hw/xen: Implement EVTCHNOP_unmask

This finally comes with a mechanism for actually injecting events into
the guest vCPU, with all the atomic-test-and-set that's involved in
setting the bit in the shinfo, then the index in the vcpu_info, and
injecting either the lapic vector as MSI, or letting KVM inject the
bare vector.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

hw/xen: Implement EVTCHNOP_close

It calls an internal close_port() helper which will also be used from
EVTCHNOP_reset and will actually do the work to disconnect/unbind a port
once any of that is actually implemented in the first place.

That in turn calls a free_port() internal function which will be in
error paths after allocation.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

hw/xen: Implement EVTCHNOP_status

This adds the basic structure for maintaining the port table and reporting
the status of ports therein.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: Add support for Xen event channel delivery to vCPU

The kvm_xen_inject_vcpu_callback_vector() function will either deliver
the per-vCPU local APIC vector (as an MSI), or just kick the vCPU out
of the kernel to trigger KVM's automatic delivery of the global vector.
Support for asserting the GSI/PCI_INTX callbacks will come later.

Also add kvm_xen_get_vcpu_info_hva() which returns the vcpu_info of
a given vCPU.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

hw/xen: Add xen_evtchn device for event channel emulation

Include basic support for setting HVM_PARAM_CALLBACK_IRQ to the global
vector method HVM_PARAM_CALLBACK_TYPE_VECTOR, which is handled in-kernel
by raising the vector whenever the vCPU's vcpu_info->evtchn_upcall_pending
flag is set.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: implement HVMOP_set_param

This is the hook for adding the HVM_PARAM_CALLBACK_IRQ parameter in a
subsequent commit.

Signed-off-by: Ankur Arora <ankur.a.arora@oracle.com>
Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
[dwmw2: Split out from another commit]
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: implement HVMOP_set_evtchn_upcall_vector

The HVMOP_set_evtchn_upcall_vector hypercall sets the per-vCPU upcall
vector, to be delivered to the local APIC just like an MSI (with an EOI).

This takes precedence over the system-wide delivery method set by the
HVMOP_set_param hypercall with HVM_PARAM_CALLBACK_IRQ. It's used by
Windows and Xen (PV shim) guests but normally not by Linux.

Signed-off-by: Ankur Arora <ankur.a.arora@oracle.com>
Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
[dwmw2: Rework for upstream kernel changes and split from HVMOP_set_param]
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: implement HYPERVISOR_event_channel_op

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
[dwmw2: Ditch event_channel_op_compat which was never available to HVM guests]
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: handle VCPUOP_register_runstate_memory_area

Allow guest to setup the vcpu runstates which is used as
steal clock.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: handle VCPUOP_register_vcpu_time_info

In order to support Linux vdso in Xen.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: handle VCPUOP_register_vcpu_info

Handle the hypercall to set a per vcpu info, and also wire up the default
vcpu_info in the shared_info page for the first 32 vCPUs.

To avoid deadlock within KVM a vCPU thread must set its *own* vcpu_info
rather than it being set from the context in which the hypercall is
invoked.

Add the vcpu_info (and default) GPA to the vmstate_x86_cpu for migration,
and restore it in kvm_arch_put_registers() appropriately.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: implement HYPERVISOR_vcpu_op

This is simply when guest tries to register a vcpu_info
and since vcpu_info placement is optional in the minimum ABI
therefore we can just fail with -ENOSYS

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: implement HYPERVISOR_hvm_op

This is when guest queries for support for HVMOP_pagetable_dying.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: implement XENMEM_add_to_physmap_batch

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: implement HYPERVISOR_memory_op

Specifically XENMEM_add_to_physmap with space XENMAPSPACE_shared_info to
allow the guest to set its shared_info page.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
[dwmw2: Use the xen_overlay device, add compat support]
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: manage and save/restore Xen guest long_mode setting

Xen will "latch" the guest's 32-bit or 64-bit ("long mode") setting when
the guest writes the MSR to fill in the hypercall page, or when the guest
sets the event channel callback in HVM_PARAM_CALLBACK_IRQ.

KVM handles the former and sets the kernel's long_mode flag accordingly.
The latter will be handled in userspace. Keep them in sync by noticing
when a hypercall is made in a mode that doesn't match qemu's idea of
the guest mode, and resyncing from the kernel. Do that same sync right
before serialization too, in case the guest has set the hypercall page
but hasn't yet made a system call.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: add pc_machine_kvm_type to initialize XEN_EMULATE mode

The xen_overlay device (and later similar devices for event channels and
grant tables) need to be instantiated. Do this from a kvm_type method on
the PC machine derivatives, since KVM is only way to support Xen emulation
for now.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

xen: Permit --xen-domid argument when accel is KVM

Signed-off-by: Paul Durrant <pdurrant@amazon.com>
Signed-off-by: David Wooodhouse <dwmw@amazon.co.uk>

hw/xen: Add xen_overlay device for emulating shared xenheap pages

For the shared info page and for grant tables, Xen shares its own pages
from the "Xen heap" to the guest. The guest requests that a given page
from a certain address space (XENMAPSPACE_shared_info, etc.) be mapped
to a given GPA using the XENMEM_add_to_physmap hypercall.

To support that in qemu when *emulating* Xen, create a memory region
(migratable) and allow it to be mapped as an overlay when requested.

Xen theoretically allows the same page to be mapped multiple times
into the guest, but that's hard to track and reinstate over migration,
so we automatically *unmap* any previous mapping when creating a new
one. This approach has been used in production with.... a non-trivial
number of guests expecting true Xen, without any problems yet being
noticed.

This adds just the shared info page for now. The grant tables will be
a larger region, and will need to be overlaid one page at a time. I
think that means I need to create separate aliases for each page of
the overall grant_frames region, so that they can be mapped individually.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: Implement SCHEDOP_poll and SCHEDOP_yield

They both do the same thing and just call sched_yield. This is enough to
stop the Linux guest panicking when running on a host kernel which doesn't
intercept SCHEDOP_poll and lets it reach userspace.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: implement HYPERVISOR_sched_op, SCHEDOP_shutdown

It allows to shutdown itself via hypercall with any of the 3 reasons:
  1) self-reboot
  2) shutdown
  3) crash

Implementing SCHEDOP_shutdown sub op let us handle crashes gracefully rather
than leading to triple faults if it remains unimplemented.

In addition, the SHUTDOWN_soft_reset reason is used for kexec, to reset
Xen shared pages and other enlightenments and leave a clean slate for the
new kernel without the hypervisor helpfully writing information at
unexpected addresses.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
[dwmw2: Ditch sched_op_compat which was never available for HVM guests,
        Add SCHEDOP_soft_reset]
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: implement HYPERVISOR_xen_version

This is just meant to serve as an example on how we can implement
hypercalls. xen_version specifically since Qemu does all kind of
feature controllability. So handling that here seems appropriate.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
[dwmw2: Implement kvm_gva_rw() safely]
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/xen: handle guest hypercalls

This means handling the new exit reason for Xen but still
crashing on purpose. As we implement each of the hypercalls
we will then return the right return code.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
[dwmw2: Add CPL to hypercall tracing, disallow hypercalls from CPL > 0]
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

xen-platform: allow its creation with XEN_EMULATE mode

The only thing we need to fix to make this build is the PIO hack which
sets the BIOS memory areas to R/W v.s. R/O. Theoretically we could hook
that up to the PAM registers on the emulated PIIX, but in practice
nobody cares, so just leave it doing nothing.

Now it builds without actual Xen, move it to CONFIG_XEN_BUS to include it
in the KVM-only builds.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

xen-platform: exclude vfio-pci from the PCI platform unplug

Such that PCI passthrough devices work for Xen emulated guests.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/kvm: Set Xen vCPU ID in KVM

There are (at least) three different vCPU ID number spaces. One is the
internal KVM vCPU index, based purely on which vCPU was chronologically
created in the kernel first. If userspace threads are all spawned and
create their KVM vCPUs in essentially random order, then the KVM indices
are basically random too.

The second number space is the APIC ID space, which is consistent and
useful for referencing vCPUs. MSIs will specify the target vCPU using
the APIC ID, for example, and the KVM Xen APIs also take an APIC ID
from userspace whenever a vCPU needs to be specified (as opposed to
just using the appropriate vCPU fd).

The third number space is not normally relevant to the kernel, and is
the ACPI/MADT/Xen CPU number which corresponds to cs->cpu_index. But
Xen timer hypercalls use it, and Xen timer hypercalls *really* want
to be accelerated in the kernel rather than handled in userspace, so
the kernel needs to be told.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/kvm: handle Xen HVM cpuid leaves

Introduce support for emulating CPUID for Xen HVM guests. It doesn't make
sense to advertise the KVM leaves to a Xen guest, so do Xen unconditionally
when the xen-version machine property is set.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
[dwmw2: Obtain xen_version from KVM property, make it automatic]
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

i386/kvm: Add xen-version KVM accelerator property and init KVM Xen support

This just initializes the basic Xen support in KVM for now. Only permitted
on TYPE_PC_MACHINE because that's where the sysbus devices for Xen heap
overlay, event channel, grant tables and other stuff will exist. There's
no point having the basic hypercall support if nothing else works.

Provide sysemu/kvm_xen.h and a kvm_xen_get_caps() which will be used
later by support devices.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

xen: Add XEN_DISABLED mode and make it default

Also set XEN_ATTACH mode in xen_init() to reflect the truth; not that
anyone ever cared before. It was *only* ever checked in xen_init_pv()
before.

Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

xen: add CONFIG_XEN_BUS and CONFIG_XEN_EMU options for Xen emulation

The XEN_EMU option will cover core Xen support in target/, which exists
only for x86 with KVM today but could theoretically also be implemented
on Arm/Aarch64 and with TCG or other accelerators (if anyone wants to
run the gauntlet of struct layout compatibility, errno mapping, and the
rest of that fui).

It will also cover the support for architecture-independent grant table
and event channel support which will be added in hw/i386/kvm/ (on the
basis that the non-KVM support is very theoretical and making it not use
KVM directly seems like gratuitous overengineering at this point).

The XEN_BUS option is for the xenfv platform support, which will now be
used both by XEN_EMU and by real Xen.

The XEN option remains dependent on the Xen runtime libraries, and covers
support for real Xen. Some code which currently resides under CONFIG_XEN
will be moving to CONFIG_XEN_BUS over time as the direct dependencies on
Xen runtime libraries are eliminated. The Xen PCI platform device will
also reside under CONFIG_XEN_BUS.

Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

include: import Xen public headers to hw/xen/interface

There's already a partial set here; update them and pull in a more
complete set.

To start with, define __XEN_TOOLS__ in hw/xen/xen.h to ensure that any
internal definitions needed by Xen toolstack libraries are present
regardless of the order in which the headers are included. A reckoning
will come later, once we make the PV backends work in emulation and
untangle the headers for Xen-native vs. generic parts.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
[dwmw2: Update to Xen public headers from 4.16.2 release, add some in io/,
        define __XEN_TOOLS__ in hw/xen/xen.h, move to hw/xen/interface/]
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Reviewed-by: Paul Durrant <paul@xen.org>

Merge tag 'buildsys-qom-qdev-ui-20230227' of https://github.com/philmd/qemu into staging

- buildsys
  - Various header cleaned up (removing pointless headers)
  - Mark various files/code user/system specific
  - Make various objects target-independent
  - Remove tswapN() calls from dump.o
  - Suggest g_assert_not_reached() instead of assert(0)

- qdev / qom
  - Replace various container_of() by QOM cast macros
  - Declare some QOM macros using OBJECT_DECLARE_TYPE()
  - Embed OHCI QOM child in SM501 chipset

- hw (ISA & IDE)
  - add some documentation, improve function names
  - un-inline, open-code few functions
  - have ISA API accessing IRQ/DMA prefer ISABus over ISADevice
  - Demote IDE subsystem maintenance to "Odd Fixes"

- ui: Improve Ctrl+Alt hint on Darwin Cocoa

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEE+qvnXhKRciHc/Wuy4+MsLN6twN4FAmP9IeAACgkQ4+MsLN6t
# wN7bdQ//SxJYJuQvqTT6s+O0LmP6NbqvhxCXX7YAwK2jCTM+zTgcqqRZCcisLQol
# 3ENu2UhnZmiLKHSOxatOVozbws08/u8Vl+WkW4UTMUb1yo5KPaPtq808Y95RdAJB
# 7D7B5juDGnFRAHXZz38zVk9uIuEkm+Po/pD0JQa+upBtAAgOJTqGavDNSR5+T0Yl
# VjGdwK0b10skPqiF6OABYoy/4IFHVJJFIbARZh+a7hrF0llsbzUts5JiYsOxEEHQ
# t3woUItdMnS1m0+Ty4AQ8m0Yv9y4HZOIzixvsZ+vChj5ariwUhL9/7wC/s/UCYEg
# gKVA5X8R6n/ME6DScK99a+CyR/MXkz70b/rOUZxoutXhV3xdh4X1stL4WN9W/m3z
# D4i4ZrUsDUcKCGWlj49of/dKbOPwk1+e/mT0oDZD6JzG0ODjfdVxvJ/JEV2iHgS3
# WqHuSKzX/20H9j7/MgfbQ0HjBFOQ8tl781vQzhD+y+cF/IiTsHhrE6esIWho4bob
# kfSdVydUWWRnBsnyGoRZXoEMX9tn+pu0nKxEDm2Bo2+jajsa0aZZPokgjxaz4MnD
# Hx+/p1E+8IuOn05JgzQSgTJmKFdSbya203tXIsTo1kL2aJTJ6QfMvgEPP/fkn+lS
# oQyVBFZmb1JDdTM1MxOncnlWLg74rp/CWEc+u5pSdbxMO/M/uac=
# =AV/+
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 27 Feb 2023 21:34:24 GMT
# gpg:                using RSA key FAABE75E12917221DCFD6BB2E3E32C2CDEADC0DE
# gpg: Good signature from "Philippe Mathieu-Daudé (F4BUG) <f4bug@amsat.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: FAAB E75E 1291 7221 DCFD  6BB2 E3E3 2C2C DEAD C0DE

* tag 'buildsys-qom-qdev-ui-20230227' of https://github.com/philmd/qemu: (125 commits)
  ui/cocoa: user friendly characters for release mouse
  dump: Add create_win_dump() stub for non-x86 targets
  dump: Simplify compiling win_dump.o by introducing win_dump_available()
  dump: Clean included headers
  dump: Replace TARGET_PAGE_SIZE -> qemu_target_page_size()
  dump: Replace tswapN() -> cpu_to_dumpN()
  hw/ide/pci: Add PCIIDEState::isa_irq[]
  hw/ide/via: Replace magic 2 value by ARRAY_SIZE / MAX_IDE_DEVS
  hw/ide/piix: Refactor pci_piix_init_ports as pci_piix_init_bus per bus
  hw/ide/piix: Pass Error* to pci_piix_init_ports() for better error msg
  hw/ide/piix: Remove unused includes
  hw/ide/pci: Unexport bmdma_active_if()
  hw/ide/ioport: Remove unnecessary includes
  hw/ide: Declare ide_get_[geometry/bios_chs_trans] in 'hw/ide/internal.h'
  hw/ide: Rename idebus_active_if() -> ide_bus_active_if()
  hw/ide: Rename ide_init2() -> ide_bus_init_output_irq()
  hw/ide: Rename ide_exec_cmd() -> ide_bus_exec_cmd()
  hw/ide: Rename ide_register_restart_cb -> ide_bus_register_restart_cb
  hw/ide: Rename ide_create_drive() -> ide_bus_create_drive()
  hw/ide: Rename ide_set_irq() -> ide_bus_set_irq()
  ...

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

ui/cocoa: user friendly characters for release mouse

While mouse is grabbed, window title contains a hint for the user what
keyboard keys to press to release the mouse. Make that hint text a bit
more user friendly for a Mac user:

 - Replace "Ctrl" and "Alt" by appropriate symbols for those keyboard
   keys typically displayed for them on a Mac (encode those symbols by
   using UTF-8 characters).

 - Drop " + " in between the keys, as that's not common on macOS for
   documenting keyboard shortcuts.

 - Convert lower case "g" to upper case "G", as that's common on macOS.

 - Add one additional space at start and end of key stroke set, to
   visually separate the key strokes from the rest of the text.

Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <E1pAClj-0003Jo-OB@lizzy.crudebyte.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

dump: Add create_win_dump() stub for non-x86 targets

Implement the non-x86 create_win_dump(). We can remove
the last TARGET_X86_64 #ifdef'ry in dump.c, which thus
becomes target-independent. Update meson accordingly.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230225094903.53167-6-philmd@linaro.org>

dump: Simplify compiling win_dump.o by introducing win_dump_available()

To make dump.c less target dependent, move the TARGET_X86_64 #ifdef'ry
from dump.c to win_dump.c (introducing a win_dump_available() method
there). By doing so we can build win_dump.c on any target, and
simplify the meson rule.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230225094903.53167-5-philmd@linaro.org>

dump: Clean included headers

"qemu/win_dump_defs.h" is only required by win_dump.c,
but win_dump.h requires "sysemu/dump.h" which declares
the DumpState type. Remove various unused headers.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230225094903.53167-4-philmd@linaro.org>

dump: Replace TARGET_PAGE_SIZE -> qemu_target_page_size()

TARGET_PAGE_SIZE is target specific. In preparation of
making dump.c target-agnostic, replace the compile-time
TARGET_PAGE_SIZE definition by runtime qemu_target_page_size().

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230225094903.53167-3-philmd@linaro.org>

dump: Replace tswapN() -> cpu_to_dumpN()

All uses of tswap in that file are wrong, and should be using
cpu_to_dumpN, which correctly tests the endianness of the output.

Reported-by: Richard Henderson <richard.henderson@linaro.org>
Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230225094903.53167-2-philmd@linaro.org>

hw/ide/pci: Add PCIIDEState::isa_irq[]

These legacy ISA IRQs allow the PIIX IDE functions to be wired up in
their south bridges and the VIA IDE functions to disuse
PCI_INTERRUPT_LINE as outlined in https://lists.nongnu.org/archive/html/qemu-devel/2020-03/msg01707.html

Suggested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230126211740.66874-7-shentey@gmail.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/ide/via: Replace magic 2 value by ARRAY_SIZE / MAX_IDE_DEVS

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Acked-by: John Snow <jsnow@redhat.com>
Message-Id: <20210511041848.2743312-5-f4bug@amsat.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/ide/piix: Refactor pci_piix_init_ports as pci_piix_init_bus per bus

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-21-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

hw/ide/piix: Pass Error* to pci_piix_init_ports() for better error msg

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-20-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

hw/ide/piix: Remove unused includes

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-19-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

hw/ide/pci: Unexport bmdma_active_if()

The function is only used inside ide/pci.c, so doesn't need to be exported.

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-18-philmd@linaro.org>

hw/ide/ioport: Remove unnecessary includes

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-17-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

hw/ide: Declare ide_get_[geometry/bios_chs_trans] in 'hw/ide/internal.h'

ide_get_geometry() and ide_get_bios_chs_trans() are only
used by the TYPE_PC_MACHINE.
"hw/ide.h" is a mixed bag of lost IDE declarations. In order
to remove this (almost) pointless header soon, move these
declarations to "hw/ide/internal.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230220091358.17038-18-philmd@linaro.org>

hw/ide: Rename idebus_active_if() -> ide_bus_active_if()

idebus_active_if() operates on a IDEBus; rename it as
ide_bus_active_if() to emphasize its first argument
is a IDEBus.

Mechanical change using:

  $ sed -i -e 's/idebus_active_if/ide_bus_active_if/g' \
        $(git grep -l idebus_active_if)

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-16-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

hw/ide: Rename ide_init2() -> ide_bus_init_output_irq()

ide_init2() initializes a IDEBus, and set its output IRQ.
To emphasize this, rename it as ide_bus_init_output_irq().

Mechanical change using:

  $ sed -i -e 's/ide_init2/ide_bus_init_output_irq/g' \
        $(git grep -l ide_init2)

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-15-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

hw/ide: Rename ide_exec_cmd() -> ide_bus_exec_cmd()

ide_exec_cmd() operates on a IDEBus; rename it as
ide_bus_exec_cmd() to emphasize its first argument
is a IDEBus.

Mechanical change using:

  $ sed -i -e 's/ide_exec_cmd/ide_bus_exec_cmd/g' \
        $(git grep -wl ide_exec_cmd)

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-14-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

hw/ide: Rename ide_register_restart_cb -> ide_bus_register_restart_cb

ide_register_restart_cb() operates on a IDEBus; rename it as
ide_bus_register_restart_cb() to emphasize its first argument
is a IDEBus.

Mechanical change using:

  $ sed -i -e 's/ide_register_restart_cb/ide_bus_register_restart_cb/g' \
    $(git grep -l ide_register_restart_cb)

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-13-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

hw/ide: Rename ide_create_drive() -> ide_bus_create_drive()

ide_create_drive() operates on a IDEBus; rename it as
ide_bus_create_drive() to emphasize its first argument
is a IDEBus.

Mechanical change using:

  $ sed -i -e 's/ide_create_drive/ide_bus_create_drive/g' \
        $(git grep -wl ide_create_drive)

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-12-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

hw/ide: Rename ide_set_irq() -> ide_bus_set_irq()

ide_set_irq() operates on a IDEBus; rename it as
ide_bus_set_irq() to emphasize its first argument
is a IDEBus.

Mechanical change using:

  $ sed -i -e 's/ide_set_irq/ide_bus_set_irq/g' \
        $(git grep -l ide_set_irq)

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-11-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

hw/ide: Un-inline ide_set_irq()

Only include "hw/irq.h" where appropriate.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-10-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

hw/ide: Include 'exec/ioport.h' instead of 'hw/isa/isa.h'

The IDEBus structure has PortioList fields, so we need its
declarations from "exec/ioport.h". "hw/isa/isa.h" is not required.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-9-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

hw/ide: Remove unused 'qapi/qapi-types-run-state.h'

Missed in commit d7458e7754 ("hw/ide/internal: Remove unused
DMARestartFunc typedef") which removed the single use of RunState.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-8-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

hw/ide/atapi: Restrict 'scsi/constants.h' inclusion

Only atapi.c requires the SCSI constants. No need to include
it in all files including "hw/ide/internal.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230215112712.23110-7-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

hw/ide/isa: Remove intermediate ISAIDEState::irq variable

The intermediate ISAIDEState::irq variable just add noise, remove it.

Message-Id: <20230215112712.23110-6-philmd@linaro.org>
Reviewed-by: Bernhard Beschow <shentey@gmail.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/ide/isa: Extract TYPE_ISA_IDE declarations to 'hw/ide/isa.h'

"hw/ide.h" is a mixed bag of lost IDE declarations.

Extract isa_ide_init() and the TYPE_ISA_IDE QOM declarations
to a new "hw/ide/isa.h" header.

Rename ISAIDEState::isairq as 'irqnum' to emphasize this is
not a qemu_irq object but the number (index) of an ISA IRQ.

Message-Id: <20230215112712.23110-5-philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Bernhard Beschow <shentey@gmail.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/ide/mmio: Extract TYPE_MMIO_IDE declarations to 'hw/ide/mmio.h'

"hw/ide.h" is a mixed bag of lost IDE declarations.

Extract mmio_ide_init_drives() and the TYPE_MMIO_IDE QOM
declarations to a new "hw/ide/mmio.h" header.

Document the SysBus interface.

Message-Id: <20230215112712.23110-4-philmd@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/ide/mmio: Use CamelCase for MMIO_IDE state name

Following docs/devel/style.rst guidelines, rename MMIOIDEState
as IdeMmioState.

Having the structure name and its typedef named equally,
we can manually convert from the old DECLARE_INSTANCE_CHECKER()
macro to the more recent OBJECT_DECLARE_SIMPLE_TYPE().

Note, due to that name mismatch, this macro wasn't automatically
converted during commit 8063396bf3 ("Use OBJECT_DECLARE_SIMPLE_TYPE
when possible").

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230220091358.17038-3-philmd@linaro.org>

hw/ide/ahci: Trace ncq write command as write instead of read

Fixes: e4baa9f00b ("AHCI: Replace DPRINTF with trace-events")
Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: John Snow <jsnow@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Message-Id: <20230217103130.42077-1-f.ebner@proxmox.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/i386/xen: Remove unused 'hw/ide.h' include from header

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Paul Durrant <paul@xen.org>
Message-Id: <20230220092707.22584-1-philmd@linaro.org>

MAINTAINERS: Mark IDE and Floppy as "Odd Fixes"

I have not been able to give these devices the love they need for a
while now. Update the maintainers file to reflect the truth of the
matter.

Signed-off-by: John Snow <jsnow@redhat.com>
Message-Id: <20230206182544.711117-1-jsnow@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

scripts/checkpatch.pl: Do not allow assert(0)

Since commit 262a69f428 ("osdep.h: Prohibit disabling assert()
in supported builds") we can not build QEMU with NDEBUG (or
G_DISABLE_ASSERT) defined, thus 'assert(0)' always aborts QEMU.

However some static analyzers / compilers doesn't notice NDEBUG
can't be defined and emit warnings if code is used after an
'assert(0)' call.

Apparently such compiler isn't as clever with G_DISABLE_ASSERT,
so we can silent these warnings by using g_assert_not_reached()
which is easier to read anyway.

In order to avoid these annoying warnings, add a checkpatch rule
to prohibit 'assert(0)'. Suggest using g_assert_not_reached()
instead. For example when reverting the previous patch we get:

  ERROR: use g_assert_not_reached() instead of assert(0)
  #21: FILE: target/ppc/dfp_helper.c:124:
  +            assert(0); /* cannot get here */

  ERROR: use g_assert_not_reached() instead of assert(0)
  #30: FILE: target/ppc/dfp_helper.c:141:
  +            assert(0); /* cannot get here */

  total: 2 errors, 0 warnings, 16 lines checked

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230221232520.14480-3-philmd@linaro.org>

block/vvfat: Remove pointless check of NDEBUG

Since commit 262a69f428 ("osdep.h: Prohibit disabling
assert() in supported builds") 'NDEBUG' can not be defined,
so '#ifndef NDEBUG' is dead code. Remove it.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230221232520.14480-5-philmd@linaro.org>

hw/ppc/sam460ex: Correctly set MAL properties

MAL properties are declared as uint8_t:

  static Property ppc4xx_mal_properties[] = {
      DEFINE_PROP_UINT8("txc-num", Ppc4xxMalState, txcnum, 0),
      DEFINE_PROP_UINT8("rxc-num", Ppc4xxMalState, rxcnum, 0),
      DEFINE_PROP_END_OF_LIST(),
  };

Correct the API use by setting the property using
qdev_prop_set_uint8(). No behavioral change.

Fixes: da116a8aab ("ppc/ppc405: QOM'ify MAL")
Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230203145536.17585-7-philmd@linaro.org>

hw/display/sm501: Add fallbacks to pixman routines

Pixman may return false if it does not have a suitable implementation.
Add fallbacks to handle such cases.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reported-by: Rene Engel <ReneEngel80@emailn.de>
Tested-by: Rene Engel <ReneEngel80@emailn.de>
Message-Id: <20ed9442a0146238254ccc340c0d1efa226c6356.1677445307.git.balaton@eik.bme.hu>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/display/sm501: Implement more 2D raster operations

Add simple implementation for two raster operations that are used by
AmigaOS which fixes graphics problems in some programs using these.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reported-by: Rene Engel <ReneEngel80@emailn.de>
Tested-by: Rene Engel <ReneEngel80@emailn.de>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <17ef3c59dc7868f75034e9ebe21e2999c8f718d4.1677445307.git.balaton@eik.bme.hu>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/display/sm501: Alias 'dma-offset' QOM property in chipset object

No need to use an intermediate 'dma-offset' property in the
chipset object. Alias the property, so when the machine (here
r2d-plus) sets the value on the chipset, it is propagated to
the OHCI object.

Note we can rename the chipset 'base' property as 'dma-offset'
since the object is a non-user-creatable sysbus type.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
Message-Id: <20230203145536.17585-12-philmd@linaro.org>

hw/display/sm501: Embed OHCI QOM child in chipset

Note this device doesn't implement unrealize().

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: BALATON Zoltan <balaton@eik.bme.hu>
Message-Id: <20230203145536.17585-11-philmd@linaro.org>

hw/usb/xhci-nec: Replace container_of() by NEC_XHCI() QOM cast macro

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230220150515.32549-8-philmd@linaro.org>

hw/usb/xhci-nec: Declare QOM macros for NEC_XHCI

NEC_XHCI is a QOM object type. Declare its macros /
typedefs using OBJECT_DECLARE_SIMPLE_TYPE().

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230220150515.32549-7-philmd@linaro.org>

hw/usb/uhci: Replace container_of() by UHCI_GET_CLASS() QOM macro

By using the QOM UHCI_GET_CLASS() cast macro we don't to
use the intermediate PCIDeviceClass variable.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230220150515.32549-6-philmd@linaro.org>

hw/usb/uhci: Declare QOM macros using OBJECT_DECLARE_TYPE()

The automatic conversion done during commit a489d1951c
("Use OBJECT_DECLARE_TYPE when possible") missed this
model because the typedefs are in a different file unit
(hcd-uhci.c) than where the DECLARE_INSTANCE_CHECKER()
is (hcd-uhci.h). Manually convert to OBJECT_DECLARE_TYPE().

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230220150515.32549-5-philmd@linaro.org>

hw/usb/ohci: Fix typo

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <03599fd4db313ac4f651cceb43340109ad6a14b8.1676916640.git.balaton@eik.bme.hu>

hw/usb/ohci: Add trace points for register access

To help debugging add trace points that print values read from or
written to the device's registers.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Message-Id: <1bb4985e5dfc1df5a290e77f76fd827ae3592ab7.1676916640.git.balaton@eik.bme.hu>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/usb/ohci: Move a function next to where it is used

The ohci_port_set_if_connected() function is only used by
ohci_port_set_status(), move next to it to have them at the same place.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <46411d4980ab0fba61ab0d2209a939fdc41eb573.1676916640.git.balaton@eik.bme.hu>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/usb/ohci: Code style fix missing braces and extra parenthesis

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <aaa3ddee99c7677d6cc137f637982e94267b99b6.1676916640.git.balaton@eik.bme.hu>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/usb/ohci: Code style fix white space errors

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <c9b99b3555dcd03194a8950b810f5e1b4b4bd5d3.1676916640.git.balaton@eik.bme.hu>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/usb/ohci: Code style fix comments

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <9b0aadedc7c4780fefdc27f14f72ac9003032fbf.1676916639.git.balaton@eik.bme.hu>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/usb/ohci: Use OHCIState type definition

Forward-define the type first, then use it for the ohci_die() handler.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230220150515.32549-3-philmd@linaro.org>

hw/usb/ohci: Include missing 'sysbus.h' header

Avoid when including "hw/usb/hcd-ohci.h":

  hw/usb/hcd-ohci.h:100:5: error: unknown type name 'SysBusDevice'
      SysBusDevice parent_obj;
      ^

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20230203113650.78146-6-philmd@linaro.org>

hw/usb/u2f: Declare QOM macros using OBJECT_DECLARE_TYPE()

hw/usb/u2f.h was added by commit 80e267f1d1 ("hw/usb: Add
U2F key base class"), almost the same time of the automatic
conversion done by commit c821774a3b ("Use OBJECT_DECLARE_TYPE
where posible"). Manually convert to OBJECT_DECLARE_TYPE().

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230220150515.32549-9-philmd@linaro.org>

hw/usb/dev-smartcard-reader: Avoid forward-declaring CCIDBus

To avoid forward-declaring CCIDBus, declare CCID_BUS QOM
definitions before its use in the USBCCIDState structure.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230220150515.32549-2-philmd@linaro.org>

hw/audio/ac97: Split off some definitions to a header

These can be shared with other AC97 implementations.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <6f0980fdc3753624be6f3935a6ab0a2dc1df4b30.1677445307.git.balaton@eik.bme.hu>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/audio/ac97: Replace container_of() by AC97() QOM cast macro

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230220131837.26292-5-philmd@linaro.org>

hw/audio/es1370: Replace container_of() by ES1370() QOM cast macro

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230220131837.26292-4-philmd@linaro.org>

hw/audio/es1370: Avoid forward-declaring ES1370State

To avoid forward-declaring ES1370State, declare ES1370 QOM
definitions before its use in the chan_bits structure.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230220131837.26292-3-philmd@linaro.org>

hw/audio/hda-codec: Avoid forward-declaring HDAAudioState

To avoid forward-declaring HDAAudioState, declare HDA_AUDIO QOM
definitions before its use in the HDAAudioStream structure.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230220131837.26292-2-philmd@linaro.org>

hw/timer/hpet: Include missing 'hw/qdev-properties.h' header

Avoid when refactoring unrelated headers:

  hw/timer/hpet.c:776:39: error: array has incomplete element type 'Property' (aka 'struct Property')
  static Property hpet_device_properties[] = {
                                        ^
  hw/timer/hpet.c:777:5: error: implicit declaration of function 'DEFINE_PROP_UINT8' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
      DEFINE_PROP_UINT8("timers", HPETState, num_timers, HPET_MIN_TIMERS),
      ^

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230215174353.37097-2-philmd@linaro.org>

hw/rtc: Rename rtc_[get|set]_memory -> mc146818rtc_[get|set]_cmos_data

rtc_get_memory() and rtc_set_memory() helpers only work with
TYPE_MC146818_RTC devices. 'memory' in their name refer to
the CMOS region. Rename them as mc146818rtc_get_cmos_data()
and mc146818rtc_set_cmos_data() to be explicit about what
they are doing.

Mechanical change doing:

  $ sed -i -e 's/rtc_set_memory/mc146818rtc_set_cmos_data/g' \
        $(git grep -wl rtc_set_memory)
  $ sed -i -e 's/rtc_get_memory/mc146818rtc_get_cmos_data/g' \
        $(git grep -wl rtc_get_memory)

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230210233116.80311-4-philmd@linaro.org>

hw/rtc/mc146818rtc: Pass MC146818RtcState instead of ISADevice argument

rtc_get_memory() and rtc_set_memory() methods can not take any
TYPE_ISA_DEVICE object. They expect a TYPE_MC146818_RTC one.

Simplify the API by passing a MC146818RtcState.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230210233116.80311-3-philmd@linaro.org>

hw/rtc/mc146818rtc: Rename RTCState -> MC146818RtcState

RTCState only represents a Motorola MC146818 model,
not any RTC chipset. Rename the structure as MC146818RtcState
using:

  $ sed -i -e s/RTCState/MC146818RtcState/g $(git grep -wl RTCState)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230210233116.80311-2-philmd@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw: Replace isa_get_irq() by isa_bus_get_irq() when ISABus is available

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230215161641.32663-4-philmd@linaro.org>

hw/isa: Factor isa_bus_get_irq() out of isa_get_irq()

isa_get_irq() was added in commit 3a38d437ca
("Add isa_reserve_irq()" Fri Aug 14 11:36:15 2009) as:

    a temporary interface to be used to allocate ISA IRQs for
    devices which have not yet been converted to qdev, and for
    special cases which are not suited for qdev conversions,
    such as the 'ferr'.

We still use it 14 years later, using the global 'isabus'
singleton. In order to get rid of such *temporary* interface,
extract isa_bus_get_irq() which can take any ISABus* object.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230215161641.32663-3-philmd@linaro.org>

hw/isa: Rename isa_get_dma() -> isa_bus_get_dma()

isa_get_dma() returns a DMA channel handler from an ISABus.
To emphasize this, rename it as isa_bus_get_dma().

Mechanical change using:

  $ sed -i -e 's/isa_get_dma/isa_bus_get_dma/g' \
        $(git grep -l isa_get_dma)

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20230215161641.32663-2-philmd@linaro.org>

hw/isa: Use isa_address_space_io() to reduce access on global 'isabus'

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230210163744.32182-11-philmd@linaro.org>

hw/isa: Rename isa_bus_irqs() -> isa_bus_register_input_irqs()

isa_bus_irqs() register an array of input IRQs on
the ISA bus. Rename it as isa_bus_register_input_irqs().

Mechanical change using:

 $ sed -i -e 's/isa_bus_irqs/isa_bus_register_input_irqs/g' \
   $(git grep -wl isa_bus_irqs)

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230210163744.32182-10-philmd@linaro.org>

hw/isa: Un-inline isa_bus_from_device()

No point in inlining isa_bus_from_device() which is only
used at device realization time.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230210163744.32182-9-philmd@linaro.org>
Reviewed-by: Bernhard Beschow <shentey@gmail.com>

hw/isa: Reorder to separate ISABus* vs ISADevice* functions

Separate functions taking an ISABus* argument versus
functions taking a ISADevice* one.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230210163744.32182-8-philmd@linaro.org>