/ppc: implement cbcdtd
  target/ppc: implement addg6s
  target/ppc: Add flag for ISA v2.06 BCDA instructions
  tests/tcg/ppc64: Add mffsce test
  target/ppc: Implement mffscdrn[i] instructions
  target/ppc: Move mffs[.] to decodetree
  target/ppc: Move mffsl to decodetree
  target/ppc: Move mffsce to decodetree
  target/ppc: Move mffscrn[i] to decodetree
  target/ppc: Fix insn32.decode style issues
  ppc/spapr: Implement H_WATCHDOG
  ppc: Define SETFIELD for the ppc target
  target/ppc: use int128.h methods in vsubcuq
  target/ppc: use int128.h methods in vsubecuq and vsubeuqm
  target/ppc: use int128.h methods in vsubuqm
  target/ppc: use int128.h methods in vaddcuq
  ...

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

target/s390x: Exit tb after executing ex_value

When EXECUTE sets ex_value to interrupt the constructed instruction,
we implicitly disable interrupts so that the value is not corrupted.
Exit to the main loop after execution, so that we re-evaluate any
pending interrupts.

Reported-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220702060228.420454-5-richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

target/s390x: Remove DISAS_PC_STALE_NOCHAIN

Replace this with a flag: exit_to_mainloop.
We can now control the exit for each of DISAS_TOO_MANY,
DISAS_PC_UPDATED, and DISAS_PC_CC_UPDATED, and fold in
the check for PER.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220702060228.420454-4-richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

target/s390x: Remove DISAS_PC_STALE

There is nothing to distinguish this from DISAS_TOO_MANY.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220702060228.420454-3-richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

target/s390x: Remove DISAS_GOTO_TB

There is nothing to distinguish this from DISAS_NORETURN.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220702060228.420454-2-richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

pc-bios/s390-ccw: Update the s390-ccw bios binaries with the virtio-blk fixes

The binaries have been recompiled with the fixes from the previous patches.

Signed-off-by: Thomas Huth <thuth@redhat.com>

pc-bios/s390-ccw/netboot.mak: Ignore Clang's warnings about GNU extensions

When compiling the s390-ccw bios with Clang (v14.0), there is currently
an unuseful warning like this:

  CC      pc-bios/s390-ccw/ipv6.o
 ../../roms/SLOF/lib/libnet/ipv6.c:447:18: warning: variable length array
  folded to constant array as an extension [-Wgnu-folding-constant]
                unsigned short raw[ip6size];
                               ^

SLOF is currently GCC-only and cannot be compiled with Clang yet, so
it is expected that such extensions sneak in there - and as long as
we don't want to compile the code with a compiler that is neither GCC
or Clang, it is also not necessary to avoid such extensions.

Thus these GNU-extension related warnings are completely useless in
the s390-ccw bios, especially in the code that is coming from SLOF,
so we should simply disable the related warnings here now.

Message-Id: <20220704111903.62400-13-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

pc-bios/s390-ccw/virtio: Remove "extern" keyword from prototypes

All the other protytpes in the headers here do not use the "extern"
keyword, so let's unify this by removing the "extern" from the misfits,
too.

Message-Id: <20220704111903.62400-12-thuth@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

pc-bios/s390-ccw/virtio-blkdev: Request the right feature bits

The virtio-blk code uses the block size and geometry fields in the
config area. According to the virtio-spec, these have to be negotiated
with the right feature bits during initialization, otherwise they
might not be available. QEMU is so far very forgiving and always
provides them, but we should not rely on this behavior, so let's
better request them properly via the VIRTIO_BLK_F_GEOMETRY and
VIRTIO_BLK_F_BLK_SIZE feature bits.

Message-Id: <20220704111903.62400-11-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

pc-bios/s390-ccw: Split virtio-scsi code from virtio_blk_setup_device()

The next patch is going to add more virtio-block specific code to
virtio_blk_setup_device(), and if the virtio-scsi code is also in
there, this is more cumbersome. And the calling function virtio_setup()
in main.c looks at the device type already anyway, so it's more
logical to separate the virtio-scsi stuff into a new function in
virtio-scsi.c instead.

Message-Id: <20220704111903.62400-10-thuth@redhat.com>
Reviewed-by: Eric Farman <farman@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

pc-bios/s390-ccw/virtio: Beautify the code for reading virtqueue configuration

It looks nicer if we separate the run_ccw() from the IPL_assert()
statement, and the error message should talk about "virtio device"
instead of "block device", since this code is nowadays used for
non-block (i.e. network) devices, too.

Message-Id: <20220704111903.62400-9-thuth@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Eric Farman <farman@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

pc-bios/s390-ccw/virtio: Read device config after feature negotiation

Feature negotiation should be done first, since some fields in the
config area can depend on the negotiated features and thus should
rather be read afterwards.

While we're at it, also adjust the error message here a little bit
(the code is nowadays used for non-block virtio devices, too).

Message-Id: <20220704111903.62400-8-thuth@redhat.com>
Reviewed-by: Eric Farman <farman@linux.ibm.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

pc-bios/s390-ccw/virtio: Set missing status bits while initializing

According chapter "3.1.1 Driver Requirements: Device Initialization"
of the Virtio specification (v1.1), a driver for a device has to set
the ACKNOWLEDGE and DRIVER bits in the status field after resetting
the device. The s390-ccw bios skipped these steps so far and seems
like QEMU never cared. Anyway, it's better to follow the spec, so
let's set these bits now in the right spots, too.

Message-Id: <20220704111903.62400-7-thuth@redhat.com>
Acked-by: Christian Borntraeger <borntraeger@linux.ibm.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Eric Farman <farman@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

pc-bios/s390-ccw/virtio-blkdev: Remove virtio_assume_scsi()

The virtio_assume_scsi() function is very questionable: First, it
is only called for virtio-blk, and not for virtio-scsi, so the naming
is already quite confusing. Second, it is called if we detected a
"invalid" IPL disk, trying to fix it by blindly setting a sector
size of 512. This of course won't work in most cases since disks
might have a different sector size for a reason.

Thus let's remove this strange function now. The calling code can
also be removed completely, since there is another spot in main.c
that does "IPL_assert(virtio_ipl_disk_is_valid(), ...)" to make
sure that we do not try to IPL from an invalid device.

Message-Id: <20220704111903.62400-6-thuth@redhat.com>
Reviewed-by: Eric Farman <farman@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

pc-bios/s390-ccw/virtio-blkdev: Simplify/fix virtio_ipl_disk_is_valid()

The s390-ccw bios fails to boot if the boot disk is a virtio-blk
disk with a sector size of 4096. For example:

 dasdfmt -b 4096 -d cdl -y -p -M quick /dev/dasdX
 fdasd -a /dev/dasdX
 install a guest onto /dev/dasdX1 using virtio-blk
 qemu-system-s390x -nographic -hda /dev/dasdX1

The bios then bails out with:

 ! Cannot read block 0 !

Looking at virtio_ipl_disk_is_valid() and especially the function
virtio_disk_is_scsi(), it does not really make sense that we expect
only such a limited disk geometry (like a block size of 512) for
our boot disks. Let's relax the check and allow everything that
remotely looks like a sane disk.

Message-Id: <20220704111903.62400-5-thuth@redhat.com>
Reviewed-by: Eric Farman <farman@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

pc-bios/s390-ccw/bootmap: Improve the guessing logic in zipl_load_vblk()

The logic of trying an final ISO or ECKD boot on virtio-block devices is
very weird: Since the geometry hardly ever matches in virtio_disk_is_scsi(),
virtio_blk_setup_device() always sets a "guessed" disk geometry via
virtio_assume_scsi() (which is certainly also wrong in a lot of cases).

zipl_load_vblk() then sees that there's been a "virtio_guessed_disk_nature"
and tries to fix up the geometry again via virtio_assume_iso9660() before
always trying to do ipl_iso_el_torito(). That's a very brain-twisting
way of attempting to boot from ISO images, which won't work anymore after
the following patches that will clean up the virtio_assume_scsi() mess
(and thus get rid of the "virtio_guessed_disk_nature" here).

Let's try a better approach instead: ISO files always have a magic
string "CD001" at offset 0x8001 (see e.g. the ECMA-119 specification)
which we can use to decide whether we should try to boot in ISO 9660
mode (which we should also try if we see a sector size of 2048).

And if we were not able to boot in ISO mode here, the final boot attempt
before panicking is to boot in ECKD mode. Since this is our last boot
attempt anyway, simply always assume the ECKD geometry here (if the sector
size was not 4096 yet), so that we also do not depend on the guessed disk
geometry from virtio_blk_setup_device() here anymore.

Message-Id: <20220704111903.62400-4-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

pc-bios/s390-ccw/virtio: Introduce a macro for the DASD block size

Use VIRTIO_DASD_DEFAULT_BLOCK_SIZE instead of the magic value 4096.

Message-Id: <20220704111903.62400-3-thuth@redhat.com>
Reviewed-by: Eric Farman <farman@linux.ibm.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

pc-bios/s390-ccw: Add a proper prototype for main()

Older versions of Clang complain if there is no prototype for main().
Add one, and while we're at it, make sure that we use the same type
for main.c and netmain.c - since the return value does not matter,
declare the return type of main() as "void".

Message-Id: <20220704111903.62400-2-thuth@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Eric Farman <farman@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

Merge tag 'm68k-for-7.1-pull-request' of https://github.com/vivier/qemu-m68k into staging

Pull request m68k 20220706

pass RNG seed via bootinfo block

# -----BEGIN PGP SIGNATURE-----
#
# iQJGBAABCAAwFiEEzS913cjjpNwuT1Fz8ww4vT8vvjwFAmLFdGkSHGxhdXJlbnRA
# dml2aWVyLmV1AAoJEPMMOL0/L748oSUP/1W3bsPkFTGi0F77RuVFo8jBdvnCznH+
# mh/9FJdILuOla2Z+N+Y8dBErp9ZPk0FNAOFboZyX6+UzS6G64nLeIqEKIK+gfoWA
# qImxlCnip3d0tS0/od+x5+wOYYXqTqerU++i/J/bACMjciAqbUB3GvZvmP32ixma
# cqkOo0jHawRx9k0tzuWcP1zgmGHfHIJQEu/GdBUKeTYIQ7P0ACK0qIijndb4kZYO
# OYuXF4TNncojzqJf6wOpvYbpifmpL2RkJmhHE2LLgIHeuZxV+9AdFmhbp7/D1kCw
# 64qRn2oQHxuHQWL87ZOUXGmARcecW58uwLgUrkXIR8ZjHu8Y4KCLfnDrVHzDBfjE
# AA8Jgbo07ASuGnc/DR9EieZNBFZT7sMK9VccUCI78G3ExR7xL0JkRGqc7nnX5WK/
# JDf9dDNVqlEbVJ5JUbQQBl80bTRf9toHlfbwbwDd8Ow5x/R1e62lXIukx4xo+oTW
# DhjYoGqF/grcUQop50jS9vU9BMICPqmUp+4ueoj08kzAa96Z5D7ptXVY5WZFlk/W
# DN5D1MYFflxt/FmSDBUpT/9u1GUYOmod3dt69x8Jk5nSkKjQECIed1kxndqIKhQI
# zmeo3MKgf2HEJ0tobOe9UzD24DrvbK9g5IPFQTZDu/6u1IzNUgGi/YZUy7N1Owup
# PG+ez9RwhSzV
# =lR7N
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 06 Jul 2022 05:09:21 PM +0530
# gpg:                using RSA key CD2F75DDC8E3A4DC2E4F5173F30C38BD3F2FBE3C
# gpg:                issuer "laurent@vivier.eu"
# gpg: Good signature from "Laurent Vivier <lvivier@redhat.com>" [undefined]
# gpg:                 aka "Laurent Vivier <laurent@vivier.eu>" [undefined]
# gpg:                 aka "Laurent Vivier (Red Hat) <lvivier@redhat.com>" [undefined]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: CD2F 75DD C8E3 A4DC 2E4F  5173 F30C 38BD 3F2F BE3C

* tag 'm68k-for-7.1-pull-request' of https://github.com/vivier/qemu-m68k:
  m68k: virt: pass RNG seed via bootinfo block
  m68k: use correct variable name in boot info string macro

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

target/ppc: Fix MPC8555 and MPC8560 core type to e500v1

Commit 80d11f4467c4 ("Add definitions for Freescale PowerPC implementations")
changed core type of MPC8555 and MPC8560 from e500v1 to e500v2.

But both MPC8555 and MPC8560 have just e500v1 cores, there are no features
of e500v2 cores. It can be verified by reading NXP documentations:
https://www.nxp.com/docs/en/data-sheet/MPC8555EEC.pdf
https://www.nxp.com/docs/en/data-sheet/MPC8560EC.pdf
https://www.nxp.com/docs/en/reference-manual/MPC8555ERM.pdf
https://www.nxp.com/docs/en/reference-manual/MPC8560RM.pdf

Therefore fix core type of MPC8555 and MPC8560 back to e500v1.

Just for completeness, here is list of all Motorola/Freescale/NXP
processors which were released and have e500v1 or e500v2 cores:

e500v1: MPC8540 MPC8541 MPC8555 MPC8560

e500v2: BSC9131 BSC9132
        C291 C292 C293
        MPC8533 MPC8535 MPC8536 MPC8543 MPC8544 MPC8545 MPC8547
        MPC8548 MPC8567 MPC8568 MPC8569 MPC8572
        P1010 P1011 P1012 P1013 P1014 P1015 P1016 P1020 P1021
        P1022 P1024 P1025 P2010 P2020

Sorted alphabetically; not by release date / generation / feature set.
All this is from public information available on NXP website.

Seems that qemu has support only for some subset of MPC85xx processors.
Historically processors with e500 cores have mpc85xx family codename and
lot of software have them in mpc85xx architecture subdirectory.

Note that GCC uses -mcpu=8540 option for specifying e500v1 core and
-mcpu=8548 option for specifying e500v2 core.

So sometimes (mpc)8540 is alias for e500v1 and (mpc)8548 is alias for
e500v2.

Fixes: 80d11f4467c4 ("Add definitions for Freescale PowerPC implementations")
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220703195029.23793-1-pali@kernel.org>
[danielhb: added more context in the commit msg]
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc/cpu-models: Remove the "default" CPU alias

QEMU emulates a *lot* of PowerPC-based machines - having a CPU
that is named "default" and cannot be used with most of those
machines sounds just wrong. Thus let's remove this old and confusing
alias now.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220705151030.662140-1-thuth@redhat.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Return default CPU for max CPU

All ppc CPUs represent hardware that exists in the real world, i.e.: we
do not have a "max" CPU with all possible emulated features enabled.
Return the default CPU type for the machine because that has greater
chance of being useful as the "max" CPU.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1038
Cc: Cédric Le Goater <clg@kaod.org>
Cc: Daniel Henrique Barboza <danielhb413@gmail.com>
Cc: Daniel P. Berrangé <berrange@redhat.com>
Cc: Greg Kurz <groug@kaod.org>
Cc: Matheus K. Ferst <matheus.ferst@eldorado.org.br>
Cc: Thomas Huth <thuth@redhat.com>
Signed-off-by: Murilo Opsfelder Araujo <muriloo@linux.ibm.com>
Signed-off-by: Fabiano Rosas <farosas@linux.ibm.com>
Reviewed-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Message-Id: <20220628205513.81917-1-muriloo@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: implement cdtbcd

Implements the Convert Declets To Binary Coded Decimal instruction.
Since libdecnumber doesn't expose the methods for direct conversion
(decDigitsFromDPD, DPD2BCD, etc), a positive decimal32 with zero
exponent is used as an intermediate value to convert the declets.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Signed-off-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Message-Id: <20220629162904.105060-12-victor.colombo@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: implement cbcdtd

Implements the Convert Binary Coded Decimal To Declets instruction.
Since libdecnumber doesn't expose the methods for direct conversion
(decDigitsToDPD, BCD2DPD, etc.), the BCD values are converted to
decimal32 format, from which the declets are extracted.

Where the behavior is undefined, we try to match the result observed in
a POWER9 DD2.3.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Signed-off-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Message-Id: <20220629162904.105060-11-victor.colombo@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: implement addg6s

Implements the following Power ISA v2.06 instruction:
addg6s: Add and Generate Sixes

Signed-off-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Signed-off-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Reviewed-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Message-Id: <20220629162904.105060-10-victor.colombo@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Add flag for ISA v2.06 BCDA instructions

Adds an insns_flags2 for the BCD assist instructions introduced in
Power ISA 2.06. These instructions are not listed in the manuals for
e5500[1] and e6500[2], so the flag is only added for POWER7/8/9/10
models.

[1] https://www.nxp.com/files-static/32bit/doc/ref_manual/EREF_RM.pdf
[2] https://www.nxp.com/docs/en/reference-manual/E6500RM.pdf

Signed-off-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Signed-off-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220629162904.105060-9-victor.colombo@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

tests/tcg/ppc64: Add mffsce test

Add mffsce test to check both the return value and the new fpscr
stored in the cpu.

Signed-off-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Reviewed-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Message-Id: <20220629162904.105060-8-victor.colombo@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Implement mffscdrn[i] instructions

Signed-off-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Reviewed-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Message-Id: <20220629162904.105060-7-victor.colombo@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Move mffs[.] to decodetree

Signed-off-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Reviewed-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Message-Id: <20220629162904.105060-6-victor.colombo@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Move mffsl to decodetree

Signed-off-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Reviewed-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Message-Id: <20220629162904.105060-5-victor.colombo@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Move mffsce to decodetree

Signed-off-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Reviewed-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Message-Id: <20220629162904.105060-4-victor.colombo@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Move mffscrn[i] to decodetree

Signed-off-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Reviewed-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Message-Id: <20220629162904.105060-3-victor.colombo@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Fix insn32.decode style issues

Some lines in insn32.decode have inconsistent alignment when compared
to others.
Fix this by changing the alignment of some lines, making it more
consistent throughout the file.

Signed-off-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220629162904.105060-2-victor.colombo@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

ppc/spapr: Implement H_WATCHDOG

The new PAPR 2.12 defines a watchdog facility managed via the new
H_WATCHDOG hypercall.

This adds H_WATCHDOG support which a proposed driver for pseries uses:
https://patchwork.ozlabs.org/project/linuxppc-dev/list/?series=303120

This was tested by running QEMU with a debug kernel and command line:
-append \
 "pseries-wdt.timeout=60 pseries-wdt.nowayout=1 pseries-wdt.action=2"

and running "echo V > /dev/watchdog0" inside the VM.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220622051008.1067464-1-aik@ozlabs.ru>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

ppc: Define SETFIELD for the ppc target

It keeps repeating, move it to the header. This uses __builtin_ffsll() to
allow using the macros in #define.

This is not using the QEMU's FIELD macros as this would require changing
all such macros found in skiboot (the PPC PowerNV firmware).

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220628080544.1509428-1-aik@ozlabs.ru>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: use int128.h methods in vsubcuq

And also move the insn to decodetree and remove the now unused
avr_qw_not, avr_qw_cmpu, and avr_qw_add methods.

Signed-off-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Reviewed-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Message-Id: <20220606150037.338931-8-matheus.ferst@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: use int128.h methods in vsubecuq and vsubeuqm

And also move the insns to decodetree.

Signed-off-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Reviewed-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Message-Id: <20220606150037.338931-7-matheus.ferst@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: use int128.h methods in vsubuqm

And also move the insn to decodetree

Signed-off-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Reviewed-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Message-Id: <20220606150037.338931-6-matheus.ferst@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: use int128.h methods in vaddcuq

And also move the insn to decodetree.

Signed-off-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Reviewed-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Message-Id: <20220606150037.338931-5-matheus.ferst@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: use int128.h methods in vaddecuq and vaddeuqm

And also move the insns to decodetree and remove the now unused
avr_qw_addc method.

Signed-off-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Reviewed-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Message-Id: <20220606150037.338931-4-matheus.ferst@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: use int128.h methods in vadduqm

And also move the insn to decodetree.

Signed-off-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Reviewed-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Message-Id: <20220606150037.338931-3-matheus.ferst@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: use int128.h methods in vpmsumd

Also drop VECTOR_FOR_INORDER_I usage since there is no need to access
the elements in any particular order, and move the instruction to
decodetree.

Signed-off-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Reviewed-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Message-Id: <20220606150037.338931-2-matheus.ferst@eldorado.org.br>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

spapr/ddw: Implement 64bit query extension

PAPR 2.8 (2018) defines an extension to return 64bit value for
the largest TCE block in "ibm,query-pe-dma-window". Recent Linux kernels
support this already.

This adds the extension and supports the older format.

This advertises a bigger window for the new format as the biggest
window with 2M pages below the start of the 64bit window as it is
the maximum we will see in practice.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220623073136.1380214-1-aik@ozlabs.ru>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

spapr/ddw: Reset DMA when the last non-default window is removed

PAPR+/LoPAPR says:
===
The platform must restore the default DMA window for the PE on a call
to the ibm,remove-pe-dma-window RTAS call when all of the following
are true:
 a. The call removes the last DMA window remaining for the PE.
 b. The DMA window being removed is not the default window

===

This resets DMA as PAPR mandates.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220622052955.1069903-1-aik@ozlabs.ru>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Change FPSCR_* to follow POWER ISA numbering convention

FPSCR_* bit values in QEMU are in the 'inverted' order from what Power
ISA defines (e.g. FPSCR.FI is bit 46 but is defined as 17 in cpu.h).
Now that PPC_BIT_NR macro was introduced to fix this situation for the
MSR bits, we can use it for the FPSCR bits too.

Also, adjust the comments to make then fit in 80 columns

Signed-off-by: Víctor Colombo <victor.colombo@eldorado.org.br>
Reviewed-by: Fabiano Rosas <farosas@linux.ibm.com>
Message-Id: <20220622193203.127698-1-victor.colombo@eldorado.org.br>
[danielhb: fixed 'exceptio' typo in target/ppc/cpu.h]
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

ppc/pnv: remove 'INTERFACE_PCIE_DEVICE' from phb4 root bus

It's unneeded. No other PCIE_BUS implements this interface.

Reviewed-by: Frederic Barrat <fbarrat@linux.ibm.com>
Fixes: 4f9924c4d4cf ("ppc/pnv: Add models for POWER9 PHB4 PCIe Host bridge")
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220621173436.165912-9-danielhb413@gmail.com>

ppc/pnv: remove 'INTERFACE_PCIE_DEVICE' from phb3 root bus

It's unneeded. No other PCIE_BUS implements this interface.

Reviewed-by: Frederic Barrat <fbarrat@linux.ibm.com>
Fixes: 9ae1329ee2fe ("ppc/pnv: Add models for POWER8 PHB3 PCIe Host bridge")
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220621173436.165912-8-danielhb413@gmail.com>

ppc/pnv: make pnv_chip_power8_pic_print_info() use chip8->phbs[]

It's inneficient to scroll all child objects when we have all PHBs
available in chip8->phbs[].

pnv_chip_power8_pic_print_info_child() ended up folded into
pic_print_info() for simplicity.

Reviewed-by: Frederic Barrat <fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220621173436.165912-7-danielhb413@gmail.com>

ppc/pnv: make pnv_ics_resend() use chip8->phbs[]

pnv_ics_resend() is scrolling through all the child objects of the chip
to search for the PHBs. It's faster and simpler to just use the phbs[]
array.

pnv_ics_resend_child() was folded into pnv_ics_resend() since it's too
simple to justify its own function.

Reviewed-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Frederic Barrat <fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220621173436.165912-6-danielhb413@gmail.com>

ppc/pnv: make pnv_ics_get() use the chip8->phbs[] array

The function is working today by getting all the child objects of the
chip, interacting with each of them to check whether the child is a PHB,
and then doing what needs to be done.

We have all the chip PHBs in the phbs[] array so interacting with all
child objects is unneeded. Open code pnv_ics_get_phb_ics() into
pnv_ics_get() and remove both pnv_ics_get_phb_ics() and the
ForeachPhb3Args struct.

Reviewed-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Frederic Barrat <fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220621173436.165912-5-danielhb413@gmail.com>

ppc/pnv: assign pnv-phb-root-port chassis/slot earlier

It is not advisable to execute an object_dynamic_cast() to poke into
bus->qbus.parent and follow it up with a C cast into the PnvPHB type we
think we got.

In fact this is not needed. There is nothing sophisticated being done
with the PHB object retrieved during root_port_realize() for both PHB3
and PHB4. We're retrieving a PHB reference just to access phb->chip_id
and phb->phb_id and use them to define the chassis/slot of the root
port.

phb->phb_id is already being passed to pnv_phb_attach_root_port() via
the 'index' parameter. Let's also add a 'chip_id' parameter to this
function and assign chassis and slot right there. This will spare us
from the hassle of accessing the PHB object inside realize().

Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Frederic Barrat <fbarrat@linux.ibm.com>
Message-Id: <20220621173436.165912-4-danielhb413@gmail.com>

ppc/pnv: attach phb3/phb4 root ports in QOM tree

At this moment we leave the pnv-phb3(4)-root-port unattached in QOM:

  /unattached (container)
(...)
    /device[2] (pnv-phb3-root-port)
      /bus master container[0] (memory-region)
      /bus master[0] (memory-region)
      /pci_bridge_io[0] (memory-region)
      /pci_bridge_io[1] (memory-region)
      /pci_bridge_mem[0] (memory-region)
      /pci_bridge_pci[0] (memory-region)
      /pci_bridge_pref_mem[0] (memory-region)
      /pci_bridge_vga_io_hi[0] (memory-region)
      /pci_bridge_vga_io_lo[0] (memory-region)
      /pci_bridge_vga_mem[0] (memory-region)
      /pcie.0 (PCIE)

Let's make changes in pnv_phb_attach_root_port() to attach the created
root ports to its corresponding PHB.

This is the result afterwards:

    /pnv-phb3[0] (pnv-phb3)
      /lsi (ics)
      /msi (phb3-msi)
      /msi32[0] (memory-region)
      /msi64[0] (memory-region)
      /pbcq (pnv-pbcq)
    (...)
      /phb3_iommu[0] (pnv-phb3-iommu-memory-region)
      /pnv-phb3-root.0 (pnv-phb3-root)
        /pnv-phb3-root-port[0] (pnv-phb3-root-port)
          /bus master container[0] (memory-region)
          /bus master[0] (memory-region)
          /pci_bridge_io[0] (memory-region)
          /pci_bridge_io[1] (memory-region)
          /pci_bridge_mem[0] (memory-region)
          /pci_bridge_pci[0] (memory-region)
          /pci_bridge_pref_mem[0] (memory-region)
          /pci_bridge_vga_io_hi[0] (memory-region)
          /pci_bridge_vga_io_lo[0] (memory-region)
          /pci_bridge_vga_mem[0] (memory-region)
          /pcie.0 (PCIE)

Reviewed-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220621173436.165912-3-danielhb413@gmail.com>

ppc/pnv: move root port attach to pnv_phb4_realize()

Creating a root port is something related to the PHB, not the PEC. It
also makes the logic more in line with what pnv-phb3 does.

Reviewed-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220621173436.165912-2-danielhb413@gmail.com>

m68k: virt: pass RNG seed via bootinfo block

This commit wires up bootinfo's RNG seed attribute so that Linux VMs can
have their RNG seeded from the earliest possible time in boot, just like
the "rng-seed" device tree property on those platforms. The link
contains the corresponding Linux patch.

Link: https://lore.kernel.org/lkml/20220626111509.330159-1-Jason@zx2c4.com/
Based-on: <20220625152318.120849-1-Jason@zx2c4.com>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Reviewed-by: Geert Uytterhoeven <geert@linux-m68k.org>
Message-Id: <20220626111804.330745-1-Jason@zx2c4.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

m68k: use correct variable name in boot info string macro

Every time this macro is used, the caller is passing in
"parameters_base", so this bug wasn't spotted. But the actual macro
variable name is "base", so use that instead.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20220625152318.120849-1-Jason@zx2c4.com>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

target/s390x/tcg: SPX: check validity of new prefix

According to the architecture, SET PREFIX must try to access the new
prefix area and recognize an addressing exception if the area is not
accessible.
For qemu this check prevents a crash in cpu_map_lowcore after an
inaccessible prefix area has been set.

Signed-off-by: Janis Schoetterl-Glausch <scgl@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20220630094340.3646279-1-scgl@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

Merge tag 'net-pull-request' of https://github.com/jasowang/qemu into staging

# -----BEGIN PGP SIGNATURE-----
# Version: GnuPG v1
#
# iQEcBAABAgAGBQJixQP1AAoJEO8Ells5jWIRAsEH/jDlZa1m8jM7vGn6VVUhC6gR
# wnwMZVVrg/N+P6614dQ6jFEhTL52AhNONCwl+nzKZDksWMtkX/Owy3DEx0HqXk3m
# DyQyI5ngBiRPOOOlgyVHMyc304TFmf9qh4Cz80o5UNsI50lpLMA/vpwgK3zz/KXZ
# V6d5C3+kL2y7l5dmHuASiUD8xRTQvDn3/mY94VNzwU36uqODwAizm6Ft6ohTkvdO
# WtPQ9AXiVUyHigCxX4r7l9vZjBw0sobVYS1TEoagB2wiEpwTnOYbPhGntBP1eMzv
# E1Wf2tJbz67SP+DHuUCrrUyFaDCVLmG7IAh44wIoK4XCGqfrjTxwUftvo1XZ6Gk=
# =yqUz
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 06 Jul 2022 09:09:33 AM +0530
# gpg:                using RSA key EF04965B398D6211
# gpg: Good signature from "Jason Wang (Jason Wang on RedHat) <jasowang@redhat.com>" [undefined]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 215D 46F4 8246 689E C77F  3562 EF04 965B 398D 6211

* tag 'net-pull-request' of https://github.com/jasowang/qemu:
  ebpf: replace deprecated bpf_program__set_socket_filter
  e1000: set RX descriptor status in a separate operation

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

ebpf: replace deprecated bpf_program__set_socket_filter

bpf_program__set_<TYPE> functions have been deprecated since libbpf 0.8.
Replace with the equivalent bpf_program__set_type call to avoid a
deprecation warning.

Signed-off-by: Haochen Tong <i@hexchain.org>
Reviewed-by: Zhang Chen <chen.zhang@intel.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>

e1000: set RX descriptor status in a separate operation

The code of setting RX descriptor status field maybe work fine in
previously, however with the update of glibc version, it shows two
issues when guest using dpdk receive packets:

  1. The dpdk has a certain probability getting wrong buffer_addr

     this impact may be not obvious, such as lost a packet once in
     a while

  2. The dpdk may consume a packet twice when scan the RX desc queue
     over again

     this impact will lead a infinite wait in Qemu, since the RDT
     (tail pointer) be inscreased to equal to RDH by unexpected,
     which regard as the RX desc queue is full

Write a whole of RX desc with DD flag on is not quite correct, because
when the underlying implementation of memcpy using XMM registers to
copy e1000_rx_desc (when AVX or something else CPU feature is usable),
the bytes order of desc writing to memory is indeterminacy

We can use full-scale test case to reproduce the issue-2 by
https://github.com/BASM/qemu_dpdk_e1000_test (thanks to Leonid Myravjev)

I also write a POC test case at https://github.com/cdkey/e1000_poc
which can reproduce both of them, and easy to verify the patch effect.

The hw watchpoint also shows that, when Qemu using XMM related instructions
writing 16 bytes e1000_rx_desc, concurrent with DPDK using movb
writing 1 byte status, the final result of writing to memory will be one
of them, if it made by Qemu which DD flag is on, DPDK will consume it
again.

Setting DD status in a separate operation, can prevent the impact of
disorder memory writing by memcpy, also avoid unexpected data when
concurrent writing status by qemu and guest dpdk.

Links: https://lore.kernel.org/qemu-devel/20200102110504.GG121208@stefanha-x1.localdomain/T/

Reported-by: Leonid Myravjev <asm@asm.pp.ru>
Cc: Stefan Hajnoczi <stefanha@gmail.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: qemu-stable@nongnu.org
Tested-by: Jing Zhang <zhangjing@sangfor.com.cn>
Reviewed-by: Frank Lee <lifan38153@sangfor.com.cn>
Signed-off-by: Ding Hui <dinghui@sangfor.com.cn>
Signed-off-by: Jason Wang <jasowang@redhat.com>

Merge tag 'pull-xen-20220705' of https://xenbits.xen.org/git-http/people/aperard/qemu-dm into staging

Xen patches

- Xen PCI passthrough fixes

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEE+AwAYwjiLP2KkueYDPVXL9f7Va8FAmLEQFMACgkQDPVXL9f7
# Va9K8ggArmxiJ+g+RQtY8/D7MXjDTA3wfUFWytCAIxarbmxR3eVf5N3DnrP8Yak2
# BX3UmCuZuwedkU11ZXKaMa9PM/tQOpZTWwai2tDuxw8AwjpW3iG4V1jXJ58BXZsM
# /81p1/1NqJhGacY1agdf4Yd0uJW+atxbo8/ejLOpmA8S9pHGDMRvSkRezrrOyQEG
# z8V9l1ZiAWeJCny98blhfY+RyJ7vj27xf0gBli/kpDwwm/wnxcD6geT3U7hNAwT6
# gcw22u3cHHf9svGUP+X2Y4XpJI5hv4ZoKwVG5ogEkyPiT4eY3ApRzYSAwMtQesqR
# rLMpkOEwiGfSJzfHAtdHVconx7Q/fw==
# =lPaS
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 05 Jul 2022 07:14:51 PM +0530
# gpg:                using RSA key F80C006308E22CFD8A92E7980CF5572FD7FB55AF
# gpg: Good signature from "Anthony PERARD <anthony.perard@gmail.com>" [unknown]
# gpg:                 aka "Anthony PERARD <anthony.perard@citrix.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: 5379 2F71 024C 600F 778A  7161 D8D5 7199 DF83 42C8
#      Subkey fingerprint: F80C 0063 08E2 2CFD 8A92  E798 0CF5 572F D7FB 55AF

* tag 'pull-xen-20220705' of https://xenbits.xen.org/git-http/people/aperard/qemu-dm:
  xen/pass-through: don't create needless register group
  xen/pass-through: merge emulated bits correctly

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

xen/pass-through: don't create needless register group

Currently we are creating a register group for the Intel IGD OpRegion
for every device we pass through, but the XEN_PCI_INTEL_OPREGION
register group is only valid for an Intel IGD. Add a check to make
sure the device is an Intel IGD and a check that the administrator has
enabled gfx_passthru in the xl domain configuration. Require both checks
to be true before creating the register group. Use the existing
is_igd_vga_passthrough() function to check for a graphics device from
any vendor and that the administrator enabled gfx_passthru in the xl
domain configuration, but further require that the vendor be Intel,
because only Intel IGD devices have an Intel OpRegion. These are the
same checks hvmloader and libxl do to determine if the Intel OpRegion
needs to be mapped into the guest's memory. Also, move the comment
about trapping 0xfc for the Intel OpRegion where it belongs after
applying this patch.

Signed-off-by: Chuck Zmudzinski <brchuckz@aol.com>
Reviewed-by: Anthony PERARD <anthony.perard@citrix.com>
Message-Id: <c76dff6369ccf2256bd9eed5141da1db767293d2.1656480662.git.brchuckz@aol.com>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

xen/pass-through: merge emulated bits correctly

In xen_pt_config_reg_init(), there is an error in the merging of the
emulated data with the host value. With the current Qemu, instead of
merging the emulated bits with the host bits as defined by emu_mask,
the emulated bits are merged with the host bits as defined by the
inverse of emu_mask. In some cases, depending on the data in the
registers on the host, the way the registers are setup, and the
initial values of the emulated bits, the end result will be that
the register is initialized with the wrong value.

To correct this error, use the XEN_PT_MERGE_VALUE macro to help ensure
the merge is done correctly.

This correction is needed to resolve Qemu project issue #1061, which
describes the failure of Xen HVM Linux guests to boot in certain
configurations with passed through PCI devices, that is, when this error
disables instead of enables the PCI_STATUS_CAP_LIST bit of the
PCI_STATUS register of a passed through PCI device, which in turn
disables the MSI-X capability of the device in Linux guests with the end
result being that the Linux guest never completes the boot process.

Fixes: 2e87512eccf3 ("xen/pt: Sync up the dev.config and data values")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1061
Buglink: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=988333
Signed-off-by: Chuck Zmudzinski <brchuckz@aol.com>
Reviewed-by: Anthony PERARD <anthony.perard@citrix.com>
Message-Id: <e4392535d8e5266063dc5461d0f1d301e3dd5951.1656522217.git.brchuckz@aol.com>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

Merge tag 'pull-request-2022-07-05' of https://gitlab.com/thuth/qemu into staging

* Fix memory leak in test-cutils
* Fix edk2/opensbi jobs to not run automatically by accident
* Improve timings in the migration qtest
* Remove libvixl disassembler
* Add ukrainian translation
* Require a recent version of libpng

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmLECEkRHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbV7lxAAmEItM6PIoW58eWPzReKVH8LE2w3UlvOZ
# JQhNgJjuN23fqjUVkcT0yCfdNCz/nKvafHnxfHQnrAXyB5V5vU8ovBgSuWK2mcmD
# NTFK+/2x5lcsyBrOe3QoeD2g1r7+Os3AYVkdnN/t2HAMLwQyaoshKaMV/UHC9O/i
# Kle1svYRNyCgyXJgxaOdbVMBSLi/L9h2R5AaG31GIi9wnf0n8HDH/ONtmeIpN09g
# BlMeZqPhGJT+tpMvviif65/Za57Y9h/r+TOgEIIs00cWmxqaBmcXXN9qog2s0n7A
# nOm3ck2lpGJCQ6+sl6/Mphyr3X6nWHsxGrLDElS0Ba5bg6T/Xqfg2pBcb81Klkjc
# QcTdFPiMxKUczgpFq326sqiaVzMgys4vwnW5iPSd5swNzrkYKADAIreki5jyM3cH
# lohBG/ruOmg5xMkX2K6pra0iOAeCz44Ku/HTREfY1CTUgEQZJY4SZrMJSnmUTnM+
# EQCkDcmOsnFDaQazneCbo18l37cXOgEhH8VoGAOqg1aRjr7TNlsJzx87PoD+9zNR
# GEh7kp18ABRGik5ZACdLQ/HhhOJa8+UWsGCwCdeBGv/TVug1Byz0OUG0PxX3X5SV
# WwubeKyZcqzoH92SQI3jZGSmuGBySy9q51T2k8FjZvaDsPiUN/MLPspNezH1qj2B
# W7qEaqIyGmo=
# =Q2vV
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 05 Jul 2022 03:15:45 PM +0530
# gpg:                using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5
# gpg:                issuer "thuth@redhat.com"
# gpg: Good signature from "Thomas Huth <th.huth@gmx.de>" [undefined]
# gpg:                 aka "Thomas Huth <thuth@redhat.com>" [undefined]
# gpg:                 aka "Thomas Huth <th.huth@posteo.de>" [unknown]
# gpg:                 aka "Thomas Huth <huth@tuxfamily.org>" [undefined]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 27B8 8847 EEE0 2501 18F3  EAB9 2ED9 D774 FE70 2DB5

* tag 'pull-request-2022-07-05' of https://gitlab.com/thuth/qemu:
  include/qemu/host-utils: Remove unused code in the *_overflow wrappers
  meson.build: Require a recent version of libpng
  po: add ukrainian translation
  disas: Remove libvixl disassembler
  tests: use consistent bandwidth/downtime limits in migration tests
  tests: increase migration test converge downtime to 30 seconds
  tests: wait for migration completion before looking for STOP event
  tests: wait max 120 seconds for migration test status changes
  gitlab-ci: Extend timeout for ubuntu-20.04-s390x-all to 75m
  gitlab: honour QEMU_CI variable in edk2/opensbi jobs
  gitlab: tweak comments in edk2/opensbi jobs
  gitlab: normalize indentation in edk2/opensbi rules
  tests/fp: Do not build softfloat3 tests if TCG is disabled
  tests: fix test-cutils leaks

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

Merge tag 'pull-la-20220705' of https://gitlab.com/rth7680/qemu into staging

Loongarch patch queue:

Build fix for --enable-debug --enable-tcg-interpreter.
Build fix for ls7a_rtc.
Clear tlb on reset.
Fixes for ipi mailboxes.
Minor tweak to scripts/qemu-binfmt-conf.

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmLEGVIdHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV+SYAgAqMk+GHMT6VQANEsk
# So58d9WCPG0XSavowl9oD4w/YSSvPZe5P4KVpJbC3WAgVwEI0RRKTX3RMAeg5z0I
# zEEzFUSplSl7cO/7vQG86JRf5C7C/n4V9Q1pQUstNnTEf1s7MdgcG9597OZbV+cF
# G5KY1RTQRUr6gpChZQSrv+6j6+aQCA5ZgNwjiVnkBjsNefz1GVFKYppanwHXmMiX
# qjxVLgZb1FwOysiKpHKObLsC9pV7ub0QKrlBBk90UyidNjXxcLvV+oQrkyaVwB0m
# UM/NN/x5Ive2dqEDfq007TXUc1RpFgwDvKU8EffavBYxx8hCed3DysroiYN+v2MK
# qoYWmg==
# =+zUy
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 05 Jul 2022 04:28:26 PM +0530
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [ultimate]

* tag 'pull-la-20220705' of https://gitlab.com/rth7680/qemu:
  hw/intc/loongarch_ipi: Fix mail send and any send function
  hw/intc/loongarch_ipi: Fix ipi device access of 64bits
  tcg/tci: Remove CONFIG_DEBUG_TCG_INTERPRETER
  scripts/qemu-binfmt-conf: Add LoongArch to qemu_get_family()
  target/loongarch: Clean up tlb when cpu reset
  hw/rtc/ls7a_rtc: Drop unused inline functions

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

hw/intc/loongarch_ipi: Fix mail send and any send function

By the document of ipi mailsend device, byte is written only when the mask bit
is 0. The original code discards mask bit and overwrite the data always, this
patch fixes the issue.

Signed-off-by: Xiaojuan Yang <yangxiaojuan@loongson.cn>
Message-Id: <20220705064901.2353349-3-yangxiaojuan@loongson.cn>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

hw/intc/loongarch_ipi: Fix ipi device access of 64bits

In general loongarch ipi device, 32bit registers is emulated, however for
anysend/mailsend device only 64bit register access is supported. So separate
the ipi memory region into two regions, including 32 bits and 64 bits.

Signed-off-by: Xiaojuan Yang <yangxiaojuan@loongson.cn>
Message-Id: <20220705064901.2353349-2-yangxiaojuan@loongson.cn>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/tci: Remove CONFIG_DEBUG_TCG_INTERPRETER

There is nothing in this environment variable that cannot
be done better with -d flags.  There is nothing special
about TCI that warrants this hack.

Moreover, it does not compile -- remove it.

Reported-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

scripts/qemu-binfmt-conf: Add LoongArch to qemu_get_family()

qemu_get_family() needs to add LoongArch support.

Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20220705065943.2353930-1-gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>