spapr: nested: Add Power11 capability support for Nested PAPR guests in TCG L0

The Power11 architected and raw mode support in Linux was merged via [1]
and the corresponding support in QEMU is waiti

spapr: nested: Add Power11 capability support for Nested PAPR guests in TCG L0

The Power11 architected and raw mode support in Linux was merged via [1]
and the corresponding support in QEMU is waiting to be added by [2]
which in V6 currently.

Add the Power11 capabilities and the required handling in TCG L0
implementation of the "Nested PAPR API".

Note: This patch is based on [2].

[1] https://lore.kernel.org/all/20240221044623.1598642-1-mpe@ellerman.id.au/
[2] https://lore.kernel.org/all/20240731055022.696051-1-adityag@linux.ibm.com/

Signed-off-by: Amit Machhiwal <amachhiw@linux.ibm.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

spapr: nested: Add support for DPDES SPR in GSB for TCG L0

The DPDES support for doorbell emulation and handling for KVM on PAPR
guests was added in Linux via [1]. Subsequently, a new GSB (Guest Sta

spapr: nested: Add support for DPDES SPR in GSB for TCG L0

The DPDES support for doorbell emulation and handling for KVM on PAPR
guests was added in Linux via [1]. Subsequently, a new GSB (Guest State
Buffer) element for DPDES was added in Linux; the same has been missing
in TCG L0 implementation. Add support for DPDES register's APIv2 GSB
element and required handling in `spapr_nested.c`.

Currently, booting a KVM guest inside a QEMU TCG guest fails with the
following crash. The crash occurs while handling the GUEST_RUN_VCPU
hcall made in TCG L0. In the hcall handling path, map_and_getset_state()
calls getset_state(), which, in turn, calls guest_state_request_check()
to validate the GSR (Guest State Request) elements. During this process,
guest_state_request_check() iterates over the GSR elements and receives
a NULL return code from guest_state_element_type_find() for the type
variable corresponding to the DPDES register's elemetn ID (which was
unknown to TCG L0). Subsequently, getset_state() returns H_P3,
ultimately leading to the hcall failure and causing the KVM guest to
crash.

KVM: unknown exit, hardware reason ffffffffffffffea
[ ... KVM register dump ... ]

Fix this by adding the required support in TCG L0 implementation of
APIv2.

[1] https://lore.kernel.org/all/20240605113913.83715-1-gautam@linux.ibm.com/

Fixes: 4a575f9a0567 ("spapr: nested: Initialize the GSB elements lookup table.")
Suggested-by: Harsh Prateek Bora <harshpb@linux.ibm.com>
Signed-off-by: Amit Machhiwal <amachhiw@linux.ibm.com>
Reviewed-by: Harsh Prateek Bora <harshpb@linux.ibm.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/spapr: remove deprecated machine pseries-2.12

Commit 0cac0f1b964 marked pseries-2.12 machines as deprecated
with reasons mentioned in its commit log.
Removing pseries-2.12 specific code with thi

ppc/spapr: remove deprecated machine pseries-2.12

Commit 0cac0f1b964 marked pseries-2.12 machines as deprecated
with reasons mentioned in its commit log.
Removing pseries-2.12 specific code with this patch.

While at it, also remove pre-3.0-migration hacks introduced for backward
compatibility which are now turned useless.

Suggested-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Harsh Prateek Bora <harshpb@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/spapr: remove deprecated machine pseries-2.9

Commit 1392617d3576 intended to tag pseries-2.1 - 2.11 machines as
deprecated with reasons mentioned in its commit log.
Removing pseries-2.9 specific

ppc/spapr: remove deprecated machine pseries-2.9

Commit 1392617d3576 intended to tag pseries-2.1 - 2.11 machines as
deprecated with reasons mentioned in its commit log.
Removing pseries-2.9 specific code with this patch for now.

While at it, also remove the pre-2.10 migration hacks which now become
obsolete.

Suggested-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Harsh Prateek Bora <harshpb@linux.ibm.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/spapr: remove deprecated machine pseries-2.5

Commit 1392617d3576 intended to tag pseries-2.1 - 2.11 machines as
deprecated with reasons mentioned in its commit log.
Removing pseries-2.5 specific

ppc/spapr: remove deprecated machine pseries-2.5

Commit 1392617d3576 intended to tag pseries-2.1 - 2.11 machines as
deprecated with reasons mentioned in its commit log.
Removing pseries-2.5 specific code with this patch for now.

Also drop sPAPRMachineClass::use_ohci_by_default which is now useless.

Suggested-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Harsh Prateek Bora <harshpb@linux.ibm.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/spapr: remove deprecated machine pseries-2.4

Commit 1392617d3576 intended to tag pseries-2.1 - 2.11 machines as
deprecated with reasons mentioned in its commit log.
Removing pseries-2.4 specific

ppc/spapr: remove deprecated machine pseries-2.4

Commit 1392617d3576 intended to tag pseries-2.1 - 2.11 machines as
deprecated with reasons mentioned in its commit log.
Removing pseries-2.4 specific code with this patch for now.

While at it, also remove SpaprMachineClass::dr_lmb_enabled which is
now turned useless.

Suggested-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Harsh Prateek Bora <harshpb@linux.ibm.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

mac_dbdma: Remove leftover `dma_memory_unmap` calls

These were passing a NULL buffer pointer unconditionally, which happens
to behave in a mostly benign way (except for the chance of an excess
memor

mac_dbdma: Remove leftover `dma_memory_unmap` calls

These were passing a NULL buffer pointer unconditionally, which happens
to behave in a mostly benign way (except for the chance of an excess
memory region unref and a bounce buffer leak). Per the function comment,
this was never meant to be accepted though, and triggers an assertion
with the "softmmu: Support concurrent bounce buffers" change.

Given that the code in question never sets up any mappings, just remove
the unnecessary dma_memory_unmap calls along with the DBDMA_io struct
fields that are now entirely unused.

Signed-off-by: Mattias Nissler <mnissler@rivosinc.com>
Message-Id: <20240916175708.1829059-1-mnissler@rivosinc.com>
Fixes: be1e343995 ("macio: switch over to new byte-aligned DMA helpers")
Reviewed-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>

show more ...

pnv/xive2: Dump more END state with 'info pic'

Additional END state 'info pic' information as added. The 'ignore',
'crowd' and 'precluded escalation control' bits of an Event Notification
Descripto

pnv/xive2: Dump more END state with 'info pic'

Additional END state 'info pic' information as added. The 'ignore',
'crowd' and 'precluded escalation control' bits of an Event Notification
Descriptor are all used when delivering an interrupt targeting a VP-group
or crowd.

Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Signed-off-by: Michael Kowal <kowal@linux.vnet.ibm.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

pnv/xive2: Move xive2_nvp_pic_print_info() to xive2.c

Moving xive2_nvp_pic_print_info() to align with the other "pic_print_info"
functions.

Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Si

pnv/xive2: Move xive2_nvp_pic_print_info() to xive2.c

Moving xive2_nvp_pic_print_info() to align with the other "pic_print_info"
functions.

Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Signed-off-by: Michael Kowal <kowal@linux.vnet.ibm.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

pnv/xive: Support cache flush and queue sync inject with notifications

Adds support for writing a completion notification byte in memory
whenever a cache flush or queue sync inject operation is requ

pnv/xive: Support cache flush and queue sync inject with notifications

Adds support for writing a completion notification byte in memory
whenever a cache flush or queue sync inject operation is requested by
software. QEMU does not cache any of the XIVE data that is in memory and
therefore it simply writes the completion notification byte at the time
that the operation is requested.

Co-authored-by: Glenn Miles <milesg@linux.vnet.ibm.com>

Signed-off-by: Glenn Miles <milesg@linux.vnet.ibm.com>
Signed-off-by: Michael Kowal <kowal@linux.vnet.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

hw/ppc: SPI controller wiring to P10 chip

In this commit, create SPI controller on p10 chip and connect cs irq.

The QOM tree of pnv-spi and seeprom are.
/machine (powernv10-machine)
/chip[0] (pow

hw/ppc: SPI controller wiring to P10 chip

In this commit, create SPI controller on p10 chip and connect cs irq.

The QOM tree of pnv-spi and seeprom are.
/machine (powernv10-machine)
/chip[0] (power10_v2.0-pnv-chip)
/pib_spic[2] (pnv-spi)
/pnv-spi-bus.2 (SSI)
/xscom-spi[0] (memory-region)

/machine (powernv10-machine)
/peripheral-anon (container)
/device[0] (25csm04)
/WP#[0] (irq)
/ssi-gpio-cs[0] (irq)

(qemu) qom-get /machine/peripheral-anon /device[76] "parent_bus"
"/machine/chip[0]/pib_spic[2]/pnv-spi-bus.2"

Signed-off-by: Chalapathi V <chalapathi.v@linux.ibm.com>
Reviewed-by: Glenn Miles <milesg@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

hw/ssi: Add SPI model

SPI controller device model supports a connection to a single SPI responder.
This provide access to SPI seeproms, TPM, flash device and an ADC controller.

All SPI function con

hw/ssi: Add SPI model

SPI controller device model supports a connection to a single SPI responder.
This provide access to SPI seeproms, TPM, flash device and an ADC controller.

All SPI function control is mapped into the SPI register space to enable full
control by firmware. In this commit SPI configuration component is modelled
which contains all SPI configuration and status registers as well as the hold
registers for data to be sent or having been received.

An existing QEMU SSI framework is used and SSI_BUS is created.

Signed-off-by: Chalapathi V <chalapathi.v@linux.ibm.com>
Reviewed-by: Caleb Schlossin <calebs@linux.vnet.ibm.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Reviewed-by: Glenn Miles <milesg@linux.ibm.com>
[np: Fix FDT macro compile for qtest]
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/pnv: Remove ppc target dependency from pnv_xscom.h

In this commit target specific dependency from include/hw/ppc/pnv_xscom.h
has been removed so that pnv_xscom.h can be included outside hw/ppc.

ppc/pnv: Remove ppc target dependency from pnv_xscom.h

In this commit target specific dependency from include/hw/ppc/pnv_xscom.h
has been removed so that pnv_xscom.h can be included outside hw/ppc.

Signed-off-by: Chalapathi V <chalapathi.v@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Reviewed-by: Caleb Schlossin <calebs@linux.vnet.ibm.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/pnv: Add an LPAR per core machine option

Recent POWER CPUs can operate in "LPAR per core" or "LPAR per thread"
modes. In per-core mode, some SPRs and IPI doorbells are shared between
threads in

ppc/pnv: Add an LPAR per core machine option

Recent POWER CPUs can operate in "LPAR per core" or "LPAR per thread"
modes. In per-core mode, some SPRs and IPI doorbells are shared between
threads in a core. In per-thread mode, supervisor and user state is
not shared between threads.

OpenPOWER systems after POWER8 use LPAR per thread mode, and it is
required for KVM. Enterprise systems use LPAR per core mode, as they
partition the machine by core.

Implement a lpar-per-core machine option for powernv machines. This
is fixed true for POWER8 machines, and defaults off for P9 and P10.

With this change, powernv8 SMT now works sufficiently to run Linux,
with a single socket. Multi-threaded KVM guests still have problems,
as does multi-socket Linux boot.

Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/pnv: Implement POWER10 PC xscom registers for direct controls

The PC unit in the processor core contains xscom registers that provide
low level status and control of the CPU.

This implements "d

ppc/pnv: Implement POWER10 PC xscom registers for direct controls

The PC unit in the processor core contains xscom registers that provide
low level status and control of the CPU.

This implements "direct controls", sufficient for skiboot firmware,
which uses it to send NMI IPIs between CPUs.

POWER10 is sufficiently different from POWER9 (particularly with respect
to QME and special wakeup) that it is not trivial to implement POWER9
support by reusing the code.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/pnv: Add a CPU nmi and resume function

Power CPUs have an execution control facility that can pause, resume,
and cause NMIs, among other things. Add a function that will nmi a CPU
and resume it

ppc/pnv: Add a CPU nmi and resume function

Power CPUs have an execution control facility that can pause, resume,
and cause NMIs, among other things. Add a function that will nmi a CPU
and resume it if it was paused, in preparation for implementing the
control facility.

Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/pnv: Add POWER10 ChipTOD quirk for big-core

POWER10 has a quirk in its ChipTOD addressing that requires the even
small-core to be selected even when programming the odd small-core.
This allows s

ppc/pnv: Add POWER10 ChipTOD quirk for big-core

POWER10 has a quirk in its ChipTOD addressing that requires the even
small-core to be selected even when programming the odd small-core.
This allows skiboot chiptod init to run in big-core mode.

Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/pnv: Add a big-core mode that joins two regular cores

POWER9 and POWER10 machines come in two variants, big-core and
small-core. Big-core machines are SMT8 from software's point of view,
but the

ppc/pnv: Add a big-core mode that joins two regular cores

POWER9 and POWER10 machines come in two variants, big-core and
small-core. Big-core machines are SMT8 from software's point of view,
but the low level platform topology ("xscom registers and pervasive
addressing"), these look more like a pair of small cores ganged
together.

Presently the way this is modelled is to create one SMT8 PnvCore and add
special cases to xscom and pervasive for big-core mode that tries to
split this into two small cores, but this is becoming too complicated to
manage.

A better approach is to create 2 core structures and ganging them
together to look like an SMT8 core in TCG. Then the xscom and pervasive
models mostly do not need to differentiate big and small core modes.

This change adds initial mode bits and QEMU topology handling to
split SMT8 cores into 2xSMT4 cores.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/pnv: Extend chip_pir class method to TIR as well

The chip_pir chip class method allows the platform to set the PIR
processor identification register. Extend this to a more general
ID function wh

ppc/pnv: Extend chip_pir class method to TIR as well

The chip_pir chip class method allows the platform to set the PIR
processor identification register. Extend this to a more general
ID function which also allows the TIR to be set. This is in
preparation for "big core", which is a more complicated topology
of cores and threads.

Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/pnv: use class attribute to limit SMT threads for different machines

Use a class attribute to specify the number of SMT threads per core
permitted for different machines, 8 for powernv8 and 4 fo

ppc/pnv: use class attribute to limit SMT threads for different machines

Use a class attribute to specify the number of SMT threads per core
permitted for different machines, 8 for powernv8 and 4 for powernv9/10.

Reviewed-by: Cédric Le Goater <clg@redhat.com>
Reviewed-by: Harsh Prateek Bora <harshpb@linux.ibm.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

target/ppc: Move SPR indirect registers into PnvCore

SPRC/SPRD were recently added to all BookS CPUs supported, but
they are only tested on POWER9 and POWER10, so restrict them to
those CPUs.

SPR i

target/ppc: Move SPR indirect registers into PnvCore

SPRC/SPRD were recently added to all BookS CPUs supported, but
they are only tested on POWER9 and POWER10, so restrict them to
those CPUs.

SPR indirect scratch registers presently replicated per-CPU like
SMT SPRs, but the PnvCore is a better place for them since they
are restricted to P9/P10.

Also add SPR indirect read access to core thread state for POWER9
since skiboot accesses that when booting to check for big-core
mode.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/pnv: Move timebase state into PnvCore

The timebase state machine is per per-core state and can be driven
by any thread in the core. It is currently implemented as a hack
where the state is in a

ppc/pnv: Move timebase state into PnvCore

The timebase state machine is per per-core state and can be driven
by any thread in the core. It is currently implemented as a hack
where the state is in a CPU structure and only thread 0's state is
accessed by the chiptod, which limits programming the timebase
side of the state machine to thread 0 of a core.

Move the state out into PnvCore and share it among all threads.

Reviewed-by: Harsh Prateek Bora <harshpb@linux.ibm.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/pnv: Add pointer from PnvCPUState to PnvCore

This helps move core state from CPU to core structures.

Reviewed-by: Cédric Le Goater <clg@redhat.com>
Reviewed-by: Harsh Prateek Bora <harshpb@linu

ppc/pnv: Add pointer from PnvCPUState to PnvCore

This helps move core state from CPU to core structures.

Reviewed-by: Cédric Le Goater <clg@redhat.com>
Reviewed-by: Harsh Prateek Bora <harshpb@linux.ibm.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/pnv: Implement ADU access to LPC space

One of the functions of the ADU is indirect memory access engines that
send and receive data via ADU registers.

This implements the ADU LPC memory access

ppc/pnv: Implement ADU access to LPC space

One of the functions of the ADU is indirect memory access engines that
send and receive data via ADU registers.

This implements the ADU LPC memory access functionality sufficiently
for IBM proprietary firmware to access the UART and print characters
to the serial port as it does on real hardware.

This requires a linkage between adu and lpc, which allows adu to
perform memory access in the lpc space.

Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...

ppc/pnv: Begin a more complete ADU LPC model for POWER9/10

This implements a framework for an ADU unit model.

The ADU unit actually implements XSCOM, which is the bridge between MMIO
and PIB. Howev

ppc/pnv: Begin a more complete ADU LPC model for POWER9/10

This implements a framework for an ADU unit model.

The ADU unit actually implements XSCOM, which is the bridge between MMIO
and PIB. However it also includes control and status registers and other
functions that are exposed as PIB (xscom) registers.

To keep things simple, pnv_xscom.c remains the XSCOM bridge
implementation, and pnv_adu.c implements the ADU registers and other
functions.

So far, just the ADU no-op registers in the pnv_xscom.c default handler
are moved over to the adu model.

Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>

show more ...