Merge tag 'powerpc-4.8-1' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux

Pull powerpc updates from Michael Ellerman:
 "Highlights:
   - PowerNV PCI hotplug support.
   - Lots more Power9 support.
   - eBPF JIT support on ppc64le.
   - Lots of cxl updates.
   - Boot code consolidation.

  Bug fixes:
   - Fix spin_unlock_wait() from Boqun Feng
   - Fix stack pointer corruption in __tm_recheckpoint() from Michael
     Neuling
   - Fix multiple bugs in memory_hotplug_max() from Bharata B Rao
   - mm: Ensure "special" zones are empty from Oliver O'Halloran
   - ftrace: Separate the heuristics for checking call sites from
     Michael Ellerman
   - modules: Never restore r2 for a mprofile-kernel style mcount() call
     from Michael Ellerman
   - Fix endianness when reading TCEs from Alexey Kardashevskiy
   - start rtasd before PCI probing from Greg Kurz
   - PCI: rpaphp: Fix slot registration for multiple slots under a PHB
     from Tyrel Datwyler
   - powerpc/mm: Add memory barrier in __hugepte_alloc() from Sukadev
     Bhattiprolu

  Cleanups & fixes:
   - Drop support for MPIC in pseries from Rashmica Gupta
   - Define and use PPC64_ELF_ABI_v2/v1 from Michael Ellerman
   - Remove unused symbols in asm-offsets.c from Rashmica Gupta
   - Fix SRIOV not building without EEH enabled from Russell Currey
   - Remove kretprobe_trampoline_holder from Thiago Jung Bauermann
   - Reduce log level of PCI I/O space warning from Benjamin
     Herrenschmidt
   - Add array bounds checking to crash_shutdown_handlers from Suraj
     Jitindar Singh
   - Avoid -maltivec when using clang integrated assembler from Anton
     Blanchard
   - Fix array overrun in ppc_rtas() syscall from Andrew Donnellan
   - Fix error return value in cmm_mem_going_offline() from Rasmus
     Villemoes
   - export cpu_to_core_id() from Mauricio Faria de Oliveira
   - Remove old symbols from defconfigs from Andrew Donnellan
   - Update obsolete comments in setup_32.c about entry conditions from
     Benjamin Herrenschmidt
   - Add comment explaining the purpose of setup_kdump_trampoline() from
     Benjamin Herrenschmidt
   - Merge the RELOCATABLE config entries for ppc32 and ppc64 from Kevin
     Hao
   - Remove RELOCATABLE_PPC32 from Kevin Hao
   - Fix .long's in tlb-radix.c to more meaningful from Balbir Singh

  Minor cleanups & fixes:
   - Andrew Donnellan, Anna-Maria Gleixner, Anton Blanchard, Benjamin
     Herrenschmidt, Bharata B Rao, Christophe Leroy, Colin Ian King,
     Geliang Tang, Greg Kurz, Madhavan Srinivasan, Michael Ellerman,
     Michael Ellerman, Stephen Rothwell, Stewart Smith.

  Freescale updates from Scott:
   - "Highlights include more 8xx optimizations, device tree updates,
     and MVME7100 support."

  PowerNV PCI hotplug from Gavin Shan:
   - PCI: Add pcibios_setup_bridge()
   - Override pcibios_setup_bridge()
   - Remove PCI_RESET_DELAY_US
   - Move pnv_pci_ioda_setup_opal_tce_kill() around
   - Increase PE# capacity
   - Allocate PE# in reverse order
   - Create PEs in pcibios_setup_bridge()
   - Setup PE for root bus
   - Extend PCI bridge resources
   - Make pnv_ioda_deconfigure_pe() visible
   - Dynamically release PE
   - Update bridge windows on PCI plug
   - Delay populating pdn
   - Support PCI slot ID
   - Use PCI slot reset infrastructure
   - Introduce pnv_pci_get_slot_id()
   - Functions to get/set PCI slot state
   - PCI/hotplug: PowerPC PowerNV PCI hotplug driver
   - Print correct PHB type names

  Power9 idle support from Shreyas B. Prabhu:
   - set power_save func after the idle states are initialized
   - Use PNV_THREAD_WINKLE macro while requesting for winkle
   - make hypervisor state restore a function
   - Rename idle_power7.S to idle_book3s.S
   - Rename reusable idle functions to hardware agnostic names
   - Make pnv_powersave_common more generic
   - abstraction for saving SPRs before entering deep idle states
   - Add platform support for stop instruction
   - cpuidle/powernv: Use CPUIDLE_STATE_MAX instead of MAX_POWERNV_IDLE_STATES
   - cpuidle/powernv: cleanup cpuidle-powernv.c
   - cpuidle/powernv: Add support for POWER ISA v3 idle states
   - Use deepest stop state when cpu is offlined

  Power9 PMU from Madhavan Srinivasan:
   - factor out power8 pmu macros and defines
   - factor out power8 pmu functions
   - factor out power8 __init_pmu code
   - Add power9 event list macros for generic and cache events
   - Power9 PMU support
   - Export Power9 generic and cache events to sysfs

  Power9 preliminary interrupt & PCI support from Benjamin Herrenschmidt:
   - Add XICS emulation APIs
   - Move a few exception common handlers to make room
   - Add support for HV virtualization interrupts
   - Add mechanism to force a replay of interrupts
   - Add ICP OPAL backend
   - Discover IODA3 PHBs
   - pci: Remove obsolete SW invalidate
   - opal: Add real mode call wrappers
   - Rename TCE invalidation calls
   - Remove SWINV constants and obsolete TCE code
   - Rework accessing the TCE invalidate register
   - Fallback to OPAL for TCE invalidations
   - Use the device-tree to get available range of M64's
   - Check status of a PHB before using it
   - pci: Don't try to allocate resources that will be reassigned

  Other Power9:
   - Send SIGBUS on unaligned copy and paste from Chris Smart
   - Large Decrementer support from Oliver O'Halloran
   - Load Monitor Register Support from Jack Miller

  Performance improvements from Anton Blanchard:
   - Avoid load hit store in __giveup_fpu() and __giveup_altivec()
   - Avoid load hit store in setup_sigcontext()
   - Remove assembly versions of strcpy, strcat, strlen and strcmp
   - Align hot loops of some string functions

  eBPF JIT from Naveen N. Rao:
   - Fix/enhance 32-bit Load Immediate implementation
   - Optimize 64-bit Immediate loads
   - Introduce rotate immediate instructions
   - A few cleanups
   - Isolate classic BPF JIT specifics into a separate header
   - Implement JIT compiler for extended BPF

  Operator Panel driver from Suraj Jitindar Singh:
   - devicetree/bindings: Add binding for operator panel on FSP machines
   - Add inline function to get rc from an ASYNC_COMP opal_msg
   - Add driver for operator panel on FSP machines

  Sparse fixes from Daniel Axtens:
   - make some things static
   - Introduce asm-prototypes.h
   - Include headers containing prototypes
   - Use #ifdef __BIG_ENDIAN__ #else for REG_BYTE
   - kvm: Clarify __user annotations
   - Pass endianness to sparse
   - Make ppc_md.{halt, restart} __noreturn

  MM fixes & cleanups from Aneesh Kumar K.V:
   - radix: Update LPCR HR bit as per ISA
   - use _raw variant of page table accessors
   - Compile out radix related functions if RADIX_MMU is disabled
   - Clear top 16 bits of va only on older cpus
   - Print formation regarding the the MMU mode
   - hash: Update SDR1 size encoding as documented in ISA 3.0
   - radix: Update PID switch sequence
   - radix: Update machine call back to support new HCALL.
   - radix: Add LPID based tlb flush helpers
   - radix: Add a kernel command line to disable radix
   - Cleanup LPCR defines

  Boot code consolidation from Benjamin Herrenschmidt:
   - Move epapr_paravirt_early_init() to early_init_devtree()
   - cell: Don't use flat device-tree after boot
   - ge_imp3a: Don't use the flat device-tree after boot
   - mpc85xx_ds: Don't use the flat device-tree after boot
   - mpc85xx_rdb: Don't use the flat device-tree after boot
   - Don't test for machine type in rtas_initialize()
   - Don't test for machine type in smp_setup_cpu_maps()
   - dt: Add of_device_compatible_match()
   - Factor do_feature_fixup calls
   - Move 64-bit feature fixup earlier
   - Move 64-bit memory reserves to setup_arch()
   - Use a cachable DART
   - Move FW feature probing out of pseries probe()
   - Put exception configuration in a common place
   - Remove early allocation of the SMU command buffer
   - Move MMU backend selection out of platform code
   - pasemi: Remove IOBMAP allocation from platform probe()
   - mm/hash: Don't use machine_is() early during boot
   - Don't test for machine type to detect HEA special case
   - pmac: Remove spurrious machine type test
   - Move hash table ops to a separate structure
   - Ensure that ppc_md is empty before probing for machine type
   - Move 64-bit probe_machine() to later in the boot process
   - Move 32-bit probe() machine to later in the boot process
   - Get rid of ppc_md.init_early()
   - Move the boot time info banner to a separate function
   - Move setting of {i,d}cache_bsize to initialize_cache_info()
   - Move the content of setup_system() to setup_arch()
   - Move cache info inits to a separate function
   - Re-order the call to smp_setup_cpu_maps()
   - Re-order setup_panic()
   - Make a few boot functions __init
   - Merge 32-bit and 64-bit setup_arch()

  Other new features:
   - tty/hvc: Use IRQF_SHARED for OPAL hvc consoles from Sam Mendoza-Jonas
   - tty/hvc: Use opal irqchip interface if available from Sam Mendoza-Jonas
   - powerpc: Add module autoloading based on CPU features from Alastair D'Silva
   - crypto: vmx - Convert to CPU feature based module autoloading from Alastair D'Silva
   - Wake up kopald polling thread before waiting for events from Benjamin Herrenschmidt
   - xmon: Dump ISA 2.06 SPRs from Michael Ellerman
   - xmon: Dump ISA 2.07 SPRs from Michael Ellerman
   - Add a parameter to disable 1TB segs from Oliver O'Halloran
   - powerpc/boot: Add OPAL console to epapr wrappers from Oliver O'Halloran
   - Assign fixed PHB number based on device-tree properties from Guilherme G. Piccoli
   - pseries: Add pseries hotplug workqueue from John Allen
   - pseries: Add support for hotplug interrupt source from John Allen
   - pseries: Use kernel hotplug queue for PowerVM hotplug events from John Allen
   - pseries: Move property cloning into its own routine from Nathan Fontenot
   - pseries: Dynamic add entires to associativity lookup array from Nathan Fontenot
   - pseries: Auto-online hotplugged memory from Nathan Fontenot
   - pseries: Remove call to memblock_add() from Nathan Fontenot

  cxl:
   - Add set and get private data to context struct from Michael Neuling
   - make base more explicitly non-modular from Paul Gortmaker
   - Use for_each_compatible_node() macro from Wei Yongjun
   - Frederic Barrat
   - Abstract the differences between the PSL and XSL
   - Make vPHB device node match adapter's
   - Philippe Bergheaud
   - Add mechanism for delivering AFU driver specific events
   - Ignore CAPI adapters misplaced in switched slots
   - Refine slice error debug messages
   - Andrew Donnellan
   - static-ify variables to fix sparse warnings
   - PCI/hotplug: pnv_php: export symbols and move struct types needed by cxl
   - PCI/hotplug: pnv_php: handle OPAL_PCI_SLOT_OFFLINE power state
   - Add cxl_check_and_switch_mode() API to switch bi-modal cards
   - remove dead Kconfig options
   - fix potential NULL dereference in free_adapter()
   - Ian Munsie
   - Update process element after allocating interrupts
   - Add support for CAPP DMA mode
   - Fix allowing bogus AFU descriptors with 0 maximum processes
   - Fix allocating a minimum of 2 pages for the SPA
   - Fix bug where AFU disable operation had no effect
   - Workaround XSL bug that does not clear the RA bit after a reset
   - Fix NULL pointer dereference on kernel contexts with no AFU interrupts
   - powerpc/powernv: Split cxl code out into a separate file
   - Add cxl_slot_is_supported API
   - Enable bus mastering for devices using CAPP DMA mode
   - Move cxl_afu_get / cxl_afu_put to base
   - Allow a default context to be associated with an external pci_dev
   - Do not create vPHB if there are no AFU configuration records
   - powerpc/powernv: Add support for the cxl kernel api on the real phb
   - Add support for using the kernel API with a real PHB
   - Add kernel APIs to get & set the max irqs per context
   - Add preliminary workaround for CX4 interrupt limitation
   - Add support for interrupts on the Mellanox CX4
   - Workaround PE=0 hardware limitation in Mellanox CX4
   - powerpc/powernv: Fix pci-cxl.c build when CONFIG_MODULES=n

  selftests:
   - Test unaligned copy and paste from Chris Smart
   - Load Monitor Register Tests from Jack Miller
   - Cyril Bur
   - exec() with suspended transaction
   - Use signed long to read perf_event_paranoid
   - Fix usage message in context_switch
   - Fix generation of vector instructions/types in context_switch
   - Michael Ellerman
   - Use "Delta" rather than "Error" in normal output
   - Import Anton's mmap & futex micro benchmarks
   - Add a test for PROT_SAO"

* tag 'powerpc-4.8-1' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux: (263 commits)
  powerpc/mm: Parenthesise IS_ENABLED() in if condition
  tty/hvc: Use opal irqchip interface if available
  tty/hvc: Use IRQF_SHARED for OPAL hvc consoles
  selftests/powerpc: exec() with suspended transaction
  powerpc: Improve comment explaining why we modify VRSAVE
  powerpc/mm: Drop unused externs for hpte_init_beat[_v3]()
  powerpc/mm: Rename hpte_init_lpar() and move the fallback to a header
  powerpc/mm: Fix build break when PPC_NATIVE=n
  crypto: vmx - Convert to CPU feature based module autoloading
  powerpc: Add module autoloading based on CPU features
  powerpc/powernv/ioda: Fix endianness when reading TCEs
  powerpc/mm: Add memory barrier in __hugepte_alloc()
  powerpc/modules: Never restore r2 for a mprofile-kernel style mcount() call
  powerpc/ftrace: Separate the heuristics for checking call sites
  powerpc: Merge 32-bit and 64-bit setup_arch()
  powerpc/64: Make a few boot functions __init
  powerpc: Re-order setup_panic()
  powerpc: Re-order the call to smp_setup_cpu_maps()
  powerpc/32: Move cache info inits to a separate function
  powerpc/64: Move the content of setup_system() to setup_arch()
  ...

16 files changed, 1177 insertions, 186 deletions

diff --git a/drivers/misc/cxl/Kconfig b/drivers/misc/cxl/Kconfig
index 8756d06e2bb8..b75cf830d08a 100644
--- a/drivers/misc/cxl/Kconfig
+++ b/drivers/misc/cxl/Kconfig
@@ -7,11 +7,7 @@ config CXL_BASE
 	default n
 	select PPC_COPRO_BASE
 
-config CXL_KERNEL_API
-	bool
-	default n
-
-config CXL_EEH
+config CXL_AFU_DRIVER_OPS
 	bool
 	default n
 
@@ -19,8 +15,7 @@ config CXL
 	tristate "Support for IBM Coherent Accelerators (CXL)"
 	depends on PPC_POWERNV && PCI_MSI && EEH
 	select CXL_BASE
-	select CXL_KERNEL_API
-	select CXL_EEH
+	select CXL_AFU_DRIVER_OPS
 	default m
 	help
 	  Select this option to enable driver support for IBM Coherent
@@ -33,3 +28,11 @@ config CXL
 	  CAPI adapters are found in POWER8 based systems.
 
 	  If unsure, say N.
+
+config CXL_BIMODAL
+	bool "Support for bi-modal CAPI cards"
+	depends on HOTPLUG_PCI_POWERNV = y && CXL || HOTPLUG_PCI_POWERNV = m && CXL = m
+	default y
+	help
+	  Select this option to enable support for bi-modal CAPI cards, such as
+	  the Mellanox CX-4.
diff --git a/drivers/misc/cxl/Makefile b/drivers/misc/cxl/Makefile
index 8a55c1aa11aa..56e9a4732ef0 100644
--- a/drivers/misc/cxl/Makefile
+++ b/drivers/misc/cxl/Makefile
@@ -3,7 +3,7 @@ ccflags-$(CONFIG_PPC_WERROR)	+= -Werror
 
 cxl-y				+= main.o file.o irq.o fault.o native.o
 cxl-y				+= context.o sysfs.o debugfs.o pci.o trace.o
-cxl-y				+= vphb.o api.o
+cxl-y				+= vphb.o phb.o api.o
 cxl-$(CONFIG_PPC_PSERIES)	+= flash.o guest.o of.o hcalls.o
 obj-$(CONFIG_CXL)		+= cxl.o
 obj-$(CONFIG_CXL_BASE)		+= base.o
diff --git a/drivers/misc/cxl/api.c b/drivers/misc/cxl/api.c
index 6d228ccd884d..f3d34b941f85 100644
--- a/drivers/misc/cxl/api.c
+++ b/drivers/misc/cxl/api.c
@@ -13,6 +13,8 @@
 #include <linux/file.h>
 #include <misc/cxl.h>
 #include <linux/fs.h>
+#include <asm/pnv-pci.h>
+#include <linux/msi.h>
 
 #include "cxl.h"
 
@@ -24,6 +26,8 @@ struct cxl_context *cxl_dev_context_init(struct pci_dev *dev)
 	int rc;
 
 	afu = cxl_pci_to_afu(dev);
+	if (IS_ERR(afu))
+		return ERR_CAST(afu);
 
 	ctx = cxl_context_alloc();
 	if (IS_ERR(ctx)) {
@@ -94,6 +98,42 @@ static irq_hw_number_t cxl_find_afu_irq(struct cxl_context *ctx, int num)
 	return 0;
 }
 
+int _cxl_next_msi_hwirq(struct pci_dev *pdev, struct cxl_context **ctx, int *afu_irq)
+{
+	if (*ctx == NULL || *afu_irq == 0) {
+		*afu_irq = 1;
+		*ctx = cxl_get_context(pdev);
+	} else {
+		(*afu_irq)++;
+		if (*afu_irq > cxl_get_max_irqs_per_process(pdev)) {
+			*ctx = list_next_entry(*ctx, extra_irq_contexts);
+			*afu_irq = 1;
+		}
+	}
+	return cxl_find_afu_irq(*ctx, *afu_irq);
+}
+/* Exported via cxl_base */
+
+int cxl_set_priv(struct cxl_context *ctx, void *priv)
+{
+	if (!ctx)
+		return -EINVAL;
+
+	ctx->priv = priv;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(cxl_set_priv);
+
+void *cxl_get_priv(struct cxl_context *ctx)
+{
+	if (!ctx)
+		return ERR_PTR(-EINVAL);
+
+	return ctx->priv;
+}
+EXPORT_SYMBOL_GPL(cxl_get_priv);
+
 int cxl_allocate_afu_irqs(struct cxl_context *ctx, int num)
 {
 	int res;
@@ -102,7 +142,10 @@ int cxl_allocate_afu_irqs(struct cxl_context *ctx, int num)
 	if (num == 0)
 		num = ctx->afu->pp_irqs;
 	res = afu_allocate_irqs(ctx, num);
-	if (!res && !cpu_has_feature(CPU_FTR_HVMODE)) {
+	if (res)
+		return res;
+
+	if (!cpu_has_feature(CPU_FTR_HVMODE)) {
 		/* In a guest, the PSL interrupt is not multiplexed. It was
 		 * allocated above, and we need to set its handler
 		 */
@@ -110,6 +153,13 @@ int cxl_allocate_afu_irqs(struct cxl_context *ctx, int num)
 		if (hwirq)
 			cxl_map_irq(ctx->afu->adapter, hwirq, cxl_ops->psl_interrupt, ctx, "psl");
 	}
+
+	if (ctx->status == STARTED) {
+		if (cxl_ops->update_ivtes)
+			cxl_ops->update_ivtes(ctx);
+		else WARN(1, "BUG: cxl_allocate_afu_irqs must be called prior to starting the context on this platform\n");
+	}
+
 	return res;
 }
 EXPORT_SYMBOL_GPL(cxl_allocate_afu_irqs);
@@ -323,6 +373,23 @@ struct cxl_context *cxl_fops_get_context(struct file *file)
 }
 EXPORT_SYMBOL_GPL(cxl_fops_get_context);
 
+void cxl_set_driver_ops(struct cxl_context *ctx,
+			struct cxl_afu_driver_ops *ops)
+{
+	WARN_ON(!ops->fetch_event || !ops->event_delivered);
+	atomic_set(&ctx->afu_driver_events, 0);
+	ctx->afu_driver_ops = ops;
+}
+EXPORT_SYMBOL_GPL(cxl_set_driver_ops);
+
+void cxl_context_events_pending(struct cxl_context *ctx,
+				unsigned int new_events)
+{
+	atomic_add(new_events, &ctx->afu_driver_events);
+	wake_up_all(&ctx->wq);
+}
+EXPORT_SYMBOL_GPL(cxl_context_events_pending);
+
 int cxl_start_work(struct cxl_context *ctx,
 		   struct cxl_ioctl_start_work *work)
 {
@@ -390,7 +457,106 @@ EXPORT_SYMBOL_GPL(cxl_perst_reloads_same_image);
 ssize_t cxl_read_adapter_vpd(struct pci_dev *dev, void *buf, size_t count)
 {
 	struct cxl_afu *afu = cxl_pci_to_afu(dev);
+	if (IS_ERR(afu))
+		return -ENODEV;
 
 	return cxl_ops->read_adapter_vpd(afu->adapter, buf, count);
 }
 EXPORT_SYMBOL_GPL(cxl_read_adapter_vpd);
+
+int cxl_set_max_irqs_per_process(struct pci_dev *dev, int irqs)
+{
+	struct cxl_afu *afu = cxl_pci_to_afu(dev);
+	if (IS_ERR(afu))
+		return -ENODEV;
+
+	if (irqs > afu->adapter->user_irqs)
+		return -EINVAL;
+
+	/* Limit user_irqs to prevent the user increasing this via sysfs */
+	afu->adapter->user_irqs = irqs;
+	afu->irqs_max = irqs;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(cxl_set_max_irqs_per_process);
+
+int cxl_get_max_irqs_per_process(struct pci_dev *dev)
+{
+	struct cxl_afu *afu = cxl_pci_to_afu(dev);
+	if (IS_ERR(afu))
+		return -ENODEV;
+
+	return afu->irqs_max;
+}
+EXPORT_SYMBOL_GPL(cxl_get_max_irqs_per_process);
+
+/*
+ * This is a special interrupt allocation routine called from the PHB's MSI
+ * setup function. When capi interrupts are allocated in this manner they must
+ * still be associated with a running context, but since the MSI APIs have no
+ * way to specify this we use the default context associated with the device.
+ *
+ * The Mellanox CX4 has a hardware limitation that restricts the maximum AFU
+ * interrupt number, so in order to overcome this their driver informs us of
+ * the restriction by setting the maximum interrupts per context, and we
+ * allocate additional contexts as necessary so that we can keep the AFU
+ * interrupt number within the supported range.
+ */
+int _cxl_cx4_setup_msi_irqs(struct pci_dev *pdev, int nvec, int type)
+{
+	struct cxl_context *ctx, *new_ctx, *default_ctx;
+	int remaining;
+	int rc;
+
+	ctx = default_ctx = cxl_get_context(pdev);
+	if (WARN_ON(!default_ctx))
+		return -ENODEV;
+
+	remaining = nvec;
+	while (remaining > 0) {
+		rc = cxl_allocate_afu_irqs(ctx, min(remaining, ctx->afu->irqs_max));
+		if (rc) {
+			pr_warn("%s: Failed to find enough free MSIs\n", pci_name(pdev));
+			return rc;
+		}
+		remaining -= ctx->afu->irqs_max;
+
+		if (ctx != default_ctx && default_ctx->status == STARTED) {
+			WARN_ON(cxl_start_context(ctx,
+				be64_to_cpu(default_ctx->elem->common.wed),
+				NULL));
+		}
+
+		if (remaining > 0) {
+			new_ctx = cxl_dev_context_init(pdev);
+			if (!new_ctx) {
+				pr_warn("%s: Failed to allocate enough contexts for MSIs\n", pci_name(pdev));
+				return -ENOSPC;
+			}
+			list_add(&new_ctx->extra_irq_contexts, &ctx->extra_irq_contexts);
+			ctx = new_ctx;
+		}
+	}
+
+	return 0;
+}
+/* Exported via cxl_base */
+
+void _cxl_cx4_teardown_msi_irqs(struct pci_dev *pdev)
+{
+	struct cxl_context *ctx, *pos, *tmp;
+
+	ctx = cxl_get_context(pdev);
+	if (WARN_ON(!ctx))
+		return;
+
+	cxl_free_afu_irqs(ctx);
+	list_for_each_entry_safe(pos, tmp, &ctx->extra_irq_contexts, extra_irq_contexts) {
+		cxl_stop_context(pos);
+		cxl_free_afu_irqs(pos);
+		list_del(&pos->extra_irq_contexts);
+		cxl_release_context(pos);
+	}
+}
+/* Exported via cxl_base */
diff --git a/drivers/misc/cxl/base.c b/drivers/misc/cxl/base.c
index 9b90ec6c07cd..cd54ce6f6230 100644
--- a/drivers/misc/cxl/base.c
+++ b/drivers/misc/cxl/base.c
@@ -54,6 +54,19 @@ static inline void cxl_calls_put(struct cxl_calls *calls) { }
 
 #endif /* CONFIG_CXL_MODULE */
 
+/* AFU refcount management */
+struct cxl_afu *cxl_afu_get(struct cxl_afu *afu)
+{
+	return (get_device(&afu->dev) == NULL) ? NULL : afu;
+}
+EXPORT_SYMBOL_GPL(cxl_afu_get);
+
+void cxl_afu_put(struct cxl_afu *afu)
+{
+	put_device(&afu->dev);
+}
+EXPORT_SYMBOL_GPL(cxl_afu_put);
+
 void cxl_slbia(struct mm_struct *mm)
 {
 	struct cxl_calls *calls;
@@ -93,9 +106,92 @@ int cxl_update_properties(struct device_node *dn,
 }
 EXPORT_SYMBOL_GPL(cxl_update_properties);
 
+/*
+ * API calls into the driver that may be called from the PHB code and must be
+ * built in.
+ */
+bool cxl_pci_associate_default_context(struct pci_dev *dev, struct cxl_afu *afu)
+{
+	bool ret;
+	struct cxl_calls *calls;
+
+	calls = cxl_calls_get();
+	if (!calls)
+		return false;
+
+	ret = calls->cxl_pci_associate_default_context(dev, afu);
+
+	cxl_calls_put(calls);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(cxl_pci_associate_default_context);
+
+void cxl_pci_disable_device(struct pci_dev *dev)
+{
+	struct cxl_calls *calls;
+
+	calls = cxl_calls_get();
+	if (!calls)
+		return;
+
+	calls->cxl_pci_disable_device(dev);
+
+	cxl_calls_put(calls);
+}
+EXPORT_SYMBOL_GPL(cxl_pci_disable_device);
+
+int cxl_next_msi_hwirq(struct pci_dev *pdev, struct cxl_context **ctx, int *afu_irq)
+{
+	int ret;
+	struct cxl_calls *calls;
+
+	calls = cxl_calls_get();
+	if (!calls)
+		return -EBUSY;
+
+	ret = calls->cxl_next_msi_hwirq(pdev, ctx, afu_irq);
+
+	cxl_calls_put(calls);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(cxl_next_msi_hwirq);
+
+int cxl_cx4_setup_msi_irqs(struct pci_dev *pdev, int nvec, int type)
+{
+	int ret;
+	struct cxl_calls *calls;
+
+	calls = cxl_calls_get();
+	if (!calls)
+		return false;
+
+	ret = calls->cxl_cx4_setup_msi_irqs(pdev, nvec, type);
+
+	cxl_calls_put(calls);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(cxl_cx4_setup_msi_irqs);
+
+void cxl_cx4_teardown_msi_irqs(struct pci_dev *pdev)
+{
+	struct cxl_calls *calls;
+
+	calls = cxl_calls_get();
+	if (!calls)
+		return;
+
+	calls->cxl_cx4_teardown_msi_irqs(pdev);
+
+	cxl_calls_put(calls);
+}
+EXPORT_SYMBOL_GPL(cxl_cx4_teardown_msi_irqs);
+
 static int __init cxl_base_init(void)
 {
-	struct device_node *np = NULL;
+	struct device_node *np;
 	struct platform_device *dev;
 	int count = 0;
 
@@ -105,8 +201,7 @@ static int __init cxl_base_init(void)
 	if (cpu_has_feature(CPU_FTR_HVMODE))
 		return 0;
 
-	while ((np = of_find_compatible_node(np, NULL,
-				     "ibm,coherent-platform-facility"))) {
+	for_each_compatible_node(np, NULL, "ibm,coherent-platform-facility") {
 		dev = of_platform_device_create(np, NULL, NULL);
 		if (dev)
 			count++;
@@ -114,5 +209,4 @@ static int __init cxl_base_init(void)
 	pr_devel("Found %d cxl device(s)\n", count);
 	return 0;
 }
-
-module_init(cxl_base_init);
+device_initcall(cxl_base_init);
diff --git a/drivers/misc/cxl/context.c b/drivers/misc/cxl/context.c
index 26d206b1d08c..bdee9a01ef35 100644
--- a/drivers/misc/cxl/context.c
+++ b/drivers/misc/cxl/context.c
@@ -67,6 +67,9 @@ int cxl_context_init(struct cxl_context *ctx, struct cxl_afu *afu, bool master,
 	ctx->pending_fault = false;
 	ctx->pending_afu_err = false;
 
+	INIT_LIST_HEAD(&ctx->irq_names);
+	INIT_LIST_HEAD(&ctx->extra_irq_contexts);
+
 	/*
 	 * When we have to destroy all contexts in cxl_context_detach_all() we
 	 * end up with afu_release_irqs() called from inside a
@@ -87,7 +90,8 @@ int cxl_context_init(struct cxl_context *ctx, struct cxl_afu *afu, bool master,
 	 */
 	mutex_lock(&afu->contexts_lock);
 	idr_preload(GFP_KERNEL);
-	i = idr_alloc(&ctx->afu->contexts_idr, ctx, 0,
+	i = idr_alloc(&ctx->afu->contexts_idr, ctx,
+		      ctx->afu->adapter->native->sl_ops->min_pe,
 		      ctx->afu->num_procs, GFP_NOWAIT);
 	idr_preload_end();
 	mutex_unlock(&afu->contexts_lock);
diff --git a/drivers/misc/cxl/cxl.h b/drivers/misc/cxl/cxl.h
index 4fe50788ff45..de090533f18c 100644
--- a/drivers/misc/cxl/cxl.h
+++ b/drivers/misc/cxl/cxl.h
@@ -24,6 +24,7 @@
 #include <asm/reg.h>
 #include <misc/cxl-base.h>
 
+#include <misc/cxl.h>
 #include <uapi/misc/cxl.h>
 
 extern uint cxl_verbose;
@@ -34,7 +35,7 @@ extern uint cxl_verbose;
  * Bump version each time a user API change is made, whether it is
  * backwards compatible ot not.
  */
-#define CXL_API_VERSION 2
+#define CXL_API_VERSION 3
 #define CXL_API_VERSION_COMPATIBLE 1
 
 /*
@@ -81,6 +82,7 @@ static const cxl_p1_reg_t CXL_PSL_TLBIA   = {0x00A8};
 static const cxl_p1_reg_t CXL_PSL_AFUSEL  = {0x00B0};
 
 /* 0x00C0:7EFF Implementation dependent area */
+/* PSL registers */
 static const cxl_p1_reg_t CXL_PSL_FIR1      = {0x0100};
 static const cxl_p1_reg_t CXL_PSL_FIR2      = {0x0108};
 static const cxl_p1_reg_t CXL_PSL_Timebase  = {0x0110};
@@ -91,6 +93,11 @@ static const cxl_p1_reg_t CXL_PSL_FIR_CNTL  = {0x0148};
 static const cxl_p1_reg_t CXL_PSL_DSNDCTL   = {0x0150};
 static const cxl_p1_reg_t CXL_PSL_SNWRALLOC = {0x0158};
 static const cxl_p1_reg_t CXL_PSL_TRACE     = {0x0170};
+/* XSL registers (Mellanox CX4) */
+static const cxl_p1_reg_t CXL_XSL_Timebase  = {0x0100};
+static const cxl_p1_reg_t CXL_XSL_TB_CTLSTAT = {0x0108};
+static const cxl_p1_reg_t CXL_XSL_FEC       = {0x0158};
+static const cxl_p1_reg_t CXL_XSL_DSNCTL    = {0x0168};
 /* 0x7F00:7FFF Reserved PCIe MSI-X Pending Bit Array area */
 /* 0x8000:FFFF Reserved PCIe MSI-X Table Area */
 
@@ -182,6 +189,18 @@ static const cxl_p2n_reg_t CXL_PSL_WED_An     = {0x0A0};
 #define CXL_PSL_ID_An_F	(1ull << (63-31))
 #define CXL_PSL_ID_An_L	(1ull << (63-30))
 
+/****** CXL_PSL_SERR_An ****************************************************/
+#define CXL_PSL_SERR_An_afuto	(1ull << (63-0))
+#define CXL_PSL_SERR_An_afudis	(1ull << (63-1))
+#define CXL_PSL_SERR_An_afuov	(1ull << (63-2))
+#define CXL_PSL_SERR_An_badsrc	(1ull << (63-3))
+#define CXL_PSL_SERR_An_badctx	(1ull << (63-4))
+#define CXL_PSL_SERR_An_llcmdis	(1ull << (63-5))
+#define CXL_PSL_SERR_An_llcmdto	(1ull << (63-6))
+#define CXL_PSL_SERR_An_afupar	(1ull << (63-7))
+#define CXL_PSL_SERR_An_afudup	(1ull << (63-8))
+#define CXL_PSL_SERR_An_AE	(1ull << (63-30))
+
 /****** CXL_PSL_SCNTL_An ****************************************************/
 #define CXL_PSL_SCNTL_An_CR          (0x1ull << (63-15))
 /* Programming Modes: */
@@ -421,18 +440,6 @@ struct cxl_afu {
 	bool enabled;
 };
 
-/* AFU refcount management */
-static inline struct cxl_afu *cxl_afu_get(struct cxl_afu *afu)
-{
-
-	return (get_device(&afu->dev) == NULL) ? NULL : afu;
-}
-
-static inline void  cxl_afu_put(struct cxl_afu *afu)
-{
-	put_device(&afu->dev);
-}
-
 
 struct cxl_irq_name {
 	struct list_head list;
@@ -477,6 +484,9 @@ struct cxl_context {
 	/* Only used in PR mode */
 	u64 process_token;
 
+	/* driver private data */
+	void *priv;
+
 	unsigned long *irq_bitmap; /* Accessed from IRQ context */
 	struct cxl_irq_ranges irqs;
 	struct list_head irq_names;
@@ -522,7 +532,36 @@ struct cxl_context {
 	bool pending_fault;
 	bool pending_afu_err;
 
+	/* Used by AFU drivers for driver specific event delivery */
+	struct cxl_afu_driver_ops *afu_driver_ops;
+	atomic_t afu_driver_events;
+
 	struct rcu_head rcu;
+
+	/*
+	 * Only used when more interrupts are allocated via
+	 * pci_enable_msix_range than are supported in the default context, to
+	 * use additional contexts to overcome the limitation. i.e. Mellanox
+	 * CX4 only:
+	 */
+	struct list_head extra_irq_contexts;
+};
+
+struct cxl_service_layer_ops {
+	int (*adapter_regs_init)(struct cxl *adapter, struct pci_dev *dev);
+	int (*afu_regs_init)(struct cxl_afu *afu);
+	int (*register_serr_irq)(struct cxl_afu *afu);
+	void (*release_serr_irq)(struct cxl_afu *afu);
+	void (*debugfs_add_adapter_sl_regs)(struct cxl *adapter, struct dentry *dir);
+	void (*debugfs_add_afu_sl_regs)(struct cxl_afu *afu, struct dentry *dir);
+	void (*psl_irq_dump_registers)(struct cxl_context *ctx);
+	void (*err_irq_dump_registers)(struct cxl *adapter);
+	void (*debugfs_stop_trace)(struct cxl *adapter);
+	void (*write_timebase_ctrl)(struct cxl *adapter);
+	u64 (*timebase_read)(struct cxl *adapter);
+	int capi_mode;
+	bool needs_reset_before_disable;
+	int min_pe;
 };
 
 struct cxl_native {
@@ -533,6 +572,7 @@ struct cxl_native {
 	irq_hw_number_t err_hwirq;
 	unsigned int err_virq;
 	u64 ps_off;
+	const struct cxl_service_layer_ops *sl_ops;
 };
 
 struct cxl_guest {
@@ -688,9 +728,21 @@ static inline u64 cxl_p2n_read(struct cxl_afu *afu, cxl_p2n_reg_t reg)
 ssize_t cxl_pci_afu_read_err_buffer(struct cxl_afu *afu, char *buf,
 				loff_t off, size_t count);
 
+/* Internal functions wrapped in cxl_base to allow PHB to call them */
+bool _cxl_pci_associate_default_context(struct pci_dev *dev, struct cxl_afu *afu);
+void _cxl_pci_disable_device(struct pci_dev *dev);
+int _cxl_next_msi_hwirq(struct pci_dev *pdev, struct cxl_context **ctx, int *afu_irq);
+int _cxl_cx4_setup_msi_irqs(struct pci_dev *pdev, int nvec, int type);
+void _cxl_cx4_teardown_msi_irqs(struct pci_dev *pdev);
 
 struct cxl_calls {
 	void (*cxl_slbia)(struct mm_struct *mm);
+	bool (*cxl_pci_associate_default_context)(struct pci_dev *dev, struct cxl_afu *afu);
+	void (*cxl_pci_disable_device)(struct pci_dev *dev);
+	int (*cxl_next_msi_hwirq)(struct pci_dev *pdev, struct cxl_context **ctx, int *afu_irq);
+	int (*cxl_cx4_setup_msi_irqs)(struct pci_dev *pdev, int nvec, int type);
+	void (*cxl_cx4_teardown_msi_irqs)(struct pci_dev *pdev);
+
 	struct module *owner;
 };
 int register_cxl_calls(struct cxl_calls *calls);
@@ -805,6 +857,11 @@ int cxl_tlb_slb_invalidate(struct cxl *adapter);
 int cxl_afu_disable(struct cxl_afu *afu);
 int cxl_psl_purge(struct cxl_afu *afu);
 
+void cxl_debugfs_add_adapter_psl_regs(struct cxl *adapter, struct dentry *dir);
+void cxl_debugfs_add_adapter_xsl_regs(struct cxl *adapter, struct dentry *dir);
+void cxl_debugfs_add_afu_psl_regs(struct cxl_afu *afu, struct dentry *dir);
+void cxl_native_psl_irq_dump_regs(struct cxl_context *ctx);
+void cxl_native_err_irq_dump_regs(struct cxl *adapter);
 void cxl_stop_trace(struct cxl *cxl);
 int cxl_pci_vphb_add(struct cxl_afu *afu);
 void cxl_pci_vphb_remove(struct cxl_afu *afu);
@@ -855,6 +912,7 @@ struct cxl_backend_ops {
 	int (*attach_process)(struct cxl_context *ctx, bool kernel,
 			u64 wed, u64 amr);
 	int (*detach_process)(struct cxl_context *ctx);
+	void (*update_ivtes)(struct cxl_context *ctx);
 	bool (*support_attributes)(const char *attr_name, enum cxl_attrs type);
 	bool (*link_ok)(struct cxl *cxl, struct cxl_afu *afu);
 	void (*release_afu)(struct device *dev);
@@ -879,4 +937,7 @@ extern const struct cxl_backend_ops *cxl_ops;
 
 /* check if the given pci_dev is on the the cxl vphb bus */
 bool cxl_pci_is_vphb_device(struct pci_dev *dev);
+
+/* decode AFU error bits in the PSL register PSL_SERR_An */
+void cxl_afu_decode_psl_serr(struct cxl_afu *afu, u64 serr);
 #endif
diff --git a/drivers/misc/cxl/debugfs.c b/drivers/misc/cxl/debugfs.c
index 5751899e0c17..ec7b8a017439 100644
--- a/drivers/misc/cxl/debugfs.c
+++ b/drivers/misc/cxl/debugfs.c
@@ -51,6 +51,19 @@ static struct dentry *debugfs_create_io_x64(const char *name, umode_t mode,
 	return debugfs_create_file(name, mode, parent, (void __force *)value, &fops_io_x64);
 }
 
+void cxl_debugfs_add_adapter_psl_regs(struct cxl *adapter, struct dentry *dir)
+{
+	debugfs_create_io_x64("fir1", S_IRUSR, dir, _cxl_p1_addr(adapter, CXL_PSL_FIR1));
+	debugfs_create_io_x64("fir2", S_IRUSR, dir, _cxl_p1_addr(adapter, CXL_PSL_FIR2));
+	debugfs_create_io_x64("fir_cntl", S_IRUSR, dir, _cxl_p1_addr(adapter, CXL_PSL_FIR_CNTL));
+	debugfs_create_io_x64("trace", S_IRUSR | S_IWUSR, dir, _cxl_p1_addr(adapter, CXL_PSL_TRACE));
+}
+
+void cxl_debugfs_add_adapter_xsl_regs(struct cxl *adapter, struct dentry *dir)
+{
+	debugfs_create_io_x64("fec", S_IRUSR, dir, _cxl_p1_addr(adapter, CXL_XSL_FEC));
+}
+
 int cxl_debugfs_adapter_add(struct cxl *adapter)
 {
 	struct dentry *dir;
@@ -65,13 +78,10 @@ int cxl_debugfs_adapter_add(struct cxl *adapter)
 		return PTR_ERR(dir);
 	adapter->debugfs = dir;
 
-	debugfs_create_io_x64("fir1",     S_IRUSR, dir, _cxl_p1_addr(adapter, CXL_PSL_FIR1));
-	debugfs_create_io_x64("fir2",     S_IRUSR, dir, _cxl_p1_addr(adapter, CXL_PSL_FIR2));
-	debugfs_create_io_x64("fir_cntl", S_IRUSR, dir, _cxl_p1_addr(adapter, CXL_PSL_FIR_CNTL));
 	debugfs_create_io_x64("err_ivte", S_IRUSR, dir, _cxl_p1_addr(adapter, CXL_PSL_ErrIVTE));
 
-	debugfs_create_io_x64("trace", S_IRUSR | S_IWUSR, dir, _cxl_p1_addr(adapter, CXL_PSL_TRACE));
-
+	if (adapter->native->sl_ops->debugfs_add_adapter_sl_regs)
+		adapter->native->sl_ops->debugfs_add_adapter_sl_regs(adapter, dir);
 	return 0;
 }
 
@@ -80,6 +90,14 @@ void cxl_debugfs_adapter_remove(struct cxl *adapter)
 	debugfs_remove_recursive(adapter->debugfs);
 }
 
+void cxl_debugfs_add_afu_psl_regs(struct cxl_afu *afu, struct dentry *dir)
+{
+	debugfs_create_io_x64("fir", S_IRUSR, dir, _cxl_p1n_addr(afu, CXL_PSL_FIR_SLICE_An));
+	debugfs_create_io_x64("serr", S_IRUSR, dir, _cxl_p1n_addr(afu, CXL_PSL_SERR_An));
+	debugfs_create_io_x64("afu_debug", S_IRUSR, dir, _cxl_p1n_addr(afu, CXL_AFU_DEBUG_An));
+	debugfs_create_io_x64("trace", S_IRUSR | S_IWUSR, dir, _cxl_p1n_addr(afu, CXL_PSL_SLICE_TRACE));
+}
+
 int cxl_debugfs_afu_add(struct cxl_afu *afu)
 {
 	struct dentry *dir;
@@ -94,18 +112,15 @@ int cxl_debugfs_afu_add(struct cxl_afu *afu)
 		return PTR_ERR(dir);
 	afu->debugfs = dir;
 
-	debugfs_create_io_x64("fir",        S_IRUSR, dir, _cxl_p1n_addr(afu, CXL_PSL_FIR_SLICE_An));
-	debugfs_create_io_x64("serr",       S_IRUSR, dir, _cxl_p1n_addr(afu, CXL_PSL_SERR_An));
-	debugfs_create_io_x64("afu_debug",  S_IRUSR, dir, _cxl_p1n_addr(afu, CXL_AFU_DEBUG_An));
 	debugfs_create_io_x64("sr",         S_IRUSR, dir, _cxl_p1n_addr(afu, CXL_PSL_SR_An));
-
 	debugfs_create_io_x64("dsisr",      S_IRUSR, dir, _cxl_p2n_addr(afu, CXL_PSL_DSISR_An));
 	debugfs_create_io_x64("dar",        S_IRUSR, dir, _cxl_p2n_addr(afu, CXL_PSL_DAR_An));
 	debugfs_create_io_x64("sstp0",      S_IRUSR, dir, _cxl_p2n_addr(afu, CXL_SSTP0_An));
 	debugfs_create_io_x64("sstp1",      S_IRUSR, dir, _cxl_p2n_addr(afu, CXL_SSTP1_An));
 	debugfs_create_io_x64("err_status", S_IRUSR, dir, _cxl_p2n_addr(afu, CXL_PSL_ErrStat_An));
 
-	debugfs_create_io_x64("trace", S_IRUSR | S_IWUSR, dir, _cxl_p1n_addr(afu, CXL_PSL_SLICE_TRACE));
+	if (afu->adapter->native->sl_ops->debugfs_add_afu_sl_regs)
+		afu->adapter->native->sl_ops->debugfs_add_afu_sl_regs(afu, dir);
 
 	return 0;
 }
diff --git a/drivers/misc/cxl/file.c b/drivers/misc/cxl/file.c
index eec468f1612f..5fb9894b157f 100644
--- a/drivers/misc/cxl/file.c
+++ b/drivers/misc/cxl/file.c
@@ -293,6 +293,17 @@ int afu_mmap(struct file *file, struct vm_area_struct *vm)
 	return cxl_context_iomap(ctx, vm);
 }
 
+static inline bool ctx_event_pending(struct cxl_context *ctx)
+{
+	if (ctx->pending_irq || ctx->pending_fault || ctx->pending_afu_err)
+		return true;
+
+	if (ctx->afu_driver_ops && atomic_read(&ctx->afu_driver_events))
+		return true;
+
+	return false;
+}
+
 unsigned int afu_poll(struct file *file, struct poll_table_struct *poll)
 {
 	struct cxl_context *ctx = file->private_data;
@@ -305,8 +316,7 @@ unsigned int afu_poll(struct file *file, struct poll_table_struct *poll)
 	pr_devel("afu_poll wait done pe: %i\n", ctx->pe);
 
 	spin_lock_irqsave(&ctx->lock, flags);
-	if (ctx->pending_irq || ctx->pending_fault ||
-	    ctx->pending_afu_err)
+	if (ctx_event_pending(ctx))
 		mask |= POLLIN | POLLRDNORM;
 	else if (ctx->status == CLOSED)
 		/* Only error on closed when there are no futher events pending
@@ -319,16 +329,46 @@ unsigned int afu_poll(struct file *file, struct poll_table_struct *poll)
 	return mask;
 }
 
-static inline int ctx_event_pending(struct cxl_context *ctx)
+static ssize_t afu_driver_event_copy(struct cxl_context *ctx,
+				     char __user *buf,
+				     struct cxl_event *event,
+				     struct cxl_event_afu_driver_reserved *pl)
 {
-	return (ctx->pending_irq || ctx->pending_fault ||
-	    ctx->pending_afu_err || (ctx->status == CLOSED));
+	/* Check event */
+	if (!pl) {
+		ctx->afu_driver_ops->event_delivered(ctx, pl, -EINVAL);
+		return -EFAULT;
+	}
+
+	/* Check event size */
+	event->header.size += pl->data_size;
+	if (event->header.size > CXL_READ_MIN_SIZE) {
+		ctx->afu_driver_ops->event_delivered(ctx, pl, -EINVAL);
+		return -EFAULT;
+	}
+
+	/* Copy event header */
+	if (copy_to_user(buf, event, sizeof(struct cxl_event_header))) {
+		ctx->afu_driver_ops->event_delivered(ctx, pl, -EFAULT);
+		return -EFAULT;
+	}
+
+	/* Copy event data */
+	buf += sizeof(struct cxl_event_header);
+	if (copy_to_user(buf, &pl->data, pl->data_size)) {
+		ctx->afu_driver_ops->event_delivered(ctx, pl, -EFAULT);
+		return -EFAULT;
+	}
+
+	ctx->afu_driver_ops->event_delivered(ctx, pl, 0); /* Success */
+	return event->header.size;
 }
 
 ssize_t afu_read(struct file *file, char __user *buf, size_t count,
 			loff_t *off)
 {
 	struct cxl_context *ctx = file->private_data;
+	struct cxl_event_afu_driver_reserved *pl = NULL;
 	struct cxl_event event;
 	unsigned long flags;
 	int rc;
@@ -344,7 +384,7 @@ ssize_t afu_read(struct file *file, char __user *buf, size_t count,
 
 	for (;;) {
 		prepare_to_wait(&ctx->wq, &wait, TASK_INTERRUPTIBLE);
-		if (ctx_event_pending(ctx))
+		if (ctx_event_pending(ctx) || (ctx->status == CLOSED))
 			break;
 
 		if (!cxl_ops->link_ok(ctx->afu->adapter, ctx->afu)) {
@@ -374,7 +414,12 @@ ssize_t afu_read(struct file *file, char __user *buf, size_t count,
 	memset(&event, 0, sizeof(event));
 	event.header.process_element = ctx->pe;
 	event.header.size = sizeof(struct cxl_event_header);
-	if (ctx->pending_irq) {
+	if (ctx->afu_driver_ops && atomic_read(&ctx->afu_driver_events)) {
+		pr_devel("afu_read delivering AFU driver specific event\n");
+		pl = ctx->afu_driver_ops->fetch_event(ctx);
+		atomic_dec(&ctx->afu_driver_events);
+		event.header.type = CXL_EVENT_AFU_DRIVER;
+	} else if (ctx->pending_irq) {
 		pr_devel("afu_read delivering AFU interrupt\n");
 		event.header.size += sizeof(struct cxl_event_afu_interrupt);
 		event.header.type = CXL_EVENT_AFU_INTERRUPT;
@@ -404,6 +449,9 @@ ssize_t afu_read(struct file *file, char __user *buf, size_t count,
 
 	spin_unlock_irqrestore(&ctx->lock, flags);
 
+	if (event.header.type == CXL_EVENT_AFU_DRIVER)
+		return afu_driver_event_copy(ctx, buf, &event, pl);
+
 	if (copy_to_user(buf, &event, event.header.size))
 		return -EFAULT;
 	return event.header.size;
@@ -558,7 +606,7 @@ int __init cxl_file_init(void)
 	 * If these change we really need to update API.  Either change some
 	 * flags or update API version number CXL_API_VERSION.
 	 */
-	BUILD_BUG_ON(CXL_API_VERSION != 2);
+	BUILD_BUG_ON(CXL_API_VERSION != 3);
 	BUILD_BUG_ON(sizeof(struct cxl_ioctl_start_work) != 64);
 	BUILD_BUG_ON(sizeof(struct cxl_event_header) != 8);
 	BUILD_BUG_ON(sizeof(struct cxl_event_afu_interrupt) != 8);
diff --git a/drivers/misc/cxl/flash.c b/drivers/misc/cxl/flash.c
index 68dd0b7da471..c63d61e17d56 100644
--- a/drivers/misc/cxl/flash.c
+++ b/drivers/misc/cxl/flash.c
@@ -24,8 +24,8 @@ struct ai_header {
 };
 
 static struct semaphore sem;
-unsigned long *buffer[CXL_AI_MAX_ENTRIES];
-struct sg_list *le;
+static unsigned long *buffer[CXL_AI_MAX_ENTRIES];
+static struct sg_list *le;
 static u64 continue_token;
 static unsigned int transfer;
 
diff --git a/drivers/misc/cxl/guest.c b/drivers/misc/cxl/guest.c
index bc8d0b9870eb..9aa58a77a24d 100644
--- a/drivers/misc/cxl/guest.c
+++ b/drivers/misc/cxl/guest.c
@@ -196,15 +196,18 @@ static irqreturn_t guest_slice_irq_err(int irq, void *data)
 {
 	struct cxl_afu *afu = data;
 	int rc;
-	u64 serr;
+	u64 serr, afu_error, dsisr;
 
-	WARN(irq, "CXL SLICE ERROR interrupt %i\n", irq);
 	rc = cxl_h_get_fn_error_interrupt(afu->guest->handle, &serr);
 	if (rc) {
 		dev_crit(&afu->dev, "Couldn't read PSL_SERR_An: %d\n", rc);
 		return IRQ_HANDLED;
 	}
-	dev_crit(&afu->dev, "PSL_SERR_An: 0x%.16llx\n", serr);
+	afu_error = cxl_p2n_read(afu, CXL_AFU_ERR_An);
+	dsisr = cxl_p2n_read(afu, CXL_PSL_DSISR_An);
+	cxl_afu_decode_psl_serr(afu, serr);
+	dev_crit(&afu->dev, "AFU_ERR_An: 0x%.16llx\n", afu_error);
+	dev_crit(&afu->dev, "PSL_DSISR_An: 0x%.16llx\n", dsisr);
 
 	rc = cxl_h_ack_fn_error_interrupt(afu->guest->handle, serr);
 	if (rc)
@@ -1052,16 +1055,18 @@ static void free_adapter(struct cxl *adapter)
 	struct irq_avail *cur;
 	int i;
 
-	if (adapter->guest->irq_avail) {
-		for (i = 0; i < adapter->guest->irq_nranges; i++) {
-			cur = &adapter->guest->irq_avail[i];
-			kfree(cur->bitmap);
+	if (adapter->guest) {
+		if (adapter->guest->irq_avail) {
+			for (i = 0; i < adapter->guest->irq_nranges; i++) {
+				cur = &adapter->guest->irq_avail[i];
+				kfree(cur->bitmap);
+			}
+			kfree(adapter->guest->irq_avail);
 		}
-		kfree(adapter->guest->irq_avail);
+		kfree(adapter->guest->status);
+		kfree(adapter->guest);
 	}
-	kfree(adapter->guest->status);
 	cxl_remove_adapter_nr(adapter);
-	kfree(adapter->guest);
 	kfree(adapter);
 }
 
@@ -1182,6 +1187,7 @@ const struct cxl_backend_ops cxl_guest_ops = {
 	.ack_irq = guest_ack_irq,
 	.attach_process = guest_attach_process,
 	.detach_process = guest_detach_process,
+	.update_ivtes = NULL,
 	.support_attributes = guest_support_attributes,
 	.link_ok = guest_link_ok,
 	.release_afu = guest_release_afu,
diff --git a/drivers/misc/cxl/irq.c b/drivers/misc/cxl/irq.c
index 8def4553acba..dec60f58a767 100644
--- a/drivers/misc/cxl/irq.c
+++ b/drivers/misc/cxl/irq.c
@@ -260,9 +260,6 @@ int afu_allocate_irqs(struct cxl_context *ctx, u32 count)
 	else
 		alloc_count = count + 1;
 
-	/* Initialize the list head to hold irq names */
-	INIT_LIST_HEAD(&ctx->irq_names);
-
 	if ((rc = cxl_ops->alloc_irq_ranges(&ctx->irqs, ctx->afu->adapter,
 							alloc_count)))
 		return rc;
@@ -374,3 +371,32 @@ void afu_release_irqs(struct cxl_context *ctx, void *cookie)
 
 	ctx->irq_count = 0;
 }
+
+void cxl_afu_decode_psl_serr(struct cxl_afu *afu, u64 serr)
+{
+	dev_crit(&afu->dev,
+		 "PSL Slice error received. Check AFU for root cause.\n");
+	dev_crit(&afu->dev, "PSL_SERR_An: 0x%016llx\n", serr);
+	if (serr & CXL_PSL_SERR_An_afuto)
+		dev_crit(&afu->dev, "AFU MMIO Timeout\n");
+	if (serr & CXL_PSL_SERR_An_afudis)
+		dev_crit(&afu->dev,
+			 "MMIO targeted Accelerator that was not enabled\n");
+	if (serr & CXL_PSL_SERR_An_afuov)
+		dev_crit(&afu->dev, "AFU CTAG Overflow\n");
+	if (serr & CXL_PSL_SERR_An_badsrc)
+		dev_crit(&afu->dev, "Bad Interrupt Source\n");
+	if (serr & CXL_PSL_SERR_An_badctx)
+		dev_crit(&afu->dev, "Bad Context Handle\n");
+	if (serr & CXL_PSL_SERR_An_llcmdis)
+		dev_crit(&afu->dev, "LLCMD to Disabled AFU\n");
+	if (serr & CXL_PSL_SERR_An_llcmdto)
+		dev_crit(&afu->dev, "LLCMD Timeout to AFU\n");
+	if (serr & CXL_PSL_SERR_An_afupar)
+		dev_crit(&afu->dev, "AFU MMIO Parity Error\n");
+	if (serr & CXL_PSL_SERR_An_afudup)
+		dev_crit(&afu->dev, "AFU MMIO Duplicate CTAG Error\n");
+	if (serr & CXL_PSL_SERR_An_AE)
+		dev_crit(&afu->dev,
+			 "AFU asserted JDONE with JERROR in AFU Directed Mode\n");
+}
diff --git a/drivers/misc/cxl/main.c b/drivers/misc/cxl/main.c
index ae68c3201156..d9be23b24aa3 100644
--- a/drivers/misc/cxl/main.c
+++ b/drivers/misc/cxl/main.c
@@ -110,6 +110,11 @@ static inline void cxl_slbia_core(struct mm_struct *mm)
 
 static struct cxl_calls cxl_calls = {
 	.cxl_slbia = cxl_slbia_core,
+	.cxl_pci_associate_default_context = _cxl_pci_associate_default_context,
+	.cxl_pci_disable_device = _cxl_pci_disable_device,
+	.cxl_next_msi_hwirq = _cxl_next_msi_hwirq,
+	.cxl_cx4_setup_msi_irqs = _cxl_cx4_setup_msi_irqs,
+	.cxl_cx4_teardown_msi_irqs = _cxl_cx4_teardown_msi_irqs,
 	.owner = THIS_MODULE,
 };
 
diff --git a/drivers/misc/cxl/native.c b/drivers/misc/cxl/native.c
index 55d8a1459f28..3bcdaee11ba1 100644
--- a/drivers/misc/cxl/native.c
+++ b/drivers/misc/cxl/native.c
@@ -21,10 +21,10 @@
 #include "cxl.h"
 #include "trace.h"
 
-static int afu_control(struct cxl_afu *afu, u64 command,
+static int afu_control(struct cxl_afu *afu, u64 command, u64 clear,
 		       u64 result, u64 mask, bool enabled)
 {
-	u64 AFU_Cntl = cxl_p2n_read(afu, CXL_AFU_Cntl_An);
+	u64 AFU_Cntl;
 	unsigned long timeout = jiffies + (HZ * CXL_TIMEOUT);
 	int rc = 0;
 
@@ -33,7 +33,8 @@ static int afu_control(struct cxl_afu *afu, u64 command,
 
 	trace_cxl_afu_ctrl(afu, command);
 
-	cxl_p2n_write(afu, CXL_AFU_Cntl_An, AFU_Cntl | command);
+	AFU_Cntl = cxl_p2n_read(afu, CXL_AFU_Cntl_An);
+	cxl_p2n_write(afu, CXL_AFU_Cntl_An, (AFU_Cntl & ~clear) | command);
 
 	AFU_Cntl = cxl_p2n_read(afu, CXL_AFU_Cntl_An);
 	while ((AFU_Cntl & mask) != result) {
@@ -54,6 +55,16 @@ static int afu_control(struct cxl_afu *afu, u64 command,
 		cpu_relax();
 		AFU_Cntl = cxl_p2n_read(afu, CXL_AFU_Cntl_An);
 	};
+
+	if (AFU_Cntl & CXL_AFU_Cntl_An_RA) {
+		/*
+		 * Workaround for a bug in the XSL used in the Mellanox CX4
+		 * that fails to clear the RA bit after an AFU reset,
+		 * preventing subsequent AFU resets from working.
+		 */
+		cxl_p2n_write(afu, CXL_AFU_Cntl_An, AFU_Cntl & ~CXL_AFU_Cntl_An_RA);
+	}
+
 	pr_devel("AFU command complete: %llx\n", command);
 	afu->enabled = enabled;
 out:
@@ -67,7 +78,7 @@ static int afu_enable(struct cxl_afu *afu)
 {
 	pr_devel("AFU enable request\n");
 
-	return afu_control(afu, CXL_AFU_Cntl_An_E,
+	return afu_control(afu, CXL_AFU_Cntl_An_E, 0,
 			   CXL_AFU_Cntl_An_ES_Enabled,
 			   CXL_AFU_Cntl_An_ES_MASK, true);
 }
@@ -76,7 +87,8 @@ int cxl_afu_disable(struct cxl_afu *afu)
 {
 	pr_devel("AFU disable request\n");
 
-	return afu_control(afu, 0, CXL_AFU_Cntl_An_ES_Disabled,
+	return afu_control(afu, 0, CXL_AFU_Cntl_An_E,
+			   CXL_AFU_Cntl_An_ES_Disabled,
 			   CXL_AFU_Cntl_An_ES_MASK, false);
 }
 
@@ -85,7 +97,7 @@ static int native_afu_reset(struct cxl_afu *afu)
 {
 	pr_devel("AFU reset request\n");
 
-	return afu_control(afu, CXL_AFU_Cntl_An_RA,
+	return afu_control(afu, CXL_AFU_Cntl_An_RA, 0,
 			   CXL_AFU_Cntl_An_RS_Complete | CXL_AFU_Cntl_An_ES_Disabled,
 			   CXL_AFU_Cntl_An_RS_MASK | CXL_AFU_Cntl_An_ES_MASK,
 			   false);
@@ -189,7 +201,7 @@ int cxl_alloc_spa(struct cxl_afu *afu)
 	unsigned spa_size;
 
 	/* Work out how many pages to allocate */
-	afu->native->spa_order = 0;
+	afu->native->spa_order = -1;
 	do {
 		afu->native->spa_order++;
 		spa_size = (1 << afu->native->spa_order) * PAGE_SIZE;
@@ -430,7 +442,6 @@ static int remove_process_element(struct cxl_context *ctx)
 	return rc;
 }
 
-
 void cxl_assign_psn_space(struct cxl_context *ctx)
 {
 	if (!ctx->afu->pp_size || ctx->master) {
@@ -507,10 +518,39 @@ static u64 calculate_sr(struct cxl_context *ctx)
 	return sr;
 }
 
+static void update_ivtes_directed(struct cxl_context *ctx)
+{
+	bool need_update = (ctx->status == STARTED);
+	int r;
+
+	if (need_update) {
+		WARN_ON(terminate_process_element(ctx));
+		WARN_ON(remove_process_element(ctx));
+	}
+
+	for (r = 0; r < CXL_IRQ_RANGES; r++) {
+		ctx->elem->ivte_offsets[r] = cpu_to_be16(ctx->irqs.offset[r]);
+		ctx->elem->ivte_ranges[r] = cpu_to_be16(ctx->irqs.range[r]);
+	}
+
+	/*
+	 * Theoretically we could use the update llcmd, instead of a
+	 * terminate/remove/add (or if an atomic update was required we could
+	 * do a suspend/update/resume), however it seems there might be issues
+	 * with the update llcmd on some cards (including those using an XSL on
+	 * an ASIC) so for now it's safest to go with the commands that are
+	 * known to work. In the future if we come across a situation where the
+	 * card may be performing transactions using the same PE while we are
+	 * doing this update we might need to revisit this.
+	 */
+	if (need_update)
+		WARN_ON(add_process_element(ctx));
+}
+
 static int attach_afu_directed(struct cxl_context *ctx, u64 wed, u64 amr)
 {
 	u32 pid;
-	int r, result;
+	int result;
 
 	cxl_assign_psn_space(ctx);
 
@@ -545,10 +585,7 @@ static int attach_afu_directed(struct cxl_context *ctx, u64 wed, u64 amr)
 		ctx->irqs.range[0] = 1;
 	}
 
-	for (r = 0; r < CXL_IRQ_RANGES; r++) {
-		ctx->elem->ivte_offsets[r] = cpu_to_be16(ctx->irqs.offset[r]);
-		ctx->elem->ivte_ranges[r] = cpu_to_be16(ctx->irqs.range[r]);
-	}
+	update_ivtes_directed(ctx);
 
 	ctx->elem->common.amr = cpu_to_be64(amr);
 	ctx->elem->common.wed = cpu_to_be64(wed);
@@ -570,7 +607,33 @@ static int deactivate_afu_directed(struct cxl_afu *afu)
 	cxl_sysfs_afu_m_remove(afu);
 	cxl_chardev_afu_remove(afu);
 
-	cxl_ops->afu_reset(afu);
+	/*
+	 * The CAIA section 2.2.1 indicates that the procedure for starting and
+	 * stopping an AFU in AFU directed mode is AFU specific, which is not
+	 * ideal since this code is generic and with one exception has no
+	 * knowledge of the AFU. This is in contrast to the procedure for
+	 * disabling a dedicated process AFU, which is documented to just
+	 * require a reset. The architecture does indicate that both an AFU
+	 * reset and an AFU disable should result in the AFU being disabled and
+	 * we do both followed by a PSL purge for safety.
+	 *
+	 * Notably we used to have some issues with the disable sequence on PSL
+	 * cards, which is why we ended up using this heavy weight procedure in
+	 * the first place, however a bug was discovered that had rendered the
+	 * disable operation ineffective, so it is conceivable that was the
+	 * sole explanation for those difficulties. Careful regression testing
+	 * is recommended if anyone attempts to remove or reorder these
+	 * operations.
+	 *
+	 * The XSL on the Mellanox CX4 behaves a little differently from the
+	 * PSL based cards and will time out an AFU reset if the AFU is still
+	 * enabled. That card is special in that we do have a means to identify
+	 * it from this code, so in that case we skip the reset and just use a
+	 * disable/purge to avoid the timeout and corresponding noise in the
+	 * kernel log.
+	 */
+	if (afu->adapter->native->sl_ops->needs_reset_before_disable)
+		cxl_ops->afu_reset(afu);
 	cxl_afu_disable(afu);
 	cxl_psl_purge(afu);
 
@@ -600,6 +663,22 @@ static int activate_dedicated_process(struct cxl_afu *afu)
 	return cxl_chardev_d_afu_add(afu);
 }
 
+static void update_ivtes_dedicated(struct cxl_context *ctx)
+{
+	struct cxl_afu *afu = ctx->afu;
+
+	cxl_p1n_write(afu, CXL_PSL_IVTE_Offset_An,
+		       (((u64)ctx->irqs.offset[0] & 0xffff) << 48) |
+		       (((u64)ctx->irqs.offset[1] & 0xffff) << 32) |
+		       (((u64)ctx->irqs.offset[2] & 0xffff) << 16) |
+			((u64)ctx->irqs.offset[3] & 0xffff));
+	cxl_p1n_write(afu, CXL_PSL_IVTE_Limit_An, (u64)
+		       (((u64)ctx->irqs.range[0] & 0xffff) << 48) |
+		       (((u64)ctx->irqs.range[1] & 0xffff) << 32) |
+		       (((u64)ctx->irqs.range[2] & 0xffff) << 16) |
+			((u64)ctx->irqs.range[3] & 0xffff));
+}
+
 static int attach_dedicated(struct cxl_context *ctx, u64 wed, u64 amr)
 {
 	struct cxl_afu *afu = ctx->afu;
@@ -618,16 +697,7 @@ static int attach_dedicated(struct cxl_context *ctx, u64 wed, u64 amr)
 
 	cxl_prefault(ctx, wed);
 
-	cxl_p1n_write(afu, CXL_PSL_IVTE_Offset_An,
-		       (((u64)ctx->irqs.offset[0] & 0xffff) << 48) |
-		       (((u64)ctx->irqs.offset[1] & 0xffff) << 32) |
-		       (((u64)ctx->irqs.offset[2] & 0xffff) << 16) |
-			((u64)ctx->irqs.offset[3] & 0xffff));
-	cxl_p1n_write(afu, CXL_PSL_IVTE_Limit_An, (u64)
-		       (((u64)ctx->irqs.range[0] & 0xffff) << 48) |
-		       (((u64)ctx->irqs.range[1] & 0xffff) << 32) |
-		       (((u64)ctx->irqs.range[2] & 0xffff) << 16) |
-			((u64)ctx->irqs.range[3] & 0xffff));
+	update_ivtes_dedicated(ctx);
 
 	cxl_p2n_write(afu, CXL_PSL_AMR_An, amr);
 
@@ -703,12 +773,37 @@ static int native_attach_process(struct cxl_context *ctx, bool kernel,
 
 static inline int detach_process_native_dedicated(struct cxl_context *ctx)
 {
+	/*
+	 * The CAIA section 2.1.1 indicates that we need to do an AFU reset to
+	 * stop the AFU in dedicated mode (we therefore do not make that
+	 * optional like we do in the afu directed path). It does not indicate
+	 * that we need to do an explicit disable (which should occur
+	 * implicitly as part of the reset) or purge, but we do these as well
+	 * to be on the safe side.
+	 *
+	 * Notably we used to have some issues with the disable sequence
+	 * (before the sequence was spelled out in the architecture) which is
+	 * why we were so heavy weight in the first place, however a bug was
+	 * discovered that had rendered the disable operation ineffective, so
+	 * it is conceivable that was the sole explanation for those
+	 * difficulties. Point is, we should be careful and do some regression
+	 * testing if we ever attempt to remove any part of this procedure.
+	 */
 	cxl_ops->afu_reset(ctx->afu);
 	cxl_afu_disable(ctx->afu);
 	cxl_psl_purge(ctx->afu);
 	return 0;
 }
 
+static void native_update_ivtes(struct cxl_context *ctx)
+{
+	if (ctx->afu->current_mode == CXL_MODE_DIRECTED)
+		return update_ivtes_directed(ctx);
+	if (ctx->afu->current_mode == CXL_MODE_DEDICATED)
+		return update_ivtes_dedicated(ctx);
+	WARN(1, "native_update_ivtes: Bad mode\n");
+}
+
 static inline int detach_process_native_afu_directed(struct cxl_context *ctx)
 {
 	if (!ctx->pe_inserted)
@@ -754,26 +849,38 @@ static int native_get_irq_info(struct cxl_afu *afu, struct cxl_irq_info *info)
 	return 0;
 }
 
-static irqreturn_t native_handle_psl_slice_error(struct cxl_context *ctx,
-						u64 dsisr, u64 errstat)
+void cxl_native_psl_irq_dump_regs(struct cxl_context *ctx)
 {
 	u64 fir1, fir2, fir_slice, serr, afu_debug;
 
 	fir1 = cxl_p1_read(ctx->afu->adapter, CXL_PSL_FIR1);
 	fir2 = cxl_p1_read(ctx->afu->adapter, CXL_PSL_FIR2);
 	fir_slice = cxl_p1n_read(ctx->afu, CXL_PSL_FIR_SLICE_An);
-	serr = cxl_p1n_read(ctx->afu, CXL_PSL_SERR_An);
 	afu_debug = cxl_p1n_read(ctx->afu, CXL_AFU_DEBUG_An);
 
-	dev_crit(&ctx->afu->dev, "PSL ERROR STATUS: 0x%016llx\n", errstat);
 	dev_crit(&ctx->afu->dev, "PSL_FIR1: 0x%016llx\n", fir1);
 	dev_crit(&ctx->afu->dev, "PSL_FIR2: 0x%016llx\n", fir2);
-	dev_crit(&ctx->afu->dev, "PSL_SERR_An: 0x%016llx\n", serr);
+	if (ctx->afu->adapter->native->sl_ops->register_serr_irq) {
+		serr = cxl_p1n_read(ctx->afu, CXL_PSL_SERR_An);
+		cxl_afu_decode_psl_serr(ctx->afu, serr);
+	}
 	dev_crit(&ctx->afu->dev, "PSL_FIR_SLICE_An: 0x%016llx\n", fir_slice);
 	dev_crit(&ctx->afu->dev, "CXL_PSL_AFU_DEBUG_An: 0x%016llx\n", afu_debug);
+}
+
+static irqreturn_t native_handle_psl_slice_error(struct cxl_context *ctx,
+						u64 dsisr, u64 errstat)
+{
+
+	dev_crit(&ctx->afu->dev, "PSL ERROR STATUS: 0x%016llx\n", errstat);
 
-	dev_crit(&ctx->afu->dev, "STOPPING CXL TRACE\n");
-	cxl_stop_trace(ctx->afu->adapter);
+	if (ctx->afu->adapter->native->sl_ops->psl_irq_dump_registers)
+		ctx->afu->adapter->native->sl_ops->psl_irq_dump_registers(ctx);
+
+	if (ctx->afu->adapter->native->sl_ops->debugfs_stop_trace) {
+		dev_crit(&ctx->afu->dev, "STOPPING CXL TRACE\n");
+		ctx->afu->adapter->native->sl_ops->debugfs_stop_trace(ctx->afu->adapter);
+	}
 
 	return cxl_ops->ack_irq(ctx, 0, errstat);
 }
@@ -849,41 +956,56 @@ void native_irq_wait(struct cxl_context *ctx)
 static irqreturn_t native_slice_irq_err(int irq, void *data)
 {
 	struct cxl_afu *afu = data;
-	u64 fir_slice, errstat, serr, afu_debug;
-
-	WARN(irq, "CXL SLICE ERROR interrupt %i\n", irq);
+	u64 fir_slice, errstat, serr, afu_debug, afu_error, dsisr;
 
+	/*
+	 * slice err interrupt is only used with full PSL (no XSL)
+	 */
 	serr = cxl_p1n_read(afu, CXL_PSL_SERR_An);
 	fir_slice = cxl_p1n_read(afu, CXL_PSL_FIR_SLICE_An);
 	errstat = cxl_p2n_read(afu, CXL_PSL_ErrStat_An);
 	afu_debug = cxl_p1n_read(afu, CXL_AFU_DEBUG_An);
-	dev_crit(&afu->dev, "PSL_SERR_An: 0x%016llx\n", serr);
+	afu_error = cxl_p2n_read(afu, CXL_AFU_ERR_An);
+	dsisr = cxl_p2n_read(afu, CXL_PSL_DSISR_An);
+	cxl_afu_decode_psl_serr(afu, serr);
 	dev_crit(&afu->dev, "PSL_FIR_SLICE_An: 0x%016llx\n", fir_slice);
 	dev_crit(&afu->dev, "CXL_PSL_ErrStat_An: 0x%016llx\n", errstat);
 	dev_crit(&afu->dev, "CXL_PSL_AFU_DEBUG_An: 0x%016llx\n", afu_debug);
+	dev_crit(&afu->dev, "AFU_ERR_An: 0x%.16llx\n", afu_error);
+	dev_crit(&afu->dev, "PSL_DSISR_An: 0x%.16llx\n", dsisr);
 
 	cxl_p1n_write(afu, CXL_PSL_SERR_An, serr);
 
 	return IRQ_HANDLED;
 }
 
+void cxl_native_err_irq_dump_regs(struct cxl *adapter)
+{
+	u64 fir1, fir2;
+
+	fir1 = cxl_p1_read(adapter, CXL_PSL_FIR1);
+	fir2 = cxl_p1_read(adapter, CXL_PSL_FIR2);
+
+	dev_crit(&adapter->dev, "PSL_FIR1: 0x%016llx\nPSL_FIR2: 0x%016llx\n", fir1, fir2);
+}
+
 static irqreturn_t native_irq_err(int irq, void *data)
 {
 	struct cxl *adapter = data;
-	u64 fir1, fir2, err_ivte;
+	u64 err_ivte;
 
 	WARN(1, "CXL ERROR interrupt %i\n", irq);
 
 	err_ivte = cxl_p1_read(adapter, CXL_PSL_ErrIVTE);
 	dev_crit(&adapter->dev, "PSL_ErrIVTE: 0x%016llx\n", err_ivte);
 
-	dev_crit(&adapter->dev, "STOPPING CXL TRACE\n");
-	cxl_stop_trace(adapter);
-
-	fir1 = cxl_p1_read(adapter, CXL_PSL_FIR1);
-	fir2 = cxl_p1_read(adapter, CXL_PSL_FIR2);
+	if (adapter->native->sl_ops->debugfs_stop_trace) {
+		dev_crit(&adapter->dev, "STOPPING CXL TRACE\n");
+		adapter->native->sl_ops->debugfs_stop_trace(adapter);
+	}
 
-	dev_crit(&adapter->dev, "PSL_FIR1: 0x%016llx\nPSL_FIR2: 0x%016llx\n", fir1, fir2);
+	if (adapter->native->sl_ops->err_irq_dump_registers)
+		adapter->native->sl_ops->err_irq_dump_registers(adapter);
 
 	return IRQ_HANDLED;
 }
@@ -1128,6 +1250,7 @@ const struct cxl_backend_ops cxl_native_ops = {
 	.irq_wait = native_irq_wait,
 	.attach_process = native_attach_process,
 	.detach_process = native_detach_process,
+	.update_ivtes = native_update_ivtes,
 	.support_attributes = native_support_attributes,
 	.link_ok = cxl_adapter_link_ok,
 	.release_afu = cxl_pci_release_afu,
diff --git a/drivers/misc/cxl/pci.c b/drivers/misc/cxl/pci.c
index a08fcc888a71..d152e2de8c93 100644
--- a/drivers/misc/cxl/pci.c
+++ b/drivers/misc/cxl/pci.c
@@ -55,6 +55,8 @@
 	pci_read_config_byte(dev, vsec + 0xa, dest)
 #define CXL_WRITE_VSEC_MODE_CONTROL(dev, vsec, val) \
 	pci_write_config_byte(dev, vsec + 0xa, val)
+#define CXL_WRITE_VSEC_MODE_CONTROL_BUS(bus, devfn, vsec, val) \
+	pci_bus_write_config_byte(bus, devfn, vsec + 0xa, val)
 #define CXL_VSEC_PROTOCOL_MASK   0xe0
 #define CXL_VSEC_PROTOCOL_1024TB 0x80
 #define CXL_VSEC_PROTOCOL_512TB  0x40
@@ -352,13 +354,10 @@ static u64 get_capp_unit_id(struct device_node *np)
 	return 0;
 }
 
-static int init_implementation_adapter_regs(struct cxl *adapter, struct pci_dev *dev)
+static int calc_capp_routing(struct pci_dev *dev, u64 *chipid, u64 *capp_unit_id)
 {
 	struct device_node *np;
 	const __be32 *prop;
-	u64 psl_dsnctl;
-	u64 chipid;
-	u64 capp_unit_id;
 
 	if (!(np = pnv_pci_get_phb_node(dev)))
 		return -ENODEV;
@@ -367,14 +366,28 @@ static int init_implementation_adapter_regs(struct cxl *adapter, struct pci_dev
 		np = of_get_next_parent(np);
 	if (!np)
 		return -ENODEV;
-	chipid = be32_to_cpup(prop);
-	capp_unit_id = get_capp_unit_id(np);
+	*chipid = be32_to_cpup(prop);
+	*capp_unit_id = get_capp_unit_id(np);
 	of_node_put(np);
-	if (!capp_unit_id) {
+	if (!*capp_unit_id) {
 		pr_err("cxl: invalid capp unit id\n");
 		return -ENODEV;
 	}
 
+	return 0;
+}
+
+static int init_implementation_adapter_psl_regs(struct cxl *adapter, struct pci_dev *dev)
+{
+	u64 psl_dsnctl;
+	u64 chipid;
+	u64 capp_unit_id;
+	int rc;
+
+	rc = calc_capp_routing(dev, &chipid, &capp_unit_id);
+	if (rc)
+		return rc;
+
 	psl_dsnctl = 0x0000900000000000ULL; /* pteupd ttype, scdone */
 	psl_dsnctl |= (0x2ULL << (63-38)); /* MMIO hang pulse: 256 us */
 	/* Tell PSL where to route data to */
@@ -393,8 +406,61 @@ static int init_implementation_adapter_regs(struct cxl *adapter, struct pci_dev
 	return 0;
 }
 
+static int init_implementation_adapter_xsl_regs(struct cxl *adapter, struct pci_dev *dev)
+{
+	u64 xsl_dsnctl;
+	u64 chipid;
+	u64 capp_unit_id;
+	int rc;
+
+	rc = calc_capp_routing(dev, &chipid, &capp_unit_id);
+	if (rc)
+		return rc;
+
+	/* Tell XSL where to route data to */
+	xsl_dsnctl = 0x0000600000000000ULL | (chipid << (63-5));
+	xsl_dsnctl |= (capp_unit_id << (63-13));
+	cxl_p1_write(adapter, CXL_XSL_DSNCTL, xsl_dsnctl);
+
+	return 0;
+}
+
+/* PSL & XSL */
+#define TBSYNC_CAL(n) (((u64)n & 0x7) << (63-3))
 #define TBSYNC_CNT(n) (((u64)n & 0x7) << (63-6))
-#define _2048_250MHZ_CYCLES 1
+/* For the PSL this is a multiple for 0 < n <= 7: */
+#define PSL_2048_250MHZ_CYCLES 1
+
+static void write_timebase_ctrl_psl(struct cxl *adapter)
+{
+	cxl_p1_write(adapter, CXL_PSL_TB_CTLSTAT,
+		     TBSYNC_CNT(2 * PSL_2048_250MHZ_CYCLES));
+}
+
+/* XSL */
+#define TBSYNC_ENA (1ULL << 63)
+/* For the XSL this is 2**n * 2000 clocks for 0 < n <= 6: */
+#define XSL_2000_CLOCKS 1
+#define XSL_4000_CLOCKS 2
+#define XSL_8000_CLOCKS 3
+
+static void write_timebase_ctrl_xsl(struct cxl *adapter)
+{
+	cxl_p1_write(adapter, CXL_XSL_TB_CTLSTAT,
+		     TBSYNC_ENA |
+		     TBSYNC_CAL(3) |
+		     TBSYNC_CNT(XSL_4000_CLOCKS));
+}
+
+static u64 timebase_read_psl(struct cxl *adapter)
+{
+	return cxl_p1_read(adapter, CXL_PSL_Timebase);
+}
+
+static u64 timebase_read_xsl(struct cxl *adapter)
+{
+	return cxl_p1_read(adapter, CXL_XSL_Timebase);
+}
 
 static void cxl_setup_psl_timebase(struct cxl *adapter, struct pci_dev *dev)
 {
@@ -421,8 +487,7 @@ static void cxl_setup_psl_timebase(struct cxl *adapter, struct pci_dev *dev)
 	 * Setup PSL Timebase Control and Status register
 	 * with the recommended Timebase Sync Count value
 	 */
-	cxl_p1_write(adapter, CXL_PSL_TB_CTLSTAT,
-		     TBSYNC_CNT(2 * _2048_250MHZ_CYCLES));
+	adapter->native->sl_ops->write_timebase_ctrl(adapter);
 
 	/* Enable PSL Timebase */
 	cxl_p1_write(adapter, CXL_PSL_Control, 0x0000000000000000);
@@ -435,7 +500,7 @@ static void cxl_setup_psl_timebase(struct cxl *adapter, struct pci_dev *dev)
 			dev_info(&dev->dev, "PSL timebase can't synchronize\n");
 			return;
 		}
-		psl_tb = cxl_p1_read(adapter, CXL_PSL_Timebase);
+		psl_tb = adapter->native->sl_ops->timebase_read(adapter);
 		delta = mftb() - psl_tb;
 		if (delta < 0)
 			delta = -delta;
@@ -445,7 +510,7 @@ static void cxl_setup_psl_timebase(struct cxl *adapter, struct pci_dev *dev)
 	return;
 }
 
-static int init_implementation_afu_regs(struct cxl_afu *afu)
+static int init_implementation_afu_psl_regs(struct cxl_afu *afu)
 {
 	/* read/write masks for this slice */
 	cxl_p1n_write(afu, CXL_PSL_APCALLOC_A, 0xFFFFFFFEFEFEFEFEULL);
@@ -551,36 +616,234 @@ static int setup_cxl_bars(struct pci_dev *dev)
 	return 0;
 }
 
-/* pciex node: ibm,opal-m64-window = <0x3d058 0x0 0x3d058 0x0 0x8 0x0>; */
-static int switch_card_to_cxl(struct pci_dev *dev)
-{
+#ifdef CONFIG_CXL_BIMODAL
+
+struct cxl_switch_work {
+	struct pci_dev *dev;
+	struct work_struct work;
 	int vsec;
+	int mode;
+};
+
+static void switch_card_to_cxl(struct work_struct *work)
+{
+	struct cxl_switch_work *switch_work =
+		container_of(work, struct cxl_switch_work, work);
+	struct pci_dev *dev = switch_work->dev;
+	struct pci_bus *bus = dev->bus;
+	struct pci_controller *hose = pci_bus_to_host(bus);
+	struct pci_dev *bridge;
+	struct pnv_php_slot *php_slot;
+	unsigned int devfn;
 	u8 val;
 	int rc;
 
-	dev_info(&dev->dev, "switch card to CXL\n");
+	dev_info(&bus->dev, "cxl: Preparing for mode switch...\n");
+	bridge = list_first_entry_or_null(&hose->bus->devices, struct pci_dev,
+					  bus_list);
+	if (!bridge) {
+		dev_WARN(&bus->dev, "cxl: Couldn't find root port!\n");
+		goto err_dev_put;
+	}
 
-	if (!(vsec = find_cxl_vsec(dev))) {
-		dev_err(&dev->dev, "ABORTING: CXL VSEC not found!\n");
+	php_slot = pnv_php_find_slot(pci_device_to_OF_node(bridge));
+	if (!php_slot) {
+		dev_err(&bus->dev, "cxl: Failed to find slot hotplug "
+			           "information. You may need to upgrade "
+			           "skiboot. Aborting.\n");
+		goto err_dev_put;
+	}
+
+	rc = CXL_READ_VSEC_MODE_CONTROL(dev, switch_work->vsec, &val);
+	if (rc) {
+		dev_err(&bus->dev, "cxl: Failed to read CAPI mode control: %i\n", rc);
+		goto err_dev_put;
+	}
+	devfn = dev->devfn;
+
+	/* Release the reference obtained in cxl_check_and_switch_mode() */
+	pci_dev_put(dev);
+
+	dev_dbg(&bus->dev, "cxl: Removing PCI devices from kernel\n");
+	pci_lock_rescan_remove();
+	pci_hp_remove_devices(bridge->subordinate);
+	pci_unlock_rescan_remove();
+
+	/* Switch the CXL protocol on the card */
+	if (switch_work->mode == CXL_BIMODE_CXL) {
+		dev_info(&bus->dev, "cxl: Switching card to CXL mode\n");
+		val &= ~CXL_VSEC_PROTOCOL_MASK;
+		val |= CXL_VSEC_PROTOCOL_256TB | CXL_VSEC_PROTOCOL_ENABLE;
+		rc = pnv_cxl_enable_phb_kernel_api(hose, true);
+		if (rc) {
+			dev_err(&bus->dev, "cxl: Failed to enable kernel API"
+				           " on real PHB, aborting\n");
+			goto err_free_work;
+		}
+	} else {
+		dev_WARN(&bus->dev, "cxl: Switching card to PCI mode not supported!\n");
+		goto err_free_work;
+	}
+
+	rc = CXL_WRITE_VSEC_MODE_CONTROL_BUS(bus, devfn, switch_work->vsec, val);
+	if (rc) {
+		dev_err(&bus->dev, "cxl: Failed to configure CXL protocol: %i\n", rc);
+		goto err_free_work;
+	}
+
+	/*
+	 * The CAIA spec (v1.1, Section 10.6 Bi-modal Device Support) states
+	 * we must wait 100ms after this mode switch before touching PCIe config
+	 * space.
+	 */
+	msleep(100);
+
+	/*
+	 * Hot reset to cause the card to come back in cxl mode. A
+	 * OPAL_RESET_PCI_LINK would be sufficient, but currently lacks support
+	 * in skiboot, so we use a hot reset instead.
+	 *
+	 * We call pci_set_pcie_reset_state() on the bridge, as a CAPI card is
+	 * guaranteed to sit directly under the root port, and setting the reset
+	 * state on a device directly under the root port is equivalent to doing
+	 * it on the root port iself.
+	 */
+	dev_info(&bus->dev, "cxl: Configuration write complete, resetting card\n");
+	pci_set_pcie_reset_state(bridge, pcie_hot_reset);
+	pci_set_pcie_reset_state(bridge, pcie_deassert_reset);
+
+	dev_dbg(&bus->dev, "cxl: Offlining slot\n");
+	rc = pnv_php_set_slot_power_state(&php_slot->slot, OPAL_PCI_SLOT_OFFLINE);
+	if (rc) {
+		dev_err(&bus->dev, "cxl: OPAL offlining call failed: %i\n", rc);
+		goto err_free_work;
+	}
+
+	dev_dbg(&bus->dev, "cxl: Onlining and probing slot\n");
+	rc = pnv_php_set_slot_power_state(&php_slot->slot, OPAL_PCI_SLOT_ONLINE);
+	if (rc) {
+		dev_err(&bus->dev, "cxl: OPAL onlining call failed: %i\n", rc);
+		goto err_free_work;
+	}
+
+	pci_lock_rescan_remove();
+	pci_hp_add_devices(bridge->subordinate);
+	pci_unlock_rescan_remove();
+
+	dev_info(&bus->dev, "cxl: CAPI mode switch completed\n");
+	kfree(switch_work);
+	return;
+
+err_dev_put:
+	/* Release the reference obtained in cxl_check_and_switch_mode() */
+	pci_dev_put(dev);
+err_free_work:
+	kfree(switch_work);
+}
+
+int cxl_check_and_switch_mode(struct pci_dev *dev, int mode, int vsec)
+{
+	struct cxl_switch_work *work;
+	u8 val;
+	int rc;
+
+	if (!cpu_has_feature(CPU_FTR_HVMODE))
 		return -ENODEV;
+
+	if (!vsec) {
+		vsec = find_cxl_vsec(dev);
+		if (!vsec) {
+			dev_info(&dev->dev, "CXL VSEC not found\n");
+			return -ENODEV;
+		}
 	}
 
-	if ((rc = CXL_READ_VSEC_MODE_CONTROL(dev, vsec, &val))) {
-		dev_err(&dev->dev, "failed to read current mode control: %i", rc);
+	rc = CXL_READ_VSEC_MODE_CONTROL(dev, vsec, &val);
+	if (rc) {
+		dev_err(&dev->dev, "Failed to read current mode control: %i", rc);
 		return rc;
 	}
-	val &= ~CXL_VSEC_PROTOCOL_MASK;
-	val |= CXL_VSEC_PROTOCOL_256TB | CXL_VSEC_PROTOCOL_ENABLE;
-	if ((rc = CXL_WRITE_VSEC_MODE_CONTROL(dev, vsec, val))) {
-		dev_err(&dev->dev, "failed to enable CXL protocol: %i", rc);
-		return rc;
+
+	if (mode == CXL_BIMODE_PCI) {
+		if (!(val & CXL_VSEC_PROTOCOL_ENABLE)) {
+			dev_info(&dev->dev, "Card is already in PCI mode\n");
+			return 0;
+		}
+		/*
+		 * TODO: Before it's safe to switch the card back to PCI mode
+		 * we need to disable the CAPP and make sure any cachelines the
+		 * card holds have been flushed out. Needs skiboot support.
+		 */
+		dev_WARN(&dev->dev, "CXL mode switch to PCI unsupported!\n");
+		return -EIO;
 	}
+
+	if (val & CXL_VSEC_PROTOCOL_ENABLE) {
+		dev_info(&dev->dev, "Card is already in CXL mode\n");
+		return 0;
+	}
+
+	dev_info(&dev->dev, "Card is in PCI mode, scheduling kernel thread "
+			    "to switch to CXL mode\n");
+
+	work = kmalloc(sizeof(struct cxl_switch_work), GFP_KERNEL);
+	if (!work)
+		return -ENOMEM;
+
+	pci_dev_get(dev);
+	work->dev = dev;
+	work->vsec = vsec;
+	work->mode = mode;
+	INIT_WORK(&work->work, switch_card_to_cxl);
+
+	schedule_work(&work->work);
+
 	/*
-	 * The CAIA spec (v0.12 11.6 Bi-modal Device Support) states
-	 * we must wait 100ms after this mode switch before touching
-	 * PCIe config space.
+	 * We return a failure now to abort the driver init. Once the
+	 * link has been cycled and the card is in cxl mode we will
+	 * come back (possibly using the generic cxl driver), but
+	 * return success as the card should then be in cxl mode.
+	 *
+	 * TODO: What if the card comes back in PCI mode even after
+	 *       the switch?  Don't want to spin endlessly.
 	 */
-	msleep(100);
+	return -EBUSY;
+}
+EXPORT_SYMBOL_GPL(cxl_check_and_switch_mode);
+
+#endif /* CONFIG_CXL_BIMODAL */
+
+static int setup_cxl_protocol_area(struct pci_dev *dev)
+{
+	u8 val;
+	int rc;
+	int vsec = find_cxl_vsec(dev);
+
+	if (!vsec) {
+		dev_info(&dev->dev, "CXL VSEC not found\n");
+		return -ENODEV;
+	}
+
+	rc = CXL_READ_VSEC_MODE_CONTROL(dev, vsec, &val);
+	if (rc) {
+		dev_err(&dev->dev, "Failed to read current mode control: %i\n", rc);
+		return rc;
+	}
+
+	if (!(val & CXL_VSEC_PROTOCOL_ENABLE)) {
+		dev_err(&dev->dev, "Card not in CAPI mode!\n");
+		return -EIO;
+	}
+
+	if ((val & CXL_VSEC_PROTOCOL_MASK) != CXL_VSEC_PROTOCOL_256TB) {
+		val &= ~CXL_VSEC_PROTOCOL_MASK;
+		val |= CXL_VSEC_PROTOCOL_256TB;
+		rc = CXL_WRITE_VSEC_MODE_CONTROL(dev, vsec, val);
+		if (rc) {
+			dev_err(&dev->dev, "Failed to set CXL protocol area: %i\n", rc);
+			return rc;
+		}
+	}
 
 	return 0;
 }
@@ -712,6 +975,21 @@ static int cxl_afu_descriptor_looks_ok(struct cxl_afu *afu)
 		}
 	}
 
+	if ((afu->modes_supported & ~CXL_MODE_DEDICATED) && afu->max_procs_virtualised == 0) {
+		/*
+		 * We could also check this for the dedicated process model
+		 * since the architecture indicates it should be set to 1, but
+		 * in that case we ignore the value and I'd rather not risk
+		 * breaking any existing dedicated process AFUs that left it as
+		 * 0 (not that I'm aware of any). It is clearly an error for an
+		 * AFU directed AFU to set this to 0, and would have previously
+		 * triggered a bug resulting in the maximum not being enforced
+		 * at all since idr_alloc treats 0 as no maximum.
+		 */
+		dev_err(&afu->dev, "AFU does not support any processes\n");
+		return -EINVAL;
+	}
+
 	return 0;
 }
 
@@ -753,11 +1031,13 @@ static int sanitise_afu_regs(struct cxl_afu *afu)
 		else
 			cxl_p2n_write(afu, CXL_PSL_TFC_An, CXL_PSL_TFC_An_A);
 	}
-	reg = cxl_p1n_read(afu, CXL_PSL_SERR_An);
-	if (reg) {
-		if (reg & ~0xffff)
-			dev_warn(&afu->dev, "AFU had pending SERR: %#016llx\n", reg);
-		cxl_p1n_write(afu, CXL_PSL_SERR_An, reg & ~0xffff);
+	if (afu->adapter->native->sl_ops->register_serr_irq) {
+		reg = cxl_p1n_read(afu, CXL_PSL_SERR_An);
+		if (reg) {
+			if (reg & ~0xffff)
+				dev_warn(&afu->dev, "AFU had pending SERR: %#016llx\n", reg);
+			cxl_p1n_write(afu, CXL_PSL_SERR_An, reg & ~0xffff);
+		}
 	}
 	reg = cxl_p2n_read(afu, CXL_PSL_ErrStat_An);
 	if (reg) {
@@ -835,11 +1115,13 @@ static int pci_configure_afu(struct cxl_afu *afu, struct cxl *adapter, struct pc
 	if ((rc = cxl_afu_descriptor_looks_ok(afu)))
 		goto err1;
 
-	if ((rc = init_implementation_afu_regs(afu)))
-		goto err1;
+	if (adapter->native->sl_ops->afu_regs_init)
+		if ((rc = adapter->native->sl_ops->afu_regs_init(afu)))
+			goto err1;
 
-	if ((rc = cxl_native_register_serr_irq(afu)))
-		goto err1;
+	if (adapter->native->sl_ops->register_serr_irq)
+		if ((rc = adapter->native->sl_ops->register_serr_irq(afu)))
+			goto err1;
 
 	if ((rc = cxl_native_register_psl_irq(afu)))
 		goto err2;
@@ -847,7 +1129,8 @@ static int pci_configure_afu(struct cxl_afu *afu, struct cxl *adapter, struct pc
 	return 0;
 
 err2:
-	cxl_native_release_serr_irq(afu);
+	if (adapter->native->sl_ops->release_serr_irq)
+		adapter->native->sl_ops->release_serr_irq(afu);
 err1:
 	pci_unmap_slice_regs(afu);
 	return rc;
@@ -856,7 +1139,8 @@ err1:
 static void pci_deconfigure_afu(struct cxl_afu *afu)
 {
 	cxl_native_release_psl_irq(afu);
-	cxl_native_release_serr_irq(afu);
+	if (afu->adapter->native->sl_ops->release_serr_irq)
+		afu->adapter->native->sl_ops->release_serr_irq(afu);
 	pci_unmap_slice_regs(afu);
 }
 
@@ -1165,7 +1449,7 @@ static int cxl_configure_adapter(struct cxl *adapter, struct pci_dev *dev)
 	if ((rc = setup_cxl_bars(dev)))
 		return rc;
 
-	if ((rc = switch_card_to_cxl(dev)))
+	if ((rc = setup_cxl_protocol_area(dev)))
 		return rc;
 
 	if ((rc = cxl_update_image_control(adapter)))
@@ -1177,10 +1461,13 @@ static int cxl_configure_adapter(struct cxl *adapter, struct pci_dev *dev)
 	if ((rc = sanitise_adapter_regs(adapter)))
 		goto err;
 
-	if ((rc = init_implementation_adapter_regs(adapter, dev)))
+	if ((rc = adapter->native->sl_ops->adapter_regs_init(adapter, dev)))
 		goto err;
 
-	if ((rc = pnv_phb_to_cxl_mode(dev, OPAL_PHB_CAPI_MODE_CAPI)))
+	/* Required for devices using CAPP DMA mode, harmless for others */
+	pci_set_master(dev);
+
+	if ((rc = pnv_phb_to_cxl_mode(dev, adapter->native->sl_ops->capi_mode)))
 		goto err;
 
 	/* If recovery happened, the last step is to turn on snooping.
@@ -1212,6 +1499,43 @@ static void cxl_deconfigure_adapter(struct cxl *adapter)
 	pci_disable_device(pdev);
 }
 
+static const struct cxl_service_layer_ops psl_ops = {
+	.adapter_regs_init = init_implementation_adapter_psl_regs,
+	.afu_regs_init = init_implementation_afu_psl_regs,
+	.register_serr_irq = cxl_native_register_serr_irq,
+	.release_serr_irq = cxl_native_release_serr_irq,
+	.debugfs_add_adapter_sl_regs = cxl_debugfs_add_adapter_psl_regs,
+	.debugfs_add_afu_sl_regs = cxl_debugfs_add_afu_psl_regs,
+	.psl_irq_dump_registers = cxl_native_psl_irq_dump_regs,
+	.err_irq_dump_registers = cxl_native_err_irq_dump_regs,
+	.debugfs_stop_trace = cxl_stop_trace,
+	.write_timebase_ctrl = write_timebase_ctrl_psl,
+	.timebase_read = timebase_read_psl,
+	.capi_mode = OPAL_PHB_CAPI_MODE_CAPI,
+	.needs_reset_before_disable = true,
+};
+
+static const struct cxl_service_layer_ops xsl_ops = {
+	.adapter_regs_init = init_implementation_adapter_xsl_regs,
+	.debugfs_add_adapter_sl_regs = cxl_debugfs_add_adapter_xsl_regs,
+	.write_timebase_ctrl = write_timebase_ctrl_xsl,
+	.timebase_read = timebase_read_xsl,
+	.capi_mode = OPAL_PHB_CAPI_MODE_DMA,
+	.min_pe = 1, /* Workaround for Mellanox CX4 HW bug */
+};
+
+static void set_sl_ops(struct cxl *adapter, struct pci_dev *dev)
+{
+	if (dev->vendor == PCI_VENDOR_ID_MELLANOX && dev->device == 0x1013) {
+		dev_info(&adapter->dev, "Device uses an XSL\n");
+		adapter->native->sl_ops = &xsl_ops;
+	} else {
+		dev_info(&adapter->dev, "Device uses a PSL\n");
+		adapter->native->sl_ops = &psl_ops;
+	}
+}
+
+
 static struct cxl *cxl_pci_init_adapter(struct pci_dev *dev)
 {
 	struct cxl *adapter;
@@ -1227,6 +1551,8 @@ static struct cxl *cxl_pci_init_adapter(struct pci_dev *dev)
 		goto err_release;
 	}
 
+	set_sl_ops(adapter, dev);
+
 	/* Set defaults for parameters which need to persist over
 	 * configure/reconfigure
 	 */
@@ -1280,6 +1606,67 @@ static void cxl_pci_remove_adapter(struct cxl *adapter)
 	device_unregister(&adapter->dev);
 }
 
+#define CXL_MAX_PCIEX_PARENT 2
+
+static int cxl_slot_is_switched(struct pci_dev *dev)
+{
+	struct device_node *np;
+	int depth = 0;
+	const __be32 *prop;
+
+	if (!(np = pci_device_to_OF_node(dev))) {
+		pr_err("cxl: np = NULL\n");
+		return -ENODEV;
+	}
+	of_node_get(np);
+	while (np) {
+		np = of_get_next_parent(np);
+		prop = of_get_property(np, "device_type", NULL);
+		if (!prop || strcmp((char *)prop, "pciex"))
+			break;
+		depth++;
+	}
+	of_node_put(np);
+	return (depth > CXL_MAX_PCIEX_PARENT);
+}
+
+bool cxl_slot_is_supported(struct pci_dev *dev, int flags)
+{
+	if (!cpu_has_feature(CPU_FTR_HVMODE))
+		return false;
+
+	if ((flags & CXL_SLOT_FLAG_DMA) && (!pvr_version_is(PVR_POWER8NVL))) {
+		/*
+		 * CAPP DMA mode is technically supported on regular P8, but
+		 * will EEH if the card attempts to access memory < 4GB, which
+		 * we cannot realistically avoid. We might be able to work
+		 * around the issue, but until then return unsupported:
+		 */
+		return false;
+	}
+
+	if (cxl_slot_is_switched(dev))
+		return false;
+
+	/*
+	 * XXX: This gets a little tricky on regular P8 (not POWER8NVL) since
+	 * the CAPP can be connected to PHB 0, 1 or 2 on a first come first
+	 * served basis, which is racy to check from here. If we need to
+	 * support this in future we might need to consider having this
+	 * function effectively reserve it ahead of time.
+	 *
+	 * Currently, the only user of this API is the Mellanox CX4, which is
+	 * only supported on P8NVL due to the above mentioned limitation of
+	 * CAPP DMA mode and therefore does not need to worry about this. If the
+	 * issue with CAPP DMA mode is later worked around on P8 we might need
+	 * to revisit this.
+	 */
+
+	return true;
+}
+EXPORT_SYMBOL_GPL(cxl_slot_is_supported);
+
+
 static int cxl_probe(struct pci_dev *dev, const struct pci_device_id *id)
 {
 	struct cxl *adapter;
@@ -1291,6 +1678,11 @@ static int cxl_probe(struct pci_dev *dev, const struct pci_device_id *id)
 		return -ENODEV;
 	}
 
+	if (cxl_slot_is_switched(dev)) {
+		dev_info(&dev->dev, "Ignoring card on incompatible PCI slot\n");
+		return -ENODEV;
+	}
+
 	if (cxl_verbose)
 		dump_cxl_config_space(dev);
 
@@ -1311,6 +1703,9 @@ static int cxl_probe(struct pci_dev *dev, const struct pci_device_id *id)
 			dev_err(&dev->dev, "AFU %i failed to start: %i\n", slice, rc);
 	}
 
+	if (pnv_pci_on_cxl_phb(dev) && adapter->slices >= 1)
+		pnv_cxl_phb_set_peer_afu(dev, adapter->afu[0]);
+
 	return 0;
 }
 
@@ -1381,6 +1776,9 @@ static pci_ers_result_t cxl_pci_error_detected(struct pci_dev *pdev,
 		 */
 		for (i = 0; i < adapter->slices; i++) {
 			afu = adapter->afu[i];
+			/* Only participate in EEH if we are on a virtual PHB */
+			if (afu->phb == NULL)
+				return PCI_ERS_RESULT_NONE;
 			cxl_vphb_error_detected(afu, state);
 		}
 		return PCI_ERS_RESULT_DISCONNECT;
diff --git a/drivers/misc/cxl/phb.c b/drivers/misc/cxl/phb.c
new file mode 100644
index 000000000000..0935d44c1770
--- /dev/null
+++ b/drivers/misc/cxl/phb.c
@@ -0,0 +1,44 @@
+/*
+ * Copyright 2014-2016 IBM Corp.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version
+ * 2 of the License, or (at your option) any later version.
+ */
+
+#include <linux/pci.h>
+#include "cxl.h"
+
+bool _cxl_pci_associate_default_context(struct pci_dev *dev, struct cxl_afu *afu)
+{
+	struct cxl_context *ctx;
+
+	/*
+	 * Allocate a context to do cxl things to. This is used for interrupts
+	 * in the peer model using a real phb, and if we eventually do DMA ops
+	 * in the virtual phb, we'll need a default context to attach them to.
+	 */
+	ctx = cxl_dev_context_init(dev);
+	if (!ctx)
+		return false;
+	dev->dev.archdata.cxl_ctx = ctx;
+
+	return (cxl_ops->afu_check_and_enable(afu) == 0);
+}
+/* exported via cxl_base */
+
+void _cxl_pci_disable_device(struct pci_dev *dev)
+{
+	struct cxl_context *ctx = cxl_get_context(dev);
+
+	if (ctx) {
+		if (ctx->status == STARTED) {
+			dev_err(&dev->dev, "Default context started\n");
+			return;
+		}
+		dev->dev.archdata.cxl_ctx = NULL;
+		cxl_release_context(ctx);
+	}
+}
+/* exported via cxl_base */
diff --git a/drivers/misc/cxl/vphb.c b/drivers/misc/cxl/vphb.c
index cdc7723b845d..dee8def1c193 100644
--- a/drivers/misc/cxl/vphb.c
+++ b/drivers/misc/cxl/vphb.c
@@ -9,6 +9,7 @@
 
 #include <linux/pci.h>
 #include <misc/cxl.h>
+#include <asm/pnv-pci.h>
 #include "cxl.h"
 
 static int cxl_dma_set_mask(struct pci_dev *pdev, u64 dma_mask)
@@ -44,7 +45,6 @@ static bool cxl_pci_enable_device_hook(struct pci_dev *dev)
 {
 	struct pci_controller *phb;
 	struct cxl_afu *afu;
-	struct cxl_context *ctx;
 
 	phb = pci_bus_to_host(dev->bus);
 	afu = (struct cxl_afu *)phb->private_data;
@@ -57,30 +57,7 @@ static bool cxl_pci_enable_device_hook(struct pci_dev *dev)
 	set_dma_ops(&dev->dev, &dma_direct_ops);
 	set_dma_offset(&dev->dev, PAGE_OFFSET);
 
-	/*
-	 * Allocate a context to do cxl things too.  If we eventually do real
-	 * DMA ops, we'll need a default context to attach them to
-	 */
-	ctx = cxl_dev_context_init(dev);
-	if (!ctx)
-		return false;
-	dev->dev.archdata.cxl_ctx = ctx;
-
-	return (cxl_ops->afu_check_and_enable(afu) == 0);
-}
-
-static void cxl_pci_disable_device(struct pci_dev *dev)
-{
-	struct cxl_context *ctx = cxl_get_context(dev);
-
-	if (ctx) {
-		if (ctx->status == STARTED) {
-			dev_err(&dev->dev, "Default context started\n");
-			return;
-		}
-		dev->dev.archdata.cxl_ctx = NULL;
-		cxl_release_context(ctx);
-	}
+	return _cxl_pci_associate_default_context(dev, afu);
 }
 
 static resource_size_t cxl_pci_window_alignment(struct pci_bus *bus,
@@ -197,8 +174,8 @@ static struct pci_controller_ops cxl_pci_controller_ops =
 {
 	.probe_mode = cxl_pci_probe_mode,
 	.enable_device_hook = cxl_pci_enable_device_hook,
-	.disable_device = cxl_pci_disable_device,
-	.release_device = cxl_pci_disable_device,
+	.disable_device = _cxl_pci_disable_device,
+	.release_device = _cxl_pci_disable_device,
 	.window_alignment = cxl_pci_window_alignment,
 	.reset_secondary_bus = cxl_pci_reset_secondary_bus,
 	.setup_msi_irqs = cxl_setup_msi_irqs,
@@ -208,20 +185,30 @@ static struct pci_controller_ops cxl_pci_controller_ops =
 
 int cxl_pci_vphb_add(struct cxl_afu *afu)
 {
-	struct pci_dev *phys_dev;
-	struct pci_controller *phb, *phys_phb;
+	struct pci_controller *phb;
 	struct device_node *vphb_dn;
 	struct device *parent;
 
-	if (cpu_has_feature(CPU_FTR_HVMODE)) {
-		phys_dev = to_pci_dev(afu->adapter->dev.parent);
-		phys_phb = pci_bus_to_host(phys_dev->bus);
-		vphb_dn = phys_phb->dn;
-		parent = &phys_dev->dev;
-	} else {
-		vphb_dn = afu->adapter->dev.parent->of_node;
-		parent = afu->adapter->dev.parent;
-	}
+	/*
+	 * If there are no AFU configuration records we won't have anything to
+	 * expose under the vPHB, so skip creating one, returning success since
+	 * this is still a valid case. This will also opt us out of EEH
+	 * handling since we won't have anything special to do if there are no
+	 * kernel drivers attached to the vPHB, and EEH handling is not yet
+	 * supported in the peer model.
+	 */
+	if (!afu->crs_num)
+		return 0;
+
+	/* The parent device is the adapter. Reuse the device node of
+	 * the adapter.
+	 * We don't seem to care what device node is used for the vPHB,
+	 * but tools such as lsvpd walk up the device parents looking
+	 * for a valid location code, so we might as well show devices
+	 * attached to the adapter as being located on that adapter.
+	 */
+	parent = afu->adapter->dev.parent;
+	vphb_dn = parent->of_node;
 
 	/* Alloc and setup PHB data structure */
 	phb = pcibios_alloc_controller(vphb_dn);
@@ -272,13 +259,18 @@ void cxl_pci_vphb_remove(struct cxl_afu *afu)
 	pcibios_free_controller(phb);
 }
 
+static bool _cxl_pci_is_vphb_device(struct pci_controller *phb)
+{
+	return (phb->ops == &cxl_pcie_pci_ops);
+}
+
 bool cxl_pci_is_vphb_device(struct pci_dev *dev)
 {
 	struct pci_controller *phb;
 
 	phb = pci_bus_to_host(dev->bus);
 
-	return (phb->ops == &cxl_pcie_pci_ops);
+	return _cxl_pci_is_vphb_device(phb);
 }
 
 struct cxl_afu *cxl_pci_to_afu(struct pci_dev *dev)
@@ -287,7 +279,13 @@ struct cxl_afu *cxl_pci_to_afu(struct pci_dev *dev)
 
 	phb = pci_bus_to_host(dev->bus);
 
-	return (struct cxl_afu *)phb->private_data;
+	if (_cxl_pci_is_vphb_device(phb))
+		return (struct cxl_afu *)phb->private_data;
+
+	if (pnv_pci_on_cxl_phb(dev))
+		return pnv_cxl_phb_to_afu(phb);
+
+	return ERR_PTR(-ENODEV);
 }
 EXPORT_SYMBOL_GPL(cxl_pci_to_afu);