Native PCI-express HotPlug support.

PCI-express HotPlug support is implemented via bits in the slot
registers of the PCI-express capability of the downstream port along
with an interrupt that triggers when bits in the slot status register
change.

This is implemented for FreeBSD by adding HotPlug support to the
PCI-PCI bridge driver which attaches to the virtual PCI-PCI bridges
representing downstream ports on HotPlug slots. The PCI-PCI bridge
driver registers an interrupt handler to receive HotPlug events. It
also uses the slot registers to determine the current HotPlug state
and drive an internal HotPlug state machine. For simplicty of
implementation, the PCI-PCI bridge device detaches and deletes the
child PCI device when a card is removed from a slot and creates and
attaches a PCI child device when a card is inserted into the slot.

The PCI-PCI bridge driver provides a bus_child_present which claims
that child devices are present on HotPlug-capable slots only when a
card is inserted. Rather than requiring a timeout in the RC for
config accesses to not-present children, the pcib_read/write_config
methods fail all requests when a card is not present (or not yet
ready).

These changes include support for various optional HotPlug
capabilities such as a power controller, mechanical latch,
electro-mechanical interlock, indicators, and an attention button.
It also includes support for devices which require waiting for
command completion events before initiating a subsequent HotPlug
command. However, it has only been tested on ExpressCard systems
which support surprise removal and have none of these optional
capabilities.

PCI-express HotPlug support is conditional on the PCI_HP option
which is enabled by default on arm64, x86, and powerpc.

Reviewed by:	adrian, imp, vangyzen (older versions)
Relnotes:	yes
Differential Revision:	https://reviews.freebsd.org/D6136

11 files changed, 572 insertions, 0 deletions

diff --git a/sys/amd64/conf/GENERIC b/sys/amd64/conf/GENERIC
index 747acc5..2859de5 100644
--- a/sys/amd64/conf/GENERIC
+++ b/sys/amd64/conf/GENERIC
@@ -102,6 +102,7 @@ device		cpufreq
 device		acpi
 options 	ACPI_DMAR
 device		pci
+options 	PCI_HP			# PCI-Express native HotPlug
 options		PCI_IOV			# PCI SR-IOV support
 
 # Floppy drives
diff --git a/sys/arm64/conf/GENERIC b/sys/arm64/conf/GENERIC
index 2b52196..9300e77 100644
--- a/sys/arm64/conf/GENERIC
+++ b/sys/arm64/conf/GENERIC
@@ -96,6 +96,7 @@ device		vtnet
 
 # Bus drivers
 device		pci
+options 	PCI_HP			# PCI-Express native HotPlug
 options 	PCI_IOV		# PCI SR-IOV support
 
 # Ethernet NICs
diff --git a/sys/conf/NOTES b/sys/conf/NOTES
index f1553bf..51e81f1 100644
--- a/sys/conf/NOTES
+++ b/sys/conf/NOTES
@@ -1411,6 +1411,7 @@ options 	MSGBUF_SIZE=40960
 # PCI bus & PCI options:
 #
 device		pci
+options 	PCI_HP			# PCI-Express native HotPlug
 options 	PCI_IOV			# PCI SR-IOV support
 
 
diff --git a/sys/conf/options b/sys/conf/options
index 4d6ee66..61bbaa7 100644
--- a/sys/conf/options
+++ b/sys/conf/options
@@ -177,6 +177,7 @@ NO_SYSCTL_DESCR	opt_global.h
 NSWBUF_MIN	opt_swap.h
 MBUF_PACKET_ZONE_DISABLE	opt_global.h
 PANIC_REBOOT_WAIT_TIME	opt_panic.h
+PCI_HP		opt_pci.h
 PCI_IOV		opt_global.h
 PPC_DEBUG	opt_ppc.h
 PPC_PROBE_CHIPSET	opt_ppc.h
diff --git a/sys/dev/acpica/acpi_pcib_acpi.c b/sys/dev/acpica/acpi_pcib_acpi.c
index e9bcbd1..506e267 100644
--- a/sys/dev/acpica/acpi_pcib_acpi.c
+++ b/sys/dev/acpica/acpi_pcib_acpi.c
@@ -29,6 +29,8 @@
 __FBSDID("$FreeBSD$");
 
 #include "opt_acpi.h"
+#include "opt_pci.h"
+
 #include <sys/param.h>
 #include <sys/bus.h>
 #include <sys/kernel.h>
@@ -312,6 +314,11 @@ acpi_pcib_osc(struct acpi_hpcib_softc *sc)
 	/* Control Field */
 	cap_set[2] = 0;
 
+#ifdef PCI_HP
+	/* Control Field: PCI Express Native Hot Plug */
+	cap_set[2] |= 0x1;
+#endif
+
 	status = acpi_EvaluateOSC(sc->ap_handle, pci_host_bridge_uuid, 1,
 	    nitems(cap_set), cap_set, cap_set, false);
 	if (ACPI_FAILURE(status)) {
diff --git a/sys/dev/pci/pci_pci.c b/sys/dev/pci/pci_pci.c
index 76643bc..e3539f5 100644
--- a/sys/dev/pci/pci_pci.c
+++ b/sys/dev/pci/pci_pci.c
@@ -35,6 +35,8 @@ __FBSDID("$FreeBSD$");
  * PCI:PCI bridge support.
  */
 
+#include "opt_pci.h"
+
 #include <sys/param.h>
 #include <sys/bus.h>
 #include <sys/kernel.h>
@@ -43,6 +45,7 @@ __FBSDID("$FreeBSD$");
 #include <sys/rman.h>
 #include <sys/sysctl.h>
 #include <sys/systm.h>
+#include <sys/taskqueue.h>
 
 #include <dev/pci/pcivar.h>
 #include <dev/pci/pcireg.h>
@@ -67,6 +70,11 @@ static int		pcib_try_enable_ari(device_t pcib, device_t dev);
 static int		pcib_ari_enabled(device_t pcib);
 static void		pcib_ari_decode_rid(device_t pcib, uint16_t rid,
 			    int *bus, int *slot, int *func);
+#ifdef PCI_HP
+static void		pcib_pcie_ab_timeout(void *arg);
+static void		pcib_pcie_cc_timeout(void *arg);
+static void		pcib_pcie_dll_timeout(void *arg);
+#endif
 
 static device_method_t pcib_methods[] = {
     /* Device interface */
@@ -78,6 +86,7 @@ static device_method_t pcib_methods[] = {
     DEVMETHOD(device_resume,		pcib_resume),
 
     /* Bus interface */
+    DEVMETHOD(bus_child_present,	pcib_child_present),
     DEVMETHOD(bus_read_ivar,		pcib_read_ivar),
     DEVMETHOD(bus_write_ivar,		pcib_write_ivar),
     DEVMETHOD(bus_alloc_resource,	pcib_alloc_resource),
@@ -839,6 +848,466 @@ pcib_set_mem_decode(struct pcib_softc *sc)
 }
 #endif
 
+#ifdef PCI_HP
+/*
+ * PCI-express HotPlug support.
+ */
+static void
+pcib_probe_hotplug(struct pcib_softc *sc)
+{
+	device_t dev;
+
+	dev = sc->dev;
+	if (pci_find_cap(dev, PCIY_EXPRESS, NULL) != 0)
+		return;
+
+	if (!(pcie_read_config(dev, PCIER_FLAGS, 2) & PCIEM_FLAGS_SLOT))
+		return;
+
+	sc->pcie_link_cap = pcie_read_config(dev, PCIER_LINK_CAP, 4);
+	sc->pcie_slot_cap = pcie_read_config(dev, PCIER_SLOT_CAP, 4);
+
+	if (sc->pcie_slot_cap & PCIEM_SLOT_CAP_HPC)
+		sc->flags |= PCIB_HOTPLUG;
+}
+
+/*
+ * Send a HotPlug command to the slot control register.  If this slot
+ * uses command completion interrupts, these updates will be buffered
+ * while a previous command is completing.
+ */
+static void
+pcib_pcie_hotplug_command(struct pcib_softc *sc, uint16_t val, uint16_t mask)
+{
+	device_t dev;
+	uint16_t ctl, new;
+
+	dev = sc->dev;
+	if (sc->pcie_slot_cap & PCIEM_SLOT_CAP_NCCS) {
+		ctl = pcie_read_config(dev, PCIER_SLOT_CTL, 2);
+		new = (ctl & ~mask) | val;
+		if (new != ctl)
+			pcie_write_config(dev, PCIER_SLOT_CTL, new, 2);
+		return;
+	}
+
+	if (sc->flags & PCIB_HOTPLUG_CMD_PENDING) {
+		sc->pcie_pending_link_ctl_val &= ~mask;
+		sc->pcie_pending_link_ctl_val |= val;
+		sc->pcie_pending_link_ctl_mask |= mask;
+	} else {
+		ctl = pcie_read_config(dev, PCIER_SLOT_CTL, 2);
+		new = (ctl & ~mask) | val;
+		if (new != ctl) {
+			pcie_write_config(dev, PCIER_SLOT_CTL, ctl, 2);
+			sc->flags |= PCIB_HOTPLUG_CMD_PENDING;
+			if (!cold)
+				callout_reset(&sc->pcie_cc_timer, hz,
+				    pcib_pcie_cc_timeout, sc);
+		}
+	}
+}
+
+static void
+pcib_pcie_hotplug_command_completed(struct pcib_softc *sc)
+{
+	device_t dev;
+	uint16_t ctl, new;
+
+	dev = sc->dev;
+
+	if (bootverbose)
+		device_printf(dev, "Command Completed\n");
+	if (!(sc->flags & PCIB_HOTPLUG_CMD_PENDING))
+		return;
+	if (sc->pcie_pending_link_ctl_mask != 0) {
+		ctl = pcie_read_config(dev, PCIER_SLOT_CTL, 2);
+		new = ctl & ~sc->pcie_pending_link_ctl_mask;
+		new |= sc->pcie_pending_link_ctl_val;
+		if (new != ctl) {
+			pcie_write_config(dev, PCIER_SLOT_CTL, ctl, 2);
+			if (!cold)
+				callout_reset(&sc->pcie_cc_timer, hz,
+				    pcib_pcie_cc_timeout, sc);
+		} else
+			sc->flags &= ~PCIB_HOTPLUG_CMD_PENDING;
+		sc->pcie_pending_link_ctl_mask = 0;
+		sc->pcie_pending_link_ctl_val = 0;
+	} else {
+		callout_stop(&sc->pcie_cc_timer);
+		sc->flags &= ~PCIB_HOTPLUG_CMD_PENDING;
+	}
+}
+
+/*
+ * Returns true if a card is fully inserted from the user's
+ * perspective.  It may not yet be ready for access, but the driver
+ * can now start enabling access if necessary.
+ */
+static bool
+pcib_hotplug_inserted(struct pcib_softc *sc)
+{
+
+	/* Pretend the card isn't present if a detach is forced. */
+	if (sc->flags & PCIB_DETACHING)
+		return (false);
+
+	/* Card must be present in the slot. */
+	if ((sc->pcie_slot_sta & PCIEM_SLOT_STA_PDS) == 0)
+		return (false);
+
+	/* A power fault implicitly turns off power to the slot. */
+	if (sc->pcie_slot_sta & PCIEM_SLOT_STA_PFD)
+		return (false);
+
+	/* If the MRL is disengaged, the slot is powered off. */
+	if (sc->pcie_slot_cap & PCIEM_SLOT_CAP_MRLSP &&
+	    (sc->pcie_slot_sta & PCIEM_SLOT_STA_MRLSS) != 0)
+		return (false);
+
+	return (true);
+}
+
+/*
+ * Returns -1 if the card is fully inserted, powered, and ready for
+ * access.  Otherwise, returns 0.
+ */
+static int
+pcib_hotplug_present(struct pcib_softc *sc)
+{
+	device_t dev;
+
+	dev = sc->dev;
+
+	/* Card must be inserted. */
+	if (!pcib_hotplug_inserted(sc))
+		return (0);
+
+	/*
+	 * Require the Electromechanical Interlock to be engaged if
+	 * present.
+	 */
+	if (sc->pcie_slot_cap & PCIEM_SLOT_CAP_EIP &&
+	    (sc->pcie_slot_sta & PCIEM_SLOT_STA_EIS) == 0)
+		return (0);
+
+	/* Require the Data Link Layer to be active. */
+	if (sc->pcie_link_cap & PCIEM_LINK_CAP_DL_ACTIVE) {
+		if (!(sc->pcie_link_sta & PCIEM_LINK_STA_DL_ACTIVE))
+			return (0);
+	}
+
+	return (-1);
+}
+
+static void
+pcib_pcie_hotplug_update(struct pcib_softc *sc, uint16_t val, uint16_t mask,
+    bool schedule_task)
+{
+	bool card_inserted;
+
+	/* Clear DETACHING if Present Detect has cleared. */
+	if ((sc->pcie_slot_sta & (PCIEM_SLOT_STA_PDC | PCIEM_SLOT_STA_PDS)) ==
+	    PCIEM_SLOT_STA_PDC)
+		sc->flags &= ~PCIB_DETACHING;
+
+	card_inserted = pcib_hotplug_inserted(sc);
+
+	/* Turn the power indicator on if a card is inserted. */
+	if (sc->pcie_slot_cap & PCIEM_SLOT_CAP_PIP) {
+		mask |= PCIEM_SLOT_CTL_PIC;
+		if (card_inserted)
+			val |= PCIEM_SLOT_CTL_PI_ON;
+		else if (sc->flags & PCIB_DETACH_PENDING)
+			val |= PCIEM_SLOT_CTL_PI_BLINK;
+		else
+			val |= PCIEM_SLOT_CTL_PI_OFF;
+	}
+
+	/* Turn the power on via the Power Controller if a card is inserted. */
+	if (sc->pcie_slot_cap & PCIEM_SLOT_CAP_PCP) {
+		mask |= PCIEM_SLOT_CTL_PCC;
+		if (card_inserted)
+			val |= PCIEM_SLOT_CTL_PC_ON;
+		else
+			val |= PCIEM_SLOT_CTL_PC_OFF;
+	}
+
+	/*
+	 * If a card is inserted, enable the Electromechanical
+	 * Interlock.  If a card is not inserted (or we are in the
+	 * process of detaching), disable the Electromechanical
+	 * Interlock.
+	 */
+	if (sc->pcie_slot_cap & PCIEM_SLOT_CAP_EIP) {
+		if (card_inserted !=
+		    !(sc->pcie_slot_sta & PCIEM_SLOT_STA_EIS)) {
+			mask |= PCIEM_SLOT_CTL_EIC;
+			val |= PCIEM_SLOT_CTL_EIC;
+		}
+	}
+
+	/*
+	 * Start a timer to see if the Data Link Layer times out.
+	 * Note that we only start the timer if Presence Detect
+	 * changed on this interrupt.  Stop any scheduled timer if
+	 * the Data Link Layer is active.
+	 */
+	if (sc->pcie_link_cap & PCIEM_LINK_CAP_DL_ACTIVE) {
+		if (card_inserted &&
+		    !(sc->pcie_link_sta & PCIEM_LINK_STA_DL_ACTIVE) &&
+		    sc->pcie_slot_sta & PCIEM_SLOT_STA_PDC) {
+			if (cold)
+				device_printf(sc->dev,
+				    "Data Link Layer inactive\n");
+			else
+				callout_reset(&sc->pcie_dll_timer, hz,
+				    pcib_pcie_dll_timeout, sc);
+		} else if (sc->pcie_link_sta & PCIEM_LINK_STA_DL_ACTIVE)
+			callout_stop(&sc->pcie_dll_timer);
+	}
+
+	pcib_pcie_hotplug_command(sc, val, mask);
+
+	/*
+	 * During attach the child "pci" device is added sychronously;
+	 * otherwise, the task is scheduled to manage the child
+	 * device.
+	 */
+	if (schedule_task &&
+	    (pcib_hotplug_present(sc) != 0) != (sc->child != NULL))
+		taskqueue_enqueue(taskqueue_thread, &sc->pcie_hp_task);
+}
+
+static void
+pcib_pcie_intr(void *arg)
+{
+	struct pcib_softc *sc;
+	device_t dev;
+
+	sc = arg;
+	dev = sc->dev;
+	sc->pcie_slot_sta = pcie_read_config(dev, PCIER_SLOT_STA, 2);
+
+	/* Clear the events just reported. */
+	pcie_write_config(dev, PCIER_SLOT_STA, sc->pcie_slot_sta, 2);
+
+	if (sc->pcie_slot_sta & PCIEM_SLOT_STA_ABP) {
+		if (sc->flags & PCIB_DETACH_PENDING) {	
+			device_printf(dev,
+			    "Attention Button Pressed: Detach Cancelled\n");
+			sc->flags &= ~PCIB_DETACH_PENDING;
+			callout_stop(&sc->pcie_ab_timer);
+		} else {
+			device_printf(dev,
+		    "Attention Button Pressed: Detaching in 5 seconds\n");
+			sc->flags |= PCIB_DETACH_PENDING;
+			callout_reset(&sc->pcie_ab_timer, 5 * hz,
+			    pcib_pcie_ab_timeout, sc);
+		}
+	}
+	if (sc->pcie_slot_sta & PCIEM_SLOT_STA_PFD)
+		device_printf(dev, "Power Fault Detected\n");
+	if (sc->pcie_slot_sta & PCIEM_SLOT_STA_MRLSC)
+		device_printf(dev, "MRL Sensor Changed to %s\n",
+		    sc->pcie_slot_sta & PCIEM_SLOT_STA_MRLSS ? "open" :
+		    "closed");
+	if (bootverbose && sc->pcie_slot_sta & PCIEM_SLOT_STA_PDC)
+		device_printf(dev, "Present Detect Changed to %s\n",
+		    sc->pcie_slot_sta & PCIEM_SLOT_STA_PDS ? "card present" :
+		    "empty");
+	if (sc->pcie_slot_sta & PCIEM_SLOT_STA_CC)
+		pcib_pcie_hotplug_command_completed(sc);
+	if (sc->pcie_slot_sta & PCIEM_SLOT_STA_DLLSC) {
+		sc->pcie_link_sta = pcie_read_config(dev, PCIER_LINK_STA, 2);
+		if (bootverbose)
+			device_printf(dev,
+			    "Data Link Layer State Changed to %s\n",
+			    sc->pcie_link_sta & PCIEM_LINK_STA_DL_ACTIVE ?
+			    "active" : "inactive");
+	}
+
+	pcib_pcie_hotplug_update(sc, 0, 0, true);
+}
+
+static void
+pcib_pcie_hotplug_task(void *context, int pending)
+{
+	struct pcib_softc *sc;
+	device_t dev;
+
+	sc = context;
+	mtx_lock(&Giant);
+	dev = sc->dev;
+	if (pcib_hotplug_present(sc) != 0) {
+		if (sc->child == NULL) {
+			sc->child = device_add_child(dev, "pci", -1);
+			bus_generic_attach(dev);
+		}
+	} else {
+		if (sc->child != NULL) {
+			if (device_delete_child(dev, sc->child) == 0)
+				sc->child = NULL;
+		}
+	}
+	mtx_unlock(&Giant);
+}
+
+static void
+pcib_pcie_ab_timeout(void *arg)
+{
+	struct pcib_softc *sc;
+	device_t dev;
+
+	sc = arg;
+	dev = sc->dev;
+	mtx_assert(&Giant, MA_OWNED);
+	if (sc->flags & PCIB_DETACH_PENDING) {
+		sc->flags |= PCIB_DETACHING;
+		sc->flags &= ~PCIB_DETACH_PENDING;
+		pcib_pcie_hotplug_update(sc, 0, 0, true);
+	}
+}
+
+static void
+pcib_pcie_cc_timeout(void *arg)
+{
+	struct pcib_softc *sc;
+	device_t dev;
+
+	sc = arg;
+	dev = sc->dev;
+	mtx_assert(&Giant, MA_OWNED);
+	if (sc->flags & PCIB_HOTPLUG_CMD_PENDING) {
+		device_printf(dev,
+		    "Hotplug Command Timed Out - forcing detach\n");
+		sc->flags &= ~(PCIB_HOTPLUG_CMD_PENDING | PCIB_DETACH_PENDING);
+		sc->flags |= PCIB_DETACHING;
+		pcib_pcie_hotplug_update(sc, 0, 0, true);
+	}
+}
+
+static void
+pcib_pcie_dll_timeout(void *arg)
+{
+	struct pcib_softc *sc;
+	device_t dev;
+	uint16_t sta;
+
+	sc = arg;
+	dev = sc->dev;
+	mtx_assert(&Giant, MA_OWNED);
+	sta = pcie_read_config(dev, PCIER_LINK_STA, 2);
+	if (!(sta & PCIEM_LINK_STA_DL_ACTIVE)) {
+		device_printf(dev,
+		    "Timed out waiting for Data Link Layer Active\n");
+		sc->flags |= PCIB_DETACHING;
+		pcib_pcie_hotplug_update(sc, 0, 0, true);
+	} else if (sta != sc->pcie_link_sta) {
+		device_printf(dev,
+		    "Missed HotPlug interrupt waiting for DLL Active\n");
+		pcib_pcie_intr(sc);
+	}
+}
+
+static int
+pcib_alloc_pcie_irq(struct pcib_softc *sc)
+{
+	device_t dev;
+	int count, error, rid;
+
+	rid = -1;
+	dev = sc->dev;
+
+	/*
+	 * For simplicity, only use MSI-X if there is a single message.
+	 * To support a device with multiple messages we would have to
+	 * use remap intr if the MSI number is not 0.
+	 */
+	count = pci_msix_count(dev);
+	if (count == 1) {
+		error = pci_alloc_msix(dev, &count);
+		if (error == 0)
+			rid = 1;
+	}
+
+	if (rid < 0 && pci_msi_count(dev) > 0) {
+		count = 1;
+		error = pci_alloc_msi(dev, &count);
+		if (error == 0)
+			rid = 1;
+	}
+
+	if (rid < 0)
+		rid = 0;
+
+	sc->pcie_irq = bus_alloc_resource_any(dev, SYS_RES_IRQ, &rid,
+	    RF_ACTIVE);
+	if (sc->pcie_irq == NULL) {
+		device_printf(dev,
+		    "Failed to allocate interrupt for PCI-e events\n");
+		if (rid > 0)
+			pci_release_msi(dev);
+		return (ENXIO);
+	}
+
+	error = bus_setup_intr(dev, sc->pcie_irq, INTR_TYPE_MISC,
+	    NULL, pcib_pcie_intr, sc, &sc->pcie_ihand);
+	if (error) {
+		device_printf(dev, "Failed to setup PCI-e interrupt handler\n");
+		bus_release_resource(dev, SYS_RES_IRQ, rid, sc->pcie_irq);
+		if (rid > 0)
+			pci_release_msi(dev);
+		return (error);
+	}
+	return (0);
+}
+
+static void
+pcib_setup_hotplug(struct pcib_softc *sc)
+{
+	device_t dev;
+	uint16_t mask, val;
+
+	dev = sc->dev;
+	callout_init(&sc->pcie_ab_timer, 0);
+	callout_init(&sc->pcie_cc_timer, 0);
+	callout_init(&sc->pcie_dll_timer, 0);
+	TASK_INIT(&sc->pcie_hp_task, 0, pcib_pcie_hotplug_task, sc);
+
+	/* Allocate IRQ. */
+	if (pcib_alloc_pcie_irq(sc) != 0)
+		return;
+
+	sc->pcie_link_sta = pcie_read_config(dev, PCIER_LINK_STA, 2);
+	sc->pcie_slot_sta = pcie_read_config(dev, PCIER_SLOT_STA, 2);
+
+	/* Enable HotPlug events. */
+	mask = PCIEM_SLOT_CTL_DLLSCE | PCIEM_SLOT_CTL_HPIE |
+	    PCIEM_SLOT_CTL_CCIE | PCIEM_SLOT_CTL_PDCE | PCIEM_SLOT_CTL_MRLSCE |
+	    PCIEM_SLOT_CTL_PFDE | PCIEM_SLOT_CTL_ABPE;
+	val = PCIEM_SLOT_CTL_PDCE | PCIEM_SLOT_CTL_HPIE;
+	if (sc->pcie_slot_cap & PCIEM_SLOT_CAP_APB)
+		val |= PCIEM_SLOT_CTL_ABPE;
+	if (sc->pcie_slot_cap & PCIEM_SLOT_CAP_PCP)
+		val |= PCIEM_SLOT_CTL_PFDE;
+	if (sc->pcie_slot_cap & PCIEM_SLOT_CAP_MRLSP)
+		val |= PCIEM_SLOT_CTL_MRLSCE;
+	if (!(sc->pcie_slot_cap & PCIEM_SLOT_CAP_NCCS))
+		val |= PCIEM_SLOT_CTL_CCIE;
+	if (sc->pcie_link_cap & PCIEM_LINK_CAP_DL_ACTIVE)
+		val |= PCIEM_SLOT_CTL_DLLSCE;
+
+	/* Turn the attention indicator off. */
+	if (sc->pcie_slot_cap & PCIEM_SLOT_CAP_AIP) {
+		mask |= PCIEM_SLOT_CTL_AIC;
+		val |= PCIEM_SLOT_CTL_AI_OFF;
+	}
+
+	pcib_pcie_hotplug_update(sc, val, mask, false);
+}
+#endif
+
 /*
  * Get current bridge configuration.
  */
@@ -1017,12 +1486,19 @@ pcib_attach_common(device_t dev)
       pci_read_config(dev, PCIR_PROGIF, 1) == PCIP_BRIDGE_PCI_SUBTRACTIVE)
 	sc->flags |= PCIB_SUBTRACTIVE;
 
+#ifdef PCI_HP
+    pcib_probe_hotplug(sc);
+#endif
 #ifdef NEW_PCIB
 #ifdef PCI_RES_BUS
     pcib_setup_secbus(dev, &sc->bus, 1);
 #endif
     pcib_probe_windows(sc);
 #endif
+#ifdef PCI_HP
+    if (sc->flags & PCIB_HOTPLUG)
+	    pcib_setup_hotplug(sc);
+#endif
     if (bootverbose) {
 	device_printf(dev, "  domain            %d\n", sc->domain);
 	device_printf(dev, "  secondary bus     %d\n", sc->bus.sec);
@@ -1074,6 +1550,17 @@ pcib_attach_common(device_t dev)
     pci_enable_busmaster(dev);
 }
 
+#ifdef PCI_HP
+static int
+pcib_present(struct pcib_softc *sc)
+{
+
+	if (sc->flags & PCIB_HOTPLUG)
+		return (pcib_hotplug_present(sc) != 0);
+	return (1);
+}
+#endif
+
 int
 pcib_attach_child(device_t dev)
 {
@@ -1085,6 +1572,13 @@ pcib_attach_child(device_t dev)
 		return(0);
 	}
 
+#ifdef PCI_HP
+	if (!pcib_present(sc)) {
+		/* An empty HotPlug slot, so don't add a PCI bus yet. */
+		return (0);
+	}
+#endif
+
 	sc->child = device_add_child(dev, "pci", -1);
 	return (bus_generic_attach(dev));
 }
@@ -1129,6 +1623,22 @@ pcib_bridge_init(device_t dev)
 }
 
 int
+pcib_child_present(device_t dev, device_t child)
+{
+#ifdef PCI_HP
+	struct pcib_softc *sc = device_get_softc(dev);
+	int retval;
+
+	retval = bus_child_present(dev);
+	if (retval != 0 && sc->flags & PCIB_HOTPLUG)
+		retval = pcib_hotplug_present(sc);
+	return (retval);
+#else
+	return (bus_child_present(dev));
+#endif
+}
+
+int
 pcib_read_ivar(device_t dev, device_t child, int which, uintptr_t *result)
 {
     struct pcib_softc	*sc = device_get_softc(dev);
@@ -1909,7 +2419,21 @@ pcib_ari_decode_rid(device_t pcib, uint16_t rid, int *bus, int *slot,
 static uint32_t
 pcib_read_config(device_t dev, u_int b, u_int s, u_int f, u_int reg, int width)
 {
+#ifdef PCI_HP
+	struct pcib_softc *sc;
 
+	sc = device_get_softc(dev);
+	if (!pcib_present(sc)) {
+		switch (width) {
+		case 2:
+			return (0xffff);
+		case 1:
+			return (0xff);
+		default:
+			return (0xffffffff);
+		}
+	}
+#endif
 	pcib_xlate_ari(dev, b, &s, &f);
 	return(PCIB_READ_CONFIG(device_get_parent(device_get_parent(dev)), b, s,
 	    f, reg, width));
@@ -1918,7 +2442,13 @@ pcib_read_config(device_t dev, u_int b, u_int s, u_int f, u_int reg, int width)
 static void
 pcib_write_config(device_t dev, u_int b, u_int s, u_int f, u_int reg, uint32_t val, int width)
 {
+#ifdef PCI_HP
+	struct pcib_softc *sc;
 
+	sc = device_get_softc(dev);
+	if (!pcib_present(sc))
+		return;
+#endif
 	pcib_xlate_ari(dev, b, &s, &f);
 	PCIB_WRITE_CONFIG(device_get_parent(device_get_parent(dev)), b, s, f,
 	    reg, val, width);
diff --git a/sys/dev/pci/pcib_private.h b/sys/dev/pci/pcib_private.h
index 93d9a71..c209162 100644
--- a/sys/dev/pci/pcib_private.h
+++ b/sys/dev/pci/pcib_private.h
@@ -33,6 +33,9 @@
 #ifndef __PCIB_PRIVATE_H__
 #define	__PCIB_PRIVATE_H__
 
+#include <sys/_callout.h>
+#include <sys/_task.h>
+
 #ifdef NEW_PCIB
 /*
  * Data structure and routines that Host to PCI bridge drivers can use
@@ -107,6 +110,10 @@ struct pcib_softc
 #define	PCIB_DISABLE_MSI	0x2
 #define	PCIB_DISABLE_MSIX	0x4
 #define	PCIB_ENABLE_ARI		0x8
+#define	PCIB_HOTPLUG		0x10
+#define	PCIB_HOTPLUG_CMD_PENDING 0x20
+#define	PCIB_DETACH_PENDING	0x40
+#define	PCIB_DETACHING		0x80
     u_int	domain;		/* domain number */
     u_int	pribus;		/* primary bus number */
     struct pcib_secbus bus;	/* secondary bus numbers */
@@ -123,6 +130,18 @@ struct pcib_softc
     uint32_t	iolimit;	/* topmost address of port window */
 #endif
     uint16_t	bridgectl;	/* bridge control register */
+    uint16_t	pcie_link_sta;
+    uint16_t	pcie_slot_sta;
+    uint32_t	pcie_link_cap;
+    uint32_t	pcie_slot_cap;
+    uint16_t	pcie_pending_link_ctl_mask;
+    uint16_t	pcie_pending_link_ctl_val;
+    struct resource *pcie_irq;
+    void	*pcie_ihand;
+    struct task	pcie_hp_task;
+    struct callout pcie_ab_timer;
+    struct callout pcie_cc_timer;
+    struct callout pcie_dll_timer;
 };
 
 #define	PCIB_SUPPORTED_ARI_VER	1
@@ -151,6 +170,7 @@ void		pcib_bridge_init(device_t dev);
 #ifdef NEW_PCIB
 const char	*pcib_child_name(device_t child);
 #endif
+int		pcib_child_present(device_t dev, device_t child);
 int		pcib_read_ivar(device_t dev, device_t child, int which, uintptr_t *result);
 int		pcib_write_ivar(device_t dev, device_t child, int which, uintptr_t value);
 struct resource *pcib_alloc_resource(device_t dev, device_t child, int type, int *rid, 
diff --git a/sys/dev/pci/pcireg.h b/sys/dev/pci/pcireg.h
index 68b5378..d463b7a 100644
--- a/sys/dev/pci/pcireg.h
+++ b/sys/dev/pci/pcireg.h
@@ -850,8 +850,16 @@
 #define	PCIEM_SLOT_CTL_CCIE		0x0010
 #define	PCIEM_SLOT_CTL_HPIE		0x0020
 #define	PCIEM_SLOT_CTL_AIC		0x00c0
+#define	PCIEM_SLOT_CTL_AI_ON		0x0040
+#define	PCIEM_SLOT_CTL_AI_BLINK		0x0080
+#define	PCIEM_SLOT_CTL_AI_OFF		0x00c0
 #define	PCIEM_SLOT_CTL_PIC		0x0300
+#define	PCIEM_SLOT_CTL_PI_ON		0x0100
+#define	PCIEM_SLOT_CTL_PI_BLINK		0x0200
+#define	PCIEM_SLOT_CTL_PI_OFF		0x0300
 #define	PCIEM_SLOT_CTL_PCC		0x0400
+#define	PCIEM_SLOT_CTL_PC_ON		0x0000
+#define	PCIEM_SLOT_CTL_PC_OFF		0x0400
 #define	PCIEM_SLOT_CTL_EIC		0x0800
 #define	PCIEM_SLOT_CTL_DLLSCE		0x1000
 #define	PCIER_SLOT_STA		0x1a
diff --git a/sys/i386/conf/GENERIC b/sys/i386/conf/GENERIC
index ee2df93..aefc507 100644
--- a/sys/i386/conf/GENERIC
+++ b/sys/i386/conf/GENERIC
@@ -101,6 +101,7 @@ device		cpufreq
 # Bus support.
 device		acpi
 device		pci
+options 	PCI_HP			# PCI-Express native HotPlug
 options		PCI_IOV			# PCI SR-IOV support
 
 # Floppy drives
diff --git a/sys/powerpc/conf/GENERIC b/sys/powerpc/conf/GENERIC
index f67e5e5..d464ac7 100644
--- a/sys/powerpc/conf/GENERIC
+++ b/sys/powerpc/conf/GENERIC
@@ -102,6 +102,7 @@ device		cpufreq
 
 # Standard busses
 device		pci
+options 	PCI_HP			# PCI-Express native HotPlug
 device		agp
 
 # ATA controllers
diff --git a/sys/powerpc/conf/GENERIC64 b/sys/powerpc/conf/GENERIC64
index 847101f..88d4811 100644
--- a/sys/powerpc/conf/GENERIC64
+++ b/sys/powerpc/conf/GENERIC64
@@ -99,6 +99,7 @@ device		cpufreq
 
 # Standard busses
 device		pci
+options 	PCI_HP			# PCI-Express native HotPlug
 device		agp
 
 # ATA controllers