target/ppc: Fix decrementer time underflow and infinite timer loop

It is possible to store a very large value to the decrementer that it
does not raise the decrementer exception so the timer is scheduled, but
the next time value wraps and is treated as in the past.

This can occur if (u64)-1 is stored on a zero-triggered exception, or
(u64)-1 is stored twice on an underflow-triggered exception, for
example.

If such a value is set in DECAR, it gets stored to the decrementer by
the timer function, which then immediately causes another timer, which
hangs QEMU.

Clamp the decrementer to the implemented width, and use that as the
value for the timer calculation, effectively preventing this overflow.

Reported-by: sdicaro@DDCI.com
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20230530131214.373524-1-npiggin@gmail.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Rework store conditional to avoid branch

Rework store conditional to avoid a branch in the success case.
Change some of the variable names and layout while here so
gen_conditional_store more closely matches gen_stqcx_.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Message-Id: <20230605025445.161932-4-npiggin@gmail.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Remove larx/stcx. memory barrier semantics

larx and stcx. are not defined to order any memory operations.
Remove the barriers.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Message-Id: <20230605025445.161932-3-npiggin@gmail.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Ensure stcx size matches larx

Differently-sized larx/stcx. pairs can succeed if the starting address
matches. Add a check to require the size of stcx. exactly match the larx
that established the reservation. Use the term "reserve_length" for this
state, which matches the terminology used in the ISA.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Message-Id: <20230605025445.161932-2-npiggin@gmail.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Fix lqarx to set cpu_reserve

lqarx does not set cpu_reserve, which causes stqcx. to never succeed.

Cc: qemu-stable@nongnu.org
Fixes: 94bf2658676 ("target/ppc: Use atomic load for LQ and LQARX")
Fixes: 57b38ffd0c6 ("target/ppc: Use tcg_gen_qemu_{ld,st}_i128 for LQARX, LQ, STQ")
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230605025445.161932-1-npiggin@gmail.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Eliminate goto in mmubooke_check_tlb()

Move out checking PID registers into a separate function which makes
mmubooke_check_tlb() simpler and avoids using goto.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <bd84d5f38af0ba2983ccd5c07635db49267c828f.1685448535.git.balaton@eik.bme.hu>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Change ppcemb_tlb_check() to return bool

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <bacd1bcbe99c07930c29a9815915da9ac75f6920.1685448535.git.balaton@eik.bme.hu>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Simplify ppcemb_tlb_search()

No nead to store return value and break from loop when we can return
directly.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <d470118c3adcbd41b1a91779f6bb7cbdb2b0d346.1685448535.git.balaton@eik.bme.hu>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Remove some unneded line breaks

Make lines shorter and fix indentation in some functions prototypes.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <70952ba2d82141db1cf5cfcf4b227402be575874.1685448535.git.balaton@eik.bme.hu>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Move ppcemb_tlb_search() to mmu_common.c

This function is the only reason why ppcemb_tlb_check() is not static
to mmu_common.c but it also better fits in mmu_common.c so move it
there.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <b64fd712a773558dea9b84945c57785546c0ae2e.1685448535.git.balaton@eik.bme.hu>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Remove "ext" parameter of ppcemb_tlb_check()

This is only used by one caller so simplify function by removing this
parameter and move the operation to the single place where it's used.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <b21f11ae20e8a8c2e8b5d943f2bff12b5356005a.1685448535.git.balaton@eik.bme.hu>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Remove single use function

The get_physical_address() function is a trivial wrapper of
get_physical_address_wtlb() that is only used once. Remove it and call
get_physical_address_wtlb() directly instead.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <302697d63d26caebefaeee1e45352145ebd0318a.1685448535.git.balaton@eik.bme.hu>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: PMU implement PERFM interrupts

The PMU raises a performance monitor exception (causing an interrupt
when MSR[EE]=1) when MMCR0[PMAO] is set, and lowers it when clear.

Wire this up and implement the interrupt delivery for books. Linux perf
record can now collect PMI-driven samples.

fire_PMC_interrupt is renamed to perfm_alert, which matches a bit closer
to the new terminology used in the ISA and distinguishes the alert
condition (e.g., counter overflow) from the PERFM (or EBB) interrupts.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20230530134313.387252-2-npiggin@gmail.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Support directed privileged doorbell interrupt (SDOOR)

BookS msgsndp instruction to self or DPDES register can cause SDOOR
interrupts which crash QEMU with exception not implemented.

Linux does not use msgsndp in SMT1, and KVM only uses DPDES to cause
doorbells when emulating a SMT guest (which is not the default), so
this has gone unnoticed.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20230530130526.372701-1-npiggin@gmail.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Fix msgclrp interrupt type

msgclrp matches msgsndp and should clear PPC_INTERRUPT_DOORBELL.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20230530130714.373215-1-npiggin@gmail.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: PMU do not clear MMCR0[FCECE] on performance monitor alert

FCECE does not get cleared according to the ISA v3.1B.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20230530134313.387252-1-npiggin@gmail.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Fix PMU hflags calculation

Some of the PMU hflags bits can go out of synch, for example a store to
MMCR0 with PMCjCE=1 fails to update hflags correctly and results in
hflags mismatch:

  qemu: fatal: TCG hflags mismatch (current:0x2408003d rebuilt:0x240a003d)

This can be reproduced by running perf on a recent machine.

Some of the fragility here is the duplication of PMU hflags calculations.
This change consolidates that in a single place to update pmu-related
hflags, to be called after a well defined state changes.

The post-load PMU update is pulled out of the MSR update because it does
not depend on the MSR value.

Fixes: 8b3d1c49a9f0 ("target/ppc: Add new PMC HFLAGS")
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20230530130447.372617-1-npiggin@gmail.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

pnv/xive2: Quiet down some error messages

When dumping the END and NVP tables ("info pic" from the HMP) on the
P10 model, we're likely to be flooded with error messages such as:

  XIVE[0] - VST: invalid NVPT entry f33800 !?

The error is printed when finding an empty VSD in an indirect
table (thus END and NVP tables with skiboot), which is going to happen
when dumping the xive state. So let's tune down those messages. They
can be re-enabled easily with a macro if needed.

Those errors were already hidden on xive/P9, for the same reason.

Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20230531150537.369350-1-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

target/ppc: Fix nested-hv HEAI delivery

ppc hypervisors turn HEAI interrupts into program interrupts injected
into the guest that executed the illegal instruction, if the hypervisor
doesn't handle it some other way.

The nested-hv implementation failed to account for this HEAI->program
conversion. The virtual hypervisor wants to see the HEAI when running
a nested guest, so that interrupt type can be returned to its KVM
caller.

Fixes: 7cebc5db2eba6 ("target/ppc: Introduce a vhyp framework for nested HV support")
Cc: balaton@eik.bme.hu
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Message-Id: <20230530132127.385001-1-npiggin@gmail.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

pnv/xive2: Handle TIMA access through all ports

The Thread Interrupt Management Area (TIMA) can be accessed through 4
ports, targeted by the address. The base address of a TIMA
is using port 0 and the other ports are 0x80 apart. Using one port or
another can be useful to balance the load on the snoop buses. With
skiboot and linux, we currently use port 0, but as it tends to be
busy, another hypervisor is using port 1 for TIMA access.

The port address bits fall in between the special op indication
bits (the 2 MSBs) and the register offset bits (the 6 LSBs). They are
"don't care" for the hardware when processing a TIMA operation. This
patch filters out those port address bits so that a TIMA operation can
be triggered using any port.

It is also true for indirect access (through the IC BAR) and it's
actually nothing new, it was already the case on P9. Which helps here,
as the TIMA handling code is common between P9 (xive) and P10 (xive2).

Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20230601121331.487207-6-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

pnv/xive2: Introduce macros to manipulate TIMA addresses

TIMA addresses are somewhat special and are split in several bit
fields with different meanings. This patch describes it and introduce
macros to more easily access the various fields.

Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20230601121331.487207-5-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

pnv/xive2: Allow writes to the Physical Thread Enable registers

Fix what was probably a silly mistake and allow to write the Physical
Thread enable registers 0 and 1. Skiboot prefers to use the ENx_SET
variant so it went unnoticed, but there's no reason to discard a write
to the full register, it is Read-Write.

Fixes: da71b7e3ed45 ("ppc/pnv: Add a XIVE2 controller to the POWER10 chip")
Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20230601121331.487207-4-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

pnv/xive2: Add definition for the ESB cache configuration register

Add basic read/write support for the ESB cache configuration register
on P10. We don't model the ESB cache in qemu so reading/writing the
register won't do anything, but it avoids logging a guest error when
skiboot configures it:

qemu-system-ppc64 -machine powernv10 ... -d guest_errors
      ...
XIVE[0] - VC: invalid read @240
XIVE[0] - VC: invalid write @240

Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20230601121331.487207-3-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

pnv/xive2: Add definition for TCTXT Config register

Add basic read/write support for the TCTXT Config register on P10. qemu
doesn't do anything with it yet, but it avoids logging a guest error
when skiboot configures the fused-core state:

qemu-system-ppc64 -machine powernv10 ... -d guest_errors
  ...
[    0.131670000,5] XIVE: [ IC 00  ] Initializing XIVE block ID 0...
XIVE[0] - TCTXT: invalid read @140
XIVE[0] - TCTXT: invalid write @140

Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20230601121331.487207-2-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

tcg/tci: Fix MemOpIdx operand index for 3-operand memops

Cut and paste error from the 4-operand memops.

Fixes: ab64da797740 ("tcg/tci: Adjust passing of MemOpIdx")
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230608162925.677598-1-richard.henderson@linaro.org>

Merge tag 'pull-maintainers-20230608' of https://gitlab.com/jraman/qemu into staging

maintainers: update maintainers list for vfio-user & multi-process QEMU

Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
# -----BEGIN PGP SIGNATURE-----
# Version: GnuPG v2.0.22 (GNU/Linux)
#
# iQIcBAABAgAGBQJkghtXAAoJEI00T2TOPz3HoS8P/3rAXQIT6GO+JcqU/FJEyG6T
# ngP38kHDtRxyAci+GyzkZEVs3NkXDPDSUQ4uqcOH89a0MOSw6jZeQb7hjLWMaJbS
# JvVXO97mY4NUSTYGTO4egyPlxnmXgaUnDjiRZSHMIxdQyOuhsstokNj5I00X2YhI
# gzoT0EEZy/y2jfgY70xZpiZz3P+JDgvbsA9JIyxUhy6+LvFdf8hdGFtezng4yBu5
# KQ1PQ8G2NaFLjI5Z2FYegWjgXyi+8Qj/wBLTmJAYJQDCL2D9WH6ACq69YXrSj0GN
# Dwakb1VCYzwFIle/L8+Gzoalk/6FISHRBErB/M84Wzvo3Q9OyFAOHJeUeIDHFKSk
# cZnJphy4MPSiT2x7KOPeJMX7lQtoLMcg545bRl3UCSl9A2S6+eI02XVpeIWC/GWz
# sOzOPlM81HKWKCnsDp5V5CIF1z/7ACK6tBzQN7jhGEVx1hDdCAG8S4J216FH4sxF
# SCcwBGYbNv5dXNJ8J9699/fvYB6f9+3xLuKCazNEMvEK+Q5vyPTHMPUvepisgMF/
# KEoXAkmH5er5RhtLusPIHDBoWH8xIMxL1401YSFeh6M0x7VV5hOyv1j1o4x3cVIz
# Ev/0tGCpiSy5K6XQLZVHUioyIIJ4UKyQlKVWpBmgftUfA6kc0WRpJe/Gh4FMXNdz
# 91Rmr56MT/GkybQhs59D
# =M2cP
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 08 Jun 2023 11:17:59 AM PDT
# gpg:                using RSA key 8D344F64CE3F3DC7
# gpg: Good signature from "Jagannathan Raman <jag.raman@oracle.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: B36D 9579 7F7B 3D4F 87A6  D8CF 8D34 4F64 CE3F 3DC7

* tag 'pull-maintainers-20230608' of https://gitlab.com/jraman/qemu:
  maintainers: update maintainers list for vfio-user & multi-process QEMU

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

maintainers: update maintainers list for vfio-user & multi-process QEMU

Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>

Merge tag 'pull-9p-20230608' of https://github.com/cschoenebeck/qemu into staging

* Fix for CVE-2023-2861.

# -----BEGIN PGP SIGNATURE-----
#
# iQJLBAABCgA1FiEEltjREM96+AhPiFkBNMK1h2Wkc5UFAmSB7yMXHHFlbXVfb3Nz
# QGNydWRlYnl0ZS5jb20ACgkQNMK1h2Wkc5XykxAAzQb+d2clDVyj3Y3UqcB/YS7X
# ijxoZph9ObweyPiP2IThjsAcvNPnVR2Bc8bgEpihRkpEYGNLicw5BSk1SjqOgZvg
# buDRc8bOvOOrKqvYEBXbzaS/OHVIdozn8h+WNjX0jSsdUd4uq9vcwX+uqshkPwl+
# L4Ipx7ChzmHpaEigkVLh1biQEkLPRCTplny5JK/ZzvAmGVaqYb1usbSx//OVu7k+
# gBuBALmvJQst3iz/1e+bmVg+JhyxRqcHfCJuuWxaOLIyiZME3ZhTn7tp+2ilivRj
# n4/AGglTAv+yaVwRi6XEca7GND23HqFs26RPGgZrIhsAkFV03Iz3IT/BJ3Psy3Qv
# 7KYE4FhhReDnNU5JNfCbNxUPWVilwLY83BXVL9I0CADbAHgTqRSnataQ/PY26VQp
# BqKJKmxjAEnmsGVZSgRuCDDOhOBlPUPMRFINCUp2b0qujsUQaV5XHUlQ3qRfjUBc
# JQCy1LrxcSINg7oTRPZczNcrb9iWtaOfD24OGGeW1O6ihCAV0CYaRSmHUhFVPOPR
# uu4LWnbSToNgfNxBXaMk3vHA0SzWxJl7zBi53GVRvn8ciiTkAPVIoZLf0W8jE47X
# 5nkzfTpNdjnQJlaKAfDx+YcAyBUPxiknJjAJmjF/mquAtW8c9XbsCVJpyUgS4Lna
# GNfRoCUHQ6+6ui+/zM0=
# =6Vxp
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 08 Jun 2023 08:09:23 AM PDT
# gpg:                using RSA key 96D8D110CF7AF8084F88590134C2B58765A47395
# gpg:                issuer "qemu_oss@crudebyte.com"
# gpg: Good signature from "Christian Schoenebeck <qemu_oss@crudebyte.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: ECAB 1A45 4014 1413 BA38  4926 30DB 47C3 A012 D5F4
#      Subkey fingerprint: 96D8 D110 CF7A F808 4F88  5901 34C2 B587 65A4 7395

* tag 'pull-9p-20230608' of https://github.com/cschoenebeck/qemu:
  9pfs: prevent opening special files (CVE-2023-2861)

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

9pfs: prevent opening special files (CVE-2023-2861)

The 9p protocol does not specifically define how server shall behave when
client tries to open a special file, however from security POV it does
make sense for 9p server to prohibit opening any special file on host side
in general. A sane Linux 9p client for instance would never attempt to
open a special file on host side, it would always handle those exclusively
on its guest side. A malicious client however could potentially escape
from the exported 9p tree by creating and opening a device file on host
side.

With QEMU this could only be exploited in the following unsafe setups:

  - Running QEMU binary as root AND 9p 'local' fs driver AND 'passthrough'
    security model.

or

  - Using 9p 'proxy' fs driver (which is running its helper daemon as
    root).

These setups were already discouraged for safety reasons before,
however for obvious reasons we are now tightening behaviour on this.

Fixes: CVE-2023-2861
Reported-by: Yanwu Shen <ywsPlz@gmail.com>
Reported-by: Jietao Xiao <shawtao1125@gmail.com>
Reported-by: Jinku Li <jkli@xidian.edu.cn>
Reported-by: Wenbo Shen <shenwenbo@zju.edu.cn>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Message-Id: <E1q6w7r-0000Q0-NM@lizzy.crudebyte.com>

Merge tag 'pull-tricore-20230607' of https://github.com/bkoppelmann/qemu into staging

- Refactor PCXI/ICR field handling in newer ISA versions
- Add simple tests written in C

# -----BEGIN PGP SIGNATURE-----
#
# iQJTBAABCgA9FiEEbmNqfoPy3Qz6bm43CtLGOWtpyhQFAmSArqkfHGtiYXN0aWFu
# QG1haWwudW5pLXBhZGVyYm9ybi5kZQAKCRAK0sY5a2nKFKKTD/0dXpexGX7K62d4
# dLZwj9AHOa/9NOD4E6+ub2UYovF7UlydSzy+mgJyxiIUGaPBHYe3dFfktHOppSwn
# OGJp9TTqeutpXYF8/zvDnmf+TDJ71DRQWM40uC0HAUXrjUO43PiK0LMh+fm5D9uG
# vRHSDGBWUNZrNxTZSj8Kx7Sb7PkqeB8qWvpIJh1AVVhIT+dyoAp5V7EkAETpwhtz
# a7qKclQpFNWdaYnthCF5wcmoMawQKMUnO96j0lQWIXfnhDP7XmrWlNM7Ry2cMPEy
# aRxAT3hTOr2dD2Cic8brF/w1NeXJjxWDz65uD7X7Rog54+SD4+SmfZiYwdAs5YMo
# 1XTkbG0qE6HwrtaO+nZDQZFc8tdvLSlDcBd/cjOonwxJyvJVX8qjR2Ufb3PSfTct
# 85R5wRBsEapSbQwicwu71fK0N8ZVkLM3fc4nFEKMxOx8I66eJm3bcTxpT/A8knEw
# OwykEVB3Xiq45JlWOV2BkyZJ1EiHeFQzZfzckW4bYFDdCnuMHlaVG9qpKEAu7mQh
# 4Ug3Y4KhYqZA0UDHG6Ik6Ms64FYU4s+zbFYic/Jhew9NC4MoMXa8oKqXJC5W7RCl
# 1HesyBPu1i/45Xk6/kneJ16YO+i4hcH+Hp/osJrRE6qFHGtbvKJ0EFy1471YHyLf
# HjRQBmsd9XW2TaYBQgST80UvCh9CkQ==
# =5m1P
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 07 Jun 2023 09:22:01 AM PDT
# gpg:                using RSA key 6E636A7E83F2DD0CFA6E6E370AD2C6396B69CA14
# gpg:                issuer "kbastian@mail.uni-paderborn.de"
# gpg: Good signature from "Bastian Koppelmann <kbastian@mail.uni-paderborn.de>" [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: 6E63 6A7E 83F2 DD0C FA6E  6E37 0AD2 C639 6B69 CA14

* tag 'pull-tricore-20230607' of https://github.com/bkoppelmann/qemu:
  tests/tcg/tricore: Add recursion test for CSAs
  target/tricore: Fix wrong PSW for call insns
  target/tricore: Refactor PCXI/ICR register fields
  tests/tcg/tricore: Add first C program
  tests/tcg/tricore: Uses label for memory addresses
  tests/tcg/tricore: Move asm tests into 'asm' directory

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

Merge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging

Build system snafus.

# -----BEGIN PGP SIGNATURE-----
#
# iQFIBAABCAAyFiEE8TM4V0tmI4mGbHaCv/vSX3jHroMFAmSAhGYUHHBib256aW5p
# QHJlZGhhdC5jb20ACgkQv/vSX3jHroM8PAf+JQXoIFJJJPQI/jqaxzXrUAbZYBD0
# YQlclLinF0Iabhm0VMR/gYLS1ozvNBzhLvV6nsX7kKegG1Zy3BpCUld+UaJCCmax
# MfnRLdSjHG/QzP3kqJW7XDwVs5MF1n5KqVzowjB6orToaN5qDvPkMNg6YIk/BvuY
# G/w0JxbKj86KASzqcHgCuP+qPTXRrLKN63MB4xos7kVZsCv1BHJls35MJrGbMKSW
# 664Ji2YOtZe5Bf3xQ+20KddU+iPtYyryrhf1SGwXQK3w2UNVze4E0ZAb2F7IET7I
# grYUNtApLVqNDnjSz+WOVI1q2mtoOLc98T4TaKNg/4Qdu/RLXcoa0//hXQ==
# =ansd
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 07 Jun 2023 06:21:42 AM PDT
# gpg:                using RSA key F13338574B662389866C7682BFFBD25F78C7AE83
# gpg:                issuer "pbonzini@redhat.com"
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>" [unknown]
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.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: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* tag 'for-upstream' of https://gitlab.com/bonzini/qemu:
  tests: fp: remove unused submodules
  configure: check for $download value properly
  meson: fix "static build" entry in summary

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

Merge tag 'pull-xen-20230607' of https://xenbits.xen.org/git-http/people/aperard/qemu-dm into staging

Xen queue

- fix for xen-block segv
- Resolve TYPE_PIIX3_XEN_DEVICE
- Xen emulation build/Coverity fixes

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEE+AwAYwjiLP2KkueYDPVXL9f7Va8FAmSAkU0ACgkQDPVXL9f7
# Va/mMQf9Hs8pFGz7qRMc8RbuBvwVNGcdqOKZN+sbBKPb2pp9X8gkP5EV5SMLF/eu
# CjaoZU+SsZcVLZ0HZ/TevAEuMrflZeeRfneJzEcl58cwOxo5l18puRwy9iDxfh6m
# goqGGxQA2OmUa5eVZ7WX2JXo0wG/RQqqc/pChKbPAsTT9/QE23irOQBdUDf7sbGP
# WFI/LoLR0c6NNbQyZNWSP0e/+es8ztq+Is7Bl6d1fdG/6YeXK2yVaro1gyMmxKAm
# EKuvI9qva2ilV5RJEc/gB/x4PuIVCPizkrbB8XClQ81Szo49x55ChPdnpT5i7Sqd
# qjFWO2plgV/gXri1/RTCzyBujeuCOA==
# =E1Qb
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 07 Jun 2023 07:16:45 AM PDT
# 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-20230607' of https://xenbits.xen.org/git-http/people/aperard/qemu-dm:
  xen-block: fix segv on unrealize
  hw/isa/piix3: Resolve redundant TYPE_PIIX3_XEN_DEVICE
  hw/isa/piix3: Resolve redundant k->config_write assignments
  hw/isa/piix3: Avoid Xen-specific variant of piix3_write_config()
  hw/isa/piix3: Wire up Xen PCI IRQ handling outside of PIIX3
  hw/isa/piix3: Reuse piix3_realize() in piix3_xen_realize()
  hw/pci/pci.c: Don't leak PCIBus::irq_count[] in pci_bus_irqs()
  include/hw/xen/xen: Rename xen_piix3_set_irq() to xen_intx_set_irq()
  hw/xen: Fix broken check for invalid state in xs_be_open()
  xen: Drop support for Xen versions below 4.7.1
  hw/xen: Fix memory leak in libxenstore_open() for Xen
  hw/xen: Simplify emulated Xen platform init

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

Merge tag 'pull-vfio-user-20230607' of https://gitlab.com/jraman/qemu into staging

vfio-user: Fix the documentation for vfio-user and multi-process QEMU

Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
# -----BEGIN PGP SIGNATURE-----
# Version: GnuPG v2.0.22 (GNU/Linux)
#
# iQIcBAABAgAGBQJkgJP9AAoJEI00T2TOPz3HSoYQAJsyb9ve1MOAaTgL6DzrOe0F
# rS9A/F+isAXNr8XENhBBPeHpmMyZLmw35G+xUuxAerZAIo+kRWrB7ab1F7VrCi16
# sGWNIZ9byjt5dTqQTX63/txmXhKnM5puJxgJ/VlEGyZwGNCLhkS8yTmlKC1n8D92
# KzP1doqvqGV8AdEOudj7sFQsN5e//qrEsXm0vkZrUlr3L44QbJxrU+525r191dpP
# qFC8+lqh6CfnGgZyUSxrnFUUEP7yNGuLsCkeoFVywWQKgR/aA0UtUS8+lQHcloAG
# 4fSDMp6dMXZDWed48ZUpd+DptdiID64WDy+v6knin24awXY91dA7gFwXMvt2+IIl
# LDr99FJ8wdZpEwwXQrUwMis4RE3E3bQw7W11hb2GEPG+K90aRKeljZZQVl3Rhs3e
# cAxY5qwTrxhwpo1erPMfv8PLqSDk5+VWS0r+whEbnpHULOppDL8fgSz77NEt0CU1
# EdMRWBs7cTf2cdGLqKusekPwDMyKgQauvvQfTBBOAvHxM7+NpJPog9APiXx4BBG9
# On+eyV23VHRDdeuXvpLJyL9OswuRc8HmXZVR/Nk0vXWGP0ZCWoTSIeZfJfgKRjdR
# 6au8m/RI88tfJHMTT1Q/MJzdX+FlFRddt3JcPOiX9rMqj9eXKIjQBzpTW6R/hKNi
# SOKYdZF06D1hs/r8De1O
# =nA+U
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 07 Jun 2023 07:28:13 AM PDT
# gpg:                using RSA key 8D344F64CE3F3DC7
# gpg: Good signature from "Jagannathan Raman <jag.raman@oracle.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: B36D 9579 7F7B 3D4F 87A6  D8CF 8D34 4F64 CE3F 3DC7

* tag 'pull-vfio-user-20230607' of https://gitlab.com/jraman/qemu:
  docs: fix multi-process QEMU documentation
  vfio-user: update comments

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

tests/tcg/tricore: Add recursion test for CSAs

Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-Id: <20230526061946.54514-7-kbastian@mail.uni-paderborn.de>

target/tricore: Fix wrong PSW for call insns

we were copying PSW into a local variable, updated PSW.CDE in the local
and never wrote it back. So when we called save_context_upper() we were
using the non-local version of PSW which did not contain the updated
PSW.CDE.

Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-Id: <20230526061946.54514-6-kbastian@mail.uni-paderborn.de>

target/tricore: Refactor PCXI/ICR register fields

starting from ISA version 1.6.1 (previously known as 1.6P/E), some
bitfields in PCXI and ICR have changed. We also refactor these
registers using the register fields API.

Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1453
Message-Id: <20230526061946.54514-5-kbastian@mail.uni-paderborn.de>

tests/tcg/tricore: Add first C program

this allows us to exercise the startup code used by GCC to call main().

Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-Id: <20230526061946.54514-4-kbastian@mail.uni-paderborn.de>

tests/tcg/tricore: Uses label for memory addresses

the linker might rearrange sections, so lets reference memory by label
name instead of addr + off.

Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-Id: <20230526061946.54514-3-kbastian@mail.uni-paderborn.de>

tests/tcg/tricore: Move asm tests into 'asm' directory

this seperates these tests from the upcoming tests written in C.
Also rename the compiled test to 'test_<foo>.asm.tst'.

Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-Id: <20230526061946.54514-2-kbastian@mail.uni-paderborn.de>

Merge tag 'pull-ci-20230607' of https://gitlab.com/rth7680/qemu into staging

Fix TCI regressions vs Int128
Fix Arm build vs --disable-tcg
Fix iotest 194.

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmSApKodHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV/0mAf/f6+JI3tF+CxyWs+J
# 5LSDn8hosJefuy+jkhSM/aPIlX5gYvmoA7S/XNGrDG0+yNS5SriZKuyt9hB/gZ5D
# JFred7xuI0RmkEX3cnqFgsrtFmOYdx6G5tt4MU25uzKFyPgYg+6hsF0fotcFCPIp
# s2XIjEc7X1hk/xr4LRRxJeRrK+ZK48sN+K9HzITclKB3v11Dxv/a0OT2kdrPvlvb
# d/yNYewZrxM86vFmlIR/dT+M/qq7ULonlnH1HuWh8IaPO0owEyJPZPFw07C3ivUi
# uIplMcqk/2Um1R8zwUgaByINB3uVQXp1PyYsbjxvS34mdLwtYXF8b+/5Ma6tc3Tb
# sWkIXg==
# =NNK9
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 07 Jun 2023 08:39:22 AM PDT
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [ultimate]

* tag 'pull-ci-20230607' of https://gitlab.com/rth7680/qemu:
  iotests: fix 194: filter out racy postcopy-active event
  gitlab: Add cross-arm64-kvm-only
  target/arm: Only include tcg/oversized-guest.h if CONFIG_TCG
  tcg/tci: Adjust call-clobbered regs for int128_t
  tcg/tci: Adjust passing of MemOpIdx

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

iotests: fix 194: filter out racy postcopy-active event

The event is racy: it will not appear in the output if bitmap is
migrated during downtime period of migration and postcopy phase is not
started.

Fixes: ae00aa239847 "iotests: 194: test also migration of dirty bitmap"
Reported-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-Id: <20230607143606.1557395-1-vsementsov@yandex-team.ru>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

gitlab: Add cross-arm64-kvm-only

We are not currently running a --disable-tcg test for arm64,
like we are for mips, ppc and s390x.  We have a job for the
native aarch64 runner, but it is not run by default and it
is not helpful for normal developer testing without access
to qemu's private runner.

Use --without-default-features to eliminate most tests.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/arm: Only include tcg/oversized-guest.h if CONFIG_TCG

Fixes the build for --disable-tcg.

This header is only needed for cross-hosting.  Without CONFIG_TCG,
we know this is an AArch64 host, CONFIG_ATOMIC64 will be set, and
the TCG_OVERSIZED_GUEST block will never be compiled.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/tci: Adjust call-clobbered regs for int128_t

We require either 2 or 4 registers to hold int128_t.
Failure to do so results in a register allocation assert.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tcg/tci: Adjust passing of MemOpIdx

Since adding MO_ATOM_MASK, the maximum MemOpIdx requires 15 bits,
which overflows the 12 bit field allocated for TCI memory ops.
Expand the field to 16 bits for 2-operand memory ops, and place
the value in TCG_REG_TMP for 3-operand memory ops (same as we
already do for 4-operand memory ops).

Cures a debug assert for aarch64, with FEAT_LSE2 enabled.

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

docs: fix multi-process QEMU documentation

Fix a typo in the system documentation for multi-process QEMU.

Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>

vfio-user: update comments

Clarify the behavior of TYPE_VFU_OBJECT when TYPE_REMOTE_MACHINE enables
the auto-shutdown property. Also, add notes to VFU_OBJECT_ERROR.

Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>

xen-block: fix segv on unrealize

Backtrace:
  qemu_lockcnt_lock (lockcnt=0xb4) at ../util/lockcnt.c:238
  aio_set_fd_handler (ctx=0x0, fd=51, is_external=true, io_read=0x0, io_write=0x0, io_poll=0x0, io_poll_ready=0x0, opaque=0x0) at ../util/aio-posix.c:119
  xen_device_unbind_event_channel (xendev=0x55c6da5b5000, channel=0x55c6da6c4c80, errp=0x7fff641ac608) at ../hw/xen/xen-bus.c:926
  xen_block_dataplane_stop (dataplane=0x55c6da6ddbe0) at ../hw/block/dataplane/xen-block.c:719
  xen_block_disconnect (xendev=0x55c6da5b5000, errp=0x0) at ../hw/block/xen-block.c:48
  xen_block_unrealize (xendev=0x55c6da5b5000) at ../hw/block/xen-block.c:154
  xen_device_unrealize (dev=0x55c6da5b5000) at ../hw/xen/xen-bus.c:956
  xen_device_exit (n=0x55c6da5b50d0, data=0x0) at ../hw/xen/xen-bus.c:985
  notifier_list_notify (list=0x55c6d91f9820 <exit_notifiers>, data=0x0) at ../util/notify.c:39
  qemu_run_exit_notifiers () at ../softmmu/runstate.c:760

Fixes: f6eac904f682 ("xen-block: implement BlockDevOps->drained_begin()")
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20230606131605.55596-1-anthony.perard@citrix.com>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

hw/isa/piix3: Resolve redundant TYPE_PIIX3_XEN_DEVICE

During the last patches, TYPE_PIIX3_XEN_DEVICE turned into a clone of
TYPE_PIIX3_DEVICE. Remove this redundancy.

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Anthony PERARD <anthony.perard@citrix.com>
Tested-by: Chuck Zmudzinski <brchuckz@aol.com>
Message-Id: <20230312120221.99183-7-shentey@gmail.com>
Message-Id: <20230403074124.3925-8-shentey@gmail.com>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

hw/isa/piix3: Resolve redundant k->config_write assignments

The previous patch unified handling of piix3_write_config() accross the
PIIX3 device models which allows for assigning k->config_write once in the
base class.

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Anthony PERARD <anthony.perard@citrix.com>
Tested-by: Chuck Zmudzinski <brchuckz@aol.com>
Message-Id: <20230312120221.99183-6-shentey@gmail.com>
Message-Id: <20230403074124.3925-7-shentey@gmail.com>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

hw/isa/piix3: Avoid Xen-specific variant of piix3_write_config()

Subscribe to pci_bus_fire_intx_routing_notifier() instead which allows for
having a common piix3_write_config() for the PIIX3 device models.

While at it, move the subscription into machine code to facilitate resolving
TYPE_PIIX3_XEN_DEVICE.

In a possible future followup, pci_bus_fire_intx_routing_notifier() could
be adjusted in such a way that subscribing to it doesn't require
knowledge of the device firing it.

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Anthony PERARD <anthony.perard@citrix.com>
Tested-by: Chuck Zmudzinski <brchuckz@aol.com>
Message-Id: <20230312120221.99183-5-shentey@gmail.com>
Message-Id: <20230403074124.3925-6-shentey@gmail.com>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

hw/isa/piix3: Wire up Xen PCI IRQ handling outside of PIIX3

xen_intx_set_irq() doesn't depend on PIIX3State. In order to resolve
TYPE_PIIX3_XEN_DEVICE and in order to make Xen agnostic about the
precise south bridge being used, set up Xen's PCI IRQ handling of PIIX3
in the board.

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Anthony PERARD <anthony.perard@citrix.com>
Tested-by: Chuck Zmudzinski <brchuckz@aol.com>
Message-Id: <20230312120221.99183-4-shentey@gmail.com>
Message-Id: <20230403074124.3925-5-shentey@gmail.com>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

hw/isa/piix3: Reuse piix3_realize() in piix3_xen_realize()

This is a preparational patch for the next one to make the following
more obvious:

First, pci_bus_irqs() is now called twice in case of Xen where the
second call overrides the pci_set_irq_fn with the Xen variant.

Second, pci_bus_set_route_irq_fn() is now also called in Xen mode.

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Anthony PERARD <anthony.perard@citrix.com>
Tested-by: Chuck Zmudzinski <brchuckz@aol.com>
Message-Id: <20230312120221.99183-3-shentey@gmail.com>
Message-Id: <20230403074124.3925-4-shentey@gmail.com>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

hw/pci/pci.c: Don't leak PCIBus::irq_count[] in pci_bus_irqs()

When calling pci_bus_irqs() multiple times on the same object without calling
pci_bus_irqs_cleanup() in between PCIBus::irq_count[] is currently leaked.
Let's fix this because Xen will do just that in a few commits, and because
calling pci_bus_irqs_cleanup() in between seems fragile and cumbersome.

Note that pci_bus_irqs_cleanup() now has to NULL irq_count such that
pci_bus_irqs() doesn't do a double free.

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Tested-by: Chuck Zmudzinski <brchuckz@aol.com>
Message-Id: <20230403074124.3925-3-shentey@gmail.com>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

include/hw/xen/xen: Rename xen_piix3_set_irq() to xen_intx_set_irq()

xen_piix3_set_irq() isn't PIIX specific: PIIX is a single PCI device
while xen_piix3_set_irq() maps multiple PCI devices to their respective
IRQs, which is board-specific. Rename xen_piix3_set_irq() to communicate
this.

Also rename XEN_PIIX_NUM_PIRQS to XEN_IOAPIC_NUM_PIRQS since the Xen's
IOAPIC rather than PIIX has this many interrupt routes.

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Anthony PERARD <anthony.perard@citrix.com>
Tested-by: Chuck Zmudzinski <brchuckz@aol.com>
Message-Id: <20230312120221.99183-2-shentey@gmail.com>
Message-Id: <20230403074124.3925-2-shentey@gmail.com>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

hw/xen: Fix broken check for invalid state in xs_be_open()

Coverity points out that if (!s && !s->impl) isn't really what we intended
to do here. CID 1508131.

Fixes: 032475127225 ("hw/xen: Add emulated implementation of XenStore operations")
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20230412185102.441523-6-dwmw2@infradead.org>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

xen: Drop support for Xen versions below 4.7.1

In restructuring to allow for internal emulation of Xen functionality,
I broke compatibility for Xen 4.6 and earlier. Fix this by explicitly
removing support for anything older than 4.7.1, which is also ancient
but it does still build, and the compatibility support for it is fairly
unintrusive.

Fixes: 15e283c5b684 ("hw/xen: Add foreignmem operations to allow redirection to internal emulation")
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>
Message-Id: <20230412185102.441523-4-dwmw2@infradead.org>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

hw/xen: Fix memory leak in libxenstore_open() for Xen

There was a superfluous allocation of the XS handle, leading to it
being leaked on both the error path and the success path (where it gets
allocated again).

Spotted by Coverity (CID 1508098).

Fixes: ba2a92db1ff6 ("hw/xen: Add xenstore operations to allow redirection to internal emulation")
Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Paul Durrant <paul@xen.org>
Message-Id: <20230412185102.441523-3-dwmw2@infradead.org>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

hw/xen: Simplify emulated Xen platform init

I initially put the basic platform init (overlay pages, grant tables,
event channels) into mc->kvm_type because that was the earliest place
that could sensibly test for xen_mode==XEN_EMULATE.

The intent was to do this early enough that we could then initialise the
XenBus and other parts which would have depended on them, from a generic
location for both Xen and KVM/Xen in the PC-specific code, as seen in
https://lore.kernel.org/qemu-devel/20230116221919.1124201-16-dwmw2@infradead.org/

However, then the Xen on Arm patches came along, and *they* wanted to
do the XenBus init from a 'generic' Xen-specific location instead:
https://lore.kernel.org/qemu-devel/20230210222729.957168-4-sstabellini@kernel.org/

Since there's no generic location that covers all three, I conceded to
do it for XEN_EMULATE mode in pc_basic_devices_init().

And now there's absolutely no point in having some of the platform init
done from pc_machine_kvm_type(); we can move it all up to live in a
single place in pc_basic_devices_init(). This has the added benefit that
we can drop the separate xen_evtchn_connect_gsis() function completely,
and pass just the system GSIs in directly to xen_evtchn_create().

While I'm at it, it does no harm to explicitly pass in the *number* of
said GSIs, because it does make me twitch a bit to pass an array of
impicit size. During the lifetime of the KVM/Xen patchset, that had
already changed (albeit just cosmetically) from GSI_NUM_PINS to
IOAPIC_NUM_PINS.

And document a bit better that this is for the *output* GSI for raising
CPU0's events when the per-CPU vector isn't available. The fact that
we create a whole set of them and then only waggle the one we're told
to, instead of having a single output and only *connecting* it to the
GSI that it should be connected to, is still non-intuitive for me.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>
Message-Id: <20230412185102.441523-2-dwmw2@infradead.org>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

tests: fp: remove unused submodules

tests/fp/berkeley-softfloat-3 and tests/fp/berkeley-testfloat-3
have been replaced by subprojects, so remove the now-unnecessary
submodules.

Reported-by: Michal Privoznik <mprivozn@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: check for $download value properly

If configure was invoked with --disable-download and git
submodules were not checked out a warning is produced and the
configure script fails. But the $download variable (which
reflects the enable/disable download argument) is checked for in
a weird fashion:

  test -f "$download" = disabled

Drop the '-f' to check for the actual value of the variable.

Fixes: 2019cabfee0 ("meson: subprojects: replace submodules with wrap files", 2023-06-06)
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

meson: fix "static build" entry in summary

Fixes: a0cbd2e8496 ("meson: use prefer_static option", 2023-05-18)
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Merge tag 'pull-target-arm-20230606' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * Support gdbstub (guest debug) in HVF
 * xnlx-versal: Support CANFD controller
 * bpim2u: New board model: Banana Pi BPI-M2 Ultra
 * Emulate FEAT_LSE2
 * allow DC CVA[D]P in user mode emulation
 * trap DCC access in user mode emulation

# -----BEGIN PGP SIGNATURE-----
#
# iQJNBAABCAA3FiEE4aXFk81BneKOgxXPPCUl7RQ2DN4FAmR/AKUZHHBldGVyLm1h
# eWRlbGxAbGluYXJvLm9yZwAKCRA8JSXtFDYM3jzIEACNepQGY44yPhrEG+wD4WAB
# fH670KI33HcsFd2rGsC369gcssQbRIW/29reOzNhRMuol+kHI6OFaONpuKSdO0Rz
# TLVIsnT2Uq8KwbYfLtDQt5knj027amPy75d4re8wIK1eZB4dOIHysqAvQrJYeync
# 9obKku8xXGLwZh/mYHoVgHcZU0cPJO9nri39n1tV3JUBsgmqEURjzbZrMcF+yMX7
# bUzOYQvC1Iedmo+aWfx43u82AlNQFz1lsqmnQj7Z5rvv0HT+BRF5WzVMP0qRh5+Z
# njkqmBH9xb9kkgeHmeMvHpWox+J+obeSmVg/4gDNlJpThmpuU0Vr7EXUN3MBQlV9
# lhyy6zrTwC/BToiQqdT2dnpao9FzXy5exfnqi/py5IuqfjAzSO+p61LlPPZ4cJri
# pCK4yq2gzQXYfrlZkUJipvRMH8Xa4IdQx+w7lXrQoJdduF4/+6aJW/GAWSu0e7eC
# zgBwaJjI7ENce8ixJnuEFUxUnaBo8dl72a0PGA1UU8PL+cJNOIpyhPk4goWQprdn
# iFF4ZnjhBRZ2gk/4HGD9u5Vo2lNqP93YS5QhkGkF+HJsBmcOZgidIUpfHhPQvvHO
# Np196T2cAETCWGV1xG4CaTpxN2ndRReq3C0/mzfhIbwhXEACtvAiSlO4KB8t6pJj
# MzinCABXHcovJbGbxZ9j6w==
# =8SdN
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 06 Jun 2023 02:47:17 AM PDT
# 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]
# gpg:                 aka "Peter Maydell <peter@archaic.org.uk>" [unknown]

* tag 'pull-target-arm-20230606' of https://git.linaro.org/people/pmaydell/qemu-arm: (42 commits)
  target/arm: trap DCC access in user mode emulation
  tests/tcg/aarch64: add DC CVA[D]P tests
  target/arm: allow DC CVA[D]P in user mode emulation
  target/arm: Enable FEAT_LSE2 for -cpu max
  tests/tcg/multiarch: Adjust sigbus.c
  tests/tcg/aarch64: Use stz2g in mte-7.c
  target/arm: Move mte check for store-exclusive
  target/arm: Relax ordered/atomic alignment checks for LSE2
  target/arm: Add SCTLR.nAA to TBFLAG_A64
  target/arm: Check alignment in helper_mte_check
  target/arm: Pass single_memop to gen_mte_checkN
  target/arm: Pass memop to gen_mte_check1*
  target/arm: Hoist finalize_memop out of do_fp_{ld, st}
  target/arm: Hoist finalize_memop out of do_gpr_{ld, st}
  target/arm: Load/store integer pair with one tcg operation
  target/arm: Sink gen_mte_check1 into load/store_exclusive
  target/arm: Use tcg_gen_qemu_{ld, st}_i128 in gen_sve_{ld, st}r
  target/arm: Use tcg_gen_qemu_st_i128 for STZG, STZ2G
  target/arm: Use tcg_gen_qemu_{st, ld}_i128 for do_fp_{st, ld}
  target/arm: Use tcg_gen_qemu_ld_i128 for LDXP
  ...

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

accel/tcg: Fix undefined shift in store_whole_le16

The computation is documented as unused in this case,
but triggers an ubsan error:

../accel/tcg/ldst_atomicity.c.inc:837:33: runtime error: shift exponent -32 is negative
SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior ../accel/tcg/ldst_atomicity.c.inc:837:33 in

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20230606171629.98157-1-richard.henderson@linaro.org>

Merge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging

* finish atomics revamp
* meson.build tweaks
* revert avocado update
* always upgrade/downgrade locally installed Python packages
* switch from submodules to subprojects
* remove --with-git= option
* rename --enable-pypi to --enable-download, control submodules and subprojects too

# -----BEGIN PGP SIGNATURE-----
#
# iQFIBAABCAAyFiEE8TM4V0tmI4mGbHaCv/vSX3jHroMFAmR/Qu8UHHBib256aW5p
# QHJlZGhhdC5jb20ACgkQv/vSX3jHroMmSwgAj5SHD8R+5D1UmptzBvI/72CfgqVv
# MJa4O1LvHwUkuSmxX1MFFhRa0mo0bu6j+bPpvJ29zKS61ybVwJl87gnsRcDAMXe7
# 08YbcG35Chox6aZxbidUQtXm18JZ3F2aMtmxUuP0PR7LDjVXLV5FsjrHTIt8KuEZ
# vUqq3IsVbc4FxCCC0ke2DzrtgpRCxYSdfPrj/t5WzAztAXId9r1zvUlCLN+FUpri
# E3KIZYpkXZyOnJQ9W30KnsZo5QtDACwlIMBK6whSdoCjyNN7TwDdhNW8QkOueNO6
# q3tLfwf5+u6uyEoaQTW+teE2oMXT8N4IJllRJj2RyQ1BFD49XhUUJmc33Q==
# =b9QD
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 06 Jun 2023 07:30:07 AM PDT
# gpg:                using RSA key F13338574B662389866C7682BFFBD25F78C7AE83
# gpg:                issuer "pbonzini@redhat.com"
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>" [unknown]
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.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: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* tag 'for-upstream' of https://gitlab.com/bonzini/qemu: (21 commits)
  configure: remove --with-git-submodules=
  build: remove git submodule handling from main makefile
  meson: subprojects: replace berkeley-{soft,test}float-3 with wraps
  pc-bios/s390-ccw: always build network bootloader
  configure: move SLOF submodule handling to pc-bios/s390-ccw
  meson: subprojects: replace submodules with wrap files
  build: log submodule update from git-submodule.sh
  git-submodule: allow partial update of .git-submodule-status
  configure: rename --enable-pypi to --enable-download, control subprojects too
  configure: remove --with-git= option
  mkvenv: always pass locally-installed packages to pip
  tests: Use separate virtual environment for avocado
  Revert "tests/requirements.txt: bump up avocado-framework version to 101.0"
  scsi/qemu-pr-helper: Drop support for 'old' libmultipath API
  meson.build: Use -Wno-undef only for SDL2 versions that need it
  meson.build: Group the audio backend entries in a separate summary section
  meson.build: Group the network backend entries in a separate summary section
  meson.build: Group the UI entries in a separate summary section
  scripts: remove dead file
  atomics: eliminate mb_read/mb_set
  ...

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

configure: remove --with-git-submodules=

Reuse --enable/--disable-download to control git submodules as well.
Adjust the error messages of git-submodule.sh to refer to the new
option.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

build: remove git submodule handling from main makefile

The only remaining user of submodules at build time is roms/SLOF,
which is handled in pc-bios/s390-ccw/Makefile.  Remove the relevant
code from the main makefile.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

meson: subprojects: replace berkeley-{soft,test}float-3 with wraps

Unlike other subprojects, these require an overlay directory to include
meson rules to build the libraries.  The rules are basically lifted
from tests/fp/meson.build, with a few changes to create platform.h
and publish a dependency.

The build defines are passed through a subproject option, and posted
back to users of the library via the dependency's compile_args.

The only remaining user of GIT_SUBMODULES and GIT_SUBMODULES_ACTION
is roms/SLOF, which is used to build pc-bios/s390-ccw.  All other
roms submodules are only present to satisfy the license on pre-built
firmware blobs.

Best reviewed with --color-moved.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

pc-bios/s390-ccw: always build network bootloader

In the beginning, the network bootloader was considered experimental and
thus optional, but it is well established nowadays and configure always
checks for roms/SLOF before compiling pc-bios/s390-ccw.

Therefore, it makes sense to always build it together with the other
part of the s390-ccw bios.

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

configure: move SLOF submodule handling to pc-bios/s390-ccw

Move the handling of the roms/SLOF submodule out of the main Makefile,
since we are going to remove submodules from the build process of QEMU.

Acked-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

meson: subprojects: replace submodules with wrap files

Compared to submodules, .wrap files have several advantages:

* option parsing and downloading is delegated to meson

* the commit is stored in a text file instead of a magic entry in the
  git tree object

* we could stop shipping external dependencies that are only used as a
  fallback, but not break compilation on platforms that lack them.
  For example it may make sense to download dtc at build time, controlled
  by --enable-download, even when building from a tarball.  Right now,
  this patch does the opposite: make-release treats dtc like libvfio-user
  (which is not stable API and therefore hasn't found its way into any
  distros) and keycodemap (which is a copylib, for better or worse).

dependency() can fall back to a wrap automatically.  However, this
is only possible for libraries that come with a .pc file, and this
is not very common for libfdt even though the upstream project in
principle provides it; it also removes the control that we provide with
--enable-fdt={system,internal}.  Therefore, the logic to pick system
vs. internal libfdt is left untouched.

--enable-fdt=git is removed; it was already a synonym for
--enable-fdt=internal.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

build: log submodule update from git-submodule.sh

Print exactly which submodules have been updated, by reusing the logic of
"git-submodule.sh validate" after executing "git submodule update --init'.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

git-submodule: allow partial update of .git-submodule-status

Allow a specific subdirectory to run git-submodule.sh with only a
subset of submodules, without removing the others from the
.git-submodule-status file.

This also allows scripts/git-submodule.sh to be more lenient:
validating an empty set of submodules is not a mistake.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: rename --enable-pypi to --enable-download, control subprojects too

The behavior of --{enable,disable}-pypi is similar to that of
-Dwrapmode={default,nodownload} respectively.  In particular,
in both cases a feature needs to be explicitly enabled for the
dependency to be downloaded.

So, use a single option to control both cases.  Now, --enable-slirp
will trigger cloning and building of libslirp if the .pc file
is not found on the machine.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: remove --with-git= option

The scenario for which --with-git= was introduced was to use a SOCKS proxy
such as tsocks.  However, this was back in 2017 when QEMU's submodules
used the git:// protocol, and it is not as important when using the
"smart HTTP" backend; for example, neither "meson subprojects download"
nor scripts/checkpatch.pl obey the GIT environment variable.

So remove the knob, but test for the presence of git in the configure and
git-submodule.sh scripts, and suggest using --with-git-submodules=validate
+ a manual invocation of git-submodule.sh when git does not work.  Hopefully
in the future the GIT environment variable will be supported by Meson.

Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

mkvenv: always pass locally-installed packages to pip

Let pip decide whether a new version should be installed or the current
one is okay.  This ensures that the virtual environment is updated
(either upgraded or downgraded) whenever a new version of a package is
requested.

The hardest part here is figuring out if a package is installed in
the venv (which also has to be done twice to account for the presence
of either setuptools in Python <3.8, or importlib in Python >=3.8).

Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Cc: John Snow <jsnow@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

tests: Use separate virtual environment for avocado

This reverts commits eea2d141179 ("Makefile: remove $(TESTS_PYTHON)",
2023-05-26) and 9c6692db550 ("tests: Use configure-provided pyvenv for
tests", 2023-05-18).

Right now, there is a conflict between wanting a ">=" constraint when
using a distro-provided package and wanting a "==" constraint when
installing Avocado from PyPI; this would provide the best of both worlds
in terms of resiliency for both distros that have required packages and
distros that don't.

The conflict is visible also for meson, where we would like to install
the latest 0.63.x version but also accept a distro 1.1.x version.
But it is worse for avocado, for two reasons:

1) we cannot use an "==" constraint to install avocado if the venv
includes a system avocado.  The distro will package plugins that have
"==" constraints on the version that is included in the distro, and, using
"pip install avocado==88.1" on a venv that includes system packages will
result in this error:

   ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
   avocado-framework-plugin-varianter-yaml-to-mux 98.0 requires avocado-framework==98.0, but you have avocado-framework 88.1 which is incompatible.
   avocado-framework-plugin-result-html 98.0 requires avocado-framework==98.0, but you have avocado-framework 88.1 which is incompatible.
   make[1]: Leaving directory '/home/berrange/src/virt/qemu/build'

2) we cannot use ">=" either if the venv does _not_ include a system
avocado, because that would result in the installation of v101.0 which
is the one we've just reverted.

So the idea is to encode the dependencies as an (acceptable, locked)
tuple, like this hypothetical TOML that would be committed inside
python/ and used by mkvenv.py:

  [meson]
  meson = { minimum = "0.63.0", install = "0.63.3", canary = "meson" }

  [docs]
  # 6.0 drops support for Python 3.7
  sphinx = { minimum = "1.6", install = "<6.0", canary = "sphinx-build" }
  sphinx_rtd_theme = { minimum = "0.5" }

  [avocado]
  avocado-framework = { minimum = "88.1", install = "88.1", canary = "avocado" }

Once this is implemented, it would also be possible to install avocado in
pyvenv/ using "mkvenv.py ensure", thus using the distro package on Fedora
and CentOS Stream (the only distros where it's available).  But until
this is implemented, keep avocado in a separate venv.  There is still the
benefit of using a single python for meson custom_targets and for sphinx.

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

Merge tag 'pull-request-2023-06-06' of https://gitlab.com/thuth/qemu into staging

* Fix emulated LCCB, LOCFHR, MXDB and MXDBR s390x instructions
* Fix the malta machine on s390x (big endian) hosts
* Emulate /proc/cpuinfo on s390x
* Remove pointless QOM casts
* Improve the inclusion logic for libkeyutils and ipmi-bt-test in meson.build

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmR+ycgRHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbWXXw//WPz3ng50KLS+M1t3/ULEjO6XkGfP2LQZ
# RsZq3hf9THFPZgcREk+6SQvttOSTuvHNfakfujS6U1Ou5thReWqLe4itFW6+hB5j
# kQ+Sm6YJ+fpezkBnSefcUoL5nA9VVKZ6KE6kxq5CUBZNoIk1sSsfrU8y8wjzW0yg
# 2nraOcG10aLpO2BfvKHVEAhJtwl9pHJsFANmHC2/h2wC9BZIAzdxiytzdcJ909gN
# AAa0hIrLK/oFgJjkSSxu+QTaVGPARXqkx5WV546F/zmDMFUWd9nrXaegwqxjgPBN
# m9Ua0SXll5hX2Z57vjJWlbTYkD+JUB22L0N7p5/xzhYRpLVSq1pdveo9psrzIC3E
# Bt7chZB58acQepJHxxa3UHDOHcnfdfaN+Dd9wD29wHr7nK8lOcsen7/7V+5YXomc
# qenkCtkpjKTl07OBxe6MDGZtPZYA8fK1CjEyYwHCe8QvxEzsyg96Bm3j4N2VPxQU
# +f/sFPX7SgogZI4mB4wdoxOF1RmQ+DXQ2tnB970txZRkmFq2jJHpW86jkkbq2Jl1
# KIjgdIXjVgy+MPtuQzO5cT+jfhGQL7FQynGXHjv/UidBid5XD3TDVNa9AthN3Mng
# +rPT90VJ7j9soMqvmNT1COSIRD+M49dQKBIQuq/gWplaTOHaAcJrCwYScwqe0u0P
# zmjCNeuPVw8=
# =dfJr
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 05 Jun 2023 10:53:12 PM PDT
# gpg:                using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5
# gpg:                issuer "thuth@redhat.com"
# gpg: Good signature from "Thomas Huth <th.huth@gmx.de>" [unknown]
# gpg:                 aka "Thomas Huth <thuth@redhat.com>" [unknown]
# gpg:                 aka "Thomas Huth <th.huth@posteo.de>" [unknown]
# gpg:                 aka "Thomas Huth <huth@tuxfamily.org>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 27B8 8847 EEE0 2501 18F3  EAB9 2ED9 D774 FE70 2DB5

* tag 'pull-request-2023-06-06' of https://gitlab.com/thuth/qemu:
  linux-user: Emulate /proc/cpuinfo on s390x
  linux-user/elfload: Introduce elf_hwcap_str() on s390x
  linux-user/elfload: Expose get_elf_hwcap() on s390x
  s390x/tcg: Fix CPU address returned by STIDP
  bulk: Remove pointless QOM casts
  scripts: Add qom-cast-macro-clean-cocci-gen.py
  hw/mips/malta: Fix the malta machine on big endian hosts
  gitlab-ci: Remove unused Python package
  tests/qtest: Run ipmi-bt-test only if CONFIG_IPMI_EXTERN is set
  tests/tcg/s390x: Test MXDB and MXDBR
  target/s390x: Fix MXDB and MXDBR
  Add conditional dependency for libkeyutils
  tests/tcg/s390x: Test single-stepping SVC
  linux-user/s390x: Fix single-stepping SVC
  tests/tcg/s390x: Test LOCFHR
  target/s390x: Fix LOCFHR taking the wrong half of R2
  tests/tcg/s390x: Test LCBB
  target/s390x: Fix LCBB overwriting the top 32 bits

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

target/arm: trap DCC access in user mode emulation

Accessing EL0-accessible Debug Communication Channel (DCC) registers in
user mode emulation is currently enabled.  However, it does not match
Linux behavior as Linux sets MDSCR_EL1.TDCC on startup to disable EL0
access to DCC (see __cpu_setup() in arch/arm64/mm/proc.S).

This patch fixes access_tdcc() to check MDSCR_EL1.TDCC for EL0 and sets
MDSCR_EL1.TDCC for user mode emulation to match Linux.

Signed-off-by: Zhuojia Shen <chaosdefinition@hotmail.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: DS7PR12MB630905198DD8E69F6817544CAC4EA@DS7PR12MB6309.namprd12.prod.outlook.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/tcg/aarch64: add DC CVA[D]P tests

Test execution of DC CVAP and DC CVADP instructions under user mode
emulation.

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

target/arm: allow DC CVA[D]P in user mode emulation

DC CVAP and DC CVADP instructions can be executed in EL0 on Linux,
either directly when SCTLR_EL1.UCI == 1 or emulated by the kernel (see
user_cache_maint_handler() in arch/arm64/kernel/traps.c).

This patch enables execution of the two instructions in user mode
emulation.

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

target/arm: Enable FEAT_LSE2 for -cpu max

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-21-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/tcg/multiarch: Adjust sigbus.c

With -cpu max and FEAT_LSE2, the __aarch64__ section will only raise
an alignment exception when the load crosses a 16-byte boundary.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-20-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/tcg/aarch64: Use stz2g in mte-7.c

We have many other instances of stg in the testsuite;
change these to provide an instance of stz2g.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-19-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Move mte check for store-exclusive

Push the mte check behind the exclusive_addr check.
Document the several ways that we are still out of spec
with this implementation.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-18-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Relax ordered/atomic alignment checks for LSE2

FEAT_LSE2 only requires that atomic operations not cross a
16-byte boundary.  Ordered operations may be completely
unaligned if SCTLR.nAA is set.

Because this alignment check is so special, do it by hand.
Make sure not to keep TCG temps live across the branch.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-17-richard.henderson@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Add SCTLR.nAA to TBFLAG_A64

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-16-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Check alignment in helper_mte_check

Fixes a bug in that with SCTLR.A set, we should raise any
alignment fault before raising any MTE check fault.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-15-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Pass single_memop to gen_mte_checkN

Pass the individual memop to gen_mte_checkN.
For the moment, do nothing with it.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-14-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Pass memop to gen_mte_check1*

Pass the completed memop to gen_mte_check1_mmuidx.
For the moment, do nothing more than extract the size.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-13-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Hoist finalize_memop out of do_fp_{ld, st}

We are going to need the complete memop beforehand,
so let's not compute it twice.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-12-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Hoist finalize_memop out of do_gpr_{ld, st}

We are going to need the complete memop beforehand,
so let's not compute it twice.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-11-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Load/store integer pair with one tcg operation

This is required for LSE2, where the pair must be treated atomically if
it does not cross a 16-byte boundary.  But it simplifies the code to do
this always.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-10-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Sink gen_mte_check1 into load/store_exclusive

No need to duplicate this check across multiple call sites.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-9-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Use tcg_gen_qemu_{ld, st}_i128 in gen_sve_{ld, st}r

Round len_align to 16 instead of 8, handling an odd 8-byte as part
of the tail.  Use MO_ATOM_NONE to indicate that all of these memory
ops have only byte atomicity.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-8-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Use tcg_gen_qemu_st_i128 for STZG, STZ2G

This fixes a bug in that these two insns should have been using atomic
16-byte stores, since MTE is ARMv8.5 and LSE2 is mandatory from ARMv8.4.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-7-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Use tcg_gen_qemu_{st, ld}_i128 for do_fp_{st, ld}

While we don't require 16-byte atomicity here, using a single larger
operation simplifies the code.  Introduce finalize_memop_asimd for this.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-6-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Use tcg_gen_qemu_ld_i128 for LDXP

While we don't require 16-byte atomicity here, using a single larger
load simplifies the code, and makes it a closer match to STXP.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-5-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Introduce finalize_memop_{atom,pair}

Let finalize_memop_atom be the new basic function, with
finalize_memop and finalize_memop_pair testing FEAT_LSE2
to apply the appropriate atomicity.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-4-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Add feature test for FEAT_LSE2

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230530191438.411344-3-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>