ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
/*
|
|
|
|
* QEMU PowerPC XIVE2 interrupt controller model (POWER10)
|
|
|
|
*
|
|
|
|
* Copyright (c) 2019-2022, IBM Corporation.
|
|
|
|
*
|
|
|
|
* This code is licensed under the GPL version 2 or later. See the
|
|
|
|
* COPYING file in the top-level directory.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "qemu/osdep.h"
|
|
|
|
#include "qemu/log.h"
|
|
|
|
#include "qapi/error.h"
|
|
|
|
#include "target/ppc/cpu.h"
|
|
|
|
#include "sysemu/cpus.h"
|
|
|
|
#include "sysemu/dma.h"
|
|
|
|
#include "monitor/monitor.h"
|
|
|
|
#include "hw/ppc/fdt.h"
|
|
|
|
#include "hw/ppc/pnv.h"
|
include/hw/ppc: Split pnv_chip.h off pnv.h
PnvChipClass, PnvChip, Pnv8Chip, Pnv9Chip, and Pnv10Chip are defined
in pnv.h. Many users of the header don't actually need them. One
instance is this inclusion loop: hw/ppc/pnv_homer.h includes
hw/ppc/pnv.h for typedef PnvChip, and vice versa for struct PnvHomer.
Similar structs live in their own headers: PnvHomerClass and PnvHomer
in pnv_homer.h, PnvLpcClass and PnvLpcController in pci_lpc.h,
PnvPsiClass, PnvPsi, Pnv8Psi, Pnv9Psi, Pnv10Psi in pnv_psi.h, ...
Move PnvChipClass, PnvChip, Pnv8Chip, Pnv9Chip, and Pnv10Chip to new
pnv_chip.h, and adjust include directives. This breaks the inclusion
loop mentioned above.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20221222104628.659681-2-armbru@redhat.com>
2022-12-22 11:46:25 +01:00
|
|
|
#include "hw/ppc/pnv_chip.h"
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
#include "hw/ppc/pnv_core.h"
|
|
|
|
#include "hw/ppc/pnv_xscom.h"
|
|
|
|
#include "hw/ppc/xive2.h"
|
|
|
|
#include "hw/ppc/pnv_xive.h"
|
|
|
|
#include "hw/ppc/xive_regs.h"
|
|
|
|
#include "hw/ppc/xive2_regs.h"
|
|
|
|
#include "hw/ppc/ppc.h"
|
|
|
|
#include "hw/qdev-properties.h"
|
|
|
|
#include "sysemu/reset.h"
|
|
|
|
|
|
|
|
#include <libfdt.h>
|
|
|
|
|
|
|
|
#include "pnv_xive2_regs.h"
|
|
|
|
|
|
|
|
#undef XIVE2_DEBUG
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Virtual structures table (VST)
|
|
|
|
*/
|
|
|
|
#define SBE_PER_BYTE 4
|
|
|
|
|
|
|
|
typedef struct XiveVstInfo {
|
|
|
|
const char *name;
|
|
|
|
uint32_t size;
|
|
|
|
uint32_t max_blocks;
|
|
|
|
} XiveVstInfo;
|
|
|
|
|
|
|
|
static const XiveVstInfo vst_infos[] = {
|
|
|
|
|
|
|
|
[VST_EAS] = { "EAT", sizeof(Xive2Eas), 16 },
|
|
|
|
[VST_ESB] = { "ESB", 1, 16 },
|
|
|
|
[VST_END] = { "ENDT", sizeof(Xive2End), 16 },
|
|
|
|
|
|
|
|
[VST_NVP] = { "NVPT", sizeof(Xive2Nvp), 16 },
|
|
|
|
[VST_NVG] = { "NVGT", sizeof(Xive2Nvgc), 16 },
|
|
|
|
[VST_NVC] = { "NVCT", sizeof(Xive2Nvgc), 16 },
|
|
|
|
|
|
|
|
[VST_IC] = { "IC", 1 /* ? */ , 16 }, /* Topology # */
|
|
|
|
[VST_SYNC] = { "SYNC", 1 /* ? */ , 16 }, /* Topology # */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This table contains the backing store pages for the interrupt
|
|
|
|
* fifos of the VC sub-engine in case of overflow.
|
|
|
|
*
|
|
|
|
* 0 - IPI,
|
|
|
|
* 1 - HWD,
|
|
|
|
* 2 - NxC,
|
|
|
|
* 3 - INT,
|
|
|
|
* 4 - OS-Queue,
|
|
|
|
* 5 - Pool-Queue,
|
|
|
|
* 6 - Hard-Queue
|
|
|
|
*/
|
|
|
|
[VST_ERQ] = { "ERQ", 1, VC_QUEUE_COUNT },
|
|
|
|
};
|
|
|
|
|
|
|
|
#define xive2_error(xive, fmt, ...) \
|
|
|
|
qemu_log_mask(LOG_GUEST_ERROR, "XIVE[%x] - " fmt "\n", \
|
|
|
|
(xive)->chip->chip_id, ## __VA_ARGS__);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* TODO: Document block id override
|
|
|
|
*/
|
|
|
|
static uint32_t pnv_xive2_block_id(PnvXive2 *xive)
|
|
|
|
{
|
|
|
|
uint8_t blk = xive->chip->chip_id;
|
|
|
|
uint64_t cfg_val = xive->cq_regs[CQ_XIVE_CFG >> 3];
|
|
|
|
|
|
|
|
if (cfg_val & CQ_XIVE_CFG_HYP_HARD_BLKID_OVERRIDE) {
|
|
|
|
blk = GETFIELD(CQ_XIVE_CFG_HYP_HARD_BLOCK_ID, cfg_val);
|
|
|
|
}
|
|
|
|
|
|
|
|
return blk;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Remote access to controllers. HW uses MMIOs. For now, a simple scan
|
|
|
|
* of the chips is good enough.
|
|
|
|
*
|
|
|
|
* TODO: Block scope support
|
|
|
|
*/
|
|
|
|
static PnvXive2 *pnv_xive2_get_remote(uint8_t blk)
|
|
|
|
{
|
|
|
|
PnvMachineState *pnv = PNV_MACHINE(qdev_get_machine());
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < pnv->num_chips; i++) {
|
|
|
|
Pnv10Chip *chip10 = PNV10_CHIP(pnv->chips[i]);
|
|
|
|
PnvXive2 *xive = &chip10->xive;
|
|
|
|
|
|
|
|
if (pnv_xive2_block_id(xive) == blk) {
|
|
|
|
return xive;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* VST accessors for ESB, EAT, ENDT, NVP
|
|
|
|
*
|
|
|
|
* Indirect VST tables are arrays of VSDs pointing to a page (of same
|
|
|
|
* size). Each page is a direct VST table.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#define XIVE_VSD_SIZE 8
|
|
|
|
|
|
|
|
/* Indirect page size can be 4K, 64K, 2M, 16M. */
|
|
|
|
static uint64_t pnv_xive2_vst_page_size_allowed(uint32_t page_shift)
|
|
|
|
{
|
|
|
|
return page_shift == 12 || page_shift == 16 ||
|
|
|
|
page_shift == 21 || page_shift == 24;
|
|
|
|
}
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_vst_addr_direct(PnvXive2 *xive, uint32_t type,
|
|
|
|
uint64_t vsd, uint32_t idx)
|
|
|
|
{
|
|
|
|
const XiveVstInfo *info = &vst_infos[type];
|
|
|
|
uint64_t vst_addr = vsd & VSD_ADDRESS_MASK;
|
|
|
|
uint64_t vst_tsize = 1ull << (GETFIELD(VSD_TSIZE, vsd) + 12);
|
|
|
|
uint32_t idx_max;
|
|
|
|
|
|
|
|
idx_max = vst_tsize / info->size - 1;
|
|
|
|
if (idx > idx_max) {
|
|
|
|
#ifdef XIVE2_DEBUG
|
|
|
|
xive2_error(xive, "VST: %s entry %x out of range [ 0 .. %x ] !?",
|
|
|
|
info->name, idx, idx_max);
|
|
|
|
#endif
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
return vst_addr + idx * info->size;
|
|
|
|
}
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_vst_addr_indirect(PnvXive2 *xive, uint32_t type,
|
|
|
|
uint64_t vsd, uint32_t idx)
|
|
|
|
{
|
|
|
|
const XiveVstInfo *info = &vst_infos[type];
|
|
|
|
uint64_t vsd_addr;
|
|
|
|
uint32_t vsd_idx;
|
|
|
|
uint32_t page_shift;
|
|
|
|
uint32_t vst_per_page;
|
|
|
|
|
|
|
|
/* Get the page size of the indirect table. */
|
|
|
|
vsd_addr = vsd & VSD_ADDRESS_MASK;
|
|
|
|
ldq_be_dma(&address_space_memory, vsd_addr, &vsd, MEMTXATTRS_UNSPECIFIED);
|
|
|
|
|
|
|
|
if (!(vsd & VSD_ADDRESS_MASK)) {
|
2023-05-31 17:05:37 +02:00
|
|
|
#ifdef XIVE2_DEBUG
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
xive2_error(xive, "VST: invalid %s entry %x !?", info->name, idx);
|
2023-05-31 17:05:37 +02:00
|
|
|
#endif
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
page_shift = GETFIELD(VSD_TSIZE, vsd) + 12;
|
|
|
|
|
|
|
|
if (!pnv_xive2_vst_page_size_allowed(page_shift)) {
|
|
|
|
xive2_error(xive, "VST: invalid %s page shift %d", info->name,
|
|
|
|
page_shift);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
vst_per_page = (1ull << page_shift) / info->size;
|
|
|
|
vsd_idx = idx / vst_per_page;
|
|
|
|
|
|
|
|
/* Load the VSD we are looking for, if not already done */
|
|
|
|
if (vsd_idx) {
|
|
|
|
vsd_addr = vsd_addr + vsd_idx * XIVE_VSD_SIZE;
|
|
|
|
ldq_be_dma(&address_space_memory, vsd_addr, &vsd,
|
|
|
|
MEMTXATTRS_UNSPECIFIED);
|
|
|
|
|
|
|
|
if (!(vsd & VSD_ADDRESS_MASK)) {
|
2023-05-31 17:05:37 +02:00
|
|
|
#ifdef XIVE2_DEBUG
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
xive2_error(xive, "VST: invalid %s entry %x !?", info->name, idx);
|
2023-05-31 17:05:37 +02:00
|
|
|
#endif
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Check that the pages have a consistent size across the
|
|
|
|
* indirect table
|
|
|
|
*/
|
|
|
|
if (page_shift != GETFIELD(VSD_TSIZE, vsd) + 12) {
|
|
|
|
xive2_error(xive, "VST: %s entry %x indirect page size differ !?",
|
|
|
|
info->name, idx);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return pnv_xive2_vst_addr_direct(xive, type, vsd, (idx % vst_per_page));
|
|
|
|
}
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_vst_addr(PnvXive2 *xive, uint32_t type, uint8_t blk,
|
|
|
|
uint32_t idx)
|
|
|
|
{
|
|
|
|
const XiveVstInfo *info = &vst_infos[type];
|
|
|
|
uint64_t vsd;
|
|
|
|
|
|
|
|
if (blk >= info->max_blocks) {
|
|
|
|
xive2_error(xive, "VST: invalid block id %d for VST %s %d !?",
|
|
|
|
blk, info->name, idx);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
vsd = xive->vsds[type][blk];
|
|
|
|
|
|
|
|
/* Remote VST access */
|
|
|
|
if (GETFIELD(VSD_MODE, vsd) == VSD_MODE_FORWARD) {
|
|
|
|
xive = pnv_xive2_get_remote(blk);
|
|
|
|
|
|
|
|
return xive ? pnv_xive2_vst_addr(xive, type, blk, idx) : 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (VSD_INDIRECT & vsd) {
|
|
|
|
return pnv_xive2_vst_addr_indirect(xive, type, vsd, idx);
|
|
|
|
}
|
|
|
|
|
|
|
|
return pnv_xive2_vst_addr_direct(xive, type, vsd, idx);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int pnv_xive2_vst_read(PnvXive2 *xive, uint32_t type, uint8_t blk,
|
|
|
|
uint32_t idx, void *data)
|
|
|
|
{
|
|
|
|
const XiveVstInfo *info = &vst_infos[type];
|
|
|
|
uint64_t addr = pnv_xive2_vst_addr(xive, type, blk, idx);
|
2023-08-29 16:32:33 +02:00
|
|
|
MemTxResult result;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
|
|
|
|
if (!addr) {
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
2023-08-29 16:32:33 +02:00
|
|
|
result = address_space_read(&address_space_memory, addr,
|
|
|
|
MEMTXATTRS_UNSPECIFIED, data,
|
|
|
|
info->size);
|
|
|
|
if (result != MEMTX_OK) {
|
|
|
|
xive2_error(xive, "VST: read failed at @0x%" HWADDR_PRIx
|
|
|
|
" for VST %s %x/%x\n", addr, info->name, blk, idx);
|
|
|
|
return -1;
|
|
|
|
}
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
#define XIVE_VST_WORD_ALL -1
|
|
|
|
|
|
|
|
static int pnv_xive2_vst_write(PnvXive2 *xive, uint32_t type, uint8_t blk,
|
|
|
|
uint32_t idx, void *data, uint32_t word_number)
|
|
|
|
{
|
|
|
|
const XiveVstInfo *info = &vst_infos[type];
|
|
|
|
uint64_t addr = pnv_xive2_vst_addr(xive, type, blk, idx);
|
2023-08-29 16:32:33 +02:00
|
|
|
MemTxResult result;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
|
|
|
|
if (!addr) {
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (word_number == XIVE_VST_WORD_ALL) {
|
2023-08-29 16:32:33 +02:00
|
|
|
result = address_space_write(&address_space_memory, addr,
|
|
|
|
MEMTXATTRS_UNSPECIFIED, data,
|
|
|
|
info->size);
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
} else {
|
2023-08-29 16:32:33 +02:00
|
|
|
result = address_space_write(&address_space_memory,
|
|
|
|
addr + word_number * 4,
|
|
|
|
MEMTXATTRS_UNSPECIFIED,
|
|
|
|
data + word_number * 4, 4);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (result != MEMTX_OK) {
|
|
|
|
xive2_error(xive, "VST: write failed at @0x%" HWADDR_PRIx
|
|
|
|
"for VST %s %x/%x\n", addr, info->name, blk, idx);
|
|
|
|
return -1;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2022-03-02 06:51:39 +01:00
|
|
|
static int pnv_xive2_get_pq(Xive2Router *xrtr, uint8_t blk, uint32_t idx,
|
|
|
|
uint8_t *pq)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(xrtr);
|
|
|
|
|
|
|
|
if (pnv_xive2_block_id(xive) != blk) {
|
|
|
|
xive2_error(xive, "VST: EAS %x is remote !?", XIVE_EAS(blk, idx));
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
*pq = xive_source_esb_get(&xive->ipi_source, idx);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int pnv_xive2_set_pq(Xive2Router *xrtr, uint8_t blk, uint32_t idx,
|
|
|
|
uint8_t *pq)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(xrtr);
|
|
|
|
|
|
|
|
if (pnv_xive2_block_id(xive) != blk) {
|
|
|
|
xive2_error(xive, "VST: EAS %x is remote !?", XIVE_EAS(blk, idx));
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
*pq = xive_source_esb_set(&xive->ipi_source, idx, *pq);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
static int pnv_xive2_get_end(Xive2Router *xrtr, uint8_t blk, uint32_t idx,
|
|
|
|
Xive2End *end)
|
|
|
|
{
|
|
|
|
return pnv_xive2_vst_read(PNV_XIVE2(xrtr), VST_END, blk, idx, end);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int pnv_xive2_write_end(Xive2Router *xrtr, uint8_t blk, uint32_t idx,
|
|
|
|
Xive2End *end, uint8_t word_number)
|
|
|
|
{
|
|
|
|
return pnv_xive2_vst_write(PNV_XIVE2(xrtr), VST_END, blk, idx, end,
|
|
|
|
word_number);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int pnv_xive2_end_update(PnvXive2 *xive)
|
|
|
|
{
|
|
|
|
uint8_t blk = GETFIELD(VC_ENDC_WATCH_BLOCK_ID,
|
|
|
|
xive->vc_regs[(VC_ENDC_WATCH0_SPEC >> 3)]);
|
|
|
|
uint32_t idx = GETFIELD(VC_ENDC_WATCH_INDEX,
|
|
|
|
xive->vc_regs[(VC_ENDC_WATCH0_SPEC >> 3)]);
|
|
|
|
int i;
|
|
|
|
uint64_t endc_watch[4];
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(endc_watch); i++) {
|
|
|
|
endc_watch[i] =
|
|
|
|
cpu_to_be64(xive->vc_regs[(VC_ENDC_WATCH0_DATA0 >> 3) + i]);
|
|
|
|
}
|
|
|
|
|
|
|
|
return pnv_xive2_vst_write(xive, VST_END, blk, idx, endc_watch,
|
|
|
|
XIVE_VST_WORD_ALL);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_end_cache_load(PnvXive2 *xive)
|
|
|
|
{
|
|
|
|
uint8_t blk = GETFIELD(VC_ENDC_WATCH_BLOCK_ID,
|
|
|
|
xive->vc_regs[(VC_ENDC_WATCH0_SPEC >> 3)]);
|
|
|
|
uint32_t idx = GETFIELD(VC_ENDC_WATCH_INDEX,
|
|
|
|
xive->vc_regs[(VC_ENDC_WATCH0_SPEC >> 3)]);
|
|
|
|
uint64_t endc_watch[4] = { 0 };
|
|
|
|
int i;
|
|
|
|
|
|
|
|
if (pnv_xive2_vst_read(xive, VST_END, blk, idx, endc_watch)) {
|
|
|
|
xive2_error(xive, "VST: no END entry %x/%x !?", blk, idx);
|
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(endc_watch); i++) {
|
|
|
|
xive->vc_regs[(VC_ENDC_WATCH0_DATA0 >> 3) + i] =
|
|
|
|
be64_to_cpu(endc_watch[i]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int pnv_xive2_get_nvp(Xive2Router *xrtr, uint8_t blk, uint32_t idx,
|
|
|
|
Xive2Nvp *nvp)
|
|
|
|
{
|
|
|
|
return pnv_xive2_vst_read(PNV_XIVE2(xrtr), VST_NVP, blk, idx, nvp);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int pnv_xive2_write_nvp(Xive2Router *xrtr, uint8_t blk, uint32_t idx,
|
|
|
|
Xive2Nvp *nvp, uint8_t word_number)
|
|
|
|
{
|
|
|
|
return pnv_xive2_vst_write(PNV_XIVE2(xrtr), VST_NVP, blk, idx, nvp,
|
|
|
|
word_number);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int pnv_xive2_nvp_update(PnvXive2 *xive)
|
|
|
|
{
|
|
|
|
uint8_t blk = GETFIELD(PC_NXC_WATCH_BLOCK_ID,
|
|
|
|
xive->pc_regs[(PC_NXC_WATCH0_SPEC >> 3)]);
|
|
|
|
uint32_t idx = GETFIELD(PC_NXC_WATCH_INDEX,
|
|
|
|
xive->pc_regs[(PC_NXC_WATCH0_SPEC >> 3)]);
|
|
|
|
int i;
|
|
|
|
uint64_t nxc_watch[4];
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(nxc_watch); i++) {
|
|
|
|
nxc_watch[i] =
|
|
|
|
cpu_to_be64(xive->pc_regs[(PC_NXC_WATCH0_DATA0 >> 3) + i]);
|
|
|
|
}
|
|
|
|
|
|
|
|
return pnv_xive2_vst_write(xive, VST_NVP, blk, idx, nxc_watch,
|
|
|
|
XIVE_VST_WORD_ALL);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_nvp_cache_load(PnvXive2 *xive)
|
|
|
|
{
|
|
|
|
uint8_t blk = GETFIELD(PC_NXC_WATCH_BLOCK_ID,
|
|
|
|
xive->pc_regs[(PC_NXC_WATCH0_SPEC >> 3)]);
|
|
|
|
uint32_t idx = GETFIELD(PC_NXC_WATCH_INDEX,
|
|
|
|
xive->pc_regs[(PC_NXC_WATCH0_SPEC >> 3)]);
|
|
|
|
uint64_t nxc_watch[4] = { 0 };
|
|
|
|
int i;
|
|
|
|
|
|
|
|
if (pnv_xive2_vst_read(xive, VST_NVP, blk, idx, nxc_watch)) {
|
|
|
|
xive2_error(xive, "VST: no NVP entry %x/%x !?", blk, idx);
|
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(nxc_watch); i++) {
|
|
|
|
xive->pc_regs[(PC_NXC_WATCH0_DATA0 >> 3) + i] =
|
|
|
|
be64_to_cpu(nxc_watch[i]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int pnv_xive2_get_eas(Xive2Router *xrtr, uint8_t blk, uint32_t idx,
|
|
|
|
Xive2Eas *eas)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(xrtr);
|
|
|
|
|
|
|
|
if (pnv_xive2_block_id(xive) != blk) {
|
|
|
|
xive2_error(xive, "VST: EAS %x is remote !?", XIVE_EAS(blk, idx));
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
return pnv_xive2_vst_read(xive, VST_EAS, blk, idx, eas);
|
|
|
|
}
|
|
|
|
|
2022-03-02 06:51:39 +01:00
|
|
|
static uint32_t pnv_xive2_get_config(Xive2Router *xrtr)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(xrtr);
|
|
|
|
uint32_t cfg = 0;
|
|
|
|
|
|
|
|
if (xive->cq_regs[CQ_XIVE_CFG >> 3] & CQ_XIVE_CFG_GEN1_TIMA_OS) {
|
|
|
|
cfg |= XIVE2_GEN1_TIMA_OS;
|
|
|
|
}
|
|
|
|
|
2022-03-02 06:51:39 +01:00
|
|
|
if (xive->cq_regs[CQ_XIVE_CFG >> 3] & CQ_XIVE_CFG_EN_VP_SAVE_RESTORE) {
|
|
|
|
cfg |= XIVE2_VP_SAVE_RESTORE;
|
|
|
|
}
|
|
|
|
|
2022-03-02 06:51:39 +01:00
|
|
|
if (GETFIELD(CQ_XIVE_CFG_HYP_HARD_RANGE,
|
|
|
|
xive->cq_regs[CQ_XIVE_CFG >> 3]) == CQ_XIVE_CFG_THREADID_8BITS) {
|
|
|
|
cfg |= XIVE2_THREADID_8BITS;
|
|
|
|
}
|
|
|
|
|
2022-03-02 06:51:39 +01:00
|
|
|
return cfg;
|
|
|
|
}
|
|
|
|
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
static bool pnv_xive2_is_cpu_enabled(PnvXive2 *xive, PowerPCCPU *cpu)
|
|
|
|
{
|
|
|
|
int pir = ppc_cpu_pir(cpu);
|
|
|
|
uint32_t fc = PNV10_PIR2FUSEDCORE(pir);
|
|
|
|
uint64_t reg = fc < 8 ? TCTXT_EN0 : TCTXT_EN1;
|
|
|
|
uint32_t bit = pir & 0x3f;
|
|
|
|
|
|
|
|
return xive->tctxt_regs[reg >> 3] & PPC_BIT(bit);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int pnv_xive2_match_nvt(XivePresenter *xptr, uint8_t format,
|
|
|
|
uint8_t nvt_blk, uint32_t nvt_idx,
|
|
|
|
bool cam_ignore, uint8_t priority,
|
|
|
|
uint32_t logic_serv, XiveTCTXMatch *match)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(xptr);
|
|
|
|
PnvChip *chip = xive->chip;
|
|
|
|
int count = 0;
|
|
|
|
int i, j;
|
2022-03-02 06:51:39 +01:00
|
|
|
bool gen1_tima_os =
|
|
|
|
xive->cq_regs[CQ_XIVE_CFG >> 3] & CQ_XIVE_CFG_GEN1_TIMA_OS;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
|
|
|
|
for (i = 0; i < chip->nr_cores; i++) {
|
|
|
|
PnvCore *pc = chip->cores[i];
|
|
|
|
CPUCore *cc = CPU_CORE(pc);
|
|
|
|
|
|
|
|
for (j = 0; j < cc->nr_threads; j++) {
|
|
|
|
PowerPCCPU *cpu = pc->threads[j];
|
|
|
|
XiveTCTX *tctx;
|
|
|
|
int ring;
|
|
|
|
|
|
|
|
if (!pnv_xive2_is_cpu_enabled(xive, cpu)) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
tctx = XIVE_TCTX(pnv_cpu_state(cpu)->intc);
|
|
|
|
|
2022-03-02 06:51:39 +01:00
|
|
|
if (gen1_tima_os) {
|
|
|
|
ring = xive_presenter_tctx_match(xptr, tctx, format, nvt_blk,
|
|
|
|
nvt_idx, cam_ignore,
|
|
|
|
logic_serv);
|
|
|
|
} else {
|
|
|
|
ring = xive2_presenter_tctx_match(xptr, tctx, format, nvt_blk,
|
|
|
|
nvt_idx, cam_ignore,
|
|
|
|
logic_serv);
|
|
|
|
}
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Save the context and follow on to catch duplicates,
|
|
|
|
* that we don't support yet.
|
|
|
|
*/
|
|
|
|
if (ring != -1) {
|
|
|
|
if (match->tctx) {
|
|
|
|
qemu_log_mask(LOG_GUEST_ERROR, "XIVE: already found a "
|
|
|
|
"thread context NVT %x/%x\n",
|
|
|
|
nvt_blk, nvt_idx);
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
match->ring = ring;
|
|
|
|
match->tctx = tctx;
|
|
|
|
count++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return count;
|
|
|
|
}
|
|
|
|
|
2023-06-22 18:25:26 +02:00
|
|
|
static uint32_t pnv_xive2_presenter_get_config(XivePresenter *xptr)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(xptr);
|
|
|
|
uint32_t cfg = 0;
|
|
|
|
|
|
|
|
if (xive->cq_regs[CQ_XIVE_CFG >> 3] & CQ_XIVE_CFG_GEN1_TIMA_OS) {
|
|
|
|
cfg |= XIVE_PRESENTER_GEN1_TIMA_OS;
|
|
|
|
}
|
|
|
|
return cfg;
|
|
|
|
}
|
|
|
|
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
static uint8_t pnv_xive2_get_block_id(Xive2Router *xrtr)
|
|
|
|
{
|
|
|
|
return pnv_xive2_block_id(PNV_XIVE2(xrtr));
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The TIMA MMIO space is shared among the chips and to identify the
|
|
|
|
* chip from which the access is being done, we extract the chip id
|
|
|
|
* from the PIR.
|
|
|
|
*/
|
|
|
|
static PnvXive2 *pnv_xive2_tm_get_xive(PowerPCCPU *cpu)
|
|
|
|
{
|
|
|
|
int pir = ppc_cpu_pir(cpu);
|
|
|
|
XivePresenter *xptr = XIVE_TCTX(pnv_cpu_state(cpu)->intc)->xptr;
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(xptr);
|
|
|
|
|
|
|
|
if (!pnv_xive2_is_cpu_enabled(xive, cpu)) {
|
|
|
|
xive2_error(xive, "IC: CPU %x is not enabled", pir);
|
|
|
|
}
|
|
|
|
return xive;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The internal sources of the interrupt controller have no knowledge
|
|
|
|
* of the XIVE2 chip on which they reside. Encode the block id in the
|
|
|
|
* source interrupt number before forwarding the source event
|
|
|
|
* notification to the Router. This is required on a multichip system.
|
|
|
|
*/
|
2022-03-02 06:51:39 +01:00
|
|
|
static void pnv_xive2_notify(XiveNotifier *xn, uint32_t srcno, bool pq_checked)
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(xn);
|
|
|
|
uint8_t blk = pnv_xive2_block_id(xive);
|
|
|
|
|
2022-03-02 06:51:39 +01:00
|
|
|
xive2_router_notify(xn, XIVE_EAS(blk, srcno), pq_checked);
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Set Translation Tables
|
|
|
|
*
|
|
|
|
* TODO add support for multiple sets
|
|
|
|
*/
|
|
|
|
static int pnv_xive2_stt_set_data(PnvXive2 *xive, uint64_t val)
|
|
|
|
{
|
|
|
|
uint8_t tsel = GETFIELD(CQ_TAR_SELECT, xive->cq_regs[CQ_TAR >> 3]);
|
|
|
|
uint8_t entry = GETFIELD(CQ_TAR_ENTRY_SELECT,
|
|
|
|
xive->cq_regs[CQ_TAR >> 3]);
|
|
|
|
|
|
|
|
switch (tsel) {
|
|
|
|
case CQ_TAR_NVPG:
|
|
|
|
case CQ_TAR_ESB:
|
|
|
|
case CQ_TAR_END:
|
|
|
|
xive->tables[tsel][entry] = val;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
xive2_error(xive, "IC: unsupported table %d", tsel);
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (xive->cq_regs[CQ_TAR >> 3] & CQ_TAR_AUTOINC) {
|
|
|
|
xive->cq_regs[CQ_TAR >> 3] = SETFIELD(CQ_TAR_ENTRY_SELECT,
|
|
|
|
xive->cq_regs[CQ_TAR >> 3], ++entry);
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* Virtual Structure Tables (VST) configuration
|
|
|
|
*/
|
|
|
|
static void pnv_xive2_vst_set_exclusive(PnvXive2 *xive, uint8_t type,
|
|
|
|
uint8_t blk, uint64_t vsd)
|
|
|
|
{
|
|
|
|
Xive2EndSource *end_xsrc = &xive->end_source;
|
|
|
|
XiveSource *xsrc = &xive->ipi_source;
|
|
|
|
const XiveVstInfo *info = &vst_infos[type];
|
|
|
|
uint32_t page_shift = GETFIELD(VSD_TSIZE, vsd) + 12;
|
|
|
|
uint64_t vst_tsize = 1ull << page_shift;
|
|
|
|
uint64_t vst_addr = vsd & VSD_ADDRESS_MASK;
|
|
|
|
|
|
|
|
/* Basic checks */
|
|
|
|
|
|
|
|
if (VSD_INDIRECT & vsd) {
|
|
|
|
if (!pnv_xive2_vst_page_size_allowed(page_shift)) {
|
|
|
|
xive2_error(xive, "VST: invalid %s page shift %d", info->name,
|
|
|
|
page_shift);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!QEMU_IS_ALIGNED(vst_addr, 1ull << page_shift)) {
|
|
|
|
xive2_error(xive, "VST: %s table address 0x%"PRIx64
|
|
|
|
" is not aligned with page shift %d",
|
|
|
|
info->name, vst_addr, page_shift);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Record the table configuration (in SRAM on HW) */
|
|
|
|
xive->vsds[type][blk] = vsd;
|
|
|
|
|
|
|
|
/* Now tune the models with the configuration provided by the FW */
|
|
|
|
|
|
|
|
switch (type) {
|
|
|
|
case VST_ESB:
|
|
|
|
/*
|
|
|
|
* Backing store pages for the source PQ bits. The model does
|
|
|
|
* not use these PQ bits backed in RAM because the XiveSource
|
|
|
|
* model has its own.
|
|
|
|
*
|
|
|
|
* If the table is direct, we can compute the number of PQ
|
|
|
|
* entries provisioned by FW (such as skiboot) and resize the
|
|
|
|
* ESB window accordingly.
|
|
|
|
*/
|
|
|
|
if (!(VSD_INDIRECT & vsd)) {
|
|
|
|
memory_region_set_size(&xsrc->esb_mmio, vst_tsize * SBE_PER_BYTE
|
|
|
|
* (1ull << xsrc->esb_shift));
|
|
|
|
}
|
|
|
|
|
|
|
|
memory_region_add_subregion(&xive->esb_mmio, 0, &xsrc->esb_mmio);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case VST_EAS: /* Nothing to be done */
|
|
|
|
break;
|
|
|
|
|
|
|
|
case VST_END:
|
|
|
|
/*
|
|
|
|
* Backing store pages for the END.
|
|
|
|
*/
|
|
|
|
if (!(VSD_INDIRECT & vsd)) {
|
|
|
|
memory_region_set_size(&end_xsrc->esb_mmio, (vst_tsize / info->size)
|
|
|
|
* (1ull << end_xsrc->esb_shift));
|
|
|
|
}
|
|
|
|
memory_region_add_subregion(&xive->end_mmio, 0, &end_xsrc->esb_mmio);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case VST_NVP: /* Not modeled */
|
|
|
|
case VST_NVG: /* Not modeled */
|
|
|
|
case VST_NVC: /* Not modeled */
|
|
|
|
case VST_IC: /* Not modeled */
|
|
|
|
case VST_SYNC: /* Not modeled */
|
|
|
|
case VST_ERQ: /* Not modeled */
|
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
g_assert_not_reached();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Both PC and VC sub-engines are configured as each use the Virtual
|
|
|
|
* Structure Tables
|
|
|
|
*/
|
|
|
|
static void pnv_xive2_vst_set_data(PnvXive2 *xive, uint64_t vsd)
|
|
|
|
{
|
|
|
|
uint8_t mode = GETFIELD(VSD_MODE, vsd);
|
|
|
|
uint8_t type = GETFIELD(VC_VSD_TABLE_SELECT,
|
|
|
|
xive->vc_regs[VC_VSD_TABLE_ADDR >> 3]);
|
|
|
|
uint8_t blk = GETFIELD(VC_VSD_TABLE_ADDRESS,
|
|
|
|
xive->vc_regs[VC_VSD_TABLE_ADDR >> 3]);
|
|
|
|
uint64_t vst_addr = vsd & VSD_ADDRESS_MASK;
|
|
|
|
|
|
|
|
if (type > VST_ERQ) {
|
|
|
|
xive2_error(xive, "VST: invalid table type %d", type);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (blk >= vst_infos[type].max_blocks) {
|
|
|
|
xive2_error(xive, "VST: invalid block id %d for"
|
|
|
|
" %s table", blk, vst_infos[type].name);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!vst_addr) {
|
|
|
|
xive2_error(xive, "VST: invalid %s table address",
|
|
|
|
vst_infos[type].name);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
switch (mode) {
|
|
|
|
case VSD_MODE_FORWARD:
|
|
|
|
xive->vsds[type][blk] = vsd;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case VSD_MODE_EXCLUSIVE:
|
|
|
|
pnv_xive2_vst_set_exclusive(xive, type, blk, vsd);
|
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
xive2_error(xive, "VST: unsupported table mode %d", mode);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* MMIO handlers
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
* IC BAR layout
|
|
|
|
*
|
|
|
|
* Page 0: Internal CQ register accesses (reads & writes)
|
|
|
|
* Page 1: Internal PC register accesses (reads & writes)
|
|
|
|
* Page 2: Internal VC register accesses (reads & writes)
|
|
|
|
* Page 3: Internal TCTXT (TIMA) reg accesses (read & writes)
|
|
|
|
* Page 4: Notify Port page (writes only, w/data),
|
|
|
|
* Page 5: Reserved
|
|
|
|
* Page 6: Sync Poll page (writes only, dataless)
|
|
|
|
* Page 7: Sync Inject page (writes only, dataless)
|
|
|
|
* Page 8: LSI Trigger page (writes only, dataless)
|
|
|
|
* Page 9: LSI SB Management page (reads & writes dataless)
|
|
|
|
* Pages 10-255: Reserved
|
|
|
|
* Pages 256-383: Direct mapped Thread Context Area (reads & writes)
|
|
|
|
* covering the 128 threads in P10.
|
|
|
|
* Pages 384-511: Reserved
|
|
|
|
*/
|
|
|
|
typedef struct PnvXive2Region {
|
|
|
|
const char *name;
|
|
|
|
uint32_t pgoff;
|
|
|
|
uint32_t pgsize;
|
|
|
|
const MemoryRegionOps *ops;
|
|
|
|
} PnvXive2Region;
|
|
|
|
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_cq_ops;
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_pc_ops;
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_vc_ops;
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_tctxt_ops;
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_notify_ops;
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_sync_ops;
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_lsi_ops;
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_tm_indirect_ops;
|
|
|
|
|
|
|
|
/* 512 pages. 4K: 2M range, 64K: 32M range */
|
|
|
|
static const PnvXive2Region pnv_xive2_ic_regions[] = {
|
|
|
|
{ "xive-ic-cq", 0, 1, &pnv_xive2_ic_cq_ops },
|
|
|
|
{ "xive-ic-vc", 1, 1, &pnv_xive2_ic_vc_ops },
|
|
|
|
{ "xive-ic-pc", 2, 1, &pnv_xive2_ic_pc_ops },
|
|
|
|
{ "xive-ic-tctxt", 3, 1, &pnv_xive2_ic_tctxt_ops },
|
|
|
|
{ "xive-ic-notify", 4, 1, &pnv_xive2_ic_notify_ops },
|
|
|
|
/* page 5 reserved */
|
|
|
|
{ "xive-ic-sync", 6, 2, &pnv_xive2_ic_sync_ops },
|
|
|
|
{ "xive-ic-lsi", 8, 2, &pnv_xive2_ic_lsi_ops },
|
|
|
|
/* pages 10-255 reserved */
|
|
|
|
{ "xive-ic-tm-indirect", 256, 128, &pnv_xive2_ic_tm_indirect_ops },
|
|
|
|
/* pages 384-511 reserved */
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* CQ operations
|
|
|
|
*/
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_ic_cq_read(void *opaque, hwaddr offset,
|
|
|
|
unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
uint32_t reg = offset >> 3;
|
|
|
|
uint64_t val = 0;
|
|
|
|
|
|
|
|
switch (offset) {
|
|
|
|
case CQ_XIVE_CAP: /* Set at reset */
|
|
|
|
case CQ_XIVE_CFG:
|
|
|
|
val = xive->cq_regs[reg];
|
|
|
|
break;
|
|
|
|
case CQ_MSGSND: /* TODO check the #cores of the machine */
|
|
|
|
val = 0xffffffff00000000;
|
|
|
|
break;
|
|
|
|
case CQ_CFG_PB_GEN:
|
|
|
|
val = CQ_CFG_PB_GEN_PB_INIT; /* TODO: fix CQ_CFG_PB_GEN default value */
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
xive2_error(xive, "CQ: invalid read @%"HWADDR_PRIx, offset);
|
|
|
|
}
|
|
|
|
|
|
|
|
return val;
|
|
|
|
}
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_bar_size(uint64_t val)
|
|
|
|
{
|
|
|
|
return 1ull << (GETFIELD(CQ_BAR_RANGE, val) + 24);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_ic_cq_write(void *opaque, hwaddr offset,
|
|
|
|
uint64_t val, unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
MemoryRegion *sysmem = get_system_memory();
|
|
|
|
uint32_t reg = offset >> 3;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
switch (offset) {
|
|
|
|
case CQ_XIVE_CFG:
|
|
|
|
case CQ_RST_CTL: /* TODO: reset all BARs */
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CQ_IC_BAR:
|
|
|
|
xive->ic_shift = val & CQ_IC_BAR_64K ? 16 : 12;
|
|
|
|
if (!(val & CQ_IC_BAR_VALID)) {
|
|
|
|
xive->ic_base = 0;
|
|
|
|
if (xive->cq_regs[reg] & CQ_IC_BAR_VALID) {
|
|
|
|
for (i = 0; i < ARRAY_SIZE(xive->ic_mmios); i++) {
|
|
|
|
memory_region_del_subregion(&xive->ic_mmio,
|
|
|
|
&xive->ic_mmios[i]);
|
|
|
|
}
|
|
|
|
memory_region_del_subregion(sysmem, &xive->ic_mmio);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
xive->ic_base = val & ~(CQ_IC_BAR_VALID | CQ_IC_BAR_64K);
|
|
|
|
if (!(xive->cq_regs[reg] & CQ_IC_BAR_VALID)) {
|
|
|
|
for (i = 0; i < ARRAY_SIZE(xive->ic_mmios); i++) {
|
|
|
|
memory_region_add_subregion(&xive->ic_mmio,
|
|
|
|
pnv_xive2_ic_regions[i].pgoff << xive->ic_shift,
|
|
|
|
&xive->ic_mmios[i]);
|
|
|
|
}
|
|
|
|
memory_region_add_subregion(sysmem, xive->ic_base,
|
|
|
|
&xive->ic_mmio);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CQ_TM_BAR:
|
|
|
|
xive->tm_shift = val & CQ_TM_BAR_64K ? 16 : 12;
|
|
|
|
if (!(val & CQ_TM_BAR_VALID)) {
|
|
|
|
xive->tm_base = 0;
|
|
|
|
if (xive->cq_regs[reg] & CQ_TM_BAR_VALID) {
|
|
|
|
memory_region_del_subregion(sysmem, &xive->tm_mmio);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
xive->tm_base = val & ~(CQ_TM_BAR_VALID | CQ_TM_BAR_64K);
|
|
|
|
if (!(xive->cq_regs[reg] & CQ_TM_BAR_VALID)) {
|
|
|
|
memory_region_add_subregion(sysmem, xive->tm_base,
|
|
|
|
&xive->tm_mmio);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CQ_ESB_BAR:
|
|
|
|
xive->esb_shift = val & CQ_BAR_64K ? 16 : 12;
|
|
|
|
if (!(val & CQ_BAR_VALID)) {
|
|
|
|
xive->esb_base = 0;
|
|
|
|
if (xive->cq_regs[reg] & CQ_BAR_VALID) {
|
|
|
|
memory_region_del_subregion(sysmem, &xive->esb_mmio);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
xive->esb_base = val & CQ_BAR_ADDR;
|
|
|
|
if (!(xive->cq_regs[reg] & CQ_BAR_VALID)) {
|
|
|
|
memory_region_set_size(&xive->esb_mmio,
|
|
|
|
pnv_xive2_bar_size(val));
|
|
|
|
memory_region_add_subregion(sysmem, xive->esb_base,
|
|
|
|
&xive->esb_mmio);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CQ_END_BAR:
|
|
|
|
xive->end_shift = val & CQ_BAR_64K ? 16 : 12;
|
|
|
|
if (!(val & CQ_BAR_VALID)) {
|
|
|
|
xive->end_base = 0;
|
|
|
|
if (xive->cq_regs[reg] & CQ_BAR_VALID) {
|
|
|
|
memory_region_del_subregion(sysmem, &xive->end_mmio);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
xive->end_base = val & CQ_BAR_ADDR;
|
|
|
|
if (!(xive->cq_regs[reg] & CQ_BAR_VALID)) {
|
|
|
|
memory_region_set_size(&xive->end_mmio,
|
|
|
|
pnv_xive2_bar_size(val));
|
|
|
|
memory_region_add_subregion(sysmem, xive->end_base,
|
|
|
|
&xive->end_mmio);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CQ_NVC_BAR:
|
|
|
|
xive->nvc_shift = val & CQ_BAR_64K ? 16 : 12;
|
|
|
|
if (!(val & CQ_BAR_VALID)) {
|
|
|
|
xive->nvc_base = 0;
|
|
|
|
if (xive->cq_regs[reg] & CQ_BAR_VALID) {
|
|
|
|
memory_region_del_subregion(sysmem, &xive->nvc_mmio);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
xive->nvc_base = val & CQ_BAR_ADDR;
|
|
|
|
if (!(xive->cq_regs[reg] & CQ_BAR_VALID)) {
|
|
|
|
memory_region_set_size(&xive->nvc_mmio,
|
|
|
|
pnv_xive2_bar_size(val));
|
|
|
|
memory_region_add_subregion(sysmem, xive->nvc_base,
|
|
|
|
&xive->nvc_mmio);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CQ_NVPG_BAR:
|
|
|
|
xive->nvpg_shift = val & CQ_BAR_64K ? 16 : 12;
|
|
|
|
if (!(val & CQ_BAR_VALID)) {
|
|
|
|
xive->nvpg_base = 0;
|
|
|
|
if (xive->cq_regs[reg] & CQ_BAR_VALID) {
|
|
|
|
memory_region_del_subregion(sysmem, &xive->nvpg_mmio);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
xive->nvpg_base = val & CQ_BAR_ADDR;
|
|
|
|
if (!(xive->cq_regs[reg] & CQ_BAR_VALID)) {
|
|
|
|
memory_region_set_size(&xive->nvpg_mmio,
|
|
|
|
pnv_xive2_bar_size(val));
|
|
|
|
memory_region_add_subregion(sysmem, xive->nvpg_base,
|
|
|
|
&xive->nvpg_mmio);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case CQ_TAR: /* Set Translation Table Address */
|
|
|
|
break;
|
|
|
|
case CQ_TDR: /* Set Translation Table Data */
|
|
|
|
pnv_xive2_stt_set_data(xive, val);
|
|
|
|
break;
|
|
|
|
case CQ_FIRMASK_OR: /* FIR error reporting */
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
xive2_error(xive, "CQ: invalid write 0x%"HWADDR_PRIx, offset);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
xive->cq_regs[reg] = val;
|
|
|
|
}
|
|
|
|
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_cq_ops = {
|
|
|
|
.read = pnv_xive2_ic_cq_read,
|
|
|
|
.write = pnv_xive2_ic_cq_write,
|
|
|
|
.endianness = DEVICE_BIG_ENDIAN,
|
|
|
|
.valid = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
.impl = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_ic_vc_read(void *opaque, hwaddr offset,
|
|
|
|
unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
uint64_t val = 0;
|
|
|
|
uint32_t reg = offset >> 3;
|
|
|
|
|
|
|
|
switch (offset) {
|
|
|
|
/*
|
|
|
|
* VSD table settings.
|
|
|
|
*/
|
|
|
|
case VC_VSD_TABLE_ADDR:
|
|
|
|
case VC_VSD_TABLE_DATA:
|
|
|
|
val = xive->vc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* ESB cache updates (not modeled)
|
|
|
|
*/
|
|
|
|
case VC_ESBC_FLUSH_CTRL:
|
|
|
|
xive->vc_regs[reg] &= ~VC_ESBC_FLUSH_CTRL_POLL_VALID;
|
|
|
|
val = xive->vc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
2023-06-01 14:13:28 +02:00
|
|
|
case VC_ESBC_CFG:
|
|
|
|
val = xive->vc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
/*
|
|
|
|
* EAS cache updates (not modeled)
|
|
|
|
*/
|
|
|
|
case VC_EASC_FLUSH_CTRL:
|
|
|
|
xive->vc_regs[reg] &= ~VC_EASC_FLUSH_CTRL_POLL_VALID;
|
|
|
|
val = xive->vc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* END cache updates
|
|
|
|
*/
|
|
|
|
case VC_ENDC_WATCH0_SPEC:
|
|
|
|
xive->vc_regs[reg] &= ~(VC_ENDC_WATCH_FULL | VC_ENDC_WATCH_CONFLICT);
|
|
|
|
val = xive->vc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
case VC_ENDC_WATCH0_DATA0:
|
|
|
|
/*
|
|
|
|
* Load DATA registers from cache with data requested by the
|
|
|
|
* SPEC register
|
|
|
|
*/
|
|
|
|
pnv_xive2_end_cache_load(xive);
|
|
|
|
val = xive->vc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
case VC_ENDC_WATCH0_DATA1 ... VC_ENDC_WATCH0_DATA3:
|
|
|
|
val = xive->vc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
case VC_ENDC_FLUSH_CTRL:
|
|
|
|
xive->vc_regs[reg] &= ~VC_ENDC_FLUSH_CTRL_POLL_VALID;
|
|
|
|
val = xive->vc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Indirect invalidation
|
|
|
|
*/
|
|
|
|
case VC_AT_MACRO_KILL_MASK:
|
|
|
|
val = xive->vc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
case VC_AT_MACRO_KILL:
|
|
|
|
xive->vc_regs[reg] &= ~VC_AT_MACRO_KILL_VALID;
|
|
|
|
val = xive->vc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Interrupt fifo overflow in memory backing store (Not modeled)
|
|
|
|
*/
|
|
|
|
case VC_QUEUES_CFG_REM0 ... VC_QUEUES_CFG_REM6:
|
|
|
|
val = xive->vc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Synchronisation
|
|
|
|
*/
|
|
|
|
case VC_ENDC_SYNC_DONE:
|
|
|
|
val = VC_ENDC_SYNC_POLL_DONE;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
xive2_error(xive, "VC: invalid read @%"HWADDR_PRIx, offset);
|
|
|
|
}
|
|
|
|
|
|
|
|
return val;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_ic_vc_write(void *opaque, hwaddr offset,
|
|
|
|
uint64_t val, unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
uint32_t reg = offset >> 3;
|
|
|
|
|
|
|
|
switch (offset) {
|
|
|
|
/*
|
|
|
|
* VSD table settings.
|
|
|
|
*/
|
|
|
|
case VC_VSD_TABLE_ADDR:
|
|
|
|
break;
|
|
|
|
case VC_VSD_TABLE_DATA:
|
|
|
|
pnv_xive2_vst_set_data(xive, val);
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* ESB cache updates (not modeled)
|
|
|
|
*/
|
|
|
|
/* case VC_ESBC_FLUSH_CTRL: */
|
|
|
|
case VC_ESBC_FLUSH_POLL:
|
|
|
|
xive->vc_regs[VC_ESBC_FLUSH_CTRL >> 3] |= VC_ESBC_FLUSH_CTRL_POLL_VALID;
|
|
|
|
/* ESB update */
|
|
|
|
break;
|
|
|
|
|
2023-06-01 14:13:28 +02:00
|
|
|
case VC_ESBC_CFG:
|
|
|
|
break;
|
|
|
|
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
/*
|
|
|
|
* EAS cache updates (not modeled)
|
|
|
|
*/
|
|
|
|
/* case VC_EASC_FLUSH_CTRL: */
|
|
|
|
case VC_EASC_FLUSH_POLL:
|
|
|
|
xive->vc_regs[VC_EASC_FLUSH_CTRL >> 3] |= VC_EASC_FLUSH_CTRL_POLL_VALID;
|
|
|
|
/* EAS update */
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* END cache updates
|
|
|
|
*/
|
|
|
|
case VC_ENDC_WATCH0_SPEC:
|
|
|
|
val &= ~VC_ENDC_WATCH_CONFLICT; /* HW will set this bit */
|
|
|
|
break;
|
|
|
|
|
|
|
|
case VC_ENDC_WATCH0_DATA1 ... VC_ENDC_WATCH0_DATA3:
|
|
|
|
break;
|
|
|
|
case VC_ENDC_WATCH0_DATA0:
|
|
|
|
/* writing to DATA0 triggers the cache write */
|
|
|
|
xive->vc_regs[reg] = val;
|
|
|
|
pnv_xive2_end_update(xive);
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
|
|
/* case VC_ENDC_FLUSH_CTRL: */
|
|
|
|
case VC_ENDC_FLUSH_POLL:
|
|
|
|
xive->vc_regs[VC_ENDC_FLUSH_CTRL >> 3] |= VC_ENDC_FLUSH_CTRL_POLL_VALID;
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Indirect invalidation
|
|
|
|
*/
|
|
|
|
case VC_AT_MACRO_KILL:
|
|
|
|
case VC_AT_MACRO_KILL_MASK:
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Interrupt fifo overflow in memory backing store (Not modeled)
|
|
|
|
*/
|
|
|
|
case VC_QUEUES_CFG_REM0 ... VC_QUEUES_CFG_REM6:
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Synchronisation
|
|
|
|
*/
|
|
|
|
case VC_ENDC_SYNC_DONE:
|
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
xive2_error(xive, "VC: invalid write @%"HWADDR_PRIx, offset);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
xive->vc_regs[reg] = val;
|
|
|
|
}
|
|
|
|
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_vc_ops = {
|
|
|
|
.read = pnv_xive2_ic_vc_read,
|
|
|
|
.write = pnv_xive2_ic_vc_write,
|
|
|
|
.endianness = DEVICE_BIG_ENDIAN,
|
|
|
|
.valid = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
.impl = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_ic_pc_read(void *opaque, hwaddr offset,
|
|
|
|
unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
uint64_t val = -1;
|
|
|
|
uint32_t reg = offset >> 3;
|
|
|
|
|
|
|
|
switch (offset) {
|
|
|
|
/*
|
|
|
|
* VSD table settings.
|
|
|
|
*/
|
|
|
|
case PC_VSD_TABLE_ADDR:
|
|
|
|
case PC_VSD_TABLE_DATA:
|
|
|
|
val = xive->pc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* cache updates
|
|
|
|
*/
|
|
|
|
case PC_NXC_WATCH0_SPEC:
|
|
|
|
xive->pc_regs[reg] &= ~(PC_NXC_WATCH_FULL | PC_NXC_WATCH_CONFLICT);
|
|
|
|
val = xive->pc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
case PC_NXC_WATCH0_DATA0:
|
|
|
|
/*
|
|
|
|
* Load DATA registers from cache with data requested by the
|
|
|
|
* SPEC register
|
|
|
|
*/
|
|
|
|
pnv_xive2_nvp_cache_load(xive);
|
|
|
|
val = xive->pc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
case PC_NXC_WATCH0_DATA1 ... PC_NXC_WATCH0_DATA3:
|
|
|
|
val = xive->pc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
case PC_NXC_FLUSH_CTRL:
|
|
|
|
xive->pc_regs[reg] &= ~PC_NXC_FLUSH_CTRL_POLL_VALID;
|
|
|
|
val = xive->pc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Indirect invalidation
|
|
|
|
*/
|
|
|
|
case PC_AT_KILL:
|
|
|
|
xive->pc_regs[reg] &= ~PC_AT_KILL_VALID;
|
|
|
|
val = xive->pc_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
xive2_error(xive, "PC: invalid read @%"HWADDR_PRIx, offset);
|
|
|
|
}
|
|
|
|
|
|
|
|
return val;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_ic_pc_write(void *opaque, hwaddr offset,
|
|
|
|
uint64_t val, unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
uint32_t reg = offset >> 3;
|
|
|
|
|
|
|
|
switch (offset) {
|
|
|
|
|
|
|
|
/*
|
|
|
|
* VSD table settings. Only taken into account in the VC
|
|
|
|
* sub-engine because the Xive2Router model combines both VC and PC
|
|
|
|
* sub-engines
|
|
|
|
*/
|
|
|
|
case PC_VSD_TABLE_ADDR:
|
|
|
|
case PC_VSD_TABLE_DATA:
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* cache updates
|
|
|
|
*/
|
|
|
|
case PC_NXC_WATCH0_SPEC:
|
|
|
|
val &= ~PC_NXC_WATCH_CONFLICT; /* HW will set this bit */
|
|
|
|
break;
|
|
|
|
|
|
|
|
case PC_NXC_WATCH0_DATA1 ... PC_NXC_WATCH0_DATA3:
|
|
|
|
break;
|
|
|
|
case PC_NXC_WATCH0_DATA0:
|
|
|
|
/* writing to DATA0 triggers the cache write */
|
|
|
|
xive->pc_regs[reg] = val;
|
|
|
|
pnv_xive2_nvp_update(xive);
|
|
|
|
break;
|
|
|
|
|
|
|
|
/* case PC_NXC_FLUSH_CTRL: */
|
|
|
|
case PC_NXC_FLUSH_POLL:
|
|
|
|
xive->pc_regs[PC_NXC_FLUSH_CTRL >> 3] |= PC_NXC_FLUSH_CTRL_POLL_VALID;
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Indirect invalidation
|
|
|
|
*/
|
|
|
|
case PC_AT_KILL:
|
|
|
|
case PC_AT_KILL_MASK:
|
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
xive2_error(xive, "PC: invalid write @%"HWADDR_PRIx, offset);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
xive->pc_regs[reg] = val;
|
|
|
|
}
|
|
|
|
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_pc_ops = {
|
|
|
|
.read = pnv_xive2_ic_pc_read,
|
|
|
|
.write = pnv_xive2_ic_pc_write,
|
|
|
|
.endianness = DEVICE_BIG_ENDIAN,
|
|
|
|
.valid = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
.impl = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_ic_tctxt_read(void *opaque, hwaddr offset,
|
|
|
|
unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
uint64_t val = -1;
|
|
|
|
uint32_t reg = offset >> 3;
|
|
|
|
|
|
|
|
switch (offset) {
|
|
|
|
/*
|
|
|
|
* XIVE2 hardware thread enablement
|
|
|
|
*/
|
|
|
|
case TCTXT_EN0:
|
|
|
|
case TCTXT_EN1:
|
|
|
|
val = xive->tctxt_regs[reg];
|
|
|
|
break;
|
|
|
|
|
|
|
|
case TCTXT_EN0_SET:
|
|
|
|
case TCTXT_EN0_RESET:
|
|
|
|
val = xive->tctxt_regs[TCTXT_EN0 >> 3];
|
|
|
|
break;
|
|
|
|
case TCTXT_EN1_SET:
|
|
|
|
case TCTXT_EN1_RESET:
|
|
|
|
val = xive->tctxt_regs[TCTXT_EN1 >> 3];
|
|
|
|
break;
|
2023-06-01 14:13:27 +02:00
|
|
|
case TCTXT_CFG:
|
|
|
|
val = xive->tctxt_regs[reg];
|
|
|
|
break;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
default:
|
|
|
|
xive2_error(xive, "TCTXT: invalid read @%"HWADDR_PRIx, offset);
|
|
|
|
}
|
|
|
|
|
|
|
|
return val;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_ic_tctxt_write(void *opaque, hwaddr offset,
|
|
|
|
uint64_t val, unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
2023-06-01 14:13:27 +02:00
|
|
|
uint32_t reg = offset >> 3;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
|
|
|
|
switch (offset) {
|
|
|
|
/*
|
|
|
|
* XIVE2 hardware thread enablement
|
|
|
|
*/
|
|
|
|
case TCTXT_EN0: /* Physical Thread Enable */
|
|
|
|
case TCTXT_EN1: /* Physical Thread Enable (fused core) */
|
2023-06-01 14:13:29 +02:00
|
|
|
xive->tctxt_regs[reg] = val;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
break;
|
|
|
|
|
|
|
|
case TCTXT_EN0_SET:
|
|
|
|
xive->tctxt_regs[TCTXT_EN0 >> 3] |= val;
|
|
|
|
break;
|
|
|
|
case TCTXT_EN1_SET:
|
|
|
|
xive->tctxt_regs[TCTXT_EN1 >> 3] |= val;
|
|
|
|
break;
|
|
|
|
case TCTXT_EN0_RESET:
|
|
|
|
xive->tctxt_regs[TCTXT_EN0 >> 3] &= ~val;
|
|
|
|
break;
|
|
|
|
case TCTXT_EN1_RESET:
|
|
|
|
xive->tctxt_regs[TCTXT_EN1 >> 3] &= ~val;
|
|
|
|
break;
|
2023-06-01 14:13:27 +02:00
|
|
|
case TCTXT_CFG:
|
|
|
|
xive->tctxt_regs[reg] = val;
|
|
|
|
break;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
default:
|
|
|
|
xive2_error(xive, "TCTXT: invalid write @%"HWADDR_PRIx, offset);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_tctxt_ops = {
|
|
|
|
.read = pnv_xive2_ic_tctxt_read,
|
|
|
|
.write = pnv_xive2_ic_tctxt_write,
|
|
|
|
.endianness = DEVICE_BIG_ENDIAN,
|
|
|
|
.valid = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
.impl = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Redirect XSCOM to MMIO handlers
|
|
|
|
*/
|
|
|
|
static uint64_t pnv_xive2_xscom_read(void *opaque, hwaddr offset,
|
|
|
|
unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
uint64_t val = -1;
|
|
|
|
uint32_t xscom_reg = offset >> 3;
|
|
|
|
uint32_t mmio_offset = (xscom_reg & 0xFF) << 3;
|
|
|
|
|
|
|
|
switch (xscom_reg) {
|
|
|
|
case 0x000 ... 0x0FF:
|
|
|
|
val = pnv_xive2_ic_cq_read(opaque, mmio_offset, size);
|
|
|
|
break;
|
|
|
|
case 0x100 ... 0x1FF:
|
|
|
|
val = pnv_xive2_ic_vc_read(opaque, mmio_offset, size);
|
|
|
|
break;
|
|
|
|
case 0x200 ... 0x2FF:
|
|
|
|
val = pnv_xive2_ic_pc_read(opaque, mmio_offset, size);
|
|
|
|
break;
|
|
|
|
case 0x300 ... 0x3FF:
|
|
|
|
val = pnv_xive2_ic_tctxt_read(opaque, mmio_offset, size);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
xive2_error(xive, "XSCOM: invalid read @%"HWADDR_PRIx, offset);
|
|
|
|
}
|
|
|
|
|
|
|
|
return val;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_xscom_write(void *opaque, hwaddr offset,
|
|
|
|
uint64_t val, unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
uint32_t xscom_reg = offset >> 3;
|
|
|
|
uint32_t mmio_offset = (xscom_reg & 0xFF) << 3;
|
|
|
|
|
|
|
|
switch (xscom_reg) {
|
|
|
|
case 0x000 ... 0x0FF:
|
|
|
|
pnv_xive2_ic_cq_write(opaque, mmio_offset, val, size);
|
|
|
|
break;
|
|
|
|
case 0x100 ... 0x1FF:
|
|
|
|
pnv_xive2_ic_vc_write(opaque, mmio_offset, val, size);
|
|
|
|
break;
|
|
|
|
case 0x200 ... 0x2FF:
|
|
|
|
pnv_xive2_ic_pc_write(opaque, mmio_offset, val, size);
|
|
|
|
break;
|
|
|
|
case 0x300 ... 0x3FF:
|
|
|
|
pnv_xive2_ic_tctxt_write(opaque, mmio_offset, val, size);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
xive2_error(xive, "XSCOM: invalid write @%"HWADDR_PRIx, offset);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static const MemoryRegionOps pnv_xive2_xscom_ops = {
|
|
|
|
.read = pnv_xive2_xscom_read,
|
|
|
|
.write = pnv_xive2_xscom_write,
|
|
|
|
.endianness = DEVICE_BIG_ENDIAN,
|
|
|
|
.valid = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
.impl = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Notify port page. The layout is compatible between 4K and 64K pages :
|
|
|
|
*
|
|
|
|
* Page 1 Notify page (writes only)
|
|
|
|
* 0x000 - 0x7FF IPI interrupt (NPU)
|
|
|
|
* 0x800 - 0xFFF HW interrupt triggers (PSI, PHB)
|
|
|
|
*/
|
|
|
|
|
|
|
|
static void pnv_xive2_ic_hw_trigger(PnvXive2 *xive, hwaddr addr,
|
|
|
|
uint64_t val)
|
|
|
|
{
|
|
|
|
uint8_t blk;
|
|
|
|
uint32_t idx;
|
|
|
|
|
|
|
|
if (val & XIVE_TRIGGER_END) {
|
|
|
|
xive2_error(xive, "IC: END trigger at @0x%"HWADDR_PRIx" data 0x%"PRIx64,
|
|
|
|
addr, val);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Forward the source event notification directly to the Router.
|
|
|
|
* The source interrupt number should already be correctly encoded
|
|
|
|
* with the chip block id by the sending device (PHB, PSI).
|
|
|
|
*/
|
|
|
|
blk = XIVE_EAS_BLOCK(val);
|
|
|
|
idx = XIVE_EAS_INDEX(val);
|
|
|
|
|
2022-03-02 06:51:39 +01:00
|
|
|
xive2_router_notify(XIVE_NOTIFIER(xive), XIVE_EAS(blk, idx),
|
|
|
|
!!(val & XIVE_TRIGGER_PQ));
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_ic_notify_write(void *opaque, hwaddr offset,
|
|
|
|
uint64_t val, unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
|
|
|
|
/* VC: IPI triggers */
|
|
|
|
switch (offset) {
|
|
|
|
case 0x000 ... 0x7FF:
|
|
|
|
/* TODO: check IPI notify sub-page routing */
|
|
|
|
pnv_xive2_ic_hw_trigger(opaque, offset, val);
|
|
|
|
break;
|
|
|
|
|
|
|
|
/* VC: HW triggers */
|
|
|
|
case 0x800 ... 0xFFF:
|
|
|
|
pnv_xive2_ic_hw_trigger(opaque, offset, val);
|
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
xive2_error(xive, "NOTIFY: invalid write @%"HWADDR_PRIx, offset);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_ic_notify_read(void *opaque, hwaddr offset,
|
|
|
|
unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
|
|
|
|
/* loads are invalid */
|
|
|
|
xive2_error(xive, "NOTIFY: invalid read @%"HWADDR_PRIx, offset);
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_notify_ops = {
|
|
|
|
.read = pnv_xive2_ic_notify_read,
|
|
|
|
.write = pnv_xive2_ic_notify_write,
|
|
|
|
.endianness = DEVICE_BIG_ENDIAN,
|
|
|
|
.valid = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
.impl = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_ic_lsi_read(void *opaque, hwaddr offset,
|
|
|
|
unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
|
|
|
|
xive2_error(xive, "LSI: invalid read @%"HWADDR_PRIx, offset);
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_ic_lsi_write(void *opaque, hwaddr offset,
|
|
|
|
uint64_t val, unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
|
|
|
|
xive2_error(xive, "LSI: invalid write @%"HWADDR_PRIx, offset);
|
|
|
|
}
|
|
|
|
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_lsi_ops = {
|
|
|
|
.read = pnv_xive2_ic_lsi_read,
|
|
|
|
.write = pnv_xive2_ic_lsi_write,
|
|
|
|
.endianness = DEVICE_BIG_ENDIAN,
|
|
|
|
.valid = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
.impl = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Sync MMIO page (write only)
|
|
|
|
*/
|
|
|
|
#define PNV_XIVE2_SYNC_IPI 0x000
|
|
|
|
#define PNV_XIVE2_SYNC_HW 0x080
|
|
|
|
#define PNV_XIVE2_SYNC_NxC 0x100
|
|
|
|
#define PNV_XIVE2_SYNC_INT 0x180
|
|
|
|
#define PNV_XIVE2_SYNC_OS_ESC 0x200
|
|
|
|
#define PNV_XIVE2_SYNC_POOL_ESC 0x280
|
|
|
|
#define PNV_XIVE2_SYNC_HARD_ESC 0x300
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_ic_sync_read(void *opaque, hwaddr offset,
|
|
|
|
unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
|
|
|
|
/* loads are invalid */
|
|
|
|
xive2_error(xive, "SYNC: invalid read @%"HWADDR_PRIx, offset);
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_ic_sync_write(void *opaque, hwaddr offset,
|
|
|
|
uint64_t val, unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
|
|
|
|
switch (offset) {
|
|
|
|
case PNV_XIVE2_SYNC_IPI:
|
|
|
|
case PNV_XIVE2_SYNC_HW:
|
|
|
|
case PNV_XIVE2_SYNC_NxC:
|
|
|
|
case PNV_XIVE2_SYNC_INT:
|
|
|
|
case PNV_XIVE2_SYNC_OS_ESC:
|
|
|
|
case PNV_XIVE2_SYNC_POOL_ESC:
|
|
|
|
case PNV_XIVE2_SYNC_HARD_ESC:
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
xive2_error(xive, "SYNC: invalid write @%"HWADDR_PRIx, offset);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_sync_ops = {
|
|
|
|
.read = pnv_xive2_ic_sync_read,
|
|
|
|
.write = pnv_xive2_ic_sync_write,
|
|
|
|
.endianness = DEVICE_BIG_ENDIAN,
|
|
|
|
.valid = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
.impl = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* When the TM direct pages of the IC controller are accessed, the
|
|
|
|
* target HW thread is deduced from the page offset.
|
|
|
|
*/
|
2022-06-02 18:53:10 +02:00
|
|
|
static uint32_t pnv_xive2_ic_tm_get_pir(PnvXive2 *xive, hwaddr offset)
|
|
|
|
{
|
|
|
|
/* On P10, the node ID shift in the PIR register is 8 bits */
|
|
|
|
return xive->chip->chip_id << 8 | offset >> xive->ic_shift;
|
|
|
|
}
|
|
|
|
|
pnv/xive2: Fix TIMA offset for indirect access
Direct TIMA operations can be done through 4 pages, each with a
different privilege level dictating what fields can be accessed. On
the other hand, indirect TIMA accesses on P10 are done through a
single page, which is the equivalent of the most privileged page of
direct TIMA accesses.
The offset in the IC bar of an indirect access specifies what hw
thread is targeted (page shift bits) and the offset in the
TIMA being accessed (the page offset bits). When the indirect
access is calling the underlying direct access functions, it is
therefore important to clearly separate the 2, as the direct functions
assume any page shift bits define the privilege ring level. For
indirect accesses, those bits must be 0. This patch fixes the offset
passed to direct TIMA functions.
It didn't matter for SMT1, as the 2 least significant bits of the page
shift are part of the hw thread ID and always 0, so the direct TIMA
functions were accessing the privilege ring 0 page. With SMT4/8, it is
no longer true.
The fix is specific to P10, as indirect TIMA access on P9 was handled
differently.
Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-ID: <20230703080858.54060-1-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2023-07-03 10:08:58 +02:00
|
|
|
static uint32_t pnv_xive2_ic_tm_get_hw_page_offset(PnvXive2 *xive,
|
|
|
|
hwaddr offset)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* Indirect TIMA accesses are similar to direct accesses for
|
|
|
|
* privilege ring 0. So remove any traces of the hw thread ID from
|
|
|
|
* the offset in the IC BAR as it could be interpreted as the ring
|
|
|
|
* privilege when calling the underlying direct access functions.
|
|
|
|
*/
|
|
|
|
return offset & ((1ull << xive->ic_shift) - 1);
|
|
|
|
}
|
|
|
|
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
static XiveTCTX *pnv_xive2_get_indirect_tctx(PnvXive2 *xive, uint32_t pir)
|
|
|
|
{
|
|
|
|
PnvChip *chip = xive->chip;
|
|
|
|
PowerPCCPU *cpu = NULL;
|
|
|
|
|
|
|
|
cpu = pnv_chip_find_cpu(chip, pir);
|
|
|
|
if (!cpu) {
|
|
|
|
xive2_error(xive, "IC: invalid PIR %x for indirect access", pir);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!pnv_xive2_is_cpu_enabled(xive, cpu)) {
|
|
|
|
xive2_error(xive, "IC: CPU %x is not enabled", pir);
|
|
|
|
}
|
|
|
|
|
|
|
|
return XIVE_TCTX(pnv_cpu_state(cpu)->intc);
|
|
|
|
}
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_ic_tm_indirect_read(void *opaque, hwaddr offset,
|
|
|
|
unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
2023-07-05 10:14:00 +02:00
|
|
|
XivePresenter *xptr = XIVE_PRESENTER(xive);
|
pnv/xive2: Fix TIMA offset for indirect access
Direct TIMA operations can be done through 4 pages, each with a
different privilege level dictating what fields can be accessed. On
the other hand, indirect TIMA accesses on P10 are done through a
single page, which is the equivalent of the most privileged page of
direct TIMA accesses.
The offset in the IC bar of an indirect access specifies what hw
thread is targeted (page shift bits) and the offset in the
TIMA being accessed (the page offset bits). When the indirect
access is calling the underlying direct access functions, it is
therefore important to clearly separate the 2, as the direct functions
assume any page shift bits define the privilege ring level. For
indirect accesses, those bits must be 0. This patch fixes the offset
passed to direct TIMA functions.
It didn't matter for SMT1, as the 2 least significant bits of the page
shift are part of the hw thread ID and always 0, so the direct TIMA
functions were accessing the privilege ring 0 page. With SMT4/8, it is
no longer true.
The fix is specific to P10, as indirect TIMA access on P9 was handled
differently.
Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-ID: <20230703080858.54060-1-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2023-07-03 10:08:58 +02:00
|
|
|
hwaddr hw_page_offset;
|
2022-06-02 18:53:10 +02:00
|
|
|
uint32_t pir;
|
|
|
|
XiveTCTX *tctx;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
uint64_t val = -1;
|
|
|
|
|
2022-06-02 18:53:10 +02:00
|
|
|
pir = pnv_xive2_ic_tm_get_pir(xive, offset);
|
pnv/xive2: Fix TIMA offset for indirect access
Direct TIMA operations can be done through 4 pages, each with a
different privilege level dictating what fields can be accessed. On
the other hand, indirect TIMA accesses on P10 are done through a
single page, which is the equivalent of the most privileged page of
direct TIMA accesses.
The offset in the IC bar of an indirect access specifies what hw
thread is targeted (page shift bits) and the offset in the
TIMA being accessed (the page offset bits). When the indirect
access is calling the underlying direct access functions, it is
therefore important to clearly separate the 2, as the direct functions
assume any page shift bits define the privilege ring level. For
indirect accesses, those bits must be 0. This patch fixes the offset
passed to direct TIMA functions.
It didn't matter for SMT1, as the 2 least significant bits of the page
shift are part of the hw thread ID and always 0, so the direct TIMA
functions were accessing the privilege ring 0 page. With SMT4/8, it is
no longer true.
The fix is specific to P10, as indirect TIMA access on P9 was handled
differently.
Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-ID: <20230703080858.54060-1-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2023-07-03 10:08:58 +02:00
|
|
|
hw_page_offset = pnv_xive2_ic_tm_get_hw_page_offset(xive, offset);
|
2022-06-02 18:53:10 +02:00
|
|
|
tctx = pnv_xive2_get_indirect_tctx(xive, pir);
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
if (tctx) {
|
2023-07-05 10:14:00 +02:00
|
|
|
val = xive_tctx_tm_read(xptr, tctx, hw_page_offset, size);
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
return val;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_ic_tm_indirect_write(void *opaque, hwaddr offset,
|
|
|
|
uint64_t val, unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
2023-07-05 10:14:00 +02:00
|
|
|
XivePresenter *xptr = XIVE_PRESENTER(xive);
|
pnv/xive2: Fix TIMA offset for indirect access
Direct TIMA operations can be done through 4 pages, each with a
different privilege level dictating what fields can be accessed. On
the other hand, indirect TIMA accesses on P10 are done through a
single page, which is the equivalent of the most privileged page of
direct TIMA accesses.
The offset in the IC bar of an indirect access specifies what hw
thread is targeted (page shift bits) and the offset in the
TIMA being accessed (the page offset bits). When the indirect
access is calling the underlying direct access functions, it is
therefore important to clearly separate the 2, as the direct functions
assume any page shift bits define the privilege ring level. For
indirect accesses, those bits must be 0. This patch fixes the offset
passed to direct TIMA functions.
It didn't matter for SMT1, as the 2 least significant bits of the page
shift are part of the hw thread ID and always 0, so the direct TIMA
functions were accessing the privilege ring 0 page. With SMT4/8, it is
no longer true.
The fix is specific to P10, as indirect TIMA access on P9 was handled
differently.
Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-ID: <20230703080858.54060-1-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2023-07-03 10:08:58 +02:00
|
|
|
hwaddr hw_page_offset;
|
2022-06-02 18:53:10 +02:00
|
|
|
uint32_t pir;
|
|
|
|
XiveTCTX *tctx;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
|
2022-06-02 18:53:10 +02:00
|
|
|
pir = pnv_xive2_ic_tm_get_pir(xive, offset);
|
pnv/xive2: Fix TIMA offset for indirect access
Direct TIMA operations can be done through 4 pages, each with a
different privilege level dictating what fields can be accessed. On
the other hand, indirect TIMA accesses on P10 are done through a
single page, which is the equivalent of the most privileged page of
direct TIMA accesses.
The offset in the IC bar of an indirect access specifies what hw
thread is targeted (page shift bits) and the offset in the
TIMA being accessed (the page offset bits). When the indirect
access is calling the underlying direct access functions, it is
therefore important to clearly separate the 2, as the direct functions
assume any page shift bits define the privilege ring level. For
indirect accesses, those bits must be 0. This patch fixes the offset
passed to direct TIMA functions.
It didn't matter for SMT1, as the 2 least significant bits of the page
shift are part of the hw thread ID and always 0, so the direct TIMA
functions were accessing the privilege ring 0 page. With SMT4/8, it is
no longer true.
The fix is specific to P10, as indirect TIMA access on P9 was handled
differently.
Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-ID: <20230703080858.54060-1-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2023-07-03 10:08:58 +02:00
|
|
|
hw_page_offset = pnv_xive2_ic_tm_get_hw_page_offset(xive, offset);
|
2022-06-02 18:53:10 +02:00
|
|
|
tctx = pnv_xive2_get_indirect_tctx(xive, pir);
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
if (tctx) {
|
2023-07-05 10:14:00 +02:00
|
|
|
xive_tctx_tm_write(xptr, tctx, hw_page_offset, val, size);
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static const MemoryRegionOps pnv_xive2_ic_tm_indirect_ops = {
|
|
|
|
.read = pnv_xive2_ic_tm_indirect_read,
|
|
|
|
.write = pnv_xive2_ic_tm_indirect_write,
|
|
|
|
.endianness = DEVICE_BIG_ENDIAN,
|
|
|
|
.valid = {
|
pnv/xive2: Allow indirect TIMA accesses of all sizes
Booting linux on the powernv10 machine logs a few errors like:
Invalid read at addr 0x38, size 1, region 'xive-ic-tm-indirect', reason: invalid size (min:8 max:8)
Invalid write at addr 0x38, size 1, region 'xive-ic-tm-indirect', reason: invalid size (min:8 max:8)
Invalid read at addr 0x38, size 1, region 'xive-ic-tm-indirect', reason: invalid size (min:8 max:8)
Those errors happen when linux is resetting XIVE. We're trying to
read/write the enablement bit for the hardware context and qemu
doesn't allow indirect TIMA accesses of less than 8 bytes. Direct TIMA
access can go through though, as well as indirect TIMA accesses on P9.
So even though there are some restrictions regarding the address/size
combinations for TIMA access, the example above is perfectly valid.
This patch lets indirect TIMA accesses of all sizes go through. The
special operations will be intercepted and the default "raw" handlers
will pick up all other requests and complain about invalid sizes as
appropriate.
Tested-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-ID: <20230626094057.1192473-1-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2023-06-26 11:40:57 +02:00
|
|
|
.min_access_size = 1,
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
.impl = {
|
pnv/xive2: Allow indirect TIMA accesses of all sizes
Booting linux on the powernv10 machine logs a few errors like:
Invalid read at addr 0x38, size 1, region 'xive-ic-tm-indirect', reason: invalid size (min:8 max:8)
Invalid write at addr 0x38, size 1, region 'xive-ic-tm-indirect', reason: invalid size (min:8 max:8)
Invalid read at addr 0x38, size 1, region 'xive-ic-tm-indirect', reason: invalid size (min:8 max:8)
Those errors happen when linux is resetting XIVE. We're trying to
read/write the enablement bit for the hardware context and qemu
doesn't allow indirect TIMA accesses of less than 8 bytes. Direct TIMA
access can go through though, as well as indirect TIMA accesses on P9.
So even though there are some restrictions regarding the address/size
combinations for TIMA access, the example above is perfectly valid.
This patch lets indirect TIMA accesses of all sizes go through. The
special operations will be intercepted and the default "raw" handlers
will pick up all other requests and complain about invalid sizes as
appropriate.
Tested-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-ID: <20230626094057.1192473-1-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2023-06-26 11:40:57 +02:00
|
|
|
.min_access_size = 1,
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* TIMA ops
|
|
|
|
*/
|
|
|
|
static void pnv_xive2_tm_write(void *opaque, hwaddr offset,
|
|
|
|
uint64_t value, unsigned size)
|
|
|
|
{
|
|
|
|
PowerPCCPU *cpu = POWERPC_CPU(current_cpu);
|
|
|
|
PnvXive2 *xive = pnv_xive2_tm_get_xive(cpu);
|
|
|
|
XiveTCTX *tctx = XIVE_TCTX(pnv_cpu_state(cpu)->intc);
|
2022-03-02 06:51:39 +01:00
|
|
|
XivePresenter *xptr = XIVE_PRESENTER(xive);
|
2023-06-01 14:13:31 +02:00
|
|
|
|
2022-03-02 06:51:39 +01:00
|
|
|
xive_tctx_tm_write(xptr, tctx, offset, value, size);
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_tm_read(void *opaque, hwaddr offset, unsigned size)
|
|
|
|
{
|
|
|
|
PowerPCCPU *cpu = POWERPC_CPU(current_cpu);
|
|
|
|
PnvXive2 *xive = pnv_xive2_tm_get_xive(cpu);
|
|
|
|
XiveTCTX *tctx = XIVE_TCTX(pnv_cpu_state(cpu)->intc);
|
2022-03-02 06:51:39 +01:00
|
|
|
XivePresenter *xptr = XIVE_PRESENTER(xive);
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
|
2022-03-02 06:51:39 +01:00
|
|
|
return xive_tctx_tm_read(xptr, tctx, offset, size);
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
static const MemoryRegionOps pnv_xive2_tm_ops = {
|
|
|
|
.read = pnv_xive2_tm_read,
|
|
|
|
.write = pnv_xive2_tm_write,
|
|
|
|
.endianness = DEVICE_BIG_ENDIAN,
|
|
|
|
.valid = {
|
|
|
|
.min_access_size = 1,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
.impl = {
|
|
|
|
.min_access_size = 1,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_nvc_read(void *opaque, hwaddr offset,
|
|
|
|
unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
|
|
|
|
xive2_error(xive, "NVC: invalid read @%"HWADDR_PRIx, offset);
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_nvc_write(void *opaque, hwaddr offset,
|
|
|
|
uint64_t val, unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
|
|
|
|
xive2_error(xive, "NVC: invalid write @%"HWADDR_PRIx, offset);
|
|
|
|
}
|
|
|
|
|
|
|
|
static const MemoryRegionOps pnv_xive2_nvc_ops = {
|
|
|
|
.read = pnv_xive2_nvc_read,
|
|
|
|
.write = pnv_xive2_nvc_write,
|
|
|
|
.endianness = DEVICE_BIG_ENDIAN,
|
|
|
|
.valid = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
.impl = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
static uint64_t pnv_xive2_nvpg_read(void *opaque, hwaddr offset,
|
|
|
|
unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
|
|
|
|
xive2_error(xive, "NVPG: invalid read @%"HWADDR_PRIx, offset);
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_nvpg_write(void *opaque, hwaddr offset,
|
|
|
|
uint64_t val, unsigned size)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(opaque);
|
|
|
|
|
|
|
|
xive2_error(xive, "NVPG: invalid write @%"HWADDR_PRIx, offset);
|
|
|
|
}
|
|
|
|
|
|
|
|
static const MemoryRegionOps pnv_xive2_nvpg_ops = {
|
|
|
|
.read = pnv_xive2_nvpg_read,
|
|
|
|
.write = pnv_xive2_nvpg_write,
|
|
|
|
.endianness = DEVICE_BIG_ENDIAN,
|
|
|
|
.valid = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
.impl = {
|
|
|
|
.min_access_size = 8,
|
|
|
|
.max_access_size = 8,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
2022-03-02 06:51:39 +01:00
|
|
|
* POWER10 default capabilities: 0x2000120076f000FC
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
*/
|
2022-03-02 06:51:39 +01:00
|
|
|
#define PNV_XIVE2_CAPABILITIES 0x2000120076f000FC
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
* POWER10 default configuration: 0x0030000033000000
|
|
|
|
*
|
|
|
|
* 8bits thread id was dropped for P10
|
|
|
|
*/
|
|
|
|
#define PNV_XIVE2_CONFIGURATION 0x0030000033000000
|
|
|
|
|
|
|
|
static void pnv_xive2_reset(void *dev)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(dev);
|
|
|
|
XiveSource *xsrc = &xive->ipi_source;
|
|
|
|
Xive2EndSource *end_xsrc = &xive->end_source;
|
|
|
|
|
|
|
|
xive->cq_regs[CQ_XIVE_CAP >> 3] = xive->capabilities;
|
|
|
|
xive->cq_regs[CQ_XIVE_CFG >> 3] = xive->config;
|
|
|
|
|
|
|
|
/* HW hardwires the #Topology of the chip in the block field */
|
|
|
|
xive->cq_regs[CQ_XIVE_CFG >> 3] |=
|
|
|
|
SETFIELD(CQ_XIVE_CFG_HYP_HARD_BLOCK_ID, 0ull, xive->chip->chip_id);
|
|
|
|
|
|
|
|
/* Set default page size to 64k */
|
|
|
|
xive->ic_shift = xive->esb_shift = xive->end_shift = 16;
|
|
|
|
xive->nvc_shift = xive->nvpg_shift = xive->tm_shift = 16;
|
|
|
|
|
|
|
|
/* Clear source MMIOs */
|
|
|
|
if (memory_region_is_mapped(&xsrc->esb_mmio)) {
|
|
|
|
memory_region_del_subregion(&xive->esb_mmio, &xsrc->esb_mmio);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (memory_region_is_mapped(&end_xsrc->esb_mmio)) {
|
|
|
|
memory_region_del_subregion(&xive->end_mmio, &end_xsrc->esb_mmio);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Maximum number of IRQs and ENDs supported by HW. Will be tuned by
|
|
|
|
* software.
|
|
|
|
*/
|
|
|
|
#define PNV_XIVE2_NR_IRQS (PNV10_XIVE2_ESB_SIZE / (1ull << XIVE_ESB_64K_2PAGE))
|
|
|
|
#define PNV_XIVE2_NR_ENDS (PNV10_XIVE2_END_SIZE / (1ull << XIVE_ESB_64K_2PAGE))
|
|
|
|
|
|
|
|
static void pnv_xive2_realize(DeviceState *dev, Error **errp)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(dev);
|
|
|
|
PnvXive2Class *pxc = PNV_XIVE2_GET_CLASS(dev);
|
|
|
|
XiveSource *xsrc = &xive->ipi_source;
|
|
|
|
Xive2EndSource *end_xsrc = &xive->end_source;
|
|
|
|
Error *local_err = NULL;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
pxc->parent_realize(dev, &local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
assert(xive->chip);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The XiveSource and Xive2EndSource objects are realized with the
|
|
|
|
* maximum allowed HW configuration. The ESB MMIO regions will be
|
|
|
|
* resized dynamically when the controller is configured by the FW
|
|
|
|
* to limit accesses to resources not provisioned.
|
|
|
|
*/
|
|
|
|
object_property_set_int(OBJECT(xsrc), "flags", XIVE_SRC_STORE_EOI,
|
|
|
|
&error_fatal);
|
|
|
|
object_property_set_int(OBJECT(xsrc), "nr-irqs", PNV_XIVE2_NR_IRQS,
|
|
|
|
&error_fatal);
|
|
|
|
object_property_set_link(OBJECT(xsrc), "xive", OBJECT(xive),
|
|
|
|
&error_fatal);
|
|
|
|
qdev_realize(DEVICE(xsrc), NULL, &local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
object_property_set_int(OBJECT(end_xsrc), "nr-ends", PNV_XIVE2_NR_ENDS,
|
|
|
|
&error_fatal);
|
|
|
|
object_property_set_link(OBJECT(end_xsrc), "xive", OBJECT(xive),
|
|
|
|
&error_abort);
|
|
|
|
qdev_realize(DEVICE(end_xsrc), NULL, &local_err);
|
|
|
|
if (local_err) {
|
|
|
|
error_propagate(errp, local_err);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* XSCOM region, used for initial configuration of the BARs */
|
|
|
|
memory_region_init_io(&xive->xscom_regs, OBJECT(dev),
|
|
|
|
&pnv_xive2_xscom_ops, xive, "xscom-xive",
|
|
|
|
PNV10_XSCOM_XIVE2_SIZE << 3);
|
|
|
|
|
|
|
|
/* Interrupt controller MMIO regions */
|
|
|
|
xive->ic_shift = 16;
|
|
|
|
memory_region_init(&xive->ic_mmio, OBJECT(dev), "xive-ic",
|
|
|
|
PNV10_XIVE2_IC_SIZE);
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(xive->ic_mmios); i++) {
|
|
|
|
memory_region_init_io(&xive->ic_mmios[i], OBJECT(dev),
|
|
|
|
pnv_xive2_ic_regions[i].ops, xive,
|
|
|
|
pnv_xive2_ic_regions[i].name,
|
|
|
|
pnv_xive2_ic_regions[i].pgsize << xive->ic_shift);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* VC MMIO regions.
|
|
|
|
*/
|
|
|
|
xive->esb_shift = 16;
|
|
|
|
xive->end_shift = 16;
|
|
|
|
memory_region_init(&xive->esb_mmio, OBJECT(xive), "xive-esb",
|
|
|
|
PNV10_XIVE2_ESB_SIZE);
|
|
|
|
memory_region_init(&xive->end_mmio, OBJECT(xive), "xive-end",
|
|
|
|
PNV10_XIVE2_END_SIZE);
|
|
|
|
|
|
|
|
/* Presenter Controller MMIO region (not modeled) */
|
|
|
|
xive->nvc_shift = 16;
|
|
|
|
xive->nvpg_shift = 16;
|
|
|
|
memory_region_init_io(&xive->nvc_mmio, OBJECT(dev),
|
|
|
|
&pnv_xive2_nvc_ops, xive,
|
|
|
|
"xive-nvc", PNV10_XIVE2_NVC_SIZE);
|
|
|
|
|
|
|
|
memory_region_init_io(&xive->nvpg_mmio, OBJECT(dev),
|
|
|
|
&pnv_xive2_nvpg_ops, xive,
|
|
|
|
"xive-nvpg", PNV10_XIVE2_NVPG_SIZE);
|
|
|
|
|
|
|
|
/* Thread Interrupt Management Area (Direct) */
|
|
|
|
xive->tm_shift = 16;
|
|
|
|
memory_region_init_io(&xive->tm_mmio, OBJECT(dev), &pnv_xive2_tm_ops,
|
|
|
|
xive, "xive-tima", PNV10_XIVE2_TM_SIZE);
|
|
|
|
|
|
|
|
qemu_register_reset(pnv_xive2_reset, dev);
|
|
|
|
}
|
|
|
|
|
|
|
|
static Property pnv_xive2_properties[] = {
|
|
|
|
DEFINE_PROP_UINT64("ic-bar", PnvXive2, ic_base, 0),
|
|
|
|
DEFINE_PROP_UINT64("esb-bar", PnvXive2, esb_base, 0),
|
|
|
|
DEFINE_PROP_UINT64("end-bar", PnvXive2, end_base, 0),
|
|
|
|
DEFINE_PROP_UINT64("nvc-bar", PnvXive2, nvc_base, 0),
|
|
|
|
DEFINE_PROP_UINT64("nvpg-bar", PnvXive2, nvpg_base, 0),
|
|
|
|
DEFINE_PROP_UINT64("tm-bar", PnvXive2, tm_base, 0),
|
|
|
|
DEFINE_PROP_UINT64("capabilities", PnvXive2, capabilities,
|
|
|
|
PNV_XIVE2_CAPABILITIES),
|
|
|
|
DEFINE_PROP_UINT64("config", PnvXive2, config,
|
|
|
|
PNV_XIVE2_CONFIGURATION),
|
|
|
|
DEFINE_PROP_LINK("chip", PnvXive2, chip, TYPE_PNV_CHIP, PnvChip *),
|
|
|
|
DEFINE_PROP_END_OF_LIST(),
|
|
|
|
};
|
|
|
|
|
|
|
|
static void pnv_xive2_instance_init(Object *obj)
|
|
|
|
{
|
|
|
|
PnvXive2 *xive = PNV_XIVE2(obj);
|
|
|
|
|
|
|
|
object_initialize_child(obj, "ipi_source", &xive->ipi_source,
|
|
|
|
TYPE_XIVE_SOURCE);
|
|
|
|
object_initialize_child(obj, "end_source", &xive->end_source,
|
|
|
|
TYPE_XIVE2_END_SOURCE);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int pnv_xive2_dt_xscom(PnvXScomInterface *dev, void *fdt,
|
|
|
|
int xscom_offset)
|
|
|
|
{
|
|
|
|
const char compat_p10[] = "ibm,power10-xive-x";
|
|
|
|
char *name;
|
|
|
|
int offset;
|
|
|
|
uint32_t reg[] = {
|
|
|
|
cpu_to_be32(PNV10_XSCOM_XIVE2_BASE),
|
|
|
|
cpu_to_be32(PNV10_XSCOM_XIVE2_SIZE)
|
|
|
|
};
|
|
|
|
|
|
|
|
name = g_strdup_printf("xive@%x", PNV10_XSCOM_XIVE2_BASE);
|
|
|
|
offset = fdt_add_subnode(fdt, xscom_offset, name);
|
|
|
|
_FDT(offset);
|
|
|
|
g_free(name);
|
|
|
|
|
|
|
|
_FDT((fdt_setprop(fdt, offset, "reg", reg, sizeof(reg))));
|
|
|
|
_FDT(fdt_setprop(fdt, offset, "compatible", compat_p10,
|
|
|
|
sizeof(compat_p10)));
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void pnv_xive2_class_init(ObjectClass *klass, void *data)
|
|
|
|
{
|
|
|
|
DeviceClass *dc = DEVICE_CLASS(klass);
|
|
|
|
PnvXScomInterfaceClass *xdc = PNV_XSCOM_INTERFACE_CLASS(klass);
|
|
|
|
Xive2RouterClass *xrc = XIVE2_ROUTER_CLASS(klass);
|
|
|
|
XiveNotifierClass *xnc = XIVE_NOTIFIER_CLASS(klass);
|
|
|
|
XivePresenterClass *xpc = XIVE_PRESENTER_CLASS(klass);
|
|
|
|
PnvXive2Class *pxc = PNV_XIVE2_CLASS(klass);
|
|
|
|
|
|
|
|
xdc->dt_xscom = pnv_xive2_dt_xscom;
|
|
|
|
|
|
|
|
dc->desc = "PowerNV XIVE2 Interrupt Controller (POWER10)";
|
|
|
|
device_class_set_parent_realize(dc, pnv_xive2_realize,
|
|
|
|
&pxc->parent_realize);
|
|
|
|
device_class_set_props(dc, pnv_xive2_properties);
|
|
|
|
|
|
|
|
xrc->get_eas = pnv_xive2_get_eas;
|
2022-03-02 06:51:39 +01:00
|
|
|
xrc->get_pq = pnv_xive2_get_pq;
|
|
|
|
xrc->set_pq = pnv_xive2_set_pq;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
xrc->get_end = pnv_xive2_get_end;
|
|
|
|
xrc->write_end = pnv_xive2_write_end;
|
|
|
|
xrc->get_nvp = pnv_xive2_get_nvp;
|
|
|
|
xrc->write_nvp = pnv_xive2_write_nvp;
|
2022-03-02 06:51:39 +01:00
|
|
|
xrc->get_config = pnv_xive2_get_config;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
xrc->get_block_id = pnv_xive2_get_block_id;
|
|
|
|
|
|
|
|
xnc->notify = pnv_xive2_notify;
|
|
|
|
|
|
|
|
xpc->match_nvt = pnv_xive2_match_nvt;
|
2023-06-22 18:25:26 +02:00
|
|
|
xpc->get_config = pnv_xive2_presenter_get_config;
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
};
|
|
|
|
|
|
|
|
static const TypeInfo pnv_xive2_info = {
|
|
|
|
.name = TYPE_PNV_XIVE2,
|
|
|
|
.parent = TYPE_XIVE2_ROUTER,
|
|
|
|
.instance_init = pnv_xive2_instance_init,
|
|
|
|
.instance_size = sizeof(PnvXive2),
|
|
|
|
.class_init = pnv_xive2_class_init,
|
|
|
|
.class_size = sizeof(PnvXive2Class),
|
|
|
|
.interfaces = (InterfaceInfo[]) {
|
|
|
|
{ TYPE_PNV_XSCOM_INTERFACE },
|
|
|
|
{ }
|
|
|
|
}
|
|
|
|
};
|
|
|
|
|
|
|
|
static void pnv_xive2_register_types(void)
|
|
|
|
{
|
|
|
|
type_register_static(&pnv_xive2_info);
|
|
|
|
}
|
|
|
|
|
|
|
|
type_init(pnv_xive2_register_types)
|
|
|
|
|
|
|
|
static void xive2_nvp_pic_print_info(Xive2Nvp *nvp, uint32_t nvp_idx,
|
|
|
|
Monitor *mon)
|
|
|
|
{
|
|
|
|
uint8_t eq_blk = xive_get_field32(NVP2_W5_VP_END_BLOCK, nvp->w5);
|
|
|
|
uint32_t eq_idx = xive_get_field32(NVP2_W5_VP_END_INDEX, nvp->w5);
|
|
|
|
|
|
|
|
if (!xive2_nvp_is_valid(nvp)) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2022-03-02 06:51:39 +01:00
|
|
|
monitor_printf(mon, " %08x end:%02x/%04x IPB:%02x",
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
nvp_idx, eq_blk, eq_idx,
|
|
|
|
xive_get_field32(NVP2_W2_IPB, nvp->w2));
|
2022-03-02 06:51:39 +01:00
|
|
|
/*
|
|
|
|
* When the NVP is HW controlled, more fields are updated
|
|
|
|
*/
|
|
|
|
if (xive2_nvp_is_hw(nvp)) {
|
|
|
|
monitor_printf(mon, " CPPR:%02x",
|
|
|
|
xive_get_field32(NVP2_W2_CPPR, nvp->w2));
|
|
|
|
if (xive2_nvp_is_co(nvp)) {
|
|
|
|
monitor_printf(mon, " CO:%04x",
|
|
|
|
xive_get_field32(NVP2_W1_CO_THRID, nvp->w1));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
monitor_printf(mon, "\n");
|
ppc/pnv: Add a XIVE2 controller to the POWER10 chip
The XIVE2 interrupt controller of the POWER10 processor follows the
same logic than on POWER9 but the HW interface has been largely
reviewed. It has a new register interface, different BARs, extra
VSDs, new layout for the XIVE2 structures, and a set of new features
which are described below.
This is a model of the POWER10 XIVE2 interrupt controller for the
PowerNV machine. It focuses primarily on the needs of the skiboot
firmware but some initial hypervisor support is implemented for KVM
use (escalation).
Support for new features will be implemented in time and will require
new support from the OS.
* XIVE2 BARS
The interrupt controller BARs have a different layout outlined below.
Each sub-engine has now own its range and the indirect TIMA access was
replaced with a set of pages, one per CPU, under the IC BAR:
- IC BAR (Interrupt Controller)
. 4 pages, one per sub-engine
. 128 indirect TIMA pages
- TM BAR (Thread Interrupt Management Area)
. 4 pages
- ESB BAR (ESB pages for IPIs)
. up to 1TB
- END BAR (ESB pages for ENDs)
. up to 2TB
- NVC BAR (Notification Virtual Crowd)
. up to 128
- NVPG BAR (Notification Virtual Process and Group)
. up to 1TB
- Direct mapped Thread Context Area (reads & writes)
OPAL does not use the grouping and crowd capability.
* Virtual Structure Tables
XIVE2 adds new tables types and also changes the field layout of the END
and NVP Virtualization Structure Descriptors.
- EAS
- END new layout
- NVT was splitted in :
. NVP (Processor), 32B
. NVG (Group), 32B
. NVC (Crowd == P9 block group) 32B
- IC for remote configuration
- SYNC for cache injection
- ERQ for event input queue
The setup is slighly different on XIVE2 because the indexing has changed
for some of the tables, block ID or the chip topology ID can be used.
* XIVE2 features
SCOM and MMIO registers have a new layout and XIVE2 adds a new global
capability and configuration registers.
The lowlevel hardware offers a set of new features among which :
- a configurable number of priorities : 1 - 8
- StoreEOI with load-after-store ordering is activated by default
- Gen2 TIMA layout
- A P9-compat mode, or Gen1, TIMA toggle bit for SW compatibility
- increase to 24bit for VP number
Other features will have some impact on the Hypervisor and guest OS
when activated, but this is not required for initial support of the
controller.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
2022-03-02 06:51:38 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If the table is direct, we can compute the number of PQ entries
|
|
|
|
* provisioned by FW.
|
|
|
|
*/
|
|
|
|
static uint32_t pnv_xive2_nr_esbs(PnvXive2 *xive)
|
|
|
|
{
|
|
|
|
uint8_t blk = pnv_xive2_block_id(xive);
|
|
|
|
uint64_t vsd = xive->vsds[VST_ESB][blk];
|
|
|
|
uint64_t vst_tsize = 1ull << (GETFIELD(VSD_TSIZE, vsd) + 12);
|
|
|
|
|
|
|
|
return VSD_INDIRECT & vsd ? 0 : vst_tsize * SBE_PER_BYTE;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Compute the number of entries per indirect subpage.
|
|
|
|
*/
|
|
|
|
static uint64_t pnv_xive2_vst_per_subpage(PnvXive2 *xive, uint32_t type)
|
|
|
|
{
|
|
|
|
uint8_t blk = pnv_xive2_block_id(xive);
|
|
|
|
uint64_t vsd = xive->vsds[type][blk];
|
|
|
|
const XiveVstInfo *info = &vst_infos[type];
|
|
|
|
uint64_t vsd_addr;
|
|
|
|
uint32_t page_shift;
|
|
|
|
|
|
|
|
/* For direct tables, fake a valid value */
|
|
|
|
if (!(VSD_INDIRECT & vsd)) {
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Get the page size of the indirect table. */
|
|
|
|
vsd_addr = vsd & VSD_ADDRESS_MASK;
|
|
|
|
ldq_be_dma(&address_space_memory, vsd_addr, &vsd, MEMTXATTRS_UNSPECIFIED);
|
|
|
|
|
|
|
|
if (!(vsd & VSD_ADDRESS_MASK)) {
|
|
|
|
#ifdef XIVE2_DEBUG
|
|
|
|
xive2_error(xive, "VST: invalid %s entry!?", info->name);
|
|
|
|
#endif
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
page_shift = GETFIELD(VSD_TSIZE, vsd) + 12;
|
|
|
|
|
|
|
|
if (!pnv_xive2_vst_page_size_allowed(page_shift)) {
|
|
|
|
xive2_error(xive, "VST: invalid %s page shift %d", info->name,
|
|
|
|
page_shift);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
return (1ull << page_shift) / info->size;
|
|
|
|
}
|
|
|
|
|
|
|
|
void pnv_xive2_pic_print_info(PnvXive2 *xive, Monitor *mon)
|
|
|
|
{
|
|
|
|
Xive2Router *xrtr = XIVE2_ROUTER(xive);
|
|
|
|
uint8_t blk = pnv_xive2_block_id(xive);
|
|
|
|
uint8_t chip_id = xive->chip->chip_id;
|
|
|
|
uint32_t srcno0 = XIVE_EAS(blk, 0);
|
|
|
|
uint32_t nr_esbs = pnv_xive2_nr_esbs(xive);
|
|
|
|
Xive2Eas eas;
|
|
|
|
Xive2End end;
|
|
|
|
Xive2Nvp nvp;
|
|
|
|
int i;
|
|
|
|
uint64_t xive_nvp_per_subpage;
|
|
|
|
|
|
|
|
monitor_printf(mon, "XIVE[%x] Source %08x .. %08x\n", blk, srcno0,
|
|
|
|
srcno0 + nr_esbs - 1);
|
|
|
|
xive_source_pic_print_info(&xive->ipi_source, srcno0, mon);
|
|
|
|
|
|
|
|
monitor_printf(mon, "XIVE[%x] EAT %08x .. %08x\n", blk, srcno0,
|
|
|
|
srcno0 + nr_esbs - 1);
|
|
|
|
for (i = 0; i < nr_esbs; i++) {
|
|
|
|
if (xive2_router_get_eas(xrtr, blk, i, &eas)) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (!xive2_eas_is_masked(&eas)) {
|
|
|
|
xive2_eas_pic_print_info(&eas, i, mon);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
monitor_printf(mon, "XIVE[%x] #%d END Escalation EAT\n", chip_id, blk);
|
|
|
|
i = 0;
|
|
|
|
while (!xive2_router_get_end(xrtr, blk, i, &end)) {
|
|
|
|
xive2_end_eas_pic_print_info(&end, i++, mon);
|
|
|
|
}
|
|
|
|
|
|
|
|
monitor_printf(mon, "XIVE[%x] #%d ENDT\n", chip_id, blk);
|
|
|
|
i = 0;
|
|
|
|
while (!xive2_router_get_end(xrtr, blk, i, &end)) {
|
|
|
|
xive2_end_pic_print_info(&end, i++, mon);
|
|
|
|
}
|
|
|
|
|
|
|
|
monitor_printf(mon, "XIVE[%x] #%d NVPT %08x .. %08x\n", chip_id, blk,
|
|
|
|
0, XIVE2_NVP_COUNT - 1);
|
|
|
|
xive_nvp_per_subpage = pnv_xive2_vst_per_subpage(xive, VST_NVP);
|
|
|
|
for (i = 0; i < XIVE2_NVP_COUNT; i += xive_nvp_per_subpage) {
|
|
|
|
while (!xive2_router_get_nvp(xrtr, blk, i, &nvp)) {
|
|
|
|
xive2_nvp_pic_print_info(&nvp, i++, mon);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|