Skip to contentSkip to conversation

RFC: standard pci bridge device

PATCH In review Michael S. Tsirkin 6 participants 26 messages Feb 13, 2012 – Feb 29, 2012

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover RFC: standard pci bridge device

Here's a new version of the patch. It works for me. Deep nesting of bridges is supported.
You need a small BIOS patch to support the OSHP method if you want hotplug to work. I will post this separately. We'd need a full ACPI driver to make hotplug work for guests without an SHPC driver (e.g. windows XP). Management support will also be needed.

One small wrinkle is that the pci_addr property wants data in a format bus:device.function which is broken as guests can change bus numbers.
For testing I used the 'addr' property which encodes slot*8+function#. We probably want to extend pci_addr in some way (e.g. :device.function ? Thoughts?).

The SHPC controller supports up to 31 devices (out of 32 slots) so slot 0 doesn't support hotplug. Non hot-pluggable devices behind the bridge don't work currectly (we'll try to unplug them) so don't do this.
For now I just blocked adding devices in slot 0, in the future it might be possible to add
a non-hotpluggable device there.

Example:

qemu-system-x86_64 -enable-kvm -m 1G
-drive file=/home/mst/rhel6.qcow2
-netdev tap,id=foo,ifname=msttap0,script=/home/mst/ifup,downscript=no,vhost=on -device pci-bridge,id=bog
-device virtio-net-pci,netdev=foo,bus=bog,addr=8

Hot-unplug currently causes qemu to crash, this happens without this patch too, so I'm not worried :)

New since v1:
hotplug support

Signature · 20 lines
MST


Michael S. Tsirkin (2):
  shpc: standard hot plug controller
  pci: add standard bridge device

 Makefile.objs       |    3 +-
 hw/pci.h            |    6 +
 hw/pci_bridge_dev.c |  136 +++++++++++
 hw/shpc.c           |  646 +++++++++++++++++++++++++++++++++++++++++++++++++++
 hw/shpc.h           |   40 ++++
 qemu-common.h       |    1 +
 6 files changed, 831 insertions(+), 1 deletions(-)
 create mode 100644 hw/pci_bridge_dev.c
 create mode 100644 hw/shpc.c
 create mode 100644 hw/shpc.h

-- 
1.7.9.111.gf3fb0
#1 lore raw

PATCH 1/2 shpc: standard hot plug controller

This adds support for SHPC interface, as defined by PCI Standard Hot-Plug Controller and Subsystem Specification, Rev 1.0
http://www.pcisig.com/specifications/conventional/pci_hot_plug/SHPC_10

Only SHPC intergrated with a PCI-to-PCI bridge is supported, SHPC integrated with a host bridge would need more work.

All main SHPC features are supported:
- MRL sensor
- Attention button
- Attention indicator
- Power indicator

Wake on hotplug and serr generation are stubbed out but unused as we don't have interfaces to generate these events ATM.

One issue that isn't completely resolved is that qemu currently expects an "eject" interface, which SHPC does not provide: it merely removes the power to device and it's up to the user to remove the device from slot. This patch works around that by ejecting the device when power is removed and power LED goes off.

TODO:
- migration support
- fix dependency on pci_internals.h

---
 Makefile.objs |    1 +
 hw/pci.h      |    6 +
 hw/shpc.c     |  646 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 hw/shpc.h     |   40 ++++
 qemu-common.h |    1 +
 5 files changed, 694 insertions(+), 0 deletions(-)
 create mode 100644 hw/shpc.c
 create mode 100644 hw/shpc.h
Makefile.objs +1 -0
index 391e524..4546477 100644
@@ -195,6 +195,7 @@ hw-obj-$(CONFIG_VIRTIO_PCI) += virtio-pci.o
 hw-obj-y += fw_cfg.o
 hw-obj-$(CONFIG_PCI) += pci.o pci_bridge.o
 hw-obj-$(CONFIG_PCI) += msix.o msi.o
+hw-obj-$(CONFIG_PCI) += shpc.o
 hw-obj-$(CONFIG_PCI) += pci_host.o pcie_host.o
 hw-obj-$(CONFIG_PCI) += ioh3420.o xio3130_upstream.o xio3130_downstream.o
 hw-obj-y += watchdog.o
hw/pci.h +6 -0
index 33b0b18..756577e 100644
@@ -125,6 +125,9 @@ enum {
     /* command register SERR bit enabled */
 #define QEMU_PCI_CAP_SERR_BITNR 4
     QEMU_PCI_CAP_SERR = (1 << QEMU_PCI_CAP_SERR_BITNR),
+    /* Standard hot plug controller. */
+#define QEMU_PCI_SHPC_BITNR 5
+    QEMU_PCI_CAP_SHPC = (1 << QEMU_PCI_SHPC_BITNR),
 };
 
 #define TYPE_PCI_DEVICE "pci-device"
@@ -229,6 +232,9 @@ struct PCIDevice {
     /* PCI Express */
     PCIExpressDevice exp;
 
+    /* SHPC */
+    SHPCDevice *shpc;
+
     /* Location of option rom */
     char *romfile;
     bool has_rom;
hw/shpc.c new file +646 -0

646 lines not on this page · raw message

hw/shpc.h new file +40 -0
new file mode 100644
index 0000000..389b178
@@ -0,0 +1,40 @@
+#ifndef SHPC_H
+#define SHPC_H
+
+#include "qemu-common.h"
+#include "memory.h"
+
+struct SHPCDevice {
+    /* Capability offset in device's config space */
+    int cap;
+
+    /* # of hot-pluggable slots */
+    int nslots;
+
+    /* SHPC WRS: working register set */
+    uint8_t *config;
+
+    /* Used to enable checks on load. Note that writable bits are
+     * never checked even if set in cmask. */
+    uint8_t *cmask;
+
+    /* Used to implement R/W bytes */
+    uint8_t *wmask;
+
+    /* Used to implement RW1C(Write 1 to Clear) bytes */
+    uint8_t *w1cmask;
+
+    /* MMIO for the SHPC BAR */
+    MemoryRegion mmio;
+
+    /* Bus controlled by this SHPC */
+    PCIBus *sec_bus;
+};
+
+void shpc_reset(PCIDevice *d);
+int shpc_bar_size(PCIDevice *dev);
+int shpc_init(PCIDevice *dev, PCIBus *sec_bus, MemoryRegion *bar, unsigned offset);
+void shpc_cleanup(PCIDevice *dev);
+void shpc_cap_write_config(PCIDevice *d, uint32_t addr, uint32_t val, int len);
+
+#endif
qemu-common.h +1 -0
index 9b997f8..4ff9c95 100644
@@ -247,6 +247,7 @@ typedef struct SSIBus SSIBus;
 typedef struct EventNotifier EventNotifier;
 typedef struct VirtIODevice VirtIODevice;
 typedef struct QEMUSGList QEMUSGList;
+typedef struct SHPCDevice SHPCDevice;
 
 typedef uint64_t pcibus_t;
 
Signature · 1 line
1.7.9.111.gf3fb0
#2 lore raw

Re: [PATCHv2-RFC 1/2] shpc: standard hot plug controller

Oh nice work.

Quoted 9 lines · Michael S. Tsirkin

This adds support for SHPC interface, as defined by PCI Standard Hot-Plug Controller and Subsystem Specification, Rev 1.0
http://www.pcisig.com/specifications/conventional/pci_hot_plug/SHPC_10

Only SHPC intergrated with a PCI-to-PCI bridge is supported, SHPC integrated with a host bridge would need more work.

All main SHPC features are supported:
- MRL sensor

Does this just report latch status? (It seems so.) Do you plan to provide interfaces to manipulate the latch?

Quoted 13 lines

- Attention button
- Attention indicator
- Power indicator

Wake on hotplug and serr generation are stubbed out but unused as we don't have interfaces to generate these events ATM.

One issue that isn't completely resolved is that qemu currently expects an "eject" interface, which SHPC does not provide: it merely removes the power to device and it's up to the user to remove the device from slot. This patch works around that by ejecting the device when power is removed and power LED goes off.

Quoted 3 lines

TODO:
- migration support
- fix dependency on pci_internals.h

If I didn't miss the code,
- QMP command for pushing attention button.
- QMP command to get LED status
- QMP events for LED on/off

thanks,

Quoted 760 lines
---
 Makefile.objs |    1 +
 hw/pci.h      |    6 +
 hw/shpc.c     |  646 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 hw/shpc.h     |   40 ++++
 qemu-common.h |    1 +
 5 files changed, 694 insertions(+), 0 deletions(-)
 create mode 100644 hw/shpc.c
 create mode 100644 hw/shpc.h
Makefile.objs +1 -0
index 391e524..4546477 100644
@@ -195,6 +195,7 @@ hw-obj-$(CONFIG_VIRTIO_PCI) += virtio-pci.o
 hw-obj-y += fw_cfg.o
 hw-obj-$(CONFIG_PCI) += pci.o pci_bridge.o
 hw-obj-$(CONFIG_PCI) += msix.o msi.o
+hw-obj-$(CONFIG_PCI) += shpc.o
 hw-obj-$(CONFIG_PCI) += pci_host.o pcie_host.o
 hw-obj-$(CONFIG_PCI) += ioh3420.o xio3130_upstream.o xio3130_downstream.o
 hw-obj-y += watchdog.o
hw/pci.h +6 -0
index 33b0b18..756577e 100644
@@ -125,6 +125,9 @@ enum {
     /* command register SERR bit enabled */
 #define QEMU_PCI_CAP_SERR_BITNR 4
     QEMU_PCI_CAP_SERR = (1 << QEMU_PCI_CAP_SERR_BITNR),
+    /* Standard hot plug controller. */
+#define QEMU_PCI_SHPC_BITNR 5
+    QEMU_PCI_CAP_SHPC = (1 << QEMU_PCI_SHPC_BITNR),
 };
 
 #define TYPE_PCI_DEVICE "pci-device"
@@ -229,6 +232,9 @@ struct PCIDevice {
     /* PCI Express */
     PCIExpressDevice exp;
 
+    /* SHPC */
+    SHPCDevice *shpc;
+
     /* Location of option rom */
     char *romfile;
     bool has_rom;
hw/shpc.c new file +646 -0
new file mode 100644
index 0000000..4baec29
@@ -0,0 +1,646 @@
+#include <strings.h>
+#include <stdint.h>
+#include "range.h"
+#include "shpc.h"
+#include "pci.h"
+#include "pci_internals.h"
+
+/* TODO: model power only and disabled slot states. */
+/* TODO: handle SERR and wakeups */
+/* TODO: consider enabling 66MHz support */
+
+/* TODO: remove fully only on state DISABLED and LED off.
+ * track state to properly record this. */
+
+/* SHPC Working Register Set */
+#define SHPC_BASE_OFFSET  0x00 /* 4 bytes */
+#define SHPC_SLOTS_33     0x04 /* 4 bytes. Also encodes PCI-X slots. */
+#define SHPC_SLOTS_66     0x08 /* 4 bytes. */
+#define SHPC_NSLOTS       0x0C /* 1 byte */
+#define SHPC_FIRST_DEV    0x0D /* 1 byte */
+#define SHPC_PHYS_SLOT    0x0E /* 2 byte */
+#define SHPC_PHYS_NUM_MAX 0x7ff
+#define SHPC_PHYS_NUM_UP  0x1000
+#define SHPC_PHYS_MRL     0x4000
+#define SHPC_PHYS_BUTTON  0x8000
+#define SHPC_SEC_BUS      0x10 /* 2 bytes */
+#define SHPC_SEC_BUS_33   0x0
+#define SHPC_SEC_BUS_66   0x1 /* Unused */
+#define SHPC_SEC_BUS_MASK 0x7
+#define SHPC_MSI_CTL      0x12 /* 1 byte */
+#define SHPC_PROG_IFC     0x13 /* 1 byte */
+#define SHPC_PROG_IFC_1_0 0x1
+#define SHPC_CMD_CODE     0x14 /* 1 byte */
+#define SHPC_CMD_TRGT     0x15 /* 1 byte */
+#define SHPC_CMD_TRGT_MIN 0x1
+#define SHPC_CMD_TRGT_MAX 0x1f
+#define SHPC_CMD_STATUS   0x16 /* 2 bytes */
+#define SHPC_CMD_STATUS_BUSY          0x1
+#define SHPC_CMD_STATUS_MRL_OPEN      0x2
+#define SHPC_CMD_STATUS_INVALID_CMD   0x4
+#define SHPC_CMD_STATUS_INVALID_MODE  0x8
+#define SHPC_INT_LOCATOR  0x18 /* 4 bytes */
+#define SHPC_INT_COMMAND  0x1
+#define SHPC_SERR_LOCATOR 0x1C /* 4 bytes */
+#define SHPC_SERR_INT     0x20 /* 4 bytes */
+#define SHPC_INT_DIS      0x1
+#define SHPC_SERR_DIS     0x2
+#define SHPC_CMD_INT_DIS  0x4
+#define SHPC_ARB_SERR_DIS 0x8
+#define SHPC_CMD_DETECTED 0x10000
+#define SHPC_ARB_DETECTED 0x20000
+ /* 4 bytes * slot # (start from 0) */
+#define SHPC_SLOT_REG(s)         (0x24 + (s) * 4)
+ /* 2 bytes */
+#define SHPC_SLOT_STATUS(s)       (0x0 + SHPC_SLOT_REG(s))
+
+/* Same slot state masks are used for command and status registers */
+#define SHPC_SLOT_STATE_MASK     0x03
+#define SHPC_SLOT_STATE_SHIFT \
+    (ffs(SHPC_SLOT_STATE_MASK) - 1)
+
+#define SHPC_STATE_NO       0x0
+#define SHPC_STATE_PWRONLY  0x1
+#define SHPC_STATE_ENABLED  0x2
+#define SHPC_STATE_DISABLED 0x3
+
+#define SHPC_SLOT_PWR_LED_MASK   0xC
+#define SHPC_SLOT_PWR_LED_SHIFT \
+    (ffs(SHPC_SLOT_PWR_LED_MASK) - 1)
+#define SHPC_SLOT_ATTN_LED_MASK  0x30
+#define SHPC_SLOT_ATTN_LED_SHIFT \
+    (ffs(SHPC_SLOT_ATTN_LED_MASK) - 1)
+
+#define SHPC_LED_NO     0x0
+#define SHPC_LED_ON     0x1
+#define SHPC_LED_BLINK  0x2
+#define SHPC_LED_OFF    0x3
+
+#define SHPC_SLOT_STATUS_PWR_FAULT      0x40
+#define SHPC_SLOT_STATUS_BUTTON         0x80
+#define SHPC_SLOT_STATUS_MRL_OPEN       0x100
+#define SHPC_SLOT_STATUS_66             0x200
+#define SHPC_SLOT_STATUS_PRSNT_MASK     0xC00
+#define SHPC_SLOT_STATUS_PRSNT_EMPTY    0x3
+#define SHPC_SLOT_STATUS_PRSNT_25W      0x1
+#define SHPC_SLOT_STATUS_PRSNT_15W      0x2
+#define SHPC_SLOT_STATUS_PRSNT_7_5W     0x0
+
+#define SHPC_SLOT_STATUS_PRSNT_PCIX     0x3000
+
+
+ /* 1 byte */
+#define SHPC_SLOT_EVENT_LATCH(s)        (0x2 + SHPC_SLOT_REG(s))
+ /* 1 byte */
+#define SHPC_SLOT_EVENT_SERR_INT_DIS(d, s) (0x3 + SHPC_SLOT_REG(s))
+#define SHPC_SLOT_EVENT_PRESENCE        0x01
+#define SHPC_SLOT_EVENT_ISOLATED_FAULT  0x02
+#define SHPC_SLOT_EVENT_BUTTON          0x04
+#define SHPC_SLOT_EVENT_MRL             0x08
+#define SHPC_SLOT_EVENT_CONNECTED_FAULT 0x10
+/* Bits below are used for Serr/Int disable only */
+#define SHPC_SLOT_EVENT_MRL_SERR_DIS    0x20
+#define SHPC_SLOT_EVENT_CONNECTED_FAULT_SERR_DIS 0x40
+
+#define SHPC_MIN_SLOTS        1
+#define SHPC_MAX_SLOTS        31
+#define SHPC_SIZEOF(d)    SHPC_SLOT_REG((d)->shpc->nslots)
+
+/* SHPC Slot identifiers */
+
+/* Hotplug supported at 31 slots out of the total 32.  We reserve slot 0,
+   and give the rest of them physical *and* pci numbers starting from 1, so
+   they match logical numbers.  Note: this means that multiple slots must have
+   different chassis number values, to make chassis+physical slot unique.
+   TODO: make this configurable? */
+#define SHPC_IDX_TO_LOGICAL(slot) ((slot) + 1)
+#define SHPC_LOGICAL_TO_IDX(target) ((target) - 1)
+#define SHPC_IDX_TO_PCI(slot) ((slot) + 1)
+#define SHPC_PCI_TO_IDX(pci_slot) ((pci_slot) - 1)
+#define SHPC_IDX_TO_PHYSICAL(slot) ((slot) + 1)
+
+static int roundup_pow_of_two(int x)
+{
+    x |= (x >> 1);
+    x |= (x >> 2);
+    x |= (x >> 4);
+    x |= (x >> 8);
+    x |= (x >> 16);
+    return x + 1;
+}
+
+static uint16_t shpc_get_status(SHPCDevice *shpc, int slot, uint16_t msk)
+{
+    uint8_t *status = shpc->config + SHPC_SLOT_STATUS(slot);
+    return (pci_get_word(status) & msk) >> (ffs(msk) - 1);
+}
+
+static void shpc_set_status(SHPCDevice *shpc,
+                            int slot, uint8_t value, uint16_t msk)
+{
+    uint8_t *status = shpc->config + SHPC_SLOT_STATUS(slot);
+    pci_word_test_and_clear_mask(status, msk);
+    pci_word_test_and_set_mask(status, value << (ffs(msk) - 1));
+}
+
+static void shpc_interrupt_update(PCIDevice *d)
+{
+    SHPCDevice *shpc = d->shpc;
+    int slot;
+    int level = 0;
+    uint32_t serr_int;
+    uint32_t int_locator = 0;
+
+    /* Update interrupt locator register */
+    for (slot = 0; slot < shpc->nslots; ++slot) {
+        uint8_t event = shpc->config[SHPC_SLOT_EVENT_LATCH(slot)];
+        uint8_t disable = shpc->config[SHPC_SLOT_EVENT_SERR_INT_DIS(d, slot)];
+        uint32_t mask = 1 << SHPC_IDX_TO_LOGICAL(slot);
+        if (event & ~disable) {
+            int_locator |= mask;
+        }
+    }
+    serr_int = pci_get_long(shpc->config + SHPC_SERR_INT);
+    if ((serr_int & SHPC_CMD_DETECTED) && !(serr_int & SHPC_CMD_INT_DIS)) {
+        int_locator |= SHPC_INT_COMMAND;
+    }
+    pci_set_long(shpc->config + SHPC_INT_LOCATOR, int_locator);
+    level = (!(serr_int & SHPC_INT_DIS) && int_locator) ? 1 : 0;
+    qemu_set_irq(d->irq[0], level);
+}
+
+static void shpc_set_sec_bus_speed(SHPCDevice *shpc, uint8_t speed)
+{
+    switch (speed) {
+    case SHPC_SEC_BUS_33:
+        shpc->config[SHPC_SEC_BUS] &= ~SHPC_SEC_BUS_MASK;
+        shpc->config[SHPC_SEC_BUS] |= speed;
+        break;
+    default:
+	pci_word_test_and_set_mask(shpc->config + SHPC_CMD_STATUS,
+				   SHPC_CMD_STATUS_INVALID_MODE);
+    }
+}
+
+void shpc_reset(PCIDevice *d)
+{
+    SHPCDevice *shpc = d->shpc;
+    int nslots = shpc->nslots;
+    int i;
+    memset(shpc->config, 0, SHPC_SIZEOF(d));
+    pci_set_byte(shpc->config + SHPC_NSLOTS, nslots);
+    pci_set_long(shpc->config + SHPC_SLOTS_33, nslots);
+    pci_set_long(shpc->config + SHPC_SLOTS_66, 0);
+    pci_set_byte(shpc->config + SHPC_FIRST_DEV, SHPC_IDX_TO_PCI(0));
+    pci_set_word(shpc->config + SHPC_PHYS_SLOT,
+		 SHPC_IDX_TO_PHYSICAL(0) |
+		 SHPC_PHYS_NUM_UP |
+		 SHPC_PHYS_MRL |
+		 SHPC_PHYS_BUTTON);
+    pci_set_long(shpc->config + SHPC_SERR_INT, SHPC_INT_DIS |
+                 SHPC_SERR_DIS |
+                 SHPC_CMD_INT_DIS |
+                 SHPC_ARB_SERR_DIS);
+    pci_set_byte(shpc->config + SHPC_PROG_IFC, SHPC_PROG_IFC_1_0);
+    pci_set_word(shpc->config + SHPC_SEC_BUS, SHPC_SEC_BUS_33);
+    for (i = 0; i < shpc->nslots; ++i) {
+        pci_set_word(shpc->config + SHPC_SLOT_EVENT_SERR_INT_DIS(d, i),
+                     SHPC_SLOT_EVENT_PRESENCE |
+                     SHPC_SLOT_EVENT_ISOLATED_FAULT |
+                     SHPC_SLOT_EVENT_BUTTON |
+                     SHPC_SLOT_EVENT_MRL |
+                     SHPC_SLOT_EVENT_CONNECTED_FAULT |
+                     SHPC_SLOT_EVENT_MRL_SERR_DIS |
+                     SHPC_SLOT_EVENT_CONNECTED_FAULT_SERR_DIS);
+        if (shpc->sec_bus->devices[PCI_DEVFN(SHPC_IDX_TO_PCI(i), 0)]) {
+            shpc_set_status(shpc, i, SHPC_STATE_ENABLED, SHPC_SLOT_STATE_MASK);
+            shpc_set_status(shpc, i, 0, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, i, SHPC_SLOT_STATUS_PRSNT_7_5W,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc_set_status(shpc, i, SHPC_LED_ON, SHPC_SLOT_PWR_LED_MASK);
+        } else {
+            shpc_set_status(shpc, i, SHPC_STATE_DISABLED, SHPC_SLOT_STATE_MASK);
+            shpc_set_status(shpc, i, 1, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, i, SHPC_SLOT_STATUS_PRSNT_EMPTY,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc_set_status(shpc, i, SHPC_LED_OFF, SHPC_SLOT_PWR_LED_MASK);
+	}
+        shpc_set_status(shpc, i, 0, SHPC_SLOT_STATUS_66);
+    }
+    shpc_set_sec_bus_speed(shpc, SHPC_SEC_BUS_33);
+    shpc_interrupt_update(d);
+}
+
+static void shpc_invalid_command(SHPCDevice *shpc)
+{
+    pci_word_test_and_set_mask(shpc->config + SHPC_CMD_STATUS,
+                               SHPC_CMD_STATUS_INVALID_CMD);
+}
+
+static void shpc_free_devices_in_slot(SHPCDevice *shpc, int slot)
+{
+    int devfn;
+    int pci_slot = SHPC_IDX_TO_PCI(slot);
+    for (devfn = PCI_DEVFN(pci_slot, 0);
+         devfn <= PCI_DEVFN(pci_slot, PCI_FUNC_MAX - 1);
+         ++devfn) {
+        PCIDevice *affected_dev = shpc->sec_bus->devices[devfn];
+        if (affected_dev) {
+            qdev_free(&affected_dev->qdev);
+        }
+    }
+}
+
+static void shpc_slot_command(SHPCDevice *shpc, uint8_t target,
+                              uint8_t state, uint8_t power, uint8_t attn)
+{
+    uint8_t current_state;
+    int slot = SHPC_LOGICAL_TO_IDX(target);
+    if (target < SHPC_CMD_TRGT_MIN || slot >= shpc->nslots) {
+        shpc_invalid_command(shpc);
+        return;
+    }
+    current_state = shpc_get_status(shpc, slot, SHPC_SLOT_STATE_MASK);
+    if (current_state == SHPC_STATE_ENABLED && state == SHPC_STATE_PWRONLY) {
+        shpc_invalid_command(shpc);
+	return;
+    }
+
+    switch (power) {
+    case SHPC_LED_NO:
+        break;
+    default:
+        /* TODO: send event to monitor */
+        shpc_set_status(shpc, slot, power, SHPC_SLOT_PWR_LED_MASK);
+    }
+    switch (attn) {
+    case SHPC_LED_NO:
+        break;
+    default:
+        /* TODO: send event to monitor */
+        shpc_set_status(shpc, slot, attn, SHPC_SLOT_ATTN_LED_MASK);
+    }
+
+    if ((current_state == SHPC_STATE_DISABLED && state == SHPC_STATE_PWRONLY) ||
+        (current_state == SHPC_STATE_DISABLED && state == SHPC_STATE_ENABLED)) {
+        shpc_set_status(shpc, slot, state, SHPC_SLOT_STATE_MASK);
+    } else if ((current_state == SHPC_STATE_ENABLED ||
+		current_state == SHPC_STATE_PWRONLY) &&
+	       state == SHPC_STATE_DISABLED) {
+        shpc_set_status(shpc, slot, state, SHPC_SLOT_STATE_MASK);
+        power = shpc_get_status(shpc, slot, SHPC_SLOT_PWR_LED_MASK);
+        /* TODO: track what monitor requested. */
+        /* Look at LED to figure out whether it's ok to remove the device. */
+        if (power == SHPC_LED_OFF) {
+            shpc_free_devices_in_slot(shpc, slot);
+            shpc_set_status(shpc, slot, 1, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_EMPTY,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_BUTTON |
+                SHPC_SLOT_EVENT_MRL |
+                SHPC_SLOT_EVENT_PRESENCE;
+        }
+    }
+}
+
+static void shpc_command(SHPCDevice *shpc)
+{
+    uint8_t code = pci_get_byte(shpc->config + SHPC_CMD_CODE);
+    uint8_t speed;
+    uint8_t target;
+    uint8_t attn;
+    uint8_t power;
+    uint8_t state;
+    int i;
+
+    /* Clear status from the previous command. */
+    pci_word_test_and_clear_mask(shpc->config + SHPC_CMD_STATUS,
+				 SHPC_CMD_STATUS_BUSY |
+				 SHPC_CMD_STATUS_MRL_OPEN |
+				 SHPC_CMD_STATUS_INVALID_CMD |
+				 SHPC_CMD_STATUS_INVALID_MODE);
+    switch (code) {
+    case 0x00 ... 0x3f:
+        target = shpc->config[SHPC_CMD_TRGT] & SHPC_CMD_TRGT_MAX;
+        state = (code & SHPC_SLOT_STATE_MASK) >> SHPC_SLOT_STATE_SHIFT;
+        power = (code & SHPC_SLOT_PWR_LED_MASK) >> SHPC_SLOT_PWR_LED_SHIFT;
+        attn = (code & SHPC_SLOT_ATTN_LED_MASK) >> SHPC_SLOT_ATTN_LED_SHIFT;
+        shpc_slot_command(shpc, target, state, power, attn);
+        break;
+    case 0x40 ... 0x47:
+        speed = code & SHPC_SEC_BUS_MASK;
+        shpc_set_sec_bus_speed(shpc, speed);
+        break;
+    case 0x48:
+        /* Power only all slots */
+        /* first verify no slots are enabled */
+        for (i = 0; i < shpc->nslots; ++i) {
+            state = shpc_get_status(shpc, i, SHPC_SLOT_STATE_MASK);
+            if (state == SHPC_STATE_ENABLED) {
+                shpc_invalid_command(shpc);
+                goto done;
+            }
+        }
+        for (i = 0; i < shpc->nslots; ++i) {
+            if (!(shpc_get_status(shpc, i, SHPC_SLOT_STATUS_MRL_OPEN))) {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_PWRONLY, SHPC_LED_ON, SHPC_LED_NO);
+            } else {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_NO, SHPC_LED_OFF, SHPC_LED_NO);
+            }
+        }
+        break;
+    case 0x49:
+        /* Enable all slots */
+        /* TODO: Spec says this shall fail if some are already enabled.
+         * This doesn't make sense - why not? a spec bug? */
+        for (i = 0; i < shpc->nslots; ++i) {
+            state = shpc_get_status(shpc, i, SHPC_SLOT_STATE_MASK);
+            if (state == SHPC_STATE_ENABLED) {
+                shpc_invalid_command(shpc);
+                goto done;
+            }
+        }
+        for (i = 0; i < shpc->nslots; ++i) {
+            if (!(shpc_get_status(shpc, i, SHPC_SLOT_STATUS_MRL_OPEN))) {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_ENABLED, SHPC_LED_ON, SHPC_LED_NO);
+            } else {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_NO, SHPC_LED_OFF, SHPC_LED_NO);
+            }
+        }
+        break;
+    default:
+        shpc_invalid_command(shpc);
+        break;
+    }
+done:
+    pci_long_test_and_set_mask(shpc->config + SHPC_SERR_INT, SHPC_CMD_DETECTED);
+}
+
+static void shpc_write(PCIDevice *d, unsigned addr, uint64_t val, int l)
+{
+    SHPCDevice *shpc = d->shpc;
+    int i;
+    if (addr >= SHPC_SIZEOF(d)) {
+        return;
+    }
+    l = MIN(l, SHPC_SIZEOF(d) - addr);
+
+    /* TODO: code duplicated from pci.c */
+    for (i = 0; i < l; val >>= 8, ++i) {
+        unsigned a = addr + i;
+        uint8_t wmask = shpc->wmask[a];
+        uint8_t w1cmask = shpc->w1cmask[a];
+        assert(!(wmask & w1cmask));
+        shpc->config[a] = (shpc->config[a] & ~wmask) | (val & wmask);
+        shpc->config[a] &= ~(val & w1cmask); /* W1C: Write 1 to Clear */
+    }
+    if (ranges_overlap(addr, l, SHPC_CMD_CODE, 2)) {
+        shpc_command(shpc);
+    }
+    shpc_interrupt_update(d);
+}
+
+static uint64_t shpc_read(PCIDevice *d, unsigned addr, int l)
+{
+    uint64_t val = 0x0;
+    if (addr >= SHPC_SIZEOF(d)) {
+        return val;
+    }
+    l = MIN(l, SHPC_SIZEOF(d) - addr);
+    memcpy(&val, d->shpc->config + addr, l);
+    return val;
+}
+
+/* SHPC Bridge Capability */
+#define SHPC_CAP_LENGTH 0x08
+#define SHPC_CAP_DWORD_SELECT 0x2 /* 1 byte */
+#define SHPC_CAP_CxP 0x3 /* 1 byte: CSP, CIP */
+#define SHPC_CAP_DWORD_DATA 0x4 /* 4 bytes */
+#define SHPC_CAP_CSP_MASK 0x4
+#define SHPC_CAP_CIP_MASK 0x8
+
+static uint8_t shpc_cap_dword(PCIDevice *d)
+{
+    return pci_get_byte(d->config + d->shpc->cap + SHPC_CAP_DWORD_SELECT);
+}
+
+/* Update dword data capability register */
+static void shpc_cap_update_dword(PCIDevice *d)
+{
+    unsigned data;
+    data = shpc_read(d, shpc_cap_dword(d) * 4, 4);
+    pci_set_long(d->config  + d->shpc->cap + SHPC_CAP_DWORD_DATA, data);
+}
+
+/* Add SHPC capability to the config space for the device. */
+static int shpc_cap_add_config(PCIDevice *d)
+{
+    uint8_t *config;
+    int config_offset;
+    config_offset = pci_add_capability(d, PCI_CAP_ID_SHPC,
+                                       0, SHPC_CAP_LENGTH);
+    if (config_offset < 0) {
+        return config_offset;
+    }
+    config = d->config + config_offset;
+
+    pci_set_byte(config + SHPC_CAP_DWORD_SELECT, 0);
+    pci_set_byte(config + SHPC_CAP_CxP, 0);
+    pci_set_long(config + SHPC_CAP_DWORD_DATA, 0);
+    d->shpc->cap = config_offset;
+    /* Make dword select and data writeable. */
+    pci_set_byte(d->wmask + config_offset + SHPC_CAP_DWORD_SELECT, 0xff);
+    pci_set_long(d->wmask + config_offset + SHPC_CAP_DWORD_DATA, 0xffffffff);
+    return 0;
+}
+
+static uint64_t shpc_mmio_read(void *opaque, target_phys_addr_t addr,
+                               unsigned size)
+{
+    return shpc_read(opaque, addr, size);
+}
+
+static void shpc_mmio_write(void *opaque, target_phys_addr_t addr,
+                            uint64_t val, unsigned size)
+{
+    shpc_write(opaque, addr, val, size);
+}
+
+static const MemoryRegionOps shpc_mmio_ops = {
+    .read = shpc_mmio_read,
+    .write = shpc_mmio_write,
+    .endianness = DEVICE_LITTLE_ENDIAN,
+    .valid = {
+        /* SHPC ECN requires dword accesses, but the original 1.0 spec doesn't.
+         * It's easier to suppport all sizes than worry about it. */
+        .min_access_size = 1,
+        .max_access_size = 4,
+    },
+};
+
+static int shpc_device_hotplug(DeviceState *qdev, PCIDevice *affected_dev,
+                               PCIHotplugState hotplug_state)
+{
+    int pci_slot = PCI_SLOT(affected_dev->devfn);
+    uint8_t state;
+    uint8_t led;
+    PCIDevice *d = DO_UPCAST(PCIDevice, qdev, qdev);
+    SHPCDevice *shpc = d->shpc;
+    int slot = SHPC_PCI_TO_IDX(pci_slot);
+    if (pci_slot < SHPC_IDX_TO_PCI(0) || slot >= shpc->nslots) {
+        error_report("Unsupported PCI slot %d for standard hotplug "
+                     "controller. Valid slots are between %d and %d.",
+                     pci_slot, SHPC_IDX_TO_PCI(0),
+                     SHPC_IDX_TO_PCI(shpc->nslots) - 1);
+        return -1;
+    }
+    /* Don't send event when device is enabled during qemu machine creation:
+     * it is present on boot, no hotplug event is necessary. We do send an
+     * event when the device is disabled later. */
+    if (hotplug_state == PCI_COLDPLUG_ENABLED) {
+        shpc_set_status(shpc, slot, 0, SHPC_SLOT_STATUS_MRL_OPEN);
+        shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_7_5W,
+                        SHPC_SLOT_STATUS_PRSNT_MASK);
+        return 0;
+    }
+    if (hotplug_state == PCI_HOTPLUG_DISABLED) {
+        shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |= SHPC_SLOT_EVENT_BUTTON;
+        state = shpc_get_status(shpc, slot, SHPC_SLOT_STATE_MASK);
+        led = shpc_get_status(shpc, slot, SHPC_SLOT_PWR_LED_MASK);
+        if (state == SHPC_STATE_DISABLED && led == SHPC_LED_OFF) {
+            shpc_free_devices_in_slot(shpc, slot);
+            shpc_set_status(shpc, slot, 1, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_EMPTY,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_MRL |
+                SHPC_SLOT_EVENT_PRESENCE;
+        }
+    } else {
+        /* This could be a cancellation of the previous removal.
+         * We check MRL state to figure out. */
+        if (shpc_get_status(shpc, slot, SHPC_SLOT_STATUS_MRL_OPEN)) {
+            shpc_set_status(shpc, slot, 0, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_7_5W,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_BUTTON |
+                SHPC_SLOT_EVENT_MRL |
+                SHPC_SLOT_EVENT_PRESENCE;
+        } else {
+            /* Press attention button to cancel removal */
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_BUTTON;
+        }
+    }
+    shpc_set_status(shpc, slot, 0, SHPC_SLOT_STATUS_66);
+    shpc_interrupt_update(d);
+    return 0;
+}
+
+/* Initialize the SHPC structure in bridge's BAR. */
+int shpc_init(PCIDevice *d, PCIBus *sec_bus, MemoryRegion *bar, unsigned offset)
+{
+    int i, ret;
+    int nslots = SHPC_MAX_SLOTS; /* TODO: qdev property? */
+    SHPCDevice *shpc = d->shpc = g_malloc0(sizeof(*d->shpc));
+    shpc->sec_bus = sec_bus;
+    ret = shpc_cap_add_config(d);
+    if (ret) {
+        g_free(d->shpc);
+        return ret;
+    }
+    if (nslots < SHPC_MIN_SLOTS) {
+        return 0;
+    }
+    if (nslots > SHPC_MAX_SLOTS ||
+        SHPC_IDX_TO_PCI(nslots) > PCI_SLOT_MAX) {
+        /* TODO: report an error mesage that makes sense. */
+        return -EINVAL;
+    }
+    shpc->nslots = nslots;
+    shpc->config = g_malloc0(SHPC_SIZEOF(d));
+    shpc->cmask = g_malloc0(SHPC_SIZEOF(d));
+    shpc->wmask = g_malloc0(SHPC_SIZEOF(d));
+    shpc->w1cmask = g_malloc0(SHPC_SIZEOF(d));
+
+    shpc_reset(d);
+
+    pci_set_long(shpc->config + SHPC_BASE_OFFSET, offset);
+
+    pci_set_byte(shpc->wmask + SHPC_CMD_CODE, 0xff);
+    pci_set_byte(shpc->wmask + SHPC_CMD_TRGT, SHPC_CMD_TRGT_MAX);
+    pci_set_byte(shpc->wmask + SHPC_CMD_TRGT, SHPC_CMD_TRGT_MAX);
+    pci_set_long(shpc->wmask + SHPC_SERR_INT,
+                 SHPC_INT_DIS |
+                 SHPC_SERR_DIS |
+                 SHPC_CMD_INT_DIS |
+                 SHPC_ARB_SERR_DIS);
+    pci_set_long(shpc->w1cmask + SHPC_SERR_INT,
+		 SHPC_CMD_DETECTED |
+                 SHPC_ARB_DETECTED);
+    for (i = 0; i < nslots; ++i) {
+	    pci_set_byte(shpc->wmask +
+			 SHPC_SLOT_EVENT_SERR_INT_DIS(d, i),
+			 SHPC_SLOT_EVENT_PRESENCE |
+			 SHPC_SLOT_EVENT_ISOLATED_FAULT |
+			 SHPC_SLOT_EVENT_BUTTON |
+			 SHPC_SLOT_EVENT_MRL |
+			 SHPC_SLOT_EVENT_CONNECTED_FAULT |
+			 SHPC_SLOT_EVENT_MRL_SERR_DIS |
+			 SHPC_SLOT_EVENT_CONNECTED_FAULT_SERR_DIS);
+	    pci_set_byte(shpc->w1cmask +
+			 SHPC_SLOT_EVENT_LATCH(i),
+			 SHPC_SLOT_EVENT_PRESENCE |
+			 SHPC_SLOT_EVENT_ISOLATED_FAULT |
+			 SHPC_SLOT_EVENT_BUTTON |
+			 SHPC_SLOT_EVENT_MRL |
+			 SHPC_SLOT_EVENT_CONNECTED_FAULT);
+    }
+
+    /* TODO: init cmask */
+    memory_region_init_io(&shpc->mmio, &shpc_mmio_ops, d, "shpc-mmio",
+                          SHPC_SIZEOF(d));
+    shpc_cap_update_dword(d);
+    memory_region_add_subregion(bar, offset, &shpc->mmio);
+    pci_bus_hotplug(sec_bus, shpc_device_hotplug, &d->qdev);
+    return 0;
+}
+
+int shpc_bar_size(PCIDevice *d)
+{
+    return roundup_pow_of_two(SHPC_SLOT_REG(SHPC_MAX_SLOTS));
+}
+
+void shpc_cleanup(PCIDevice *d)
+{
+    SHPCDevice *shpc = d->shpc;
+    /* TODO: cleanup config space changes? */
+    g_free(shpc->config);
+    g_free(shpc->cmask);
+    g_free(shpc->wmask);
+    g_free(shpc->w1cmask);
+    memory_region_destroy(&shpc->mmio);
+    g_free(shpc);
+}
+
+void shpc_cap_write_config(PCIDevice *d, uint32_t addr, uint32_t val, int l)
+{
+    if (!ranges_overlap(addr, l, d->shpc->cap, SHPC_CAP_LENGTH)) {
+        return;
+    }
+    fprintf(stderr, "%s: 0x%x 0x%x %d\n", __func__, addr, val, l);
+    if (ranges_overlap(addr, l, d->shpc->cap + SHPC_CAP_DWORD_DATA, 4)) {
+        unsigned dword_data;
+        dword_data = pci_get_long(d->shpc->config + d->shpc->cap
+                                  + SHPC_CAP_DWORD_DATA);
+        shpc_write(d, shpc_cap_dword(d) * 4, dword_data, 4);
+    }
+    /* Update cap dword data in case guest is going to read it. */
+    shpc_cap_update_dword(d);
+}
hw/shpc.h new file +40 -0
new file mode 100644
index 0000000..389b178
@@ -0,0 +1,40 @@
+#ifndef SHPC_H
+#define SHPC_H
+
+#include "qemu-common.h"
+#include "memory.h"
+
+struct SHPCDevice {
+    /* Capability offset in device's config space */
+    int cap;
+
+    /* # of hot-pluggable slots */
+    int nslots;
+
+    /* SHPC WRS: working register set */
+    uint8_t *config;
+
+    /* Used to enable checks on load. Note that writable bits are
+     * never checked even if set in cmask. */
+    uint8_t *cmask;
+
+    /* Used to implement R/W bytes */
+    uint8_t *wmask;
+
+    /* Used to implement RW1C(Write 1 to Clear) bytes */
+    uint8_t *w1cmask;
+
+    /* MMIO for the SHPC BAR */
+    MemoryRegion mmio;
+
+    /* Bus controlled by this SHPC */
+    PCIBus *sec_bus;
+};
+
+void shpc_reset(PCIDevice *d);
+int shpc_bar_size(PCIDevice *dev);
+int shpc_init(PCIDevice *dev, PCIBus *sec_bus, MemoryRegion *bar, unsigned offset);
+void shpc_cleanup(PCIDevice *dev);
+void shpc_cap_write_config(PCIDevice *d, uint32_t addr, uint32_t val, int len);
+
+#endif
qemu-common.h +1 -0
index 9b997f8..4ff9c95 100644
@@ -247,6 +247,7 @@ typedef struct SSIBus SSIBus;
 typedef struct EventNotifier EventNotifier;
 typedef struct VirtIODevice VirtIODevice;
 typedef struct QEMUSGList QEMUSGList;
+typedef struct SHPCDevice SHPCDevice;
 
 typedef uint64_t pcibus_t;
 
Signature · 1 line
1.7.9.111.gf3fb0
Signature · 1 line
yamahata
#3 lore raw

Re: [PATCHv2-RFC 1/2] shpc: standard hot plug controller

Quoted 7 lines · Isaku Yamahata

Oh nice work.

Quoted 4 lines · Michael S. Tsirkin

This adds support for SHPC interface, as defined by PCI Standard Hot-Plug Controller and Subsystem Specification, Rev 1.0
http://www.pcisig.com/specifications/conventional/pci_hot_plug/SHPC_10

Quoted 7 lines
Quoted 5 lines

Only SHPC intergrated with a PCI-to-PCI bridge is supported, SHPC integrated with a host bridge would need more work.

All main SHPC features are supported:
- MRL sensor

Does this just report latch status? (It seems so.)

What happens is that adding a device closes the latch, removing a device opens the latch. This simplifies the number of supported configurations significantly.

Quoted 1 line

Do you plan to provide interfaces to manipulate the latch?

I didn't plan to do this, and this is non-trivial. Do you just want this for empty slots? And why?

Quoted 14 lines
Quoted 14 lines

- Attention button
- Attention indicator
- Power indicator

Wake on hotplug and serr generation are stubbed out but unused as we don't have interfaces to generate these events ATM.

One issue that isn't completely resolved is that qemu currently expects an "eject" interface, which SHPC does not provide: it merely removes the power to device and it's up to the user to remove the device from slot. This patch works around that by ejecting the device when power is removed and power LED goes off.

Quoted 7 lines
Quoted 3 lines

TODO:
- migration support
- fix dependency on pci_internals.h

If I didn't miss the code,
- QMP command for pushing attention button.
- QMP command to get LED status

It's easy to add these, so I'd accept such a patch, but I wonder why.

Quoted 1 line

- QMP events for LED on/off

There's also blink :)

Quoted 2 lines

thanks,

I'm concerned that a guest can flood the management with such events. It's better to send a single "LED change" event, then we can suppress further events until next "get LED status" command.

Quoted 763 lines
Quoted 760 lines
---
 Makefile.objs |    1 +
 hw/pci.h      |    6 +
 hw/shpc.c     |  646 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 hw/shpc.h     |   40 ++++
 qemu-common.h |    1 +
 5 files changed, 694 insertions(+), 0 deletions(-)
 create mode 100644 hw/shpc.c
 create mode 100644 hw/shpc.h
Makefile.objs +1 -0
index 391e524..4546477 100644
@@ -195,6 +195,7 @@ hw-obj-$(CONFIG_VIRTIO_PCI) += virtio-pci.o
 hw-obj-y += fw_cfg.o
 hw-obj-$(CONFIG_PCI) += pci.o pci_bridge.o
 hw-obj-$(CONFIG_PCI) += msix.o msi.o
+hw-obj-$(CONFIG_PCI) += shpc.o
 hw-obj-$(CONFIG_PCI) += pci_host.o pcie_host.o
 hw-obj-$(CONFIG_PCI) += ioh3420.o xio3130_upstream.o xio3130_downstream.o
 hw-obj-y += watchdog.o
hw/pci.h +6 -0
index 33b0b18..756577e 100644
@@ -125,6 +125,9 @@ enum {
     /* command register SERR bit enabled */
 #define QEMU_PCI_CAP_SERR_BITNR 4
     QEMU_PCI_CAP_SERR = (1 << QEMU_PCI_CAP_SERR_BITNR),
+    /* Standard hot plug controller. */
+#define QEMU_PCI_SHPC_BITNR 5
+    QEMU_PCI_CAP_SHPC = (1 << QEMU_PCI_SHPC_BITNR),
 };
 
 #define TYPE_PCI_DEVICE "pci-device"
@@ -229,6 +232,9 @@ struct PCIDevice {
     /* PCI Express */
     PCIExpressDevice exp;
 
+    /* SHPC */
+    SHPCDevice *shpc;
+
     /* Location of option rom */
     char *romfile;
     bool has_rom;
hw/shpc.c new file +646 -0
new file mode 100644
index 0000000..4baec29
@@ -0,0 +1,646 @@
+#include <strings.h>
+#include <stdint.h>
+#include "range.h"
+#include "shpc.h"
+#include "pci.h"
+#include "pci_internals.h"
+
+/* TODO: model power only and disabled slot states. */
+/* TODO: handle SERR and wakeups */
+/* TODO: consider enabling 66MHz support */
+
+/* TODO: remove fully only on state DISABLED and LED off.
+ * track state to properly record this. */
+
+/* SHPC Working Register Set */
+#define SHPC_BASE_OFFSET  0x00 /* 4 bytes */
+#define SHPC_SLOTS_33     0x04 /* 4 bytes. Also encodes PCI-X slots. */
+#define SHPC_SLOTS_66     0x08 /* 4 bytes. */
+#define SHPC_NSLOTS       0x0C /* 1 byte */
+#define SHPC_FIRST_DEV    0x0D /* 1 byte */
+#define SHPC_PHYS_SLOT    0x0E /* 2 byte */
+#define SHPC_PHYS_NUM_MAX 0x7ff
+#define SHPC_PHYS_NUM_UP  0x1000
+#define SHPC_PHYS_MRL     0x4000
+#define SHPC_PHYS_BUTTON  0x8000
+#define SHPC_SEC_BUS      0x10 /* 2 bytes */
+#define SHPC_SEC_BUS_33   0x0
+#define SHPC_SEC_BUS_66   0x1 /* Unused */
+#define SHPC_SEC_BUS_MASK 0x7
+#define SHPC_MSI_CTL      0x12 /* 1 byte */
+#define SHPC_PROG_IFC     0x13 /* 1 byte */
+#define SHPC_PROG_IFC_1_0 0x1
+#define SHPC_CMD_CODE     0x14 /* 1 byte */
+#define SHPC_CMD_TRGT     0x15 /* 1 byte */
+#define SHPC_CMD_TRGT_MIN 0x1
+#define SHPC_CMD_TRGT_MAX 0x1f
+#define SHPC_CMD_STATUS   0x16 /* 2 bytes */
+#define SHPC_CMD_STATUS_BUSY          0x1
+#define SHPC_CMD_STATUS_MRL_OPEN      0x2
+#define SHPC_CMD_STATUS_INVALID_CMD   0x4
+#define SHPC_CMD_STATUS_INVALID_MODE  0x8
+#define SHPC_INT_LOCATOR  0x18 /* 4 bytes */
+#define SHPC_INT_COMMAND  0x1
+#define SHPC_SERR_LOCATOR 0x1C /* 4 bytes */
+#define SHPC_SERR_INT     0x20 /* 4 bytes */
+#define SHPC_INT_DIS      0x1
+#define SHPC_SERR_DIS     0x2
+#define SHPC_CMD_INT_DIS  0x4
+#define SHPC_ARB_SERR_DIS 0x8
+#define SHPC_CMD_DETECTED 0x10000
+#define SHPC_ARB_DETECTED 0x20000
+ /* 4 bytes * slot # (start from 0) */
+#define SHPC_SLOT_REG(s)         (0x24 + (s) * 4)
+ /* 2 bytes */
+#define SHPC_SLOT_STATUS(s)       (0x0 + SHPC_SLOT_REG(s))
+
+/* Same slot state masks are used for command and status registers */
+#define SHPC_SLOT_STATE_MASK     0x03
+#define SHPC_SLOT_STATE_SHIFT \
+    (ffs(SHPC_SLOT_STATE_MASK) - 1)
+
+#define SHPC_STATE_NO       0x0
+#define SHPC_STATE_PWRONLY  0x1
+#define SHPC_STATE_ENABLED  0x2
+#define SHPC_STATE_DISABLED 0x3
+
+#define SHPC_SLOT_PWR_LED_MASK   0xC
+#define SHPC_SLOT_PWR_LED_SHIFT \
+    (ffs(SHPC_SLOT_PWR_LED_MASK) - 1)
+#define SHPC_SLOT_ATTN_LED_MASK  0x30
+#define SHPC_SLOT_ATTN_LED_SHIFT \
+    (ffs(SHPC_SLOT_ATTN_LED_MASK) - 1)
+
+#define SHPC_LED_NO     0x0
+#define SHPC_LED_ON     0x1
+#define SHPC_LED_BLINK  0x2
+#define SHPC_LED_OFF    0x3
+
+#define SHPC_SLOT_STATUS_PWR_FAULT      0x40
+#define SHPC_SLOT_STATUS_BUTTON         0x80
+#define SHPC_SLOT_STATUS_MRL_OPEN       0x100
+#define SHPC_SLOT_STATUS_66             0x200
+#define SHPC_SLOT_STATUS_PRSNT_MASK     0xC00
+#define SHPC_SLOT_STATUS_PRSNT_EMPTY    0x3
+#define SHPC_SLOT_STATUS_PRSNT_25W      0x1
+#define SHPC_SLOT_STATUS_PRSNT_15W      0x2
+#define SHPC_SLOT_STATUS_PRSNT_7_5W     0x0
+
+#define SHPC_SLOT_STATUS_PRSNT_PCIX     0x3000
+
+
+ /* 1 byte */
+#define SHPC_SLOT_EVENT_LATCH(s)        (0x2 + SHPC_SLOT_REG(s))
+ /* 1 byte */
+#define SHPC_SLOT_EVENT_SERR_INT_DIS(d, s) (0x3 + SHPC_SLOT_REG(s))
+#define SHPC_SLOT_EVENT_PRESENCE        0x01
+#define SHPC_SLOT_EVENT_ISOLATED_FAULT  0x02
+#define SHPC_SLOT_EVENT_BUTTON          0x04
+#define SHPC_SLOT_EVENT_MRL             0x08
+#define SHPC_SLOT_EVENT_CONNECTED_FAULT 0x10
+/* Bits below are used for Serr/Int disable only */
+#define SHPC_SLOT_EVENT_MRL_SERR_DIS    0x20
+#define SHPC_SLOT_EVENT_CONNECTED_FAULT_SERR_DIS 0x40
+
+#define SHPC_MIN_SLOTS        1
+#define SHPC_MAX_SLOTS        31
+#define SHPC_SIZEOF(d)    SHPC_SLOT_REG((d)->shpc->nslots)
+
+/* SHPC Slot identifiers */
+
+/* Hotplug supported at 31 slots out of the total 32.  We reserve slot 0,
+   and give the rest of them physical *and* pci numbers starting from 1, so
+   they match logical numbers.  Note: this means that multiple slots must have
+   different chassis number values, to make chassis+physical slot unique.
+   TODO: make this configurable? */
+#define SHPC_IDX_TO_LOGICAL(slot) ((slot) + 1)
+#define SHPC_LOGICAL_TO_IDX(target) ((target) - 1)
+#define SHPC_IDX_TO_PCI(slot) ((slot) + 1)
+#define SHPC_PCI_TO_IDX(pci_slot) ((pci_slot) - 1)
+#define SHPC_IDX_TO_PHYSICAL(slot) ((slot) + 1)
+
+static int roundup_pow_of_two(int x)
+{
+    x |= (x >> 1);
+    x |= (x >> 2);
+    x |= (x >> 4);
+    x |= (x >> 8);
+    x |= (x >> 16);
+    return x + 1;
+}
+
+static uint16_t shpc_get_status(SHPCDevice *shpc, int slot, uint16_t msk)
+{
+    uint8_t *status = shpc->config + SHPC_SLOT_STATUS(slot);
+    return (pci_get_word(status) & msk) >> (ffs(msk) - 1);
+}
+
+static void shpc_set_status(SHPCDevice *shpc,
+                            int slot, uint8_t value, uint16_t msk)
+{
+    uint8_t *status = shpc->config + SHPC_SLOT_STATUS(slot);
+    pci_word_test_and_clear_mask(status, msk);
+    pci_word_test_and_set_mask(status, value << (ffs(msk) - 1));
+}
+
+static void shpc_interrupt_update(PCIDevice *d)
+{
+    SHPCDevice *shpc = d->shpc;
+    int slot;
+    int level = 0;
+    uint32_t serr_int;
+    uint32_t int_locator = 0;
+
+    /* Update interrupt locator register */
+    for (slot = 0; slot < shpc->nslots; ++slot) {
+        uint8_t event = shpc->config[SHPC_SLOT_EVENT_LATCH(slot)];
+        uint8_t disable = shpc->config[SHPC_SLOT_EVENT_SERR_INT_DIS(d, slot)];
+        uint32_t mask = 1 << SHPC_IDX_TO_LOGICAL(slot);
+        if (event & ~disable) {
+            int_locator |= mask;
+        }
+    }
+    serr_int = pci_get_long(shpc->config + SHPC_SERR_INT);
+    if ((serr_int & SHPC_CMD_DETECTED) && !(serr_int & SHPC_CMD_INT_DIS)) {
+        int_locator |= SHPC_INT_COMMAND;
+    }
+    pci_set_long(shpc->config + SHPC_INT_LOCATOR, int_locator);
+    level = (!(serr_int & SHPC_INT_DIS) && int_locator) ? 1 : 0;
+    qemu_set_irq(d->irq[0], level);
+}
+
+static void shpc_set_sec_bus_speed(SHPCDevice *shpc, uint8_t speed)
+{
+    switch (speed) {
+    case SHPC_SEC_BUS_33:
+        shpc->config[SHPC_SEC_BUS] &= ~SHPC_SEC_BUS_MASK;
+        shpc->config[SHPC_SEC_BUS] |= speed;
+        break;
+    default:
+	pci_word_test_and_set_mask(shpc->config + SHPC_CMD_STATUS,
+				   SHPC_CMD_STATUS_INVALID_MODE);
+    }
+}
+
+void shpc_reset(PCIDevice *d)
+{
+    SHPCDevice *shpc = d->shpc;
+    int nslots = shpc->nslots;
+    int i;
+    memset(shpc->config, 0, SHPC_SIZEOF(d));
+    pci_set_byte(shpc->config + SHPC_NSLOTS, nslots);
+    pci_set_long(shpc->config + SHPC_SLOTS_33, nslots);
+    pci_set_long(shpc->config + SHPC_SLOTS_66, 0);
+    pci_set_byte(shpc->config + SHPC_FIRST_DEV, SHPC_IDX_TO_PCI(0));
+    pci_set_word(shpc->config + SHPC_PHYS_SLOT,
+		 SHPC_IDX_TO_PHYSICAL(0) |
+		 SHPC_PHYS_NUM_UP |
+		 SHPC_PHYS_MRL |
+		 SHPC_PHYS_BUTTON);
+    pci_set_long(shpc->config + SHPC_SERR_INT, SHPC_INT_DIS |
+                 SHPC_SERR_DIS |
+                 SHPC_CMD_INT_DIS |
+                 SHPC_ARB_SERR_DIS);
+    pci_set_byte(shpc->config + SHPC_PROG_IFC, SHPC_PROG_IFC_1_0);
+    pci_set_word(shpc->config + SHPC_SEC_BUS, SHPC_SEC_BUS_33);
+    for (i = 0; i < shpc->nslots; ++i) {
+        pci_set_word(shpc->config + SHPC_SLOT_EVENT_SERR_INT_DIS(d, i),
+                     SHPC_SLOT_EVENT_PRESENCE |
+                     SHPC_SLOT_EVENT_ISOLATED_FAULT |
+                     SHPC_SLOT_EVENT_BUTTON |
+                     SHPC_SLOT_EVENT_MRL |
+                     SHPC_SLOT_EVENT_CONNECTED_FAULT |
+                     SHPC_SLOT_EVENT_MRL_SERR_DIS |
+                     SHPC_SLOT_EVENT_CONNECTED_FAULT_SERR_DIS);
+        if (shpc->sec_bus->devices[PCI_DEVFN(SHPC_IDX_TO_PCI(i), 0)]) {
+            shpc_set_status(shpc, i, SHPC_STATE_ENABLED, SHPC_SLOT_STATE_MASK);
+            shpc_set_status(shpc, i, 0, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, i, SHPC_SLOT_STATUS_PRSNT_7_5W,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc_set_status(shpc, i, SHPC_LED_ON, SHPC_SLOT_PWR_LED_MASK);
+        } else {
+            shpc_set_status(shpc, i, SHPC_STATE_DISABLED, SHPC_SLOT_STATE_MASK);
+            shpc_set_status(shpc, i, 1, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, i, SHPC_SLOT_STATUS_PRSNT_EMPTY,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc_set_status(shpc, i, SHPC_LED_OFF, SHPC_SLOT_PWR_LED_MASK);
+	}
+        shpc_set_status(shpc, i, 0, SHPC_SLOT_STATUS_66);
+    }
+    shpc_set_sec_bus_speed(shpc, SHPC_SEC_BUS_33);
+    shpc_interrupt_update(d);
+}
+
+static void shpc_invalid_command(SHPCDevice *shpc)
+{
+    pci_word_test_and_set_mask(shpc->config + SHPC_CMD_STATUS,
+                               SHPC_CMD_STATUS_INVALID_CMD);
+}
+
+static void shpc_free_devices_in_slot(SHPCDevice *shpc, int slot)
+{
+    int devfn;
+    int pci_slot = SHPC_IDX_TO_PCI(slot);
+    for (devfn = PCI_DEVFN(pci_slot, 0);
+         devfn <= PCI_DEVFN(pci_slot, PCI_FUNC_MAX - 1);
+         ++devfn) {
+        PCIDevice *affected_dev = shpc->sec_bus->devices[devfn];
+        if (affected_dev) {
+            qdev_free(&affected_dev->qdev);
+        }
+    }
+}
+
+static void shpc_slot_command(SHPCDevice *shpc, uint8_t target,
+                              uint8_t state, uint8_t power, uint8_t attn)
+{
+    uint8_t current_state;
+    int slot = SHPC_LOGICAL_TO_IDX(target);
+    if (target < SHPC_CMD_TRGT_MIN || slot >= shpc->nslots) {
+        shpc_invalid_command(shpc);
+        return;
+    }
+    current_state = shpc_get_status(shpc, slot, SHPC_SLOT_STATE_MASK);
+    if (current_state == SHPC_STATE_ENABLED && state == SHPC_STATE_PWRONLY) {
+        shpc_invalid_command(shpc);
+	return;
+    }
+
+    switch (power) {
+    case SHPC_LED_NO:
+        break;
+    default:
+        /* TODO: send event to monitor */
+        shpc_set_status(shpc, slot, power, SHPC_SLOT_PWR_LED_MASK);
+    }
+    switch (attn) {
+    case SHPC_LED_NO:
+        break;
+    default:
+        /* TODO: send event to monitor */
+        shpc_set_status(shpc, slot, attn, SHPC_SLOT_ATTN_LED_MASK);
+    }
+
+    if ((current_state == SHPC_STATE_DISABLED && state == SHPC_STATE_PWRONLY) ||
+        (current_state == SHPC_STATE_DISABLED && state == SHPC_STATE_ENABLED)) {
+        shpc_set_status(shpc, slot, state, SHPC_SLOT_STATE_MASK);
+    } else if ((current_state == SHPC_STATE_ENABLED ||
+		current_state == SHPC_STATE_PWRONLY) &&
+	       state == SHPC_STATE_DISABLED) {
+        shpc_set_status(shpc, slot, state, SHPC_SLOT_STATE_MASK);
+        power = shpc_get_status(shpc, slot, SHPC_SLOT_PWR_LED_MASK);
+        /* TODO: track what monitor requested. */
+        /* Look at LED to figure out whether it's ok to remove the device. */
+        if (power == SHPC_LED_OFF) {
+            shpc_free_devices_in_slot(shpc, slot);
+            shpc_set_status(shpc, slot, 1, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_EMPTY,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_BUTTON |
+                SHPC_SLOT_EVENT_MRL |
+                SHPC_SLOT_EVENT_PRESENCE;
+        }
+    }
+}
+
+static void shpc_command(SHPCDevice *shpc)
+{
+    uint8_t code = pci_get_byte(shpc->config + SHPC_CMD_CODE);
+    uint8_t speed;
+    uint8_t target;
+    uint8_t attn;
+    uint8_t power;
+    uint8_t state;
+    int i;
+
+    /* Clear status from the previous command. */
+    pci_word_test_and_clear_mask(shpc->config + SHPC_CMD_STATUS,
+				 SHPC_CMD_STATUS_BUSY |
+				 SHPC_CMD_STATUS_MRL_OPEN |
+				 SHPC_CMD_STATUS_INVALID_CMD |
+				 SHPC_CMD_STATUS_INVALID_MODE);
+    switch (code) {
+    case 0x00 ... 0x3f:
+        target = shpc->config[SHPC_CMD_TRGT] & SHPC_CMD_TRGT_MAX;
+        state = (code & SHPC_SLOT_STATE_MASK) >> SHPC_SLOT_STATE_SHIFT;
+        power = (code & SHPC_SLOT_PWR_LED_MASK) >> SHPC_SLOT_PWR_LED_SHIFT;
+        attn = (code & SHPC_SLOT_ATTN_LED_MASK) >> SHPC_SLOT_ATTN_LED_SHIFT;
+        shpc_slot_command(shpc, target, state, power, attn);
+        break;
+    case 0x40 ... 0x47:
+        speed = code & SHPC_SEC_BUS_MASK;
+        shpc_set_sec_bus_speed(shpc, speed);
+        break;
+    case 0x48:
+        /* Power only all slots */
+        /* first verify no slots are enabled */
+        for (i = 0; i < shpc->nslots; ++i) {
+            state = shpc_get_status(shpc, i, SHPC_SLOT_STATE_MASK);
+            if (state == SHPC_STATE_ENABLED) {
+                shpc_invalid_command(shpc);
+                goto done;
+            }
+        }
+        for (i = 0; i < shpc->nslots; ++i) {
+            if (!(shpc_get_status(shpc, i, SHPC_SLOT_STATUS_MRL_OPEN))) {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_PWRONLY, SHPC_LED_ON, SHPC_LED_NO);
+            } else {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_NO, SHPC_LED_OFF, SHPC_LED_NO);
+            }
+        }
+        break;
+    case 0x49:
+        /* Enable all slots */
+        /* TODO: Spec says this shall fail if some are already enabled.
+         * This doesn't make sense - why not? a spec bug? */
+        for (i = 0; i < shpc->nslots; ++i) {
+            state = shpc_get_status(shpc, i, SHPC_SLOT_STATE_MASK);
+            if (state == SHPC_STATE_ENABLED) {
+                shpc_invalid_command(shpc);
+                goto done;
+            }
+        }
+        for (i = 0; i < shpc->nslots; ++i) {
+            if (!(shpc_get_status(shpc, i, SHPC_SLOT_STATUS_MRL_OPEN))) {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_ENABLED, SHPC_LED_ON, SHPC_LED_NO);
+            } else {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_NO, SHPC_LED_OFF, SHPC_LED_NO);
+            }
+        }
+        break;
+    default:
+        shpc_invalid_command(shpc);
+        break;
+    }
+done:
+    pci_long_test_and_set_mask(shpc->config + SHPC_SERR_INT, SHPC_CMD_DETECTED);
+}
+
+static void shpc_write(PCIDevice *d, unsigned addr, uint64_t val, int l)
+{
+    SHPCDevice *shpc = d->shpc;
+    int i;
+    if (addr >= SHPC_SIZEOF(d)) {
+        return;
+    }
+    l = MIN(l, SHPC_SIZEOF(d) - addr);
+
+    /* TODO: code duplicated from pci.c */
+    for (i = 0; i < l; val >>= 8, ++i) {
+        unsigned a = addr + i;
+        uint8_t wmask = shpc->wmask[a];
+        uint8_t w1cmask = shpc->w1cmask[a];
+        assert(!(wmask & w1cmask));
+        shpc->config[a] = (shpc->config[a] & ~wmask) | (val & wmask);
+        shpc->config[a] &= ~(val & w1cmask); /* W1C: Write 1 to Clear */
+    }
+    if (ranges_overlap(addr, l, SHPC_CMD_CODE, 2)) {
+        shpc_command(shpc);
+    }
+    shpc_interrupt_update(d);
+}
+
+static uint64_t shpc_read(PCIDevice *d, unsigned addr, int l)
+{
+    uint64_t val = 0x0;
+    if (addr >= SHPC_SIZEOF(d)) {
+        return val;
+    }
+    l = MIN(l, SHPC_SIZEOF(d) - addr);
+    memcpy(&val, d->shpc->config + addr, l);
+    return val;
+}
+
+/* SHPC Bridge Capability */
+#define SHPC_CAP_LENGTH 0x08
+#define SHPC_CAP_DWORD_SELECT 0x2 /* 1 byte */
+#define SHPC_CAP_CxP 0x3 /* 1 byte: CSP, CIP */
+#define SHPC_CAP_DWORD_DATA 0x4 /* 4 bytes */
+#define SHPC_CAP_CSP_MASK 0x4
+#define SHPC_CAP_CIP_MASK 0x8
+
+static uint8_t shpc_cap_dword(PCIDevice *d)
+{
+    return pci_get_byte(d->config + d->shpc->cap + SHPC_CAP_DWORD_SELECT);
+}
+
+/* Update dword data capability register */
+static void shpc_cap_update_dword(PCIDevice *d)
+{
+    unsigned data;
+    data = shpc_read(d, shpc_cap_dword(d) * 4, 4);
+    pci_set_long(d->config  + d->shpc->cap + SHPC_CAP_DWORD_DATA, data);
+}
+
+/* Add SHPC capability to the config space for the device. */
+static int shpc_cap_add_config(PCIDevice *d)
+{
+    uint8_t *config;
+    int config_offset;
+    config_offset = pci_add_capability(d, PCI_CAP_ID_SHPC,
+                                       0, SHPC_CAP_LENGTH);
+    if (config_offset < 0) {
+        return config_offset;
+    }
+    config = d->config + config_offset;
+
+    pci_set_byte(config + SHPC_CAP_DWORD_SELECT, 0);
+    pci_set_byte(config + SHPC_CAP_CxP, 0);
+    pci_set_long(config + SHPC_CAP_DWORD_DATA, 0);
+    d->shpc->cap = config_offset;
+    /* Make dword select and data writeable. */
+    pci_set_byte(d->wmask + config_offset + SHPC_CAP_DWORD_SELECT, 0xff);
+    pci_set_long(d->wmask + config_offset + SHPC_CAP_DWORD_DATA, 0xffffffff);
+    return 0;
+}
+
+static uint64_t shpc_mmio_read(void *opaque, target_phys_addr_t addr,
+                               unsigned size)
+{
+    return shpc_read(opaque, addr, size);
+}
+
+static void shpc_mmio_write(void *opaque, target_phys_addr_t addr,
+                            uint64_t val, unsigned size)
+{
+    shpc_write(opaque, addr, val, size);
+}
+
+static const MemoryRegionOps shpc_mmio_ops = {
+    .read = shpc_mmio_read,
+    .write = shpc_mmio_write,
+    .endianness = DEVICE_LITTLE_ENDIAN,
+    .valid = {
+        /* SHPC ECN requires dword accesses, but the original 1.0 spec doesn't.
+         * It's easier to suppport all sizes than worry about it. */
+        .min_access_size = 1,
+        .max_access_size = 4,
+    },
+};
+
+static int shpc_device_hotplug(DeviceState *qdev, PCIDevice *affected_dev,
+                               PCIHotplugState hotplug_state)
+{
+    int pci_slot = PCI_SLOT(affected_dev->devfn);
+    uint8_t state;
+    uint8_t led;
+    PCIDevice *d = DO_UPCAST(PCIDevice, qdev, qdev);
+    SHPCDevice *shpc = d->shpc;
+    int slot = SHPC_PCI_TO_IDX(pci_slot);
+    if (pci_slot < SHPC_IDX_TO_PCI(0) || slot >= shpc->nslots) {
+        error_report("Unsupported PCI slot %d for standard hotplug "
+                     "controller. Valid slots are between %d and %d.",
+                     pci_slot, SHPC_IDX_TO_PCI(0),
+                     SHPC_IDX_TO_PCI(shpc->nslots) - 1);
+        return -1;
+    }
+    /* Don't send event when device is enabled during qemu machine creation:
+     * it is present on boot, no hotplug event is necessary. We do send an
+     * event when the device is disabled later. */
+    if (hotplug_state == PCI_COLDPLUG_ENABLED) {
+        shpc_set_status(shpc, slot, 0, SHPC_SLOT_STATUS_MRL_OPEN);
+        shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_7_5W,
+                        SHPC_SLOT_STATUS_PRSNT_MASK);
+        return 0;
+    }
+    if (hotplug_state == PCI_HOTPLUG_DISABLED) {
+        shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |= SHPC_SLOT_EVENT_BUTTON;
+        state = shpc_get_status(shpc, slot, SHPC_SLOT_STATE_MASK);
+        led = shpc_get_status(shpc, slot, SHPC_SLOT_PWR_LED_MASK);
+        if (state == SHPC_STATE_DISABLED && led == SHPC_LED_OFF) {
+            shpc_free_devices_in_slot(shpc, slot);
+            shpc_set_status(shpc, slot, 1, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_EMPTY,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_MRL |
+                SHPC_SLOT_EVENT_PRESENCE;
+        }
+    } else {
+        /* This could be a cancellation of the previous removal.
+         * We check MRL state to figure out. */
+        if (shpc_get_status(shpc, slot, SHPC_SLOT_STATUS_MRL_OPEN)) {
+            shpc_set_status(shpc, slot, 0, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_7_5W,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_BUTTON |
+                SHPC_SLOT_EVENT_MRL |
+                SHPC_SLOT_EVENT_PRESENCE;
+        } else {
+            /* Press attention button to cancel removal */
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_BUTTON;
+        }
+    }
+    shpc_set_status(shpc, slot, 0, SHPC_SLOT_STATUS_66);
+    shpc_interrupt_update(d);
+    return 0;
+}
+
+/* Initialize the SHPC structure in bridge's BAR. */
+int shpc_init(PCIDevice *d, PCIBus *sec_bus, MemoryRegion *bar, unsigned offset)
+{
+    int i, ret;
+    int nslots = SHPC_MAX_SLOTS; /* TODO: qdev property? */
+    SHPCDevice *shpc = d->shpc = g_malloc0(sizeof(*d->shpc));
+    shpc->sec_bus = sec_bus;
+    ret = shpc_cap_add_config(d);
+    if (ret) {
+        g_free(d->shpc);
+        return ret;
+    }
+    if (nslots < SHPC_MIN_SLOTS) {
+        return 0;
+    }
+    if (nslots > SHPC_MAX_SLOTS ||
+        SHPC_IDX_TO_PCI(nslots) > PCI_SLOT_MAX) {
+        /* TODO: report an error mesage that makes sense. */
+        return -EINVAL;
+    }
+    shpc->nslots = nslots;
+    shpc->config = g_malloc0(SHPC_SIZEOF(d));
+    shpc->cmask = g_malloc0(SHPC_SIZEOF(d));
+    shpc->wmask = g_malloc0(SHPC_SIZEOF(d));
+    shpc->w1cmask = g_malloc0(SHPC_SIZEOF(d));
+
+    shpc_reset(d);
+
+    pci_set_long(shpc->config + SHPC_BASE_OFFSET, offset);
+
+    pci_set_byte(shpc->wmask + SHPC_CMD_CODE, 0xff);
+    pci_set_byte(shpc->wmask + SHPC_CMD_TRGT, SHPC_CMD_TRGT_MAX);
+    pci_set_byte(shpc->wmask + SHPC_CMD_TRGT, SHPC_CMD_TRGT_MAX);
+    pci_set_long(shpc->wmask + SHPC_SERR_INT,
+                 SHPC_INT_DIS |
+                 SHPC_SERR_DIS |
+                 SHPC_CMD_INT_DIS |
+                 SHPC_ARB_SERR_DIS);
+    pci_set_long(shpc->w1cmask + SHPC_SERR_INT,
+		 SHPC_CMD_DETECTED |
+                 SHPC_ARB_DETECTED);
+    for (i = 0; i < nslots; ++i) {
+	    pci_set_byte(shpc->wmask +
+			 SHPC_SLOT_EVENT_SERR_INT_DIS(d, i),
+			 SHPC_SLOT_EVENT_PRESENCE |
+			 SHPC_SLOT_EVENT_ISOLATED_FAULT |
+			 SHPC_SLOT_EVENT_BUTTON |
+			 SHPC_SLOT_EVENT_MRL |
+			 SHPC_SLOT_EVENT_CONNECTED_FAULT |
+			 SHPC_SLOT_EVENT_MRL_SERR_DIS |
+			 SHPC_SLOT_EVENT_CONNECTED_FAULT_SERR_DIS);
+	    pci_set_byte(shpc->w1cmask +
+			 SHPC_SLOT_EVENT_LATCH(i),
+			 SHPC_SLOT_EVENT_PRESENCE |
+			 SHPC_SLOT_EVENT_ISOLATED_FAULT |
+			 SHPC_SLOT_EVENT_BUTTON |
+			 SHPC_SLOT_EVENT_MRL |
+			 SHPC_SLOT_EVENT_CONNECTED_FAULT);
+    }
+
+    /* TODO: init cmask */
+    memory_region_init_io(&shpc->mmio, &shpc_mmio_ops, d, "shpc-mmio",
+                          SHPC_SIZEOF(d));
+    shpc_cap_update_dword(d);
+    memory_region_add_subregion(bar, offset, &shpc->mmio);
+    pci_bus_hotplug(sec_bus, shpc_device_hotplug, &d->qdev);
+    return 0;
+}
+
+int shpc_bar_size(PCIDevice *d)
+{
+    return roundup_pow_of_two(SHPC_SLOT_REG(SHPC_MAX_SLOTS));
+}
+
+void shpc_cleanup(PCIDevice *d)
+{
+    SHPCDevice *shpc = d->shpc;
+    /* TODO: cleanup config space changes? */
+    g_free(shpc->config);
+    g_free(shpc->cmask);
+    g_free(shpc->wmask);
+    g_free(shpc->w1cmask);
+    memory_region_destroy(&shpc->mmio);
+    g_free(shpc);
+}
+
+void shpc_cap_write_config(PCIDevice *d, uint32_t addr, uint32_t val, int l)
+{
+    if (!ranges_overlap(addr, l, d->shpc->cap, SHPC_CAP_LENGTH)) {
+        return;
+    }
+    fprintf(stderr, "%s: 0x%x 0x%x %d\n", __func__, addr, val, l);
+    if (ranges_overlap(addr, l, d->shpc->cap + SHPC_CAP_DWORD_DATA, 4)) {
+        unsigned dword_data;
+        dword_data = pci_get_long(d->shpc->config + d->shpc->cap
+                                  + SHPC_CAP_DWORD_DATA);
+        shpc_write(d, shpc_cap_dword(d) * 4, dword_data, 4);
+    }
+    /* Update cap dword data in case guest is going to read it. */
+    shpc_cap_update_dword(d);
+}
hw/shpc.h new file +40 -0
new file mode 100644
index 0000000..389b178
@@ -0,0 +1,40 @@
+#ifndef SHPC_H
+#define SHPC_H
+
+#include "qemu-common.h"
+#include "memory.h"
+
+struct SHPCDevice {
+    /* Capability offset in device's config space */
+    int cap;
+
+    /* # of hot-pluggable slots */
+    int nslots;
+
+    /* SHPC WRS: working register set */
+    uint8_t *config;
+
+    /* Used to enable checks on load. Note that writable bits are
+     * never checked even if set in cmask. */
+    uint8_t *cmask;
+
+    /* Used to implement R/W bytes */
+    uint8_t *wmask;
+
+    /* Used to implement RW1C(Write 1 to Clear) bytes */
+    uint8_t *w1cmask;
+
+    /* MMIO for the SHPC BAR */
+    MemoryRegion mmio;
+
+    /* Bus controlled by this SHPC */
+    PCIBus *sec_bus;
+};
+
+void shpc_reset(PCIDevice *d);
+int shpc_bar_size(PCIDevice *dev);
+int shpc_init(PCIDevice *dev, PCIBus *sec_bus, MemoryRegion *bar, unsigned offset);
+void shpc_cleanup(PCIDevice *dev);
+void shpc_cap_write_config(PCIDevice *d, uint32_t addr, uint32_t val, int len);
+
+#endif
qemu-common.h +1 -0
index 9b997f8..4ff9c95 100644
@@ -247,6 +247,7 @@ typedef struct SSIBus SSIBus;
 typedef struct EventNotifier EventNotifier;
 typedef struct VirtIODevice VirtIODevice;
 typedef struct QEMUSGList QEMUSGList;
+typedef struct SHPCDevice SHPCDevice;
 
 typedef uint64_t pcibus_t;
 
Signature · 1 line
1.7.9.111.gf3fb0
Signature · 1 line
yamahata
#4 lore raw

Re: [PATCHv2-RFC 1/2] shpc: standard hot plug controller

Quoted 21 lines · Michael S. Tsirkin
Quoted 7 lines · Isaku Yamahata

Oh nice work.

Quoted 4 lines · Michael S. Tsirkin

This adds support for SHPC interface, as defined by PCI Standard Hot-Plug Controller and Subsystem Specification, Rev 1.0
http://www.pcisig.com/specifications/conventional/pci_hot_plug/SHPC_10

Quoted 7 lines
Quoted 5 lines

Only SHPC intergrated with a PCI-to-PCI bridge is supported, SHPC integrated with a host bridge would need more work.

All main SHPC features are supported:
- MRL sensor

Does this just report latch status? (It seems so.)

What happens is that adding a device closes the latch, removing a device opens the latch. This simplifies the number of supported configurations significantly.

Quoted 4 lines
Quoted 1 line

Do you plan to provide interfaces to manipulate the latch?

I didn't plan to do this, and this is non-trivial. Do you just want this for empty slots? And why?

No, I just wondered your plan.

Quoted 17 lines
Quoted 17 lines
Quoted 16 lines

- Attention button
- Attention indicator
- Power indicator

Wake on hotplug and serr generation are stubbed out but unused as we don't have interfaces to generate these events ATM.

One issue that isn't completely resolved is that qemu currently expects an "eject" interface, which SHPC does not provide: it merely removes the power to device and it's up to the user to remove the device from slot. This patch works around that by ejecting the device when power is removed and power LED goes off.

TODO:
- migration support
- fix dependency on pci_internals.h

Quoted 6 lines
Quoted 3 lines

If I didn't miss the code,
- QMP command for pushing attention button.
- QMP command to get LED status

It's easy to add these, so I'd accept such a patch, but I wonder why.

My concern is how libvirt/virt-manger (or other UI) presents slot status to operators/users.

Quoted 10 lines
Quoted 1 line

- QMP events for LED on/off

There's also blink :)

Quoted 2 lines

thanks,

I'm concerned that a guest can flood the management with such events. It's better to send a single "LED change" event, then we can suppress further events until next "get LED status" command.

Makes sense.

Quoted 765 lines
Quoted 764 lines
Quoted 761 lines
---
 Makefile.objs |    1 +
 hw/pci.h      |    6 +
 hw/shpc.c     |  646 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 hw/shpc.h     |   40 ++++
 qemu-common.h |    1 +
 5 files changed, 694 insertions(+), 0 deletions(-)
 create mode 100644 hw/shpc.c
 create mode 100644 hw/shpc.h
Makefile.objs +1 -0
index 391e524..4546477 100644
@@ -195,6 +195,7 @@ hw-obj-$(CONFIG_VIRTIO_PCI) += virtio-pci.o
 hw-obj-y += fw_cfg.o
 hw-obj-$(CONFIG_PCI) += pci.o pci_bridge.o
 hw-obj-$(CONFIG_PCI) += msix.o msi.o
+hw-obj-$(CONFIG_PCI) += shpc.o
 hw-obj-$(CONFIG_PCI) += pci_host.o pcie_host.o
 hw-obj-$(CONFIG_PCI) += ioh3420.o xio3130_upstream.o xio3130_downstream.o
 hw-obj-y += watchdog.o
hw/pci.h +6 -0
index 33b0b18..756577e 100644
@@ -125,6 +125,9 @@ enum {
     /* command register SERR bit enabled */
 #define QEMU_PCI_CAP_SERR_BITNR 4
     QEMU_PCI_CAP_SERR = (1 << QEMU_PCI_CAP_SERR_BITNR),
+    /* Standard hot plug controller. */
+#define QEMU_PCI_SHPC_BITNR 5
+    QEMU_PCI_CAP_SHPC = (1 << QEMU_PCI_SHPC_BITNR),
 };
 
 #define TYPE_PCI_DEVICE "pci-device"
@@ -229,6 +232,9 @@ struct PCIDevice {
     /* PCI Express */
     PCIExpressDevice exp;
 
+    /* SHPC */
+    SHPCDevice *shpc;
+
     /* Location of option rom */
     char *romfile;
     bool has_rom;
hw/shpc.c new file +646 -0
new file mode 100644
index 0000000..4baec29
@@ -0,0 +1,646 @@
+#include <strings.h>
+#include <stdint.h>
+#include "range.h"
+#include "shpc.h"
+#include "pci.h"
+#include "pci_internals.h"
+
+/* TODO: model power only and disabled slot states. */
+/* TODO: handle SERR and wakeups */
+/* TODO: consider enabling 66MHz support */
+
+/* TODO: remove fully only on state DISABLED and LED off.
+ * track state to properly record this. */
+
+/* SHPC Working Register Set */
+#define SHPC_BASE_OFFSET  0x00 /* 4 bytes */
+#define SHPC_SLOTS_33     0x04 /* 4 bytes. Also encodes PCI-X slots. */
+#define SHPC_SLOTS_66     0x08 /* 4 bytes. */
+#define SHPC_NSLOTS       0x0C /* 1 byte */
+#define SHPC_FIRST_DEV    0x0D /* 1 byte */
+#define SHPC_PHYS_SLOT    0x0E /* 2 byte */
+#define SHPC_PHYS_NUM_MAX 0x7ff
+#define SHPC_PHYS_NUM_UP  0x1000
+#define SHPC_PHYS_MRL     0x4000
+#define SHPC_PHYS_BUTTON  0x8000
+#define SHPC_SEC_BUS      0x10 /* 2 bytes */
+#define SHPC_SEC_BUS_33   0x0
+#define SHPC_SEC_BUS_66   0x1 /* Unused */
+#define SHPC_SEC_BUS_MASK 0x7
+#define SHPC_MSI_CTL      0x12 /* 1 byte */
+#define SHPC_PROG_IFC     0x13 /* 1 byte */
+#define SHPC_PROG_IFC_1_0 0x1
+#define SHPC_CMD_CODE     0x14 /* 1 byte */
+#define SHPC_CMD_TRGT     0x15 /* 1 byte */
+#define SHPC_CMD_TRGT_MIN 0x1
+#define SHPC_CMD_TRGT_MAX 0x1f
+#define SHPC_CMD_STATUS   0x16 /* 2 bytes */
+#define SHPC_CMD_STATUS_BUSY          0x1
+#define SHPC_CMD_STATUS_MRL_OPEN      0x2
+#define SHPC_CMD_STATUS_INVALID_CMD   0x4
+#define SHPC_CMD_STATUS_INVALID_MODE  0x8
+#define SHPC_INT_LOCATOR  0x18 /* 4 bytes */
+#define SHPC_INT_COMMAND  0x1
+#define SHPC_SERR_LOCATOR 0x1C /* 4 bytes */
+#define SHPC_SERR_INT     0x20 /* 4 bytes */
+#define SHPC_INT_DIS      0x1
+#define SHPC_SERR_DIS     0x2
+#define SHPC_CMD_INT_DIS  0x4
+#define SHPC_ARB_SERR_DIS 0x8
+#define SHPC_CMD_DETECTED 0x10000
+#define SHPC_ARB_DETECTED 0x20000
+ /* 4 bytes * slot # (start from 0) */
+#define SHPC_SLOT_REG(s)         (0x24 + (s) * 4)
+ /* 2 bytes */
+#define SHPC_SLOT_STATUS(s)       (0x0 + SHPC_SLOT_REG(s))
+
+/* Same slot state masks are used for command and status registers */
+#define SHPC_SLOT_STATE_MASK     0x03
+#define SHPC_SLOT_STATE_SHIFT \
+    (ffs(SHPC_SLOT_STATE_MASK) - 1)
+
+#define SHPC_STATE_NO       0x0
+#define SHPC_STATE_PWRONLY  0x1
+#define SHPC_STATE_ENABLED  0x2
+#define SHPC_STATE_DISABLED 0x3
+
+#define SHPC_SLOT_PWR_LED_MASK   0xC
+#define SHPC_SLOT_PWR_LED_SHIFT \
+    (ffs(SHPC_SLOT_PWR_LED_MASK) - 1)
+#define SHPC_SLOT_ATTN_LED_MASK  0x30
+#define SHPC_SLOT_ATTN_LED_SHIFT \
+    (ffs(SHPC_SLOT_ATTN_LED_MASK) - 1)
+
+#define SHPC_LED_NO     0x0
+#define SHPC_LED_ON     0x1
+#define SHPC_LED_BLINK  0x2
+#define SHPC_LED_OFF    0x3
+
+#define SHPC_SLOT_STATUS_PWR_FAULT      0x40
+#define SHPC_SLOT_STATUS_BUTTON         0x80
+#define SHPC_SLOT_STATUS_MRL_OPEN       0x100
+#define SHPC_SLOT_STATUS_66             0x200
+#define SHPC_SLOT_STATUS_PRSNT_MASK     0xC00
+#define SHPC_SLOT_STATUS_PRSNT_EMPTY    0x3
+#define SHPC_SLOT_STATUS_PRSNT_25W      0x1
+#define SHPC_SLOT_STATUS_PRSNT_15W      0x2
+#define SHPC_SLOT_STATUS_PRSNT_7_5W     0x0
+
+#define SHPC_SLOT_STATUS_PRSNT_PCIX     0x3000
+
+
+ /* 1 byte */
+#define SHPC_SLOT_EVENT_LATCH(s)        (0x2 + SHPC_SLOT_REG(s))
+ /* 1 byte */
+#define SHPC_SLOT_EVENT_SERR_INT_DIS(d, s) (0x3 + SHPC_SLOT_REG(s))
+#define SHPC_SLOT_EVENT_PRESENCE        0x01
+#define SHPC_SLOT_EVENT_ISOLATED_FAULT  0x02
+#define SHPC_SLOT_EVENT_BUTTON          0x04
+#define SHPC_SLOT_EVENT_MRL             0x08
+#define SHPC_SLOT_EVENT_CONNECTED_FAULT 0x10
+/* Bits below are used for Serr/Int disable only */
+#define SHPC_SLOT_EVENT_MRL_SERR_DIS    0x20
+#define SHPC_SLOT_EVENT_CONNECTED_FAULT_SERR_DIS 0x40
+
+#define SHPC_MIN_SLOTS        1
+#define SHPC_MAX_SLOTS        31
+#define SHPC_SIZEOF(d)    SHPC_SLOT_REG((d)->shpc->nslots)
+
+/* SHPC Slot identifiers */
+
+/* Hotplug supported at 31 slots out of the total 32.  We reserve slot 0,
+   and give the rest of them physical *and* pci numbers starting from 1, so
+   they match logical numbers.  Note: this means that multiple slots must have
+   different chassis number values, to make chassis+physical slot unique.
+   TODO: make this configurable? */
+#define SHPC_IDX_TO_LOGICAL(slot) ((slot) + 1)
+#define SHPC_LOGICAL_TO_IDX(target) ((target) - 1)
+#define SHPC_IDX_TO_PCI(slot) ((slot) + 1)
+#define SHPC_PCI_TO_IDX(pci_slot) ((pci_slot) - 1)
+#define SHPC_IDX_TO_PHYSICAL(slot) ((slot) + 1)
+
+static int roundup_pow_of_two(int x)
+{
+    x |= (x >> 1);
+    x |= (x >> 2);
+    x |= (x >> 4);
+    x |= (x >> 8);
+    x |= (x >> 16);
+    return x + 1;
+}
+
+static uint16_t shpc_get_status(SHPCDevice *shpc, int slot, uint16_t msk)
+{
+    uint8_t *status = shpc->config + SHPC_SLOT_STATUS(slot);
+    return (pci_get_word(status) & msk) >> (ffs(msk) - 1);
+}
+
+static void shpc_set_status(SHPCDevice *shpc,
+                            int slot, uint8_t value, uint16_t msk)
+{
+    uint8_t *status = shpc->config + SHPC_SLOT_STATUS(slot);
+    pci_word_test_and_clear_mask(status, msk);
+    pci_word_test_and_set_mask(status, value << (ffs(msk) - 1));
+}
+
+static void shpc_interrupt_update(PCIDevice *d)
+{
+    SHPCDevice *shpc = d->shpc;
+    int slot;
+    int level = 0;
+    uint32_t serr_int;
+    uint32_t int_locator = 0;
+
+    /* Update interrupt locator register */
+    for (slot = 0; slot < shpc->nslots; ++slot) {
+        uint8_t event = shpc->config[SHPC_SLOT_EVENT_LATCH(slot)];
+        uint8_t disable = shpc->config[SHPC_SLOT_EVENT_SERR_INT_DIS(d, slot)];
+        uint32_t mask = 1 << SHPC_IDX_TO_LOGICAL(slot);
+        if (event & ~disable) {
+            int_locator |= mask;
+        }
+    }
+    serr_int = pci_get_long(shpc->config + SHPC_SERR_INT);
+    if ((serr_int & SHPC_CMD_DETECTED) && !(serr_int & SHPC_CMD_INT_DIS)) {
+        int_locator |= SHPC_INT_COMMAND;
+    }
+    pci_set_long(shpc->config + SHPC_INT_LOCATOR, int_locator);
+    level = (!(serr_int & SHPC_INT_DIS) && int_locator) ? 1 : 0;
+    qemu_set_irq(d->irq[0], level);
+}
+
+static void shpc_set_sec_bus_speed(SHPCDevice *shpc, uint8_t speed)
+{
+    switch (speed) {
+    case SHPC_SEC_BUS_33:
+        shpc->config[SHPC_SEC_BUS] &= ~SHPC_SEC_BUS_MASK;
+        shpc->config[SHPC_SEC_BUS] |= speed;
+        break;
+    default:
+	pci_word_test_and_set_mask(shpc->config + SHPC_CMD_STATUS,
+				   SHPC_CMD_STATUS_INVALID_MODE);
+    }
+}
+
+void shpc_reset(PCIDevice *d)
+{
+    SHPCDevice *shpc = d->shpc;
+    int nslots = shpc->nslots;
+    int i;
+    memset(shpc->config, 0, SHPC_SIZEOF(d));
+    pci_set_byte(shpc->config + SHPC_NSLOTS, nslots);
+    pci_set_long(shpc->config + SHPC_SLOTS_33, nslots);
+    pci_set_long(shpc->config + SHPC_SLOTS_66, 0);
+    pci_set_byte(shpc->config + SHPC_FIRST_DEV, SHPC_IDX_TO_PCI(0));
+    pci_set_word(shpc->config + SHPC_PHYS_SLOT,
+		 SHPC_IDX_TO_PHYSICAL(0) |
+		 SHPC_PHYS_NUM_UP |
+		 SHPC_PHYS_MRL |
+		 SHPC_PHYS_BUTTON);
+    pci_set_long(shpc->config + SHPC_SERR_INT, SHPC_INT_DIS |
+                 SHPC_SERR_DIS |
+                 SHPC_CMD_INT_DIS |
+                 SHPC_ARB_SERR_DIS);
+    pci_set_byte(shpc->config + SHPC_PROG_IFC, SHPC_PROG_IFC_1_0);
+    pci_set_word(shpc->config + SHPC_SEC_BUS, SHPC_SEC_BUS_33);
+    for (i = 0; i < shpc->nslots; ++i) {
+        pci_set_word(shpc->config + SHPC_SLOT_EVENT_SERR_INT_DIS(d, i),
+                     SHPC_SLOT_EVENT_PRESENCE |
+                     SHPC_SLOT_EVENT_ISOLATED_FAULT |
+                     SHPC_SLOT_EVENT_BUTTON |
+                     SHPC_SLOT_EVENT_MRL |
+                     SHPC_SLOT_EVENT_CONNECTED_FAULT |
+                     SHPC_SLOT_EVENT_MRL_SERR_DIS |
+                     SHPC_SLOT_EVENT_CONNECTED_FAULT_SERR_DIS);
+        if (shpc->sec_bus->devices[PCI_DEVFN(SHPC_IDX_TO_PCI(i), 0)]) {
+            shpc_set_status(shpc, i, SHPC_STATE_ENABLED, SHPC_SLOT_STATE_MASK);
+            shpc_set_status(shpc, i, 0, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, i, SHPC_SLOT_STATUS_PRSNT_7_5W,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc_set_status(shpc, i, SHPC_LED_ON, SHPC_SLOT_PWR_LED_MASK);
+        } else {
+            shpc_set_status(shpc, i, SHPC_STATE_DISABLED, SHPC_SLOT_STATE_MASK);
+            shpc_set_status(shpc, i, 1, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, i, SHPC_SLOT_STATUS_PRSNT_EMPTY,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc_set_status(shpc, i, SHPC_LED_OFF, SHPC_SLOT_PWR_LED_MASK);
+	}
+        shpc_set_status(shpc, i, 0, SHPC_SLOT_STATUS_66);
+    }
+    shpc_set_sec_bus_speed(shpc, SHPC_SEC_BUS_33);
+    shpc_interrupt_update(d);
+}
+
+static void shpc_invalid_command(SHPCDevice *shpc)
+{
+    pci_word_test_and_set_mask(shpc->config + SHPC_CMD_STATUS,
+                               SHPC_CMD_STATUS_INVALID_CMD);
+}
+
+static void shpc_free_devices_in_slot(SHPCDevice *shpc, int slot)
+{
+    int devfn;
+    int pci_slot = SHPC_IDX_TO_PCI(slot);
+    for (devfn = PCI_DEVFN(pci_slot, 0);
+         devfn <= PCI_DEVFN(pci_slot, PCI_FUNC_MAX - 1);
+         ++devfn) {
+        PCIDevice *affected_dev = shpc->sec_bus->devices[devfn];
+        if (affected_dev) {
+            qdev_free(&affected_dev->qdev);
+        }
+    }
+}
+
+static void shpc_slot_command(SHPCDevice *shpc, uint8_t target,
+                              uint8_t state, uint8_t power, uint8_t attn)
+{
+    uint8_t current_state;
+    int slot = SHPC_LOGICAL_TO_IDX(target);
+    if (target < SHPC_CMD_TRGT_MIN || slot >= shpc->nslots) {
+        shpc_invalid_command(shpc);
+        return;
+    }
+    current_state = shpc_get_status(shpc, slot, SHPC_SLOT_STATE_MASK);
+    if (current_state == SHPC_STATE_ENABLED && state == SHPC_STATE_PWRONLY) {
+        shpc_invalid_command(shpc);
+	return;
+    }
+
+    switch (power) {
+    case SHPC_LED_NO:
+        break;
+    default:
+        /* TODO: send event to monitor */
+        shpc_set_status(shpc, slot, power, SHPC_SLOT_PWR_LED_MASK);
+    }
+    switch (attn) {
+    case SHPC_LED_NO:
+        break;
+    default:
+        /* TODO: send event to monitor */
+        shpc_set_status(shpc, slot, attn, SHPC_SLOT_ATTN_LED_MASK);
+    }
+
+    if ((current_state == SHPC_STATE_DISABLED && state == SHPC_STATE_PWRONLY) ||
+        (current_state == SHPC_STATE_DISABLED && state == SHPC_STATE_ENABLED)) {
+        shpc_set_status(shpc, slot, state, SHPC_SLOT_STATE_MASK);
+    } else if ((current_state == SHPC_STATE_ENABLED ||
+		current_state == SHPC_STATE_PWRONLY) &&
+	       state == SHPC_STATE_DISABLED) {
+        shpc_set_status(shpc, slot, state, SHPC_SLOT_STATE_MASK);
+        power = shpc_get_status(shpc, slot, SHPC_SLOT_PWR_LED_MASK);
+        /* TODO: track what monitor requested. */
+        /* Look at LED to figure out whether it's ok to remove the device. */
+        if (power == SHPC_LED_OFF) {
+            shpc_free_devices_in_slot(shpc, slot);
+            shpc_set_status(shpc, slot, 1, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_EMPTY,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_BUTTON |
+                SHPC_SLOT_EVENT_MRL |
+                SHPC_SLOT_EVENT_PRESENCE;
+        }
+    }
+}
+
+static void shpc_command(SHPCDevice *shpc)
+{
+    uint8_t code = pci_get_byte(shpc->config + SHPC_CMD_CODE);
+    uint8_t speed;
+    uint8_t target;
+    uint8_t attn;
+    uint8_t power;
+    uint8_t state;
+    int i;
+
+    /* Clear status from the previous command. */
+    pci_word_test_and_clear_mask(shpc->config + SHPC_CMD_STATUS,
+				 SHPC_CMD_STATUS_BUSY |
+				 SHPC_CMD_STATUS_MRL_OPEN |
+				 SHPC_CMD_STATUS_INVALID_CMD |
+				 SHPC_CMD_STATUS_INVALID_MODE);
+    switch (code) {
+    case 0x00 ... 0x3f:
+        target = shpc->config[SHPC_CMD_TRGT] & SHPC_CMD_TRGT_MAX;
+        state = (code & SHPC_SLOT_STATE_MASK) >> SHPC_SLOT_STATE_SHIFT;
+        power = (code & SHPC_SLOT_PWR_LED_MASK) >> SHPC_SLOT_PWR_LED_SHIFT;
+        attn = (code & SHPC_SLOT_ATTN_LED_MASK) >> SHPC_SLOT_ATTN_LED_SHIFT;
+        shpc_slot_command(shpc, target, state, power, attn);
+        break;
+    case 0x40 ... 0x47:
+        speed = code & SHPC_SEC_BUS_MASK;
+        shpc_set_sec_bus_speed(shpc, speed);
+        break;
+    case 0x48:
+        /* Power only all slots */
+        /* first verify no slots are enabled */
+        for (i = 0; i < shpc->nslots; ++i) {
+            state = shpc_get_status(shpc, i, SHPC_SLOT_STATE_MASK);
+            if (state == SHPC_STATE_ENABLED) {
+                shpc_invalid_command(shpc);
+                goto done;
+            }
+        }
+        for (i = 0; i < shpc->nslots; ++i) {
+            if (!(shpc_get_status(shpc, i, SHPC_SLOT_STATUS_MRL_OPEN))) {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_PWRONLY, SHPC_LED_ON, SHPC_LED_NO);
+            } else {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_NO, SHPC_LED_OFF, SHPC_LED_NO);
+            }
+        }
+        break;
+    case 0x49:
+        /* Enable all slots */
+        /* TODO: Spec says this shall fail if some are already enabled.
+         * This doesn't make sense - why not? a spec bug? */
+        for (i = 0; i < shpc->nslots; ++i) {
+            state = shpc_get_status(shpc, i, SHPC_SLOT_STATE_MASK);
+            if (state == SHPC_STATE_ENABLED) {
+                shpc_invalid_command(shpc);
+                goto done;
+            }
+        }
+        for (i = 0; i < shpc->nslots; ++i) {
+            if (!(shpc_get_status(shpc, i, SHPC_SLOT_STATUS_MRL_OPEN))) {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_ENABLED, SHPC_LED_ON, SHPC_LED_NO);
+            } else {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_NO, SHPC_LED_OFF, SHPC_LED_NO);
+            }
+        }
+        break;
+    default:
+        shpc_invalid_command(shpc);
+        break;
+    }
+done:
+    pci_long_test_and_set_mask(shpc->config + SHPC_SERR_INT, SHPC_CMD_DETECTED);
+}
+
+static void shpc_write(PCIDevice *d, unsigned addr, uint64_t val, int l)
+{
+    SHPCDevice *shpc = d->shpc;
+    int i;
+    if (addr >= SHPC_SIZEOF(d)) {
+        return;
+    }
+    l = MIN(l, SHPC_SIZEOF(d) - addr);
+
+    /* TODO: code duplicated from pci.c */
+    for (i = 0; i < l; val >>= 8, ++i) {
+        unsigned a = addr + i;
+        uint8_t wmask = shpc->wmask[a];
+        uint8_t w1cmask = shpc->w1cmask[a];
+        assert(!(wmask & w1cmask));
+        shpc->config[a] = (shpc->config[a] & ~wmask) | (val & wmask);
+        shpc->config[a] &= ~(val & w1cmask); /* W1C: Write 1 to Clear */
+    }
+    if (ranges_overlap(addr, l, SHPC_CMD_CODE, 2)) {
+        shpc_command(shpc);
+    }
+    shpc_interrupt_update(d);
+}
+
+static uint64_t shpc_read(PCIDevice *d, unsigned addr, int l)
+{
+    uint64_t val = 0x0;
+    if (addr >= SHPC_SIZEOF(d)) {
+        return val;
+    }
+    l = MIN(l, SHPC_SIZEOF(d) - addr);
+    memcpy(&val, d->shpc->config + addr, l);
+    return val;
+}
+
+/* SHPC Bridge Capability */
+#define SHPC_CAP_LENGTH 0x08
+#define SHPC_CAP_DWORD_SELECT 0x2 /* 1 byte */
+#define SHPC_CAP_CxP 0x3 /* 1 byte: CSP, CIP */
+#define SHPC_CAP_DWORD_DATA 0x4 /* 4 bytes */
+#define SHPC_CAP_CSP_MASK 0x4
+#define SHPC_CAP_CIP_MASK 0x8
+
+static uint8_t shpc_cap_dword(PCIDevice *d)
+{
+    return pci_get_byte(d->config + d->shpc->cap + SHPC_CAP_DWORD_SELECT);
+}
+
+/* Update dword data capability register */
+static void shpc_cap_update_dword(PCIDevice *d)
+{
+    unsigned data;
+    data = shpc_read(d, shpc_cap_dword(d) * 4, 4);
+    pci_set_long(d->config  + d->shpc->cap + SHPC_CAP_DWORD_DATA, data);
+}
+
+/* Add SHPC capability to the config space for the device. */
+static int shpc_cap_add_config(PCIDevice *d)
+{
+    uint8_t *config;
+    int config_offset;
+    config_offset = pci_add_capability(d, PCI_CAP_ID_SHPC,
+                                       0, SHPC_CAP_LENGTH);
+    if (config_offset < 0) {
+        return config_offset;
+    }
+    config = d->config + config_offset;
+
+    pci_set_byte(config + SHPC_CAP_DWORD_SELECT, 0);
+    pci_set_byte(config + SHPC_CAP_CxP, 0);
+    pci_set_long(config + SHPC_CAP_DWORD_DATA, 0);
+    d->shpc->cap = config_offset;
+    /* Make dword select and data writeable. */
+    pci_set_byte(d->wmask + config_offset + SHPC_CAP_DWORD_SELECT, 0xff);
+    pci_set_long(d->wmask + config_offset + SHPC_CAP_DWORD_DATA, 0xffffffff);
+    return 0;
+}
+
+static uint64_t shpc_mmio_read(void *opaque, target_phys_addr_t addr,
+                               unsigned size)
+{
+    return shpc_read(opaque, addr, size);
+}
+
+static void shpc_mmio_write(void *opaque, target_phys_addr_t addr,
+                            uint64_t val, unsigned size)
+{
+    shpc_write(opaque, addr, val, size);
+}
+
+static const MemoryRegionOps shpc_mmio_ops = {
+    .read = shpc_mmio_read,
+    .write = shpc_mmio_write,
+    .endianness = DEVICE_LITTLE_ENDIAN,
+    .valid = {
+        /* SHPC ECN requires dword accesses, but the original 1.0 spec doesn't.
+         * It's easier to suppport all sizes than worry about it. */
+        .min_access_size = 1,
+        .max_access_size = 4,
+    },
+};
+
+static int shpc_device_hotplug(DeviceState *qdev, PCIDevice *affected_dev,
+                               PCIHotplugState hotplug_state)
+{
+    int pci_slot = PCI_SLOT(affected_dev->devfn);
+    uint8_t state;
+    uint8_t led;
+    PCIDevice *d = DO_UPCAST(PCIDevice, qdev, qdev);
+    SHPCDevice *shpc = d->shpc;
+    int slot = SHPC_PCI_TO_IDX(pci_slot);
+    if (pci_slot < SHPC_IDX_TO_PCI(0) || slot >= shpc->nslots) {
+        error_report("Unsupported PCI slot %d for standard hotplug "
+                     "controller. Valid slots are between %d and %d.",
+                     pci_slot, SHPC_IDX_TO_PCI(0),
+                     SHPC_IDX_TO_PCI(shpc->nslots) - 1);
+        return -1;
+    }
+    /* Don't send event when device is enabled during qemu machine creation:
+     * it is present on boot, no hotplug event is necessary. We do send an
+     * event when the device is disabled later. */
+    if (hotplug_state == PCI_COLDPLUG_ENABLED) {
+        shpc_set_status(shpc, slot, 0, SHPC_SLOT_STATUS_MRL_OPEN);
+        shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_7_5W,
+                        SHPC_SLOT_STATUS_PRSNT_MASK);
+        return 0;
+    }
+    if (hotplug_state == PCI_HOTPLUG_DISABLED) {
+        shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |= SHPC_SLOT_EVENT_BUTTON;
+        state = shpc_get_status(shpc, slot, SHPC_SLOT_STATE_MASK);
+        led = shpc_get_status(shpc, slot, SHPC_SLOT_PWR_LED_MASK);
+        if (state == SHPC_STATE_DISABLED && led == SHPC_LED_OFF) {
+            shpc_free_devices_in_slot(shpc, slot);
+            shpc_set_status(shpc, slot, 1, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_EMPTY,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_MRL |
+                SHPC_SLOT_EVENT_PRESENCE;
+        }
+    } else {
+        /* This could be a cancellation of the previous removal.
+         * We check MRL state to figure out. */
+        if (shpc_get_status(shpc, slot, SHPC_SLOT_STATUS_MRL_OPEN)) {
+            shpc_set_status(shpc, slot, 0, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_7_5W,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_BUTTON |
+                SHPC_SLOT_EVENT_MRL |
+                SHPC_SLOT_EVENT_PRESENCE;
+        } else {
+            /* Press attention button to cancel removal */
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_BUTTON;
+        }
+    }
+    shpc_set_status(shpc, slot, 0, SHPC_SLOT_STATUS_66);
+    shpc_interrupt_update(d);
+    return 0;
+}
+
+/* Initialize the SHPC structure in bridge's BAR. */
+int shpc_init(PCIDevice *d, PCIBus *sec_bus, MemoryRegion *bar, unsigned offset)
+{
+    int i, ret;
+    int nslots = SHPC_MAX_SLOTS; /* TODO: qdev property? */
+    SHPCDevice *shpc = d->shpc = g_malloc0(sizeof(*d->shpc));
+    shpc->sec_bus = sec_bus;
+    ret = shpc_cap_add_config(d);
+    if (ret) {
+        g_free(d->shpc);
+        return ret;
+    }
+    if (nslots < SHPC_MIN_SLOTS) {
+        return 0;
+    }
+    if (nslots > SHPC_MAX_SLOTS ||
+        SHPC_IDX_TO_PCI(nslots) > PCI_SLOT_MAX) {
+        /* TODO: report an error mesage that makes sense. */
+        return -EINVAL;
+    }
+    shpc->nslots = nslots;
+    shpc->config = g_malloc0(SHPC_SIZEOF(d));
+    shpc->cmask = g_malloc0(SHPC_SIZEOF(d));
+    shpc->wmask = g_malloc0(SHPC_SIZEOF(d));
+    shpc->w1cmask = g_malloc0(SHPC_SIZEOF(d));
+
+    shpc_reset(d);
+
+    pci_set_long(shpc->config + SHPC_BASE_OFFSET, offset);
+
+    pci_set_byte(shpc->wmask + SHPC_CMD_CODE, 0xff);
+    pci_set_byte(shpc->wmask + SHPC_CMD_TRGT, SHPC_CMD_TRGT_MAX);
+    pci_set_byte(shpc->wmask + SHPC_CMD_TRGT, SHPC_CMD_TRGT_MAX);
+    pci_set_long(shpc->wmask + SHPC_SERR_INT,
+                 SHPC_INT_DIS |
+                 SHPC_SERR_DIS |
+                 SHPC_CMD_INT_DIS |
+                 SHPC_ARB_SERR_DIS);
+    pci_set_long(shpc->w1cmask + SHPC_SERR_INT,
+		 SHPC_CMD_DETECTED |
+                 SHPC_ARB_DETECTED);
+    for (i = 0; i < nslots; ++i) {
+	    pci_set_byte(shpc->wmask +
+			 SHPC_SLOT_EVENT_SERR_INT_DIS(d, i),
+			 SHPC_SLOT_EVENT_PRESENCE |
+			 SHPC_SLOT_EVENT_ISOLATED_FAULT |
+			 SHPC_SLOT_EVENT_BUTTON |
+			 SHPC_SLOT_EVENT_MRL |
+			 SHPC_SLOT_EVENT_CONNECTED_FAULT |
+			 SHPC_SLOT_EVENT_MRL_SERR_DIS |
+			 SHPC_SLOT_EVENT_CONNECTED_FAULT_SERR_DIS);
+	    pci_set_byte(shpc->w1cmask +
+			 SHPC_SLOT_EVENT_LATCH(i),
+			 SHPC_SLOT_EVENT_PRESENCE |
+			 SHPC_SLOT_EVENT_ISOLATED_FAULT |
+			 SHPC_SLOT_EVENT_BUTTON |
+			 SHPC_SLOT_EVENT_MRL |
+			 SHPC_SLOT_EVENT_CONNECTED_FAULT);
+    }
+
+    /* TODO: init cmask */
+    memory_region_init_io(&shpc->mmio, &shpc_mmio_ops, d, "shpc-mmio",
+                          SHPC_SIZEOF(d));
+    shpc_cap_update_dword(d);
+    memory_region_add_subregion(bar, offset, &shpc->mmio);
+    pci_bus_hotplug(sec_bus, shpc_device_hotplug, &d->qdev);
+    return 0;
+}
+
+int shpc_bar_size(PCIDevice *d)
+{
+    return roundup_pow_of_two(SHPC_SLOT_REG(SHPC_MAX_SLOTS));
+}
+
+void shpc_cleanup(PCIDevice *d)
+{
+    SHPCDevice *shpc = d->shpc;
+    /* TODO: cleanup config space changes? */
+    g_free(shpc->config);
+    g_free(shpc->cmask);
+    g_free(shpc->wmask);
+    g_free(shpc->w1cmask);
+    memory_region_destroy(&shpc->mmio);
+    g_free(shpc);
+}
+
+void shpc_cap_write_config(PCIDevice *d, uint32_t addr, uint32_t val, int l)
+{
+    if (!ranges_overlap(addr, l, d->shpc->cap, SHPC_CAP_LENGTH)) {
+        return;
+    }
+    fprintf(stderr, "%s: 0x%x 0x%x %d\n", __func__, addr, val, l);
+    if (ranges_overlap(addr, l, d->shpc->cap + SHPC_CAP_DWORD_DATA, 4)) {
+        unsigned dword_data;
+        dword_data = pci_get_long(d->shpc->config + d->shpc->cap
+                                  + SHPC_CAP_DWORD_DATA);
+        shpc_write(d, shpc_cap_dword(d) * 4, dword_data, 4);
+    }
+    /* Update cap dword data in case guest is going to read it. */
+    shpc_cap_update_dword(d);
+}
hw/shpc.h new file +40 -0
new file mode 100644
index 0000000..389b178
@@ -0,0 +1,40 @@
+#ifndef SHPC_H
+#define SHPC_H
+
+#include "qemu-common.h"
+#include "memory.h"
+
+struct SHPCDevice {
+    /* Capability offset in device's config space */
+    int cap;
+
+    /* # of hot-pluggable slots */
+    int nslots;
+
+    /* SHPC WRS: working register set */
+    uint8_t *config;
+
+    /* Used to enable checks on load. Note that writable bits are
+     * never checked even if set in cmask. */
+    uint8_t *cmask;
+
+    /* Used to implement R/W bytes */
+    uint8_t *wmask;
+
+    /* Used to implement RW1C(Write 1 to Clear) bytes */
+    uint8_t *w1cmask;
+
+    /* MMIO for the SHPC BAR */
+    MemoryRegion mmio;
+
+    /* Bus controlled by this SHPC */
+    PCIBus *sec_bus;
+};
+
+void shpc_reset(PCIDevice *d);
+int shpc_bar_size(PCIDevice *dev);
+int shpc_init(PCIDevice *dev, PCIBus *sec_bus, MemoryRegion *bar, unsigned offset);
+void shpc_cleanup(PCIDevice *dev);
+void shpc_cap_write_config(PCIDevice *d, uint32_t addr, uint32_t val, int len);
+
+#endif
qemu-common.h +1 -0
index 9b997f8..4ff9c95 100644
@@ -247,6 +247,7 @@ typedef struct SSIBus SSIBus;
 typedef struct EventNotifier EventNotifier;
 typedef struct VirtIODevice VirtIODevice;
 typedef struct QEMUSGList QEMUSGList;
+typedef struct SHPCDevice SHPCDevice;
 
 typedef uint64_t pcibus_t;
 
Signature · 1 line
1.7.9.111.gf3fb0
Signature · 1 line
yamahata
Signature · 1 line
yamahata
#5 lore raw

Re: [PATCHv2-RFC 1/2] shpc: standard hot plug controller

Quoted 51 lines · Isaku Yamahata
Quoted 21 lines · Michael S. Tsirkin
Quoted 7 lines · Isaku Yamahata

Oh nice work.

Quoted 4 lines · Michael S. Tsirkin

This adds support for SHPC interface, as defined by PCI Standard Hot-Plug Controller and Subsystem Specification, Rev 1.0
http://www.pcisig.com/specifications/conventional/pci_hot_plug/SHPC_10

Quoted 7 lines
Quoted 5 lines

Only SHPC intergrated with a PCI-to-PCI bridge is supported, SHPC integrated with a host bridge would need more work.

All main SHPC features are supported:
- MRL sensor

Does this just report latch status? (It seems so.)

What happens is that adding a device closes the latch, removing a device opens the latch. This simplifies the number of supported configurations significantly.

Quoted 4 lines
Quoted 1 line

Do you plan to provide interfaces to manipulate the latch?

I didn't plan to do this, and this is non-trivial. Do you just want this for empty slots? And why?

No, I just wondered your plan.

Quoted 21 lines
Quoted 20 lines
Quoted 13 lines

- Attention button
- Attention indicator
- Power indicator

Wake on hotplug and serr generation are stubbed out but unused as we don't have interfaces to generate these events ATM.

One issue that isn't completely resolved is that qemu currently expects an "eject" interface, which SHPC does not provide: it merely removes the power to device and it's up to the user to remove the device from slot. This patch works around that by ejecting the device when power is removed and power LED goes off.

Quoted 3 lines

TODO:
- migration support
- fix dependency on pci_internals.h

If I didn't miss the code,
- QMP command for pushing attention button.
- QMP command to get LED status

Quoted 5 lines
Quoted 2 lines

It's easy to add these, so I'd accept such a patch, but I wonder why.

My concern is how libvirt/virt-manger (or other UI) presents slot status to operators/users.

They currently present free/busy status just by looking at info pci. Maybe that is enough.

My concern is rather with the eject hack above: the add/delete API maps reasonably to _EJ0 interface, but isn't generic enough for SHPC. We'll need a better API for that.

Quoted 781 lines
Quoted 10 lines
Quoted 1 line

- QMP events for LED on/off

There's also blink :)

Quoted 2 lines

thanks,

I'm concerned that a guest can flood the management with such events. It's better to send a single "LED change" event, then we can suppress further events until next "get LED status" command.

Makes sense.

Quoted 765 lines
Quoted 764 lines
Quoted 761 lines
---
 Makefile.objs |    1 +
 hw/pci.h      |    6 +
 hw/shpc.c     |  646 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 hw/shpc.h     |   40 ++++
 qemu-common.h |    1 +
 5 files changed, 694 insertions(+), 0 deletions(-)
 create mode 100644 hw/shpc.c
 create mode 100644 hw/shpc.h
Makefile.objs +1 -0
index 391e524..4546477 100644
@@ -195,6 +195,7 @@ hw-obj-$(CONFIG_VIRTIO_PCI) += virtio-pci.o
 hw-obj-y += fw_cfg.o
 hw-obj-$(CONFIG_PCI) += pci.o pci_bridge.o
 hw-obj-$(CONFIG_PCI) += msix.o msi.o
+hw-obj-$(CONFIG_PCI) += shpc.o
 hw-obj-$(CONFIG_PCI) += pci_host.o pcie_host.o
 hw-obj-$(CONFIG_PCI) += ioh3420.o xio3130_upstream.o xio3130_downstream.o
 hw-obj-y += watchdog.o
hw/pci.h +6 -0
index 33b0b18..756577e 100644
@@ -125,6 +125,9 @@ enum {
     /* command register SERR bit enabled */
 #define QEMU_PCI_CAP_SERR_BITNR 4
     QEMU_PCI_CAP_SERR = (1 << QEMU_PCI_CAP_SERR_BITNR),
+    /* Standard hot plug controller. */
+#define QEMU_PCI_SHPC_BITNR 5
+    QEMU_PCI_CAP_SHPC = (1 << QEMU_PCI_SHPC_BITNR),
 };
 
 #define TYPE_PCI_DEVICE "pci-device"
@@ -229,6 +232,9 @@ struct PCIDevice {
     /* PCI Express */
     PCIExpressDevice exp;
 
+    /* SHPC */
+    SHPCDevice *shpc;
+
     /* Location of option rom */
     char *romfile;
     bool has_rom;
hw/shpc.c new file +646 -0
new file mode 100644
index 0000000..4baec29
@@ -0,0 +1,646 @@
+#include <strings.h>
+#include <stdint.h>
+#include "range.h"
+#include "shpc.h"
+#include "pci.h"
+#include "pci_internals.h"
+
+/* TODO: model power only and disabled slot states. */
+/* TODO: handle SERR and wakeups */
+/* TODO: consider enabling 66MHz support */
+
+/* TODO: remove fully only on state DISABLED and LED off.
+ * track state to properly record this. */
+
+/* SHPC Working Register Set */
+#define SHPC_BASE_OFFSET  0x00 /* 4 bytes */
+#define SHPC_SLOTS_33     0x04 /* 4 bytes. Also encodes PCI-X slots. */
+#define SHPC_SLOTS_66     0x08 /* 4 bytes. */
+#define SHPC_NSLOTS       0x0C /* 1 byte */
+#define SHPC_FIRST_DEV    0x0D /* 1 byte */
+#define SHPC_PHYS_SLOT    0x0E /* 2 byte */
+#define SHPC_PHYS_NUM_MAX 0x7ff
+#define SHPC_PHYS_NUM_UP  0x1000
+#define SHPC_PHYS_MRL     0x4000
+#define SHPC_PHYS_BUTTON  0x8000
+#define SHPC_SEC_BUS      0x10 /* 2 bytes */
+#define SHPC_SEC_BUS_33   0x0
+#define SHPC_SEC_BUS_66   0x1 /* Unused */
+#define SHPC_SEC_BUS_MASK 0x7
+#define SHPC_MSI_CTL      0x12 /* 1 byte */
+#define SHPC_PROG_IFC     0x13 /* 1 byte */
+#define SHPC_PROG_IFC_1_0 0x1
+#define SHPC_CMD_CODE     0x14 /* 1 byte */
+#define SHPC_CMD_TRGT     0x15 /* 1 byte */
+#define SHPC_CMD_TRGT_MIN 0x1
+#define SHPC_CMD_TRGT_MAX 0x1f
+#define SHPC_CMD_STATUS   0x16 /* 2 bytes */
+#define SHPC_CMD_STATUS_BUSY          0x1
+#define SHPC_CMD_STATUS_MRL_OPEN      0x2
+#define SHPC_CMD_STATUS_INVALID_CMD   0x4
+#define SHPC_CMD_STATUS_INVALID_MODE  0x8
+#define SHPC_INT_LOCATOR  0x18 /* 4 bytes */
+#define SHPC_INT_COMMAND  0x1
+#define SHPC_SERR_LOCATOR 0x1C /* 4 bytes */
+#define SHPC_SERR_INT     0x20 /* 4 bytes */
+#define SHPC_INT_DIS      0x1
+#define SHPC_SERR_DIS     0x2
+#define SHPC_CMD_INT_DIS  0x4
+#define SHPC_ARB_SERR_DIS 0x8
+#define SHPC_CMD_DETECTED 0x10000
+#define SHPC_ARB_DETECTED 0x20000
+ /* 4 bytes * slot # (start from 0) */
+#define SHPC_SLOT_REG(s)         (0x24 + (s) * 4)
+ /* 2 bytes */
+#define SHPC_SLOT_STATUS(s)       (0x0 + SHPC_SLOT_REG(s))
+
+/* Same slot state masks are used for command and status registers */
+#define SHPC_SLOT_STATE_MASK     0x03
+#define SHPC_SLOT_STATE_SHIFT \
+    (ffs(SHPC_SLOT_STATE_MASK) - 1)
+
+#define SHPC_STATE_NO       0x0
+#define SHPC_STATE_PWRONLY  0x1
+#define SHPC_STATE_ENABLED  0x2
+#define SHPC_STATE_DISABLED 0x3
+
+#define SHPC_SLOT_PWR_LED_MASK   0xC
+#define SHPC_SLOT_PWR_LED_SHIFT \
+    (ffs(SHPC_SLOT_PWR_LED_MASK) - 1)
+#define SHPC_SLOT_ATTN_LED_MASK  0x30
+#define SHPC_SLOT_ATTN_LED_SHIFT \
+    (ffs(SHPC_SLOT_ATTN_LED_MASK) - 1)
+
+#define SHPC_LED_NO     0x0
+#define SHPC_LED_ON     0x1
+#define SHPC_LED_BLINK  0x2
+#define SHPC_LED_OFF    0x3
+
+#define SHPC_SLOT_STATUS_PWR_FAULT      0x40
+#define SHPC_SLOT_STATUS_BUTTON         0x80
+#define SHPC_SLOT_STATUS_MRL_OPEN       0x100
+#define SHPC_SLOT_STATUS_66             0x200
+#define SHPC_SLOT_STATUS_PRSNT_MASK     0xC00
+#define SHPC_SLOT_STATUS_PRSNT_EMPTY    0x3
+#define SHPC_SLOT_STATUS_PRSNT_25W      0x1
+#define SHPC_SLOT_STATUS_PRSNT_15W      0x2
+#define SHPC_SLOT_STATUS_PRSNT_7_5W     0x0
+
+#define SHPC_SLOT_STATUS_PRSNT_PCIX     0x3000
+
+
+ /* 1 byte */
+#define SHPC_SLOT_EVENT_LATCH(s)        (0x2 + SHPC_SLOT_REG(s))
+ /* 1 byte */
+#define SHPC_SLOT_EVENT_SERR_INT_DIS(d, s) (0x3 + SHPC_SLOT_REG(s))
+#define SHPC_SLOT_EVENT_PRESENCE        0x01
+#define SHPC_SLOT_EVENT_ISOLATED_FAULT  0x02
+#define SHPC_SLOT_EVENT_BUTTON          0x04
+#define SHPC_SLOT_EVENT_MRL             0x08
+#define SHPC_SLOT_EVENT_CONNECTED_FAULT 0x10
+/* Bits below are used for Serr/Int disable only */
+#define SHPC_SLOT_EVENT_MRL_SERR_DIS    0x20
+#define SHPC_SLOT_EVENT_CONNECTED_FAULT_SERR_DIS 0x40
+
+#define SHPC_MIN_SLOTS        1
+#define SHPC_MAX_SLOTS        31
+#define SHPC_SIZEOF(d)    SHPC_SLOT_REG((d)->shpc->nslots)
+
+/* SHPC Slot identifiers */
+
+/* Hotplug supported at 31 slots out of the total 32.  We reserve slot 0,
+   and give the rest of them physical *and* pci numbers starting from 1, so
+   they match logical numbers.  Note: this means that multiple slots must have
+   different chassis number values, to make chassis+physical slot unique.
+   TODO: make this configurable? */
+#define SHPC_IDX_TO_LOGICAL(slot) ((slot) + 1)
+#define SHPC_LOGICAL_TO_IDX(target) ((target) - 1)
+#define SHPC_IDX_TO_PCI(slot) ((slot) + 1)
+#define SHPC_PCI_TO_IDX(pci_slot) ((pci_slot) - 1)
+#define SHPC_IDX_TO_PHYSICAL(slot) ((slot) + 1)
+
+static int roundup_pow_of_two(int x)
+{
+    x |= (x >> 1);
+    x |= (x >> 2);
+    x |= (x >> 4);
+    x |= (x >> 8);
+    x |= (x >> 16);
+    return x + 1;
+}
+
+static uint16_t shpc_get_status(SHPCDevice *shpc, int slot, uint16_t msk)
+{
+    uint8_t *status = shpc->config + SHPC_SLOT_STATUS(slot);
+    return (pci_get_word(status) & msk) >> (ffs(msk) - 1);
+}
+
+static void shpc_set_status(SHPCDevice *shpc,
+                            int slot, uint8_t value, uint16_t msk)
+{
+    uint8_t *status = shpc->config + SHPC_SLOT_STATUS(slot);
+    pci_word_test_and_clear_mask(status, msk);
+    pci_word_test_and_set_mask(status, value << (ffs(msk) - 1));
+}
+
+static void shpc_interrupt_update(PCIDevice *d)
+{
+    SHPCDevice *shpc = d->shpc;
+    int slot;
+    int level = 0;
+    uint32_t serr_int;
+    uint32_t int_locator = 0;
+
+    /* Update interrupt locator register */
+    for (slot = 0; slot < shpc->nslots; ++slot) {
+        uint8_t event = shpc->config[SHPC_SLOT_EVENT_LATCH(slot)];
+        uint8_t disable = shpc->config[SHPC_SLOT_EVENT_SERR_INT_DIS(d, slot)];
+        uint32_t mask = 1 << SHPC_IDX_TO_LOGICAL(slot);
+        if (event & ~disable) {
+            int_locator |= mask;
+        }
+    }
+    serr_int = pci_get_long(shpc->config + SHPC_SERR_INT);
+    if ((serr_int & SHPC_CMD_DETECTED) && !(serr_int & SHPC_CMD_INT_DIS)) {
+        int_locator |= SHPC_INT_COMMAND;
+    }
+    pci_set_long(shpc->config + SHPC_INT_LOCATOR, int_locator);
+    level = (!(serr_int & SHPC_INT_DIS) && int_locator) ? 1 : 0;
+    qemu_set_irq(d->irq[0], level);
+}
+
+static void shpc_set_sec_bus_speed(SHPCDevice *shpc, uint8_t speed)
+{
+    switch (speed) {
+    case SHPC_SEC_BUS_33:
+        shpc->config[SHPC_SEC_BUS] &= ~SHPC_SEC_BUS_MASK;
+        shpc->config[SHPC_SEC_BUS] |= speed;
+        break;
+    default:
+	pci_word_test_and_set_mask(shpc->config + SHPC_CMD_STATUS,
+				   SHPC_CMD_STATUS_INVALID_MODE);
+    }
+}
+
+void shpc_reset(PCIDevice *d)
+{
+    SHPCDevice *shpc = d->shpc;
+    int nslots = shpc->nslots;
+    int i;
+    memset(shpc->config, 0, SHPC_SIZEOF(d));
+    pci_set_byte(shpc->config + SHPC_NSLOTS, nslots);
+    pci_set_long(shpc->config + SHPC_SLOTS_33, nslots);
+    pci_set_long(shpc->config + SHPC_SLOTS_66, 0);
+    pci_set_byte(shpc->config + SHPC_FIRST_DEV, SHPC_IDX_TO_PCI(0));
+    pci_set_word(shpc->config + SHPC_PHYS_SLOT,
+		 SHPC_IDX_TO_PHYSICAL(0) |
+		 SHPC_PHYS_NUM_UP |
+		 SHPC_PHYS_MRL |
+		 SHPC_PHYS_BUTTON);
+    pci_set_long(shpc->config + SHPC_SERR_INT, SHPC_INT_DIS |
+                 SHPC_SERR_DIS |
+                 SHPC_CMD_INT_DIS |
+                 SHPC_ARB_SERR_DIS);
+    pci_set_byte(shpc->config + SHPC_PROG_IFC, SHPC_PROG_IFC_1_0);
+    pci_set_word(shpc->config + SHPC_SEC_BUS, SHPC_SEC_BUS_33);
+    for (i = 0; i < shpc->nslots; ++i) {
+        pci_set_word(shpc->config + SHPC_SLOT_EVENT_SERR_INT_DIS(d, i),
+                     SHPC_SLOT_EVENT_PRESENCE |
+                     SHPC_SLOT_EVENT_ISOLATED_FAULT |
+                     SHPC_SLOT_EVENT_BUTTON |
+                     SHPC_SLOT_EVENT_MRL |
+                     SHPC_SLOT_EVENT_CONNECTED_FAULT |
+                     SHPC_SLOT_EVENT_MRL_SERR_DIS |
+                     SHPC_SLOT_EVENT_CONNECTED_FAULT_SERR_DIS);
+        if (shpc->sec_bus->devices[PCI_DEVFN(SHPC_IDX_TO_PCI(i), 0)]) {
+            shpc_set_status(shpc, i, SHPC_STATE_ENABLED, SHPC_SLOT_STATE_MASK);
+            shpc_set_status(shpc, i, 0, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, i, SHPC_SLOT_STATUS_PRSNT_7_5W,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc_set_status(shpc, i, SHPC_LED_ON, SHPC_SLOT_PWR_LED_MASK);
+        } else {
+            shpc_set_status(shpc, i, SHPC_STATE_DISABLED, SHPC_SLOT_STATE_MASK);
+            shpc_set_status(shpc, i, 1, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, i, SHPC_SLOT_STATUS_PRSNT_EMPTY,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc_set_status(shpc, i, SHPC_LED_OFF, SHPC_SLOT_PWR_LED_MASK);
+	}
+        shpc_set_status(shpc, i, 0, SHPC_SLOT_STATUS_66);
+    }
+    shpc_set_sec_bus_speed(shpc, SHPC_SEC_BUS_33);
+    shpc_interrupt_update(d);
+}
+
+static void shpc_invalid_command(SHPCDevice *shpc)
+{
+    pci_word_test_and_set_mask(shpc->config + SHPC_CMD_STATUS,
+                               SHPC_CMD_STATUS_INVALID_CMD);
+}
+
+static void shpc_free_devices_in_slot(SHPCDevice *shpc, int slot)
+{
+    int devfn;
+    int pci_slot = SHPC_IDX_TO_PCI(slot);
+    for (devfn = PCI_DEVFN(pci_slot, 0);
+         devfn <= PCI_DEVFN(pci_slot, PCI_FUNC_MAX - 1);
+         ++devfn) {
+        PCIDevice *affected_dev = shpc->sec_bus->devices[devfn];
+        if (affected_dev) {
+            qdev_free(&affected_dev->qdev);
+        }
+    }
+}
+
+static void shpc_slot_command(SHPCDevice *shpc, uint8_t target,
+                              uint8_t state, uint8_t power, uint8_t attn)
+{
+    uint8_t current_state;
+    int slot = SHPC_LOGICAL_TO_IDX(target);
+    if (target < SHPC_CMD_TRGT_MIN || slot >= shpc->nslots) {
+        shpc_invalid_command(shpc);
+        return;
+    }
+    current_state = shpc_get_status(shpc, slot, SHPC_SLOT_STATE_MASK);
+    if (current_state == SHPC_STATE_ENABLED && state == SHPC_STATE_PWRONLY) {
+        shpc_invalid_command(shpc);
+	return;
+    }
+
+    switch (power) {
+    case SHPC_LED_NO:
+        break;
+    default:
+        /* TODO: send event to monitor */
+        shpc_set_status(shpc, slot, power, SHPC_SLOT_PWR_LED_MASK);
+    }
+    switch (attn) {
+    case SHPC_LED_NO:
+        break;
+    default:
+        /* TODO: send event to monitor */
+        shpc_set_status(shpc, slot, attn, SHPC_SLOT_ATTN_LED_MASK);
+    }
+
+    if ((current_state == SHPC_STATE_DISABLED && state == SHPC_STATE_PWRONLY) ||
+        (current_state == SHPC_STATE_DISABLED && state == SHPC_STATE_ENABLED)) {
+        shpc_set_status(shpc, slot, state, SHPC_SLOT_STATE_MASK);
+    } else if ((current_state == SHPC_STATE_ENABLED ||
+		current_state == SHPC_STATE_PWRONLY) &&
+	       state == SHPC_STATE_DISABLED) {
+        shpc_set_status(shpc, slot, state, SHPC_SLOT_STATE_MASK);
+        power = shpc_get_status(shpc, slot, SHPC_SLOT_PWR_LED_MASK);
+        /* TODO: track what monitor requested. */
+        /* Look at LED to figure out whether it's ok to remove the device. */
+        if (power == SHPC_LED_OFF) {
+            shpc_free_devices_in_slot(shpc, slot);
+            shpc_set_status(shpc, slot, 1, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_EMPTY,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_BUTTON |
+                SHPC_SLOT_EVENT_MRL |
+                SHPC_SLOT_EVENT_PRESENCE;
+        }
+    }
+}
+
+static void shpc_command(SHPCDevice *shpc)
+{
+    uint8_t code = pci_get_byte(shpc->config + SHPC_CMD_CODE);
+    uint8_t speed;
+    uint8_t target;
+    uint8_t attn;
+    uint8_t power;
+    uint8_t state;
+    int i;
+
+    /* Clear status from the previous command. */
+    pci_word_test_and_clear_mask(shpc->config + SHPC_CMD_STATUS,
+				 SHPC_CMD_STATUS_BUSY |
+				 SHPC_CMD_STATUS_MRL_OPEN |
+				 SHPC_CMD_STATUS_INVALID_CMD |
+				 SHPC_CMD_STATUS_INVALID_MODE);
+    switch (code) {
+    case 0x00 ... 0x3f:
+        target = shpc->config[SHPC_CMD_TRGT] & SHPC_CMD_TRGT_MAX;
+        state = (code & SHPC_SLOT_STATE_MASK) >> SHPC_SLOT_STATE_SHIFT;
+        power = (code & SHPC_SLOT_PWR_LED_MASK) >> SHPC_SLOT_PWR_LED_SHIFT;
+        attn = (code & SHPC_SLOT_ATTN_LED_MASK) >> SHPC_SLOT_ATTN_LED_SHIFT;
+        shpc_slot_command(shpc, target, state, power, attn);
+        break;
+    case 0x40 ... 0x47:
+        speed = code & SHPC_SEC_BUS_MASK;
+        shpc_set_sec_bus_speed(shpc, speed);
+        break;
+    case 0x48:
+        /* Power only all slots */
+        /* first verify no slots are enabled */
+        for (i = 0; i < shpc->nslots; ++i) {
+            state = shpc_get_status(shpc, i, SHPC_SLOT_STATE_MASK);
+            if (state == SHPC_STATE_ENABLED) {
+                shpc_invalid_command(shpc);
+                goto done;
+            }
+        }
+        for (i = 0; i < shpc->nslots; ++i) {
+            if (!(shpc_get_status(shpc, i, SHPC_SLOT_STATUS_MRL_OPEN))) {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_PWRONLY, SHPC_LED_ON, SHPC_LED_NO);
+            } else {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_NO, SHPC_LED_OFF, SHPC_LED_NO);
+            }
+        }
+        break;
+    case 0x49:
+        /* Enable all slots */
+        /* TODO: Spec says this shall fail if some are already enabled.
+         * This doesn't make sense - why not? a spec bug? */
+        for (i = 0; i < shpc->nslots; ++i) {
+            state = shpc_get_status(shpc, i, SHPC_SLOT_STATE_MASK);
+            if (state == SHPC_STATE_ENABLED) {
+                shpc_invalid_command(shpc);
+                goto done;
+            }
+        }
+        for (i = 0; i < shpc->nslots; ++i) {
+            if (!(shpc_get_status(shpc, i, SHPC_SLOT_STATUS_MRL_OPEN))) {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_ENABLED, SHPC_LED_ON, SHPC_LED_NO);
+            } else {
+                shpc_slot_command(shpc, i + SHPC_CMD_TRGT_MIN,
+                                  SHPC_STATE_NO, SHPC_LED_OFF, SHPC_LED_NO);
+            }
+        }
+        break;
+    default:
+        shpc_invalid_command(shpc);
+        break;
+    }
+done:
+    pci_long_test_and_set_mask(shpc->config + SHPC_SERR_INT, SHPC_CMD_DETECTED);
+}
+
+static void shpc_write(PCIDevice *d, unsigned addr, uint64_t val, int l)
+{
+    SHPCDevice *shpc = d->shpc;
+    int i;
+    if (addr >= SHPC_SIZEOF(d)) {
+        return;
+    }
+    l = MIN(l, SHPC_SIZEOF(d) - addr);
+
+    /* TODO: code duplicated from pci.c */
+    for (i = 0; i < l; val >>= 8, ++i) {
+        unsigned a = addr + i;
+        uint8_t wmask = shpc->wmask[a];
+        uint8_t w1cmask = shpc->w1cmask[a];
+        assert(!(wmask & w1cmask));
+        shpc->config[a] = (shpc->config[a] & ~wmask) | (val & wmask);
+        shpc->config[a] &= ~(val & w1cmask); /* W1C: Write 1 to Clear */
+    }
+    if (ranges_overlap(addr, l, SHPC_CMD_CODE, 2)) {
+        shpc_command(shpc);
+    }
+    shpc_interrupt_update(d);
+}
+
+static uint64_t shpc_read(PCIDevice *d, unsigned addr, int l)
+{
+    uint64_t val = 0x0;
+    if (addr >= SHPC_SIZEOF(d)) {
+        return val;
+    }
+    l = MIN(l, SHPC_SIZEOF(d) - addr);
+    memcpy(&val, d->shpc->config + addr, l);
+    return val;
+}
+
+/* SHPC Bridge Capability */
+#define SHPC_CAP_LENGTH 0x08
+#define SHPC_CAP_DWORD_SELECT 0x2 /* 1 byte */
+#define SHPC_CAP_CxP 0x3 /* 1 byte: CSP, CIP */
+#define SHPC_CAP_DWORD_DATA 0x4 /* 4 bytes */
+#define SHPC_CAP_CSP_MASK 0x4
+#define SHPC_CAP_CIP_MASK 0x8
+
+static uint8_t shpc_cap_dword(PCIDevice *d)
+{
+    return pci_get_byte(d->config + d->shpc->cap + SHPC_CAP_DWORD_SELECT);
+}
+
+/* Update dword data capability register */
+static void shpc_cap_update_dword(PCIDevice *d)
+{
+    unsigned data;
+    data = shpc_read(d, shpc_cap_dword(d) * 4, 4);
+    pci_set_long(d->config  + d->shpc->cap + SHPC_CAP_DWORD_DATA, data);
+}
+
+/* Add SHPC capability to the config space for the device. */
+static int shpc_cap_add_config(PCIDevice *d)
+{
+    uint8_t *config;
+    int config_offset;
+    config_offset = pci_add_capability(d, PCI_CAP_ID_SHPC,
+                                       0, SHPC_CAP_LENGTH);
+    if (config_offset < 0) {
+        return config_offset;
+    }
+    config = d->config + config_offset;
+
+    pci_set_byte(config + SHPC_CAP_DWORD_SELECT, 0);
+    pci_set_byte(config + SHPC_CAP_CxP, 0);
+    pci_set_long(config + SHPC_CAP_DWORD_DATA, 0);
+    d->shpc->cap = config_offset;
+    /* Make dword select and data writeable. */
+    pci_set_byte(d->wmask + config_offset + SHPC_CAP_DWORD_SELECT, 0xff);
+    pci_set_long(d->wmask + config_offset + SHPC_CAP_DWORD_DATA, 0xffffffff);
+    return 0;
+}
+
+static uint64_t shpc_mmio_read(void *opaque, target_phys_addr_t addr,
+                               unsigned size)
+{
+    return shpc_read(opaque, addr, size);
+}
+
+static void shpc_mmio_write(void *opaque, target_phys_addr_t addr,
+                            uint64_t val, unsigned size)
+{
+    shpc_write(opaque, addr, val, size);
+}
+
+static const MemoryRegionOps shpc_mmio_ops = {
+    .read = shpc_mmio_read,
+    .write = shpc_mmio_write,
+    .endianness = DEVICE_LITTLE_ENDIAN,
+    .valid = {
+        /* SHPC ECN requires dword accesses, but the original 1.0 spec doesn't.
+         * It's easier to suppport all sizes than worry about it. */
+        .min_access_size = 1,
+        .max_access_size = 4,
+    },
+};
+
+static int shpc_device_hotplug(DeviceState *qdev, PCIDevice *affected_dev,
+                               PCIHotplugState hotplug_state)
+{
+    int pci_slot = PCI_SLOT(affected_dev->devfn);
+    uint8_t state;
+    uint8_t led;
+    PCIDevice *d = DO_UPCAST(PCIDevice, qdev, qdev);
+    SHPCDevice *shpc = d->shpc;
+    int slot = SHPC_PCI_TO_IDX(pci_slot);
+    if (pci_slot < SHPC_IDX_TO_PCI(0) || slot >= shpc->nslots) {
+        error_report("Unsupported PCI slot %d for standard hotplug "
+                     "controller. Valid slots are between %d and %d.",
+                     pci_slot, SHPC_IDX_TO_PCI(0),
+                     SHPC_IDX_TO_PCI(shpc->nslots) - 1);
+        return -1;
+    }
+    /* Don't send event when device is enabled during qemu machine creation:
+     * it is present on boot, no hotplug event is necessary. We do send an
+     * event when the device is disabled later. */
+    if (hotplug_state == PCI_COLDPLUG_ENABLED) {
+        shpc_set_status(shpc, slot, 0, SHPC_SLOT_STATUS_MRL_OPEN);
+        shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_7_5W,
+                        SHPC_SLOT_STATUS_PRSNT_MASK);
+        return 0;
+    }
+    if (hotplug_state == PCI_HOTPLUG_DISABLED) {
+        shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |= SHPC_SLOT_EVENT_BUTTON;
+        state = shpc_get_status(shpc, slot, SHPC_SLOT_STATE_MASK);
+        led = shpc_get_status(shpc, slot, SHPC_SLOT_PWR_LED_MASK);
+        if (state == SHPC_STATE_DISABLED && led == SHPC_LED_OFF) {
+            shpc_free_devices_in_slot(shpc, slot);
+            shpc_set_status(shpc, slot, 1, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_EMPTY,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_MRL |
+                SHPC_SLOT_EVENT_PRESENCE;
+        }
+    } else {
+        /* This could be a cancellation of the previous removal.
+         * We check MRL state to figure out. */
+        if (shpc_get_status(shpc, slot, SHPC_SLOT_STATUS_MRL_OPEN)) {
+            shpc_set_status(shpc, slot, 0, SHPC_SLOT_STATUS_MRL_OPEN);
+            shpc_set_status(shpc, slot, SHPC_SLOT_STATUS_PRSNT_7_5W,
+                            SHPC_SLOT_STATUS_PRSNT_MASK);
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_BUTTON |
+                SHPC_SLOT_EVENT_MRL |
+                SHPC_SLOT_EVENT_PRESENCE;
+        } else {
+            /* Press attention button to cancel removal */
+            shpc->config[SHPC_SLOT_EVENT_LATCH(slot)] |=
+                SHPC_SLOT_EVENT_BUTTON;
+        }
+    }
+    shpc_set_status(shpc, slot, 0, SHPC_SLOT_STATUS_66);
+    shpc_interrupt_update(d);
+    return 0;
+}
+
+/* Initialize the SHPC structure in bridge's BAR. */
+int shpc_init(PCIDevice *d, PCIBus *sec_bus, MemoryRegion *bar, unsigned offset)
+{
+    int i, ret;
+    int nslots = SHPC_MAX_SLOTS; /* TODO: qdev property? */
+    SHPCDevice *shpc = d->shpc = g_malloc0(sizeof(*d->shpc));
+    shpc->sec_bus = sec_bus;
+    ret = shpc_cap_add_config(d);
+    if (ret) {
+        g_free(d->shpc);
+        return ret;
+    }
+    if (nslots < SHPC_MIN_SLOTS) {
+        return 0;
+    }
+    if (nslots > SHPC_MAX_SLOTS ||
+        SHPC_IDX_TO_PCI(nslots) > PCI_SLOT_MAX) {
+        /* TODO: report an error mesage that makes sense. */
+        return -EINVAL;
+    }
+    shpc->nslots = nslots;
+    shpc->config = g_malloc0(SHPC_SIZEOF(d));
+    shpc->cmask = g_malloc0(SHPC_SIZEOF(d));
+    shpc->wmask = g_malloc0(SHPC_SIZEOF(d));
+    shpc->w1cmask = g_malloc0(SHPC_SIZEOF(d));
+
+    shpc_reset(d);
+
+    pci_set_long(shpc->config + SHPC_BASE_OFFSET, offset);
+
+    pci_set_byte(shpc->wmask + SHPC_CMD_CODE, 0xff);
+    pci_set_byte(shpc->wmask + SHPC_CMD_TRGT, SHPC_CMD_TRGT_MAX);
+    pci_set_byte(shpc->wmask + SHPC_CMD_TRGT, SHPC_CMD_TRGT_MAX);
+    pci_set_long(shpc->wmask + SHPC_SERR_INT,
+                 SHPC_INT_DIS |
+                 SHPC_SERR_DIS |
+                 SHPC_CMD_INT_DIS |
+                 SHPC_ARB_SERR_DIS);
+    pci_set_long(shpc->w1cmask + SHPC_SERR_INT,
+		 SHPC_CMD_DETECTED |
+                 SHPC_ARB_DETECTED);
+    for (i = 0; i < nslots; ++i) {
+	    pci_set_byte(shpc->wmask +
+			 SHPC_SLOT_EVENT_SERR_INT_DIS(d, i),
+			 SHPC_SLOT_EVENT_PRESENCE |
+			 SHPC_SLOT_EVENT_ISOLATED_FAULT |
+			 SHPC_SLOT_EVENT_BUTTON |
+			 SHPC_SLOT_EVENT_MRL |
+			 SHPC_SLOT_EVENT_CONNECTED_FAULT |
+			 SHPC_SLOT_EVENT_MRL_SERR_DIS |
+			 SHPC_SLOT_EVENT_CONNECTED_FAULT_SERR_DIS);
+	    pci_set_byte(shpc->w1cmask +
+			 SHPC_SLOT_EVENT_LATCH(i),
+			 SHPC_SLOT_EVENT_PRESENCE |
+			 SHPC_SLOT_EVENT_ISOLATED_FAULT |
+			 SHPC_SLOT_EVENT_BUTTON |
+			 SHPC_SLOT_EVENT_MRL |
+			 SHPC_SLOT_EVENT_CONNECTED_FAULT);
+    }
+
+    /* TODO: init cmask */
+    memory_region_init_io(&shpc->mmio, &shpc_mmio_ops, d, "shpc-mmio",
+                          SHPC_SIZEOF(d));
+    shpc_cap_update_dword(d);
+    memory_region_add_subregion(bar, offset, &shpc->mmio);
+    pci_bus_hotplug(sec_bus, shpc_device_hotplug, &d->qdev);
+    return 0;
+}
+
+int shpc_bar_size(PCIDevice *d)
+{
+    return roundup_pow_of_two(SHPC_SLOT_REG(SHPC_MAX_SLOTS));
+}
+
+void shpc_cleanup(PCIDevice *d)
+{
+    SHPCDevice *shpc = d->shpc;
+    /* TODO: cleanup config space changes? */
+    g_free(shpc->config);
+    g_free(shpc->cmask);
+    g_free(shpc->wmask);
+    g_free(shpc->w1cmask);
+    memory_region_destroy(&shpc->mmio);
+    g_free(shpc);
+}
+
+void shpc_cap_write_config(PCIDevice *d, uint32_t addr, uint32_t val, int l)
+{
+    if (!ranges_overlap(addr, l, d->shpc->cap, SHPC_CAP_LENGTH)) {
+        return;
+    }
+    fprintf(stderr, "%s: 0x%x 0x%x %d\n", __func__, addr, val, l);
+    if (ranges_overlap(addr, l, d->shpc->cap + SHPC_CAP_DWORD_DATA, 4)) {
+        unsigned dword_data;
+        dword_data = pci_get_long(d->shpc->config + d->shpc->cap
+                                  + SHPC_CAP_DWORD_DATA);
+        shpc_write(d, shpc_cap_dword(d) * 4, dword_data, 4);
+    }
+    /* Update cap dword data in case guest is going to read it. */
+    shpc_cap_update_dword(d);
+}
hw/shpc.h new file +40 -0
new file mode 100644
index 0000000..389b178
@@ -0,0 +1,40 @@
+#ifndef SHPC_H
+#define SHPC_H
+
+#include "qemu-common.h"
+#include "memory.h"
+
+struct SHPCDevice {
+    /* Capability offset in device's config space */
+    int cap;
+
+    /* # of hot-pluggable slots */
+    int nslots;
+
+    /* SHPC WRS: working register set */
+    uint8_t *config;
+
+    /* Used to enable checks on load. Note that writable bits are
+     * never checked even if set in cmask. */
+    uint8_t *cmask;
+
+    /* Used to implement R/W bytes */
+    uint8_t *wmask;
+
+    /* Used to implement RW1C(Write 1 to Clear) bytes */
+    uint8_t *w1cmask;
+
+    /* MMIO for the SHPC BAR */
+    MemoryRegion mmio;
+
+    /* Bus controlled by this SHPC */
+    PCIBus *sec_bus;
+};
+
+void shpc_reset(PCIDevice *d);
+int shpc_bar_size(PCIDevice *dev);
+int shpc_init(PCIDevice *dev, PCIBus *sec_bus, MemoryRegion *bar, unsigned offset);
+void shpc_cleanup(PCIDevice *dev);
+void shpc_cap_write_config(PCIDevice *d, uint32_t addr, uint32_t val, int len);
+
+#endif
qemu-common.h +1 -0
index 9b997f8..4ff9c95 100644
@@ -247,6 +247,7 @@ typedef struct SSIBus SSIBus;
 typedef struct EventNotifier EventNotifier;
 typedef struct VirtIODevice VirtIODevice;
 typedef struct QEMUSGList QEMUSGList;
+typedef struct SHPCDevice SHPCDevice;
 
 typedef uint64_t pcibus_t;
 
Signature · 1 line
1.7.9.111.gf3fb0
Signature · 1 line
yamahata
Signature · 1 line
yamahata
#6 lore raw

Re: [PATCHv2-RFC 1/2] shpc: standard hot plug controller

In reply to Michael S. Tsirkin
Quoted 3 lines · Michael S. Tsirkin

TODO:
- migration support
- fix dependency on pci_internals.h

fix checkpatch warnings

cheers,
Gerd

#7 lore raw

PATCH 2/2 pci: add standard bridge device

In reply to Michael S. Tsirkin

This adds support for a standard pci to pci bridge, enabling support for more than 32 PCI devices in the system. Device hotplug is supported by means of SHPC controller. For guests with an SHPC driver, this allows robust hotplug and even hotplug of nested bridges, up to 31 devices per bridge.

TODO:
- chassis capability support
- migration support
- remove dependency on pci_internals.h

---
 Makefile.objs       |    2 +-
 hw/pci_bridge_dev.c |  136 +++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 137 insertions(+), 1 deletions(-)
 create mode 100644 hw/pci_bridge_dev.c
Makefile.objs +1 -1
index 4546477..e89112c 100644
@@ -193,7 +193,7 @@ hw-obj-$(CONFIG_VIRTIO) += virtio-console.o
 hw-obj-y += usb-libhw.o
 hw-obj-$(CONFIG_VIRTIO_PCI) += virtio-pci.o
 hw-obj-y += fw_cfg.o
-hw-obj-$(CONFIG_PCI) += pci.o pci_bridge.o
+hw-obj-$(CONFIG_PCI) += pci.o pci_bridge.o pci_bridge_dev.o
 hw-obj-$(CONFIG_PCI) += msix.o msi.o
 hw-obj-$(CONFIG_PCI) += shpc.o
 hw-obj-$(CONFIG_PCI) += pci_host.o pcie_host.o
hw/pci_bridge_dev.c new file +136 -0

136 lines not on this page · raw message

Signature · 1 line
1.7.9.111.gf3fb0
#8 lore raw

Re: [PATCHv2-RFC 2/2] pci: add standard bridge device

Hi,

hw/pci_bridge_dev.c line 57

Quoted 6 lines
hw/pci_bridge_dev.c new file +6 -0
+    /* If we don't specify the name, the bus will be addressed as <id>.0, where
+     * id is the parent id.  But it seems more natural to address the bus using
+     * the parent device name. */
+    if (dev->qdev.id && *dev->qdev.id) {
+        br->bus_name = dev->qdev.id;
+    }

That makes the bridge behave different than everybody else. Not a good idea IMHO.

cheers,
Gerd

#9 lore raw

Re: [PATCHv2-RFC 2/2] pci: add standard bridge device

Quoted 9 lines · Gerd Hoffmann

Hi,

hw/pci_bridge_dev.c line 57

Quoted 6 lines
hw/pci_bridge_dev.c new file +6 -0
+    /* If we don't specify the name, the bus will be addressed as <id>.0, where
+     * id is the parent id.  But it seems more natural to address the bus using
+     * the parent device name. */
+    if (dev->qdev.id && *dev->qdev.id) {
+        br->bus_name = dev->qdev.id;
+    }
Quoted 5 lines

That makes the bridge behave different than everybody else. Not a good idea IMHO.

cheers,
Gerd

Everybody else has names built up according to an undocumented scheme which no one can figure out without reading code, so no one uses them. We need to fix that, but there is, generally, no need for these names so it stayed low priority.

With the bridge people must use the id to connect devices to it, so name must be a sane one.

Signature · 1 line
MST
#10 lore raw

Re: [Qemu-devel] [PATCHv2-RFC 2/2] pci: add standard bridge device

Quoted 16 lines · Michael S. Tsirkin
Quoted 9 lines · Gerd Hoffmann

Hi,

hw/pci_bridge_dev.c line 57

Quoted 6 lines
hw/pci_bridge_dev.c new file +6 -0
+    /* If we don't specify the name, the bus will be addressed as <id>.0, where
+     * id is the parent id.  But it seems more natural to address the bus using
+     * the parent device name. */
+    if (dev->qdev.id && *dev->qdev.id) {
+        br->bus_name = dev->qdev.id;
+    }
Quoted 5 lines

That makes the bridge behave different than everybody else. Not a good idea IMHO.

cheers,
Gerd

Quoted 7 lines

Everybody else has names built up according to an undocumented scheme which no one can figure out without reading code, so no one uses them. We need to fix that, but there is, generally, no need for these names so it stayed low priority.

With the bridge people must use the id to connect devices to it, so name must be a sane one.

I just sent the patch making bus id for bridges follow the value set by the user.
That will make the bridge behave in the same way as everybody else :)

Quoted 2 lines
Signature · 1 line
MST
#11 lore raw

Re: [PATCHv2-RFC 2/2] pci: add standard bridge device

In reply to Michael S. Tsirkin
Quoted 6 lines · Michael S. Tsirkin

This adds support for a standard pci to pci bridge, enabling support for more than 32 PCI devices in the system. Device hotplug is supported by means of SHPC controller. For guests with an SHPC driver, this allows robust hotplug and even hotplug of nested bridges, up to 31 devices per bridge.

This seems to not support 64bit prefetchable memory windows, at least linux doesn't think it does, lspci looks like this:

00:10.0 PCI bridge: Red Hat, Inc. Device 0001 (prog-if 00 [Normal decode])
Physical Slot: 16
Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop-
ParErr- Stepping- SERR- FastB2B- DisINTx-
Status: Cap+ 66MHz+ UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort-
<TAbort- <MAbort- >SERR- <PERR- INTx-

        Latency: 0
        Region 0: Memory at f6126000 (32-bit, non-prefetchable) [size=256]
        Bus: primary=00, secondary=01, subordinate=01, sec-latency=0
        Memory behind bridge: f6000000-f60fffff
        Prefetchable memory behind bridge: f8000000-fbffffff
        Secondary status: 66MHz+ FastB2B+ ParErr- DEVSEL=fast >TAbort-

<TAbort- <MAbort- <SERR- <PERR-
BridgeCtl: Parity- SERR- NoISA- VGA- MAbort- >Reset- FastB2B-
PriDiscTmr- SecDiscTmr- DiscTmrStat- DiscTmrSERREn-
Capabilities: [40] Hot-plug capable
Kernel modules: shpchp

Intentional?

cheers,
Gerd

#12 lore raw

Re: [Qemu-devel] [PATCHv2-RFC 2/2] pci: add standard bridge device

Quoted 29 lines · Gerd Hoffmann
Quoted 6 lines · Michael S. Tsirkin

This adds support for a standard pci to pci bridge, enabling support for more than 32 PCI devices in the system. Device hotplug is supported by means of SHPC controller. For guests with an SHPC driver, this allows robust hotplug and even hotplug of nested bridges, up to 31 devices per bridge.

This seems to not support 64bit prefetchable memory windows, at least linux doesn't think it does, lspci looks like this:

00:10.0 PCI bridge: Red Hat, Inc. Device 0001 (prog-if 00 [Normal decode])
Physical Slot: 16
Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop-
ParErr- Stepping- SERR- FastB2B- DisINTx-
Status: Cap+ 66MHz+ UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort-
<TAbort- <MAbort- >SERR- <PERR- INTx-

        Latency: 0
        Region 0: Memory at f6126000 (32-bit, non-prefetchable) [size=256]
        Bus: primary=00, secondary=01, subordinate=01, sec-latency=0
        Memory behind bridge: f6000000-f60fffff
        Prefetchable memory behind bridge: f8000000-fbffffff
        Secondary status: 66MHz+ FastB2B+ ParErr- DEVSEL=fast >TAbort-

<TAbort- <MAbort- <SERR- <PERR-
BridgeCtl: Parity- SERR- NoISA- VGA- MAbort- >Reset- FastB2B-
PriDiscTmr- SecDiscTmr- DiscTmrStat- DiscTmrSERREn-
Capabilities: [40] Hot-plug capable
Kernel modules: shpchp

Quoted 4 lines

Intentional?

cheers,
Gerd

I'll need to check. v3 I am sending out has this code unchanged. What in the above tells you that 64 bit windows are not supported? BAR0 is 32 bit non prefetcheable. As far as I can see linux driver takes no precautions against access combining that can happen with prefetcheable BARs, so non-prefetcheable seems safer. I could make it 64 bit I guess, I just heard that there is some known bug being worked around in
memory region code triggered somehow by 64 bit, and waiting for the dust to settle.

Signature · 1 line
MST
#13 lore raw

Re: [PATCHv2-RFC 2/2] pci: add standard bridge device

Hi,

Quoted 4 lines
Quoted 4 lines

This seems to not support 64bit prefetchable memory windows, at least linux doesn't think it does, lspci looks like this:

00:10.0 PCI bridge: Red Hat, Inc. Device 0001 (prog-if 00 [Normal decode])

[ ... ]

Quoted 4 lines
Quoted 2 lines

Memory behind bridge: f6000000-f60fffff Prefetchable memory behind bridge: f8000000-fbffffff

What in the above tells you that 64 bit windows are not supported?

lspci prints 64bit addresses then, like this:

00:1c.0 PCI bridge: Intel Corporation 82801I (ICH9 Family) PCI Express Port 1 (rev 03) (prog-if 00 [Normal decode])
[ ... ]

        I/O behind bridge: 00008000-00008fff
        Memory behind bridge: c0000000-c01fffff
        Prefetchable memory behind bridge: 00000000c0200000-00000000c03fffff
Quoted 4 lines

BAR0 is 32 bit non prefetcheable. As far as I can see linux driver takes no precautions against access combining that can happen with prefetcheable BARs, so non-prefetcheable seems safer.

I'm not talking about bar #0, but about the prefetchable memory window for devices behind the bridge.

cheers,
Gerd

#14 lore raw

RFC PATCH 1/1 seabios: add OSHP method stub

In reply to Michael S. Tsirkin

To allow guests to load the native SHPC driver for a bridge, we must declare an OSHP method for the appropriate device which lets the OS take control of the SHPC.
As we don't access SHPC at the moment, we don't need to do anything - just report success.

---

src/ssdt-pcihp.dsl +1 -0
index 442e7a8..3f50169 100644
@@ -24,6 +24,7 @@ DefinitionBlock ("ssdt-pcihp.aml", "SSDT", 0x01, "BXPC", "BXSSDTPCIHP", 0x1)
            ACPI_EXTRACT_METHOD_STRING aml_ej0_name      \
            Method (_EJ0, 1) { Return(PCEJ(0x##slot)) }  \
            Name (_SUN, 0x##slot)                        \
+           Method (OSHP, 1) { Return(0x0) }  \
         }
 
         hotplug_slot(03)
#15 lore raw

Re: [PATCH RFC] seabios: add OSHP method stub

Quoted 6 lines · Michael S. Tsirkin

To allow guests to load the native SHPC driver for a bridge, we must declare an OSHP method for the appropriate device which lets the OS take control of the SHPC.
As we don't access SHPC at the moment, we don't need to do anything - just report success.

The patch is fine with me, but since this is really qemu/kvm specific, please provide an ack from one of the qemu/kvm maintainers.

-Kevin

#16 lore raw

Re: [PATCH RFC] seabios: add OSHP method stub

Quoted 8 lines · Kevin O'Connor
Quoted 6 lines · Michael S. Tsirkin

To allow guests to load the native SHPC driver for a bridge, we must declare an OSHP method for the appropriate device which lets the OS take control of the SHPC.
As we don't access SHPC at the moment, we don't need to do anything - just report success.

Quoted 4 lines

The patch is fine with me, but since this is really qemu/kvm specific, please provide an ack from one of the qemu/kvm maintainers.

-Kevin

I expect no problem with this,
though I'm wondering what makes it qemu specific.

#17 lore raw

Re: [PATCH RFC] seabios: add OSHP method stub

Quoted 9 lines · Michael S. Tsirkin
Quoted 8 lines · Kevin O'Connor
Quoted 6 lines · Michael S. Tsirkin

To allow guests to load the native SHPC driver for a bridge, we must declare an OSHP method for the appropriate device which lets the OS take control of the SHPC.
As we don't access SHPC at the moment, we don't need to do anything - just report success.

Quoted 7 lines
Quoted 4 lines

The patch is fine with me, but since this is really qemu/kvm specific, please provide an ack from one of the qemu/kvm maintainers.

-Kevin

I expect no problem with this,
though I'm wondering what makes it qemu specific.

Only kvm/qemu use the ACPI tables in seabios.

In a nutshell, I don't know what a SHPC is (nor OSHP), so I'm looking for an additional Ack.

-Kevin

#18 lore raw

Re: [PATCH RFC] seabios: add OSHP method stub

Quoted 18 lines · Kevin O'Connor
Quoted 9 lines · Michael S. Tsirkin
Quoted 8 lines · Kevin O'Connor
Quoted 6 lines · Michael S. Tsirkin

To allow guests to load the native SHPC driver for a bridge, we must declare an OSHP method for the appropriate device which lets the OS take control of the SHPC.
As we don't access SHPC at the moment, we don't need to do anything - just report success.

Quoted 7 lines
Quoted 4 lines

The patch is fine with me, but since this is really qemu/kvm specific, please provide an ack from one of the qemu/kvm maintainers.

-Kevin

I expect no problem with this,
though I'm wondering what makes it qemu specific.

Quoted 6 lines

Only kvm/qemu use the ACPI tables in seabios.

In a nutshell, I don't know what a SHPC is (nor OSHP), so I'm looking for an additional Ack.

-Kevin

No problem, I'll get an Ack :)
Meanwhile - here's a summary, as far as I understand it.

Originally PCI SIG only defined the electrical
and mechanical requirements from hotplug, no standard software interface. So it needed ACPI to drive device-specific registers to actually do hotplug.
At some point PCISIG defined standard interfaces
for PCI hotplug. There are two of them: standard
hot plug controller (SHPC) for PCI and PCIE hotplug
for Express.

Now an OS can have a standard driver and use it
to activate hotplug functionality. This is OS hotplug (OSHP).

But what about older OSes that do not have this driver? ACPI can support these by driving the SHPC interface itself. This will work but then it can conflict with the new driver in the OS.

To solve the problem, OS tells ACPI to leave SHPC alone. There are two things OS tries to activate to do this, in this order:
1. _OSC for global switch to native SHPC driver
2. OSHP for switch on the given bridge device

If none of the above are present, or if they fail, OS assumes ACPI wants control of the SHPC registers and won't touch them.

Hope above makes sense.

Signature · 1 line
MST
#19 lore raw

Re: [PATCH RFC] seabios: add OSHP method stub

Quoted 15 lines
Quoted 2 lines

In a nutshell, I don't know what a SHPC is (nor OSHP), so I'm looking for an additional Ack.

No problem, I'll get an Ack :)
Meanwhile - here's a summary, as far as I understand it.

Originally PCI SIG only defined the electrical
and mechanical requirements from hotplug, no standard software interface. So it needed ACPI to drive device-specific registers to actually do hotplug.
At some point PCISIG defined standard interfaces
for PCI hotplug. There are two of them: standard
hot plug controller (SHPC) for PCI and PCIE hotplug
for Express.

Quoted 2 lines

Now an OS can have a standard driver and use it
to activate hotplug functionality. This is OS hotplug (OSHP).

So presumably this will work on targets that don't have ACPI? Assuming a competent guest OS of course. Have you tested this?

Paul

#20 lore raw

Re: [PATCH RFC] seabios: add OSHP method stub

Quoted 15 lines · Paul Brook
Quoted 15 lines
Quoted 2 lines

In a nutshell, I don't know what a SHPC is (nor OSHP), so I'm looking for an additional Ack.

No problem, I'll get an Ack :)
Meanwhile - here's a summary, as far as I understand it.

Originally PCI SIG only defined the electrical
and mechanical requirements from hotplug, no standard software interface. So it needed ACPI to drive device-specific registers to actually do hotplug.
At some point PCISIG defined standard interfaces
for PCI hotplug. There are two of them: standard
hot plug controller (SHPC) for PCI and PCIE hotplug
for Express.

Quoted 7 lines
Quoted 2 lines

Now an OS can have a standard driver and use it
to activate hotplug functionality. This is OS hotplug (OSHP).

So presumably this will work on targets that don't have ACPI? Assuming a competent guest OS of course. Have you tested this?

Paul

This being the qemu side of things? I run Linux and verified that it calls OSHP and afterwards, runs the native driver and handles hotplug/unplug without invoking ACPI at all.

It seems that at least the SHPC driver in linux doesn't work if you don't have an acpi table with the OSHP method - not many people run with acpi=off nowdays, so it's probably just a bug.
I'll check how hard it is to fix this.

Signature · 1 line
MST
#21 lore raw

Re: [Qemu-devel] [PATCH RFC] seabios: add OSHP method stub

Quoted 10 lines
Quoted 5 lines
Quoted 2 lines

Now an OS can have a standard driver and use it
to activate hotplug functionality. This is OS hotplug (OSHP).

So presumably this will work on targets that don't have ACPI? Assuming a competent guest OS of course. Have you tested this?

This being the qemu side of things? I run Linux and verified that it calls OSHP and afterwards, runs the native driver and handles hotplug/unplug without invoking ACPI at all.

I mean using your shiny new hotplug PCI-PCI bridge on arm/ppc/mips targets (i.e anything other than x86 PC). From your description it sounds like it *should* work.

Quoted 5 lines

It seems that at least the SHPC driver in linux doesn't work if you don't have an acpi table with the OSHP method - not many people run with acpi=off nowdays, so it's probably just a bug.
I'll check how hard it is to fix this.

Targets other than x86 don't have ACPI to start with.

Paul

#22 lore raw

Re: [Qemu-devel] [PATCH RFC] seabios: add OSHP method stub

Quoted 21 lines · Paul Brook
Quoted 10 lines
Quoted 5 lines
Quoted 2 lines

Now an OS can have a standard driver and use it
to activate hotplug functionality. This is OS hotplug (OSHP).

So presumably this will work on targets that don't have ACPI? Assuming a competent guest OS of course. Have you tested this?

This being the qemu side of things? I run Linux and verified that it calls OSHP and afterwards, runs the native driver and handles hotplug/unplug without invoking ACPI at all.

I mean using your shiny new hotplug PCI-PCI bridge on arm/ppc/mips targets (i.e anything other than x86 PC). From your description it sounds like it *should* work.

Quoted 5 lines

It seems that at least the SHPC driver in linux doesn't work if you don't have an acpi table with the OSHP method - not many people run with acpi=off nowdays, so it's probably just a bug.
I'll check how hard it is to fix this.

Quoted 3 lines

Targets other than x86 don't have ACPI to start with.

Paul

So

#ifdef CONFIG_ACPI
#include <linux/pci-acpi.h>
static inline int get_hp_hw_control_from_firmware(struct pci_dev *dev)

{
        u32 flags = OSC_SHPC_NATIVE_HP_CONTROL;
        return acpi_get_hp_hw_control_from_firmware(dev, flags);
}

#else
#define get_hp_hw_control_from_firmware(dev) (0)
#endif

So if you build your guest without acpi, things should work fine.

Signature · 1 line
MMST
#23 lore raw

Re: [SeaBIOS] [PATCH RFC] seabios: add OSHP method stub

In reply to Michael S. Tsirkin

src/ssdt-pcihp.dsl DefinitionBlock ("ssdt-pcihp.aml", "SSDT", 0x01, "BXPC", "BXSSDTPCIHP", 0x1) · line 28

Quoted 10 lines
src/ssdt-pcihp.dsl +1 -0
index 442e7a8..3f50169 100644
@@ -24,6 +24,7 @@ DefinitionBlock ("ssdt-pcihp.aml", "SSDT", 0x01, "BXPC", "BXSSDTPCIHP", 0x1)
            ACPI_EXTRACT_METHOD_STRING aml_ej0_name      \
            Method (_EJ0, 1) { Return(PCEJ(0x##slot)) }  \
            Name (_SUN, 0x##slot)                        \
+           Method (OSHP, 1) { Return(0x0) }  \
         }

Linux kernel (kernel-3.2.7-1.fc16.x86_64) complains:

ACPI Warning: For \_SB_.PCI0.S10_.OSHP: Insufficient arguments - needs 1, found 0 (20110623/nspredef-321)

cheers,
Gerd

#24 lore raw
In reply to Michael S. Tsirkin

At 02/13/2012 05:15 PM, Michael S. Tsirkin Wrote:

Quoted 8 lines

Here's a new version of the patch. It works for me. Deep nesting of bridges is supported.
You need a small BIOS patch to support the OSHP method if you want hotplug to work. I will post this separately. We'd need a full ACPI driver to make hotplug work for guests without an SHPC driver (e.g. windows XP). Management support will also be needed.

Quoted 7 lines

One small wrinkle is that the pci_addr property wants data in a format bus:device.function which is broken as guests can change bus numbers.
For testing I used the 'addr' property which encodes slot*8+function#. We probably want to extend pci_addr in some way (e.g. :device.function ? Thoughts?).

What about using id+device(slot)+function to set the address?

Quoted 20 lines

The SHPC controller supports up to 31 devices (out of 32 slots) so slot 0 doesn't support hotplug. Non hot-pluggable devices behind the bridge don't work currectly (we'll try to unplug them) so don't do this.
For now I just blocked adding devices in slot 0, in the future it might be possible to add
a non-hotpluggable device there.

Example:

qemu-system-x86_64 -enable-kvm -m 1G
-drive file=/home/mst/rhel6.qcow2
-netdev tap,id=foo,ifname=msttap0,script=/home/mst/ifup,downscript=no,vhost=on

(quoted diff) +0 -2
-device pci-bridge,id=bog
-device virtio-net-pci,netdev=foo,bus=bog,addr=8
Quoted 2 lines

Hot-unplug currently causes qemu to crash, this happens without this patch too, so I'm not worried :)

How to trigger this bug without this patch?

Thanks
Wen Congyang

Quoted 4 lines

New since v1:
hotplug support

#25 lore raw
Quoted 17 lines · Wen Congyang

At 02/13/2012 05:15 PM, Michael S. Tsirkin Wrote:

Quoted 8 lines

Here's a new version of the patch. It works for me. Deep nesting of bridges is supported.
You need a small BIOS patch to support the OSHP method if you want hotplug to work. I will post this separately. We'd need a full ACPI driver to make hotplug work for guests without an SHPC driver (e.g. windows XP). Management support will also be needed.

Quoted 7 lines

One small wrinkle is that the pci_addr property wants data in a format bus:device.function which is broken as guests can change bus numbers.
For testing I used the 'addr' property which encodes slot*8+function#. We probably want to extend pci_addr in some way (e.g. :device.function ? Thoughts?).

Quoted 1 line

What about using id+device(slot)+function to set the address?

That's exactly what this patch does: addr encodes
slot+function.
I was asking about a friendlier format for this.

Quoted 20 lines
Quoted 20 lines

The SHPC controller supports up to 31 devices (out of 32 slots) so slot 0 doesn't support hotplug. Non hot-pluggable devices behind the bridge don't work currectly (we'll try to unplug them) so don't do this.
For now I just blocked adding devices in slot 0, in the future it might be possible to add
a non-hotpluggable device there.

Example:

qemu-system-x86_64 -enable-kvm -m 1G
-drive file=/home/mst/rhel6.qcow2
-netdev tap,id=foo,ifname=msttap0,script=/home/mst/ifup,downscript=no,vhost=on

(quoted diff) +0 -2
-device pci-bridge,id=bog
-device virtio-net-pci,netdev=foo,bus=bog,addr=8
Quoted 7 lines
Quoted 2 lines

Hot-unplug currently causes qemu to crash, this happens without this patch too, so I'm not worried :)

How to trigger this bug without this patch?

Thanks
Wen Congyang

start with
qemu-system-x86_64 -enable-kvm -m 1G
-drive file=/home/mst/rhel6.qcow2
-netdev tap,id=foo,ifname=msttap0,script=/home/mst/ifup,downscript=no,vhost=on

next do:
device_add virtio-net-pci,netdev=foo,id=bla
<wait a bit for guest to notice the device>
device_del bla
<wait for device to go away>
and it will crash on next malloc, to trigger
malloc give another command, e.g.
info pci

Quoted 4 lines
Quoted 4 lines

New since v1:
hotplug support