Merge tag 's390-6.5-1' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux

Pull s390 updates from Alexander Gordeev:

- Fix the style of protected key API driver source: use x-mas tree for
all local variable declarations

- Rework protected key API driver to not use the struct pkey_protkey
and pkey_clrkey anymore. Both structures have a fixed size buffer,
but with the support of ECC protected key these buffers are not big
enough. Use dynamic buffers internally and transparently for
userspace

- Add support for a new 'non CCA clear key token' with ECC clear keys
supported: ECC P256, ECC P384, ECC P521, ECC ED25519 and ECC ED448.
This makes it possible to derive a protected key from the ECC clear
key input via PKEY_KBLOB2PROTK3 ioctl, while currently the only way
to derive is via PCKMO instruction

- The s390 PMU of PAI crypto and extension 1 NNPA counters use atomic_t
for reference counting. Replace this with the proper data type
refcount_t

- Select ARCH_SUPPORTS_INT128, but limit this to clang for now, since
gcc generates inefficient code, which may lead to stack overflows

- Replace one-element array with flexible-array member in struct
vfio_ccw_parent and refactor the rest of the code accordingly. Also,
prefer struct_size() over sizeof() open- coded versions

- Introduce OS_INFO_FLAGS_ENTRY pointing to a flags field and
OS_INFO_FLAG_REIPL_CLEAR flag that informs a dumper whether the
system memory should be cleared or not once dumped

- Fix a hang when a user attempts to remove a VFIO-AP mediated device
attached to a guest: add VFIO_DEVICE_GET_IRQ_INFO and
VFIO_DEVICE_SET_IRQS IOCTLs and wire up the VFIO bus driver callback
to request a release of the device

- Fix calculation for R_390_GOTENT relocations for modules

- Allow any user space process with CAP_PERFMON capability read and
display the CPU Measurement facility counter sets

- Rework large statically-defined per-CPU cpu_cf_events data structure
and replace it with dynamically allocated structures created when a
perf_event_open() system call is invoked or /dev/hwctr device is
accessed

* tag 's390-6.5-1' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
s390/cpum_cf: rework PER_CPU_DEFINE of struct cpu_cf_events
s390/cpum_cf: open access to hwctr device for CAP_PERFMON privileged process
s390/module: fix rela calculation for R_390_GOTENT
s390/vfio-ap: wire in the vfio_device_ops request callback
s390/vfio-ap: realize the VFIO_DEVICE_SET_IRQS ioctl
s390/vfio-ap: realize the VFIO_DEVICE_GET_IRQ_INFO ioctl
s390/pkey: add support for ecc clear key
s390/pkey: do not use struct pkey_protkey
s390/pkey: introduce reverse x-mas trees
s390/zcore: conditionally clear memory on reipl
s390/ipl: add REIPL_CLEAR flag to os_info
vfio/ccw: use struct_size() helper
vfio/ccw: replace one-element array with flexible-array member
s390: select ARCH_SUPPORTS_INT128
s390/pai_ext: replace atomic_t with refcount_t
s390/pai_crypto: replace atomic_t with refcount_t

arch/s390/Kconfig

···
 	select ARCH_SUPPORTS_ATOMIC_RMW
 	select ARCH_SUPPORTS_DEBUG_PAGEALLOC
 	select ARCH_SUPPORTS_HUGETLBFS
+	select ARCH_SUPPORTS_INT128 if CC_HAS_INT128 && CC_IS_CLANG
 	select ARCH_SUPPORTS_NUMA_BALANCING
 	select ARCH_SUPPORTS_PER_VMA_LOCK
 	select ARCH_USE_BUILTIN_BSWAP

arch/s390/crypto/paes_s390.c

···
  * s390 implementation of the AES Cipher Algorithm with protected keys.
  *
  * s390 Version:
- *   Copyright IBM Corp. 2017,2020
+ *   Copyright IBM Corp. 2017, 2023
  *   Author(s): Martin Schwidefsky <[e-mail]>
  *		Harald Freudenberger <[e-mail]>
  */
···
 		if (i > 0 && ret == -EAGAIN && in_task())
 			if (msleep_interruptible(1000))
 				return -EINTR;
-		ret = pkey_keyblob2pkey(kb->key, kb->keylen, pk);
+		ret = pkey_keyblob2pkey(kb->key, kb->keylen,
+					pk->protkey, &pk->len, &pk->type);
 		if (ret == 0)
 			break;
 	}
···
 	int ret;
 	struct pkey_protkey pkey;
 
+	pkey.len = sizeof(pkey.protkey);
 	ret = __paes_keyblob2pkey(&ctx->kb, &pkey);
 	if (ret)
 		return ret;
···
 static inline int __xts_paes_convert_key(struct s390_pxts_ctx *ctx)
 {
 	struct pkey_protkey pkey0, pkey1;
+
+	pkey0.len = sizeof(pkey0.protkey);
+	pkey1.len = sizeof(pkey1.protkey);
 
 	if (__paes_keyblob2pkey(&ctx->kb[0], &pkey0) ||
 	    __paes_keyblob2pkey(&ctx->kb[1], &pkey1))

arch/s390/include/asm/asm-prototypes.h

···
 #include <asm/fpu/api.h>
 #include <asm-generic/asm-prototypes.h>
 
+__int128_t __ashlti3(__int128_t a, int b);
+__int128_t __ashrti3(__int128_t a, int b);
+__int128_t __lshrti3(__int128_t a, int b);
+
 #endif /* _ASM_S390_PROTOTYPES_H */

arch/s390/include/asm/cpacf.h

···
 /*
  * CP Assist for Cryptographic Functions (CPACF)
  *
- * Copyright IBM Corp. 2003, 2017
+ * Copyright IBM Corp. 2003, 2023
  * Author(s): Thomas Spatzier
  *	      Jan Glauber
  *	      Harald Freudenberger ([e-mail])
···
 #define CPACF_PCKMO_ENC_AES_128_KEY	0x12
 #define CPACF_PCKMO_ENC_AES_192_KEY	0x13
 #define CPACF_PCKMO_ENC_AES_256_KEY	0x14
+#define CPACF_PCKMO_ENC_ECC_P256_KEY	0x20
+#define CPACF_PCKMO_ENC_ECC_P384_KEY	0x21
+#define CPACF_PCKMO_ENC_ECC_P521_KEY	0x22
+#define CPACF_PCKMO_ENC_ECC_ED25519_KEY	0x28
+#define CPACF_PCKMO_ENC_ECC_ED448_KEY	0x29
 
 /*
  * Function codes for the PRNO (PERFORM RANDOM NUMBER OPERATION)

arch/s390/include/asm/os_info.h

···
 
 #define OS_INFO_VMCOREINFO	0
 #define OS_INFO_REIPL_BLOCK	1
+#define OS_INFO_FLAGS_ENTRY	2
+
+#define OS_INFO_FLAG_REIPL_CLEAR	(1UL << 0)
 
 struct os_info_entry {
 	u64	addr;
···
 	u16	version_minor;
 	u64	crashkernel_addr;
 	u64	crashkernel_size;
-	struct os_info_entry entry[2];
-	u8	reserved[4024];
+	struct os_info_entry entry[3];
+	u8	reserved[4004];
 } __packed;
 
 void os_info_init(void);

arch/s390/include/asm/pkey.h

···
 /*
  * Kernelspace interface to the pkey device driver
  *
- * Copyright IBM Corp. 2016,2019
+ * Copyright IBM Corp. 2016, 2023
  *
  * Author: Harald Freudenberger <[e-mail]>
  *
···
  * @return 0 on success, negative errno value on failure
  */
 int pkey_keyblob2pkey(const u8 *key, u32 keylen,
-		      struct pkey_protkey *protkey);
+		      u8 *protkey, u32 *protkeylen, u32 *protkeytype);
 
 #endif /* _KAPI_PKEY_H */

arch/s390/include/uapi/asm/pkey.h

···
 /*
  * Userspace interface to the pkey device driver
  *
- * Copyright IBM Corp. 2017, 2019
+ * Copyright IBM Corp. 2017, 2023
  *
  * Author: Harald Freudenberger <[e-mail]>
  *
···
 #define MINKEYBLOBSIZE	SECKEYBLOBSIZE
 
 /* defines for the type field within the pkey_protkey struct */
-#define PKEY_KEYTYPE_AES_128		      1
-#define PKEY_KEYTYPE_AES_192		      2
-#define PKEY_KEYTYPE_AES_256		      3
-#define PKEY_KEYTYPE_ECC		      4
+#define PKEY_KEYTYPE_AES_128		1
+#define PKEY_KEYTYPE_AES_192		2
+#define PKEY_KEYTYPE_AES_256		3
+#define PKEY_KEYTYPE_ECC		4
+#define PKEY_KEYTYPE_ECC_P256		5
+#define PKEY_KEYTYPE_ECC_P384		6
+#define PKEY_KEYTYPE_ECC_P521		7
+#define PKEY_KEYTYPE_ECC_ED25519	8
+#define PKEY_KEYTYPE_ECC_ED448		9
 
 /* the newer ioctls use a pkey_key_type enum for type information */
 enum pkey_key_type {

arch/s390/kernel/ipl.c

···
 static bool reipl_ccw_clear;
 static bool reipl_eckd_clear;
 
+static unsigned long os_info_flags;
+
 static inline int __diag308(unsigned long subcode, unsigned long addr)
 {
 	union register_pair r1;
···
 	struct lowcore *abs_lc;
 	unsigned int csum;
 
+	/*
+	 * Set REIPL_CLEAR flag in os_info flags entry indicating
+	 * 'clear' sysfs attribute has been set on the panicked system
+	 * for specified reipl type.
+	 * Always set for IPL_TYPE_NSS and IPL_TYPE_UNKNOWN.
+	 */
+	if ((reipl_type == IPL_TYPE_CCW && reipl_ccw_clear) ||
+	    (reipl_type == IPL_TYPE_ECKD && reipl_eckd_clear) ||
+	    (reipl_type == IPL_TYPE_FCP && reipl_fcp_clear) ||
+	    (reipl_type == IPL_TYPE_NVME && reipl_nvme_clear) ||
+	    reipl_type == IPL_TYPE_NSS ||
+	    reipl_type == IPL_TYPE_UNKNOWN)
+		os_info_flags |= OS_INFO_FLAG_REIPL_CLEAR;
+	os_info_entry_add(OS_INFO_FLAGS_ENTRY, &os_info_flags, sizeof(os_info_flags));
 	csum = (__force unsigned int)
 	       csum_partial(reipl_block_actual, reipl_block_actual->hdr.len, 0);
 	abs_lc = get_abs_lowcore();

arch/s390/kernel/module.c

···
 			rc = apply_rela_bits(loc, val, 0, 64, 0, write);
 		else if (r_type == R_390_GOTENT ||
 			 r_type == R_390_GOTPLTENT) {
-			val += (Elf_Addr) me->mem[MOD_TEXT].base - loc;
+			val += (Elf_Addr)me->mem[MOD_TEXT].base +
+				me->arch.got_offset - loc;
 			rc = apply_rela_bits(loc, val, 1, 32, 1, write);
 		}
 		break;

arch/s390/kernel/perf_cpum_cf.c

···
 }
 
 struct cpu_cf_events {
+	refcount_t refcnt;		/* Reference count */
 	atomic_t		ctr_set[CPUMF_CTR_SET_MAX];
 	u64			state;		/* For perf_event_open SVC */
 	u64			dev_state;	/* For /dev/hwctr */
···
 	unsigned int sets;		/* # Counter set saved in memory */
 };
 
-/* Per-CPU event structure for the counter facility */
-static DEFINE_PER_CPU(struct cpu_cf_events, cpu_cf_events);
-
 static unsigned int cfdiag_cpu_speed;	/* CPU speed for CF_DIAG trailer */
 static debug_info_t *cf_dbg;
 
···
  * of counter sets. Extract this information at device driver initialization.
  */
 static struct cpumf_ctr_info	cpumf_ctr_info;
+
+struct cpu_cf_ptr {
+	struct cpu_cf_events *cpucf;
+};
+
+static struct cpu_cf_root {		/* Anchor to per CPU data */
+	refcount_t refcnt;		/* Overall active events */
+	struct cpu_cf_ptr __percpu *cfptr;
+} cpu_cf_root;
+
+/*
+ * Serialize event initialization and event removal. Both are called from
+ * user space in task context with perf_event_open() and close()
+ * system calls.
+ *
+ * This mutex serializes functions cpum_cf_alloc_cpu() called at event
+ * initialization via cpumf_pmu_event_init() and function cpum_cf_free_cpu()
+ * called at event removal via call back function hw_perf_event_destroy()
+ * when the event is deleted. They are serialized to enforce correct
+ * bookkeeping of pointer and reference counts anchored by
+ * struct cpu_cf_root and the access to cpu_cf_root::refcnt and the
+ * per CPU pointers stored in cpu_cf_root::cfptr.
+ */
+static DEFINE_MUTEX(pmc_reserve_mutex);
+
+/*
+ * Get pointer to per-cpu structure.
+ *
+ * Function get_cpu_cfhw() is called from
+ * - cfset_copy_all(): This function is protected by cpus_read_lock(), so
+ *   CPU hot plug remove can not happen. Event removal requires a close()
+ *   first.
+ *
+ * Function this_cpu_cfhw() is called from perf common code functions:
+ * - pmu_{en|dis}able(), pmu_{add|del}()and pmu_{start|stop}():
+ *   All functions execute with interrupts disabled on that particular CPU.
+ * - cfset_ioctl_{on|off}, cfset_cpu_read(): see comment cfset_copy_all().
+ *
+ * Therefore it is safe to access the CPU specific pointer to the event.
+ */
+static struct cpu_cf_events *get_cpu_cfhw(int cpu)
+{
+	struct cpu_cf_ptr __percpu *p = cpu_cf_root.cfptr;
+
+	if (p) {
+		struct cpu_cf_ptr *q = per_cpu_ptr(p, cpu);
+
+		return q->cpucf;
+	}
+	return NULL;
+}
+
+static struct cpu_cf_events *this_cpu_cfhw(void)
+{
+	return get_cpu_cfhw(smp_processor_id());
+}
+
+/* Disable counter sets on dedicated CPU */
+static void cpum_cf_reset_cpu(void *flags)
+{
+	lcctl(0);
+}
+
+/* Free per CPU data when the last event is removed. */
+static void cpum_cf_free_root(void)
+{
+	if (!refcount_dec_and_test(&cpu_cf_root.refcnt))
+		return;
+	free_percpu(cpu_cf_root.cfptr);
+	cpu_cf_root.cfptr = NULL;
+	irq_subclass_unregister(IRQ_SUBCLASS_MEASUREMENT_ALERT);
+	on_each_cpu(cpum_cf_reset_cpu, NULL, 1);
+	debug_sprintf_event(cf_dbg, 4, "%s2 root.refcnt %u cfptr %px\n",
+			    __func__, refcount_read(&cpu_cf_root.refcnt),
+			    cpu_cf_root.cfptr);
+}
+
+/*
+ * On initialization of first event also allocate per CPU data dynamically.
+ * Start with an array of pointers, the array size is the maximum number of
+ * CPUs possible, which might be larger than the number of CPUs currently
+ * online.
+ */
+static int cpum_cf_alloc_root(void)
+{
+	int rc = 0;
+
+	if (refcount_inc_not_zero(&cpu_cf_root.refcnt))
+		return rc;
+
+	/* The memory is already zeroed. */
+	cpu_cf_root.cfptr = alloc_percpu(struct cpu_cf_ptr);
+	if (cpu_cf_root.cfptr) {
+		refcount_set(&cpu_cf_root.refcnt, 1);
+		on_each_cpu(cpum_cf_reset_cpu, NULL, 1);
+		irq_subclass_register(IRQ_SUBCLASS_MEASUREMENT_ALERT);
+	} else {
+		rc = -ENOMEM;
+	}
+
+	return rc;
+}
+
+/* Free CPU counter data structure for a PMU */
+static void cpum_cf_free_cpu(int cpu)
+{
+	struct cpu_cf_events *cpuhw;
+	struct cpu_cf_ptr *p;
+
+	mutex_lock(&pmc_reserve_mutex);
+	/*
+	 * When invoked via CPU hotplug handler, there might be no events
+	 * installed or that particular CPU might not have an
+	 * event installed. This anchor pointer can be NULL!
+	 */
+	if (!cpu_cf_root.cfptr)
+		goto out;
+	p = per_cpu_ptr(cpu_cf_root.cfptr, cpu);
+	cpuhw = p->cpucf;
+	/*
+	 * Might be zero when called from CPU hotplug handler and no event
+	 * installed on that CPU, but on different CPUs.
+	 */
+	if (!cpuhw)
+		goto out;
+
+	if (refcount_dec_and_test(&cpuhw->refcnt)) {
+		kfree(cpuhw);
+		p->cpucf = NULL;
+	}
+	cpum_cf_free_root();
+out:
+	mutex_unlock(&pmc_reserve_mutex);
+}
+
+/* Allocate CPU counter data structure for a PMU. Called under mutex lock. */
+static int cpum_cf_alloc_cpu(int cpu)
+{
+	struct cpu_cf_events *cpuhw;
+	struct cpu_cf_ptr *p;
+	int rc;
+
+	mutex_lock(&pmc_reserve_mutex);
+	rc = cpum_cf_alloc_root();
+	if (rc)
+		goto unlock;
+	p = per_cpu_ptr(cpu_cf_root.cfptr, cpu);
+	cpuhw = p->cpucf;
+
+	if (!cpuhw) {
+		cpuhw = kzalloc(sizeof(*cpuhw), GFP_KERNEL);
+		if (cpuhw) {
+			p->cpucf = cpuhw;
+			refcount_set(&cpuhw->refcnt, 1);
+		} else {
+			rc = -ENOMEM;
+		}
+	} else {
+		refcount_inc(&cpuhw->refcnt);
+	}
+	if (rc) {
+		/*
+		 * Error in allocation of event, decrement anchor. Since
+		 * cpu_cf_event in not created, its destroy() function is not
+		 * invoked. Adjust the reference counter for the anchor.
+		 */
+		cpum_cf_free_root();
+	}
+unlock:
+	mutex_unlock(&pmc_reserve_mutex);
+	return rc;
+}
+
+/*
+ * Create/delete per CPU data structures for /dev/hwctr interface and events
+ * created by perf_event_open().
+ * If cpu is -1, track task on all available CPUs. This requires
+ * allocation of hardware data structures for all CPUs. This setup handles
+ * perf_event_open() with task context and /dev/hwctr interface.
+ * If cpu is non-zero install event on this CPU only. This setup handles
+ * perf_event_open() with CPU context.
+ */
+static int cpum_cf_alloc(int cpu)
+{
+	cpumask_var_t mask;
+	int rc;
+
+	if (cpu == -1) {
+		if (!zalloc_cpumask_var(&mask, GFP_KERNEL))
+			return -ENOMEM;
+		for_each_online_cpu(cpu) {
+			rc = cpum_cf_alloc_cpu(cpu);
+			if (rc) {
+				for_each_cpu(cpu, mask)
+					cpum_cf_free_cpu(cpu);
+				break;
+			}
+			cpumask_set_cpu(cpu, mask);
+		}
+		free_cpumask_var(mask);
+	} else {
+		rc = cpum_cf_alloc_cpu(cpu);
+	}
+	return rc;
+}
+
+static void cpum_cf_free(int cpu)
+{
+	if (cpu == -1) {
+		for_each_online_cpu(cpu)
+			cpum_cf_free_cpu(cpu);
+	} else {
+		cpum_cf_free_cpu(cpu);
+	}
+}
 
 #define	CF_DIAG_CTRSET_DEF		0xfeef	/* Counter set header mark */
 						/* interval in seconds */
···
  */
 static void cpumf_pmu_enable(struct pmu *pmu)
 {
-	struct cpu_cf_events *cpuhw = this_cpu_ptr(&cpu_cf_events);
+	struct cpu_cf_events *cpuhw = this_cpu_cfhw();
 	int err;
 
-	if (cpuhw->flags & PMU_F_ENABLED)
+	if (!cpuhw || (cpuhw->flags & PMU_F_ENABLED))
 		return;
 
 	err = lcctl(cpuhw->state | cpuhw->dev_state);
···
  */
 static void cpumf_pmu_disable(struct pmu *pmu)
 {
-	struct cpu_cf_events *cpuhw = this_cpu_ptr(&cpu_cf_events);
-	int err;
+	struct cpu_cf_events *cpuhw = this_cpu_cfhw();
 	u64 inactive;
+	int err;
 
-	if (!(cpuhw->flags & PMU_F_ENABLED))
+	if (!cpuhw || !(cpuhw->flags & PMU_F_ENABLED))
 		return;
 
 	inactive = cpuhw->state & ~((1 << CPUMF_LCCTL_ENABLE_SHIFT) - 1);
···
 		cpuhw->flags &= ~PMU_F_ENABLED;
 }
 
-#define PMC_INIT      0UL
-#define PMC_RELEASE   1UL
-
-static void cpum_cf_setup_cpu(void *flags)
-{
-	struct cpu_cf_events *cpuhw = this_cpu_ptr(&cpu_cf_events);
-
-	switch ((unsigned long)flags) {
-	case PMC_INIT:
-		cpuhw->flags |= PMU_F_RESERVED;
-		break;
-
-	case PMC_RELEASE:
-		cpuhw->flags &= ~PMU_F_RESERVED;
-		break;
-	}
-
-	/* Disable CPU counter sets */
-	lcctl(0);
-	debug_sprintf_event(cf_dbg, 5, "%s flags %#x flags %#x state %#llx\n",
-			    __func__, *(int *)flags, cpuhw->flags,
-			    cpuhw->state);
-}
-
-/* Initialize the CPU-measurement counter facility */
-static int __kernel_cpumcf_begin(void)
-{
-	on_each_cpu(cpum_cf_setup_cpu, (void *)PMC_INIT, 1);
-	irq_subclass_register(IRQ_SUBCLASS_MEASUREMENT_ALERT);
-
-	return 0;
-}
-
-/* Release the CPU-measurement counter facility */
-static void __kernel_cpumcf_end(void)
-{
-	on_each_cpu(cpum_cf_setup_cpu, (void *)PMC_RELEASE, 1);
-	irq_subclass_unregister(IRQ_SUBCLASS_MEASUREMENT_ALERT);
-}
-
-/* Number of perf events counting hardware events */
-static atomic_t num_events = ATOMIC_INIT(0);
-/* Used to avoid races in calling reserve/release_cpumf_hardware */
-static DEFINE_MUTEX(pmc_reserve_mutex);
-
 /* Release the PMU if event is the last perf event */
 static void hw_perf_event_destroy(struct perf_event *event)
 {
-	mutex_lock(&pmc_reserve_mutex);
-	if (atomic_dec_return(&num_events) == 0)
-		__kernel_cpumcf_end();
-	mutex_unlock(&pmc_reserve_mutex);
+	cpum_cf_free(event->cpu);
 }
 
 /* CPUMF <-> perf event mappings for kernel+userspace (basic set) */
···
 	[PERF_COUNT_HW_BRANCH_MISSES]	    = -1,
 	[PERF_COUNT_HW_BUS_CYCLES]	    = -1,
 };
-
-static void cpumf_hw_inuse(void)
-{
-	mutex_lock(&pmc_reserve_mutex);
-	if (atomic_inc_return(&num_events) == 1)
-		__kernel_cpumcf_begin();
-	mutex_unlock(&pmc_reserve_mutex);
-}
 
 static int is_userspace_event(u64 ev)
 {
···
 	}
 
 	/* Initialize for using the CPU-measurement counter facility */
-	cpumf_hw_inuse();
+	if (cpum_cf_alloc(event->cpu))
+		return -ENOMEM;
 	event->destroy = hw_perf_event_destroy;
 
 	/*
···
 
 static void cpumf_pmu_start(struct perf_event *event, int flags)
 {
-	struct cpu_cf_events *cpuhw = this_cpu_ptr(&cpu_cf_events);
+	struct cpu_cf_events *cpuhw = this_cpu_cfhw();
 	struct hw_perf_event *hwc = &event->hw;
 	int i;
 
···
 
 static void cpumf_pmu_stop(struct perf_event *event, int flags)
 {
-	struct cpu_cf_events *cpuhw = this_cpu_ptr(&cpu_cf_events);
+	struct cpu_cf_events *cpuhw = this_cpu_cfhw();
 	struct hw_perf_event *hwc = &event->hw;
 	int i;
 
···
 						      false);
 			if (cfdiag_diffctr(cpuhw, event->hw.config_base))
 				cfdiag_push_sample(event, cpuhw);
-		} else if (cpuhw->flags & PMU_F_RESERVED) {
-			/* Only update when PMU not hotplugged off */
+		} else {
 			hw_perf_event_update(event);
 		}
 		hwc->state |= PERF_HES_UPTODATE;
···
 
 static int cpumf_pmu_add(struct perf_event *event, int flags)
 {
-	struct cpu_cf_events *cpuhw = this_cpu_ptr(&cpu_cf_events);
+	struct cpu_cf_events *cpuhw = this_cpu_cfhw();
 
 	ctr_set_enable(&cpuhw->state, event->hw.config_base);
 	event->hw.state = PERF_HES_UPTODATE | PERF_HES_STOPPED;
···
 
 static void cpumf_pmu_del(struct perf_event *event, int flags)
 {
-	struct cpu_cf_events *cpuhw = this_cpu_ptr(&cpu_cf_events);
+	struct cpu_cf_events *cpuhw = this_cpu_cfhw();
 	int i;
 
 	cpumf_pmu_stop(event, PERF_EF_UPDATE);
···
 	.read	      = cpumf_pmu_read,
 };
 
-static int cpum_cf_setup(unsigned int cpu, unsigned long flags)
-{
-	local_irq_disable();
-	cpum_cf_setup_cpu((void *)flags);
-	local_irq_enable();
-	return 0;
-}
+static struct cfset_session {		/* CPUs and counter set bit mask */
+	struct list_head head;		/* Head of list of active processes */
+} cfset_session = {
+	.head = LIST_HEAD_INIT(cfset_session.head)
+};
 
+static refcount_t cfset_opencnt = REFCOUNT_INIT(0);	/* Access count */
+/*
+ * Synchronize access to device /dev/hwc. This mutex protects against
+ * concurrent access to functions cfset_open() and cfset_release().
+ * Same for CPU hotplug add and remove events triggering
+ * cpum_cf_online_cpu() and cpum_cf_offline_cpu().
+ * It also serializes concurrent device ioctl access from multiple
+ * processes accessing /dev/hwc.
+ *
+ * The mutex protects concurrent access to the /dev/hwctr session management
+ * struct cfset_session and reference counting variable cfset_opencnt.
+ */
+static DEFINE_MUTEX(cfset_ctrset_mutex);
+
+/*
+ * CPU hotplug handles only /dev/hwctr device.
+ * For perf_event_open() the CPU hotplug handling is done on kernel common
+ * code:
+ * - CPU add: Nothing is done since a file descriptor can not be created
+ *   and returned to the user.
+ * - CPU delete: Handled by common code via pmu_disable(), pmu_stop() and
+ *   pmu_delete(). The event itself is removed when the file descriptor is
+ *   closed.
+ */
 static int cfset_online_cpu(unsigned int cpu);
+
 static int cpum_cf_online_cpu(unsigned int cpu)
 {
-	debug_sprintf_event(cf_dbg, 4, "%s cpu %d in_irq %ld\n", __func__,
-			    cpu, in_interrupt());
-	cpum_cf_setup(cpu, PMC_INIT);
-	return cfset_online_cpu(cpu);
+	int rc = 0;
+
+	debug_sprintf_event(cf_dbg, 4, "%s cpu %d root.refcnt %d "
+			    "opencnt %d\n", __func__, cpu,
+			    refcount_read(&cpu_cf_root.refcnt),
+			    refcount_read(&cfset_opencnt));
+	/*
+	 * Ignore notification for perf_event_open().
+	 * Handle only /dev/hwctr device sessions.
+	 */
+	mutex_lock(&cfset_ctrset_mutex);
+	if (refcount_read(&cfset_opencnt)) {
+		rc = cpum_cf_alloc_cpu(cpu);
+		if (!rc)
+			cfset_online_cpu(cpu);
+	}
+	mutex_unlock(&cfset_ctrset_mutex);
+	return rc;
 }
 
 static int cfset_offline_cpu(unsigned int cpu);
+
 static int cpum_cf_offline_cpu(unsigned int cpu)
 {
-	debug_sprintf_event(cf_dbg, 4, "%s cpu %d\n", __func__, cpu);
-	cfset_offline_cpu(cpu);
-	return cpum_cf_setup(cpu, PMC_RELEASE);
+	debug_sprintf_event(cf_dbg, 4, "%s cpu %d root.refcnt %d opencnt %d\n",
+			    __func__, cpu, refcount_read(&cpu_cf_root.refcnt),
+			    refcount_read(&cfset_opencnt));
+	/*
+	 * During task exit processing of grouped perf events triggered by CPU
+	 * hotplug processing, pmu_disable() is called as part of perf context
+	 * removal process. Therefore do not trigger event removal now for
+	 * perf_event_open() created events. Perf common code triggers event
+	 * destruction when the event file descriptor is closed.
+	 *
+	 * Handle only /dev/hwctr device sessions.
+	 */
+	mutex_lock(&cfset_ctrset_mutex);
+	if (refcount_read(&cfset_opencnt)) {
+		cfset_offline_cpu(cpu);
+		cpum_cf_free_cpu(cpu);
+	}
+	mutex_unlock(&cfset_ctrset_mutex);
+	return 0;
 }
 
 /* Return true if store counter set multiple instruction is available */
···
 		return;
 
 	inc_irq_stat(IRQEXT_CMC);
-	cpuhw = this_cpu_ptr(&cpu_cf_events);
 
 	/*
 	 * Measurement alerts are shared and might happen when the PMU
 	 * is not reserved.  Ignore these alerts in this case.
 	 */
-	if (!(cpuhw->flags & PMU_F_RESERVED))
+	cpuhw = this_cpu_cfhw();
+	if (!cpuhw)
 		return;
 
 	/* counter authorization change alert */
···
  * counter set via normal file operations.
  */
 
-static atomic_t cfset_opencnt = ATOMIC_INIT(0);		/* Access count */
-static DEFINE_MUTEX(cfset_ctrset_mutex);/* Synchronize access to hardware */
 struct cfset_call_on_cpu_parm {		/* Parm struct for smp_call_on_cpu */
 	unsigned int sets;		/* Counter set bit mask */
 	atomic_t cpus_ack;		/* # CPUs successfully executed func */
-};
-
-static struct cfset_session {		/* CPUs and counter set bit mask */
-	struct list_head head;		/* Head of list of active processes */
-} cfset_session = {
-	.head = LIST_HEAD_INIT(cfset_session.head)
 };
 
 struct cfset_request {			/* CPUs and counter set bit mask */
···
 /* Stop all counter sets via ioctl interface */
 static void cfset_ioctl_off(void *parm)
 {
-	struct cpu_cf_events *cpuhw = this_cpu_ptr(&cpu_cf_events);
+	struct cpu_cf_events *cpuhw = this_cpu_cfhw();
 	struct cfset_call_on_cpu_parm *p = parm;
 	int rc;
 
-	/* Check if any counter set used by /dev/hwc */
+	/* Check if any counter set used by /dev/hwctr */
 	for (rc = CPUMF_CTR_SET_BASIC; rc < CPUMF_CTR_SET_MAX; ++rc)
 		if ((p->sets & cpumf_ctr_ctl[rc])) {
 			if (!atomic_dec_return(&cpuhw->ctr_set[rc])) {
···
 /* Start counter sets on particular CPU */
 static void cfset_ioctl_on(void *parm)
 {
-	struct cpu_cf_events *cpuhw = this_cpu_ptr(&cpu_cf_events);
+	struct cpu_cf_events *cpuhw = this_cpu_cfhw();
 	struct cfset_call_on_cpu_parm *p = parm;
 	int rc;
 
···
 
 static void cfset_release_cpu(void *p)
 {
-	struct cpu_cf_events *cpuhw = this_cpu_ptr(&cpu_cf_events);
+	struct cpu_cf_events *cpuhw = this_cpu_cfhw();
 	int rc;
 
 	debug_sprintf_event(cf_dbg, 4, "%s state %#llx dev_state %#llx\n",
···
 		kfree(file->private_data);
 		file->private_data = NULL;
 	}
-	if (!atomic_dec_return(&cfset_opencnt))
+	if (refcount_dec_and_test(&cfset_opencnt)) {	/* Last close */
 		on_each_cpu(cfset_release_cpu, NULL, 1);
+		cpum_cf_free(-1);
+	}
 	mutex_unlock(&cfset_ctrset_mutex);
-
-	hw_perf_event_destroy(NULL);
 	return 0;
 }
 
+/*
+ * Open via /dev/hwctr device. Allocate all per CPU resources on the first
+ * open of the device. The last close releases all per CPU resources.
+ * Parallel perf_event_open system calls also use per CPU resources.
+ * These invocations are handled via reference counting on the per CPU data
+ * structures.
+ */
 static int cfset_open(struct inode *inode, struct file *file)
 {
-	if (!capable(CAP_SYS_ADMIN))
+	int rc = 0;
+
+	if (!perfmon_capable())
 		return -EPERM;
+	file->private_data = NULL;
+
 	mutex_lock(&cfset_ctrset_mutex);
-	if (atomic_inc_return(&cfset_opencnt) == 1)
-		cfset_session_init();
+	if (!refcount_inc_not_zero(&cfset_opencnt)) {	/* First open */
+		rc = cpum_cf_alloc(-1);
+		if (!rc) {
+			cfset_session_init();
+			refcount_set(&cfset_opencnt, 1);
+		}
+	}
 	mutex_unlock(&cfset_ctrset_mutex);
 
-	cpumf_hw_inuse();
-	file->private_data = NULL;
 	/* nonseekable_open() never fails */
-	return nonseekable_open(inode, file);
+	return rc ?: nonseekable_open(inode, file);
 }
 
 static int cfset_all_start(struct cfset_request *req)
···
 	ctrset_read = (struct s390_ctrset_read __user *)arg;
 	uptr = ctrset_read->data;
 	for_each_cpu(cpu, mask) {
-		struct cpu_cf_events *cpuhw = per_cpu_ptr(&cpu_cf_events, cpu);
+		struct cpu_cf_events *cpuhw = get_cpu_cfhw(cpu);
 		struct s390_ctrset_cpudata __user *ctrset_cpudata;
 
 		ctrset_cpudata = uptr;
···
 /* Read all counter sets. */
 static void cfset_cpu_read(void *parm)
 {
-	struct cpu_cf_events *cpuhw = this_cpu_ptr(&cpu_cf_events);
+	struct cpu_cf_events *cpuhw = this_cpu_cfhw();
 	struct cfset_call_on_cpu_parm *p = parm;
 	int set, set_size;
 	size_t space;
···
 			cpuhw->used += space;
 			cpuhw->sets += 1;
 		}
+		debug_sprintf_event(cf_dbg, 4, "%s sets %d used %zd\n", __func__,
+				    cpuhw->sets, cpuhw->used);
 	}
-	debug_sprintf_event(cf_dbg, 4, "%s sets %d used %zd\n", __func__,
-			    cpuhw->sets, cpuhw->used);
 }
 
 static int cfset_all_read(unsigned long arg, struct cfset_request *req)
···
 	.name	= S390_HWCTR_DEVICE,
 	.minor	= MISC_DYNAMIC_MINOR,
 	.fops	= &cfset_fops,
+	.mode	= 0666,
 };
 
 /* Hotplug add of a CPU. Scan through all active processes and add
···
 	struct cfset_call_on_cpu_parm p;
 	struct cfset_request *rp;
 
-	mutex_lock(&cfset_ctrset_mutex);
 	if (!list_empty(&cfset_session.head)) {
 		list_for_each_entry(rp, &cfset_session.head, node) {
 			p.sets = rp->ctrset;
···
 			cpumask_set_cpu(cpu, &rp->mask);
 		}
 	}
-	mutex_unlock(&cfset_ctrset_mutex);
 	return 0;
 }
 
 /* Hotplug remove of a CPU. Scan through all active processes and clear
  * that CPU from the list of CPUs supplied with ioctl(..., START, ...).
+ * Adjust reference counts.
  */
 static int cfset_offline_cpu(unsigned int cpu)
 {
 	struct cfset_call_on_cpu_parm p;
 	struct cfset_request *rp;
 
-	mutex_lock(&cfset_ctrset_mutex);
 	if (!list_empty(&cfset_session.head)) {
 		list_for_each_entry(rp, &cfset_session.head, node) {
 			p.sets = rp->ctrset;
···
 			cpumask_clear_cpu(cpu, &rp->mask);
 		}
 	}
-	mutex_unlock(&cfset_ctrset_mutex);
 	return 0;
 }
 
···
 	}
 
 	/* Initialize for using the CPU-measurement counter facility */
-	cpumf_hw_inuse();
+	if (cpum_cf_alloc(event->cpu))
+		return -ENOMEM;
 	event->destroy = hw_perf_event_destroy;
 
 	err = cfdiag_event_init2(event);

arch/s390/kernel/perf_pai_crypto.c

···
 	unsigned long *page;		/* Page for CPU to store counters */
 	struct pai_userdata *save;	/* Page to store no-zero counters */
 	unsigned int active_events;	/* # of PAI crypto users */
-	unsigned int refcnt;		/* Reference count mapped buffers */
+	refcount_t refcnt;		/* Reference count mapped buffers */
 	enum paievt_mode mode;		/* Type of event */
 	struct perf_event *event;	/* Perf event for sampling */
 };
···
 	static_branch_dec(&pai_key);
 	mutex_lock(&pai_reserve_mutex);
 	debug_sprintf_event(cfm_dbg, 5, "%s event %#llx cpu %d users %d"
-			    " mode %d refcnt %d\n", __func__,
+			    " mode %d refcnt %u\n", __func__,
 			    event->attr.config, event->cpu,
-			    cpump->active_events, cpump->mode, cpump->refcnt);
-	if (!--cpump->refcnt) {
+			    cpump->active_events, cpump->mode,
+			    refcount_read(&cpump->refcnt));
+	if (refcount_dec_and_test(&cpump->refcnt)) {
 		debug_sprintf_event(cfm_dbg, 4, "%s page %#lx save %p\n",
 				    __func__, (unsigned long)cpump->page,
 				    cpump->save);
···
 	/* Allocate memory for counter page and counter extraction.
 	 * Only the first counting event has to allocate a page.
 	 */
-	if (cpump->page)
+	if (cpump->page) {
+		refcount_inc(&cpump->refcnt);
 		goto unlock;
+	}
 
 	rc = -ENOMEM;
 	cpump->page = (unsigned long *)get_zeroed_page(GFP_KERNEL);
···
 		goto unlock;
 	}
 	rc = 0;
+	refcount_set(&cpump->refcnt, 1);
 
 unlock:
 	/* If rc is non-zero, do not set mode and reference count */
 	if (!rc) {
-		cpump->refcnt++;
 		cpump->mode = a->sample_period ? PAI_MODE_SAMPLING
 					       : PAI_MODE_COUNTING;
 	}
 	debug_sprintf_event(cfm_dbg, 5, "%s sample_period %#llx users %d"
-			    " mode %d refcnt %d page %#lx save %p rc %d\n",
+			    " mode %d refcnt %u page %#lx save %p rc %d\n",
 			    __func__, a->sample_period, cpump->active_events,
-			    cpump->mode, cpump->refcnt,
+			    cpump->mode, refcount_read(&cpump->refcnt),
 			    (unsigned long)cpump->page, cpump->save, rc);
 	mutex_unlock(&pai_reserve_mutex);
 	return rc;

arch/s390/kernel/perf_pai_ext.c

···
 	struct pai_userdata *save;	/* Area to store non-zero counters */
 	enum paievt_mode mode;		/* Type of event */
 	unsigned int active_events;	/* # of PAI Extension users */
-	unsigned int refcnt;
+	refcount_t refcnt;
 	struct perf_event *event;	/* Perf event for sampling */
 	struct paiext_cb *paiext_cb;	/* PAI extension control block area */
 };
···
 };
 
 static struct paiext_root {		/* Anchor to per CPU data */
-	int refcnt;			/* Overall active events */
+	refcount_t refcnt;		/* Overall active events */
 	struct paiext_mapptr __percpu *mapptr;
 } paiext_root;
 
 /* Free per CPU data when the last event is removed. */
 static void paiext_root_free(void)
 {
-	if (!--paiext_root.refcnt) {
+	if (refcount_dec_and_test(&paiext_root.refcnt)) {
 		free_percpu(paiext_root.mapptr);
 		paiext_root.mapptr = NULL;
 	}
···
  */
 static int paiext_root_alloc(void)
 {
-	if (++paiext_root.refcnt == 1) {
+	if (!refcount_inc_not_zero(&paiext_root.refcnt)) {
 		/* The memory is already zeroed. */
 		paiext_root.mapptr = alloc_percpu(struct paiext_mapptr);
 		if (!paiext_root.mapptr) {
···
 			 */
 			return -ENOMEM;
 		}
+		refcount_set(&paiext_root.refcnt, 1);
 	}
 	return 0;
 }
···
 
 	mutex_lock(&paiext_reserve_mutex);
 	cpump->event = NULL;
-	if (!--cpump->refcnt)		/* Last reference gone */
+	if (refcount_dec_and_test(&cpump->refcnt))	/* Last reference gone */
 		paiext_free(mp);
 	paiext_root_free();
 	mutex_unlock(&paiext_reserve_mutex);
···
 		rc = -ENOMEM;
 		cpump = kzalloc(sizeof(*cpump), GFP_KERNEL);
 		if (!cpump)
-			goto unlock;
+			goto undo;
 
 		/* Allocate memory for counter area and counter extraction.
 		 * These are
···
 					     GFP_KERNEL);
 		if (!cpump->save || !cpump->area || !cpump->paiext_cb) {
 			paiext_free(mp);
-			goto unlock;
+			goto undo;
 		}
+		refcount_set(&cpump->refcnt, 1);
 		cpump->mode = a->sample_period ? PAI_MODE_SAMPLING
 					       : PAI_MODE_COUNTING;
 	} else {
···
 		if (cpump->mode == PAI_MODE_SAMPLING ||
 		    (cpump->mode == PAI_MODE_COUNTING && a->sample_period)) {
 			rc = -EBUSY;
-			goto unlock;
+			goto undo;
 		}
+		refcount_inc(&cpump->refcnt);
 	}
 
 	rc = 0;
 	cpump->event = event;
-	++cpump->refcnt;
 
-unlock:
+undo:
 	if (rc) {
 		/* Error in allocation of event, decrement anchor. Since
 		 * the event in not created, its destroy() function is never
···
 		 */
 		paiext_root_free();
 	}
+unlock:
 	mutex_unlock(&paiext_reserve_mutex);
 	/* If rc is non-zero, no increment of counter/sampler was done. */
 	return rc;

arch/s390/lib/Makefile

···
 # Makefile for s390-specific library files..
 #
 
-lib-y += delay.o string.o uaccess.o find.o spinlock.o
+lib-y += delay.o string.o uaccess.o find.o spinlock.o tishift.o
 obj-y += mem.o xor.o
 lib-$(CONFIG_KPROBES) += probes.o
 lib-$(CONFIG_UPROBES) += probes.o

arch/s390/lib/tishift.S

···
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#include <linux/linkage.h>
+#include <asm/nospec-insn.h>
+#include <asm/export.h>
+
+	.section .noinstr.text, "ax"
+
+	GEN_BR_THUNK %r14
+
+SYM_FUNC_START(__ashlti3)
+	lmg	%r0,%r1,0(%r3)
+	cije	%r4,0,1f
+	lhi	%r3,64
+	sr	%r3,%r4
+	jnh	0f
+	srlg	%r3,%r1,0(%r3)
+	sllg	%r0,%r0,0(%r4)
+	sllg	%r1,%r1,0(%r4)
+	ogr	%r0,%r3
+	j	1f
+0:	sllg	%r0,%r1,-64(%r4)
+	lghi	%r1,0
+1:	stmg	%r0,%r1,0(%r2)
+	BR_EX	%r14
+SYM_FUNC_END(__ashlti3)
+EXPORT_SYMBOL(__ashlti3)
+
+SYM_FUNC_START(__ashrti3)
+	lmg	%r0,%r1,0(%r3)
+	cije	%r4,0,1f
+	lhi	%r3,64
+	sr	%r3,%r4
+	jnh	0f
+	sllg	%r3,%r0,0(%r3)
+	srlg	%r1,%r1,0(%r4)
+	srag	%r0,%r0,0(%r4)
+	ogr	%r1,%r3
+	j	1f
+0:	srag	%r1,%r0,-64(%r4)
+	srag	%r0,%r0,63
+1:	stmg	%r0,%r1,0(%r2)
+	BR_EX	%r14
+SYM_FUNC_END(__ashrti3)
+EXPORT_SYMBOL(__ashrti3)
+
+SYM_FUNC_START(__lshrti3)
+	lmg	%r0,%r1,0(%r3)
+	cije	%r4,0,1f
+	lhi	%r3,64
+	sr	%r3,%r4
+	jnh	0f
+	sllg	%r3,%r0,0(%r3)
+	srlg	%r1,%r1,0(%r4)
+	srlg	%r0,%r0,0(%r4)
+	ogr	%r1,%r3
+	j	1f
+0:	srlg	%r1,%r0,-64(%r4)
+	lghi	%r0,0
+1:	stmg	%r0,%r1,0(%r2)
+	BR_EX	%r14
+SYM_FUNC_END(__lshrti3)
+EXPORT_SYMBOL(__lshrti3)

drivers/s390/char/zcore.c

···
 static struct dentry *zcore_reipl_file;
 static struct dentry *zcore_hsa_file;
 static struct ipl_parameter_block *zcore_ipl_block;
+static unsigned long os_info_flags;
 
 static DEFINE_MUTEX(hsa_buf_mutex);
 static char hsa_buf[PAGE_SIZE] __aligned(PAGE_SIZE);
···
 {
 	if (zcore_ipl_block) {
 		diag308(DIAG308_SET, zcore_ipl_block);
-		diag308(DIAG308_LOAD_CLEAR, NULL);
+		if (os_info_flags & OS_INFO_FLAG_REIPL_CLEAR)
+			diag308(DIAG308_LOAD_CLEAR, NULL);
+		/* Use special diag308 subcode for CCW normal ipl */
+		if (zcore_ipl_block->pb0_hdr.pbt == IPL_PBT_CCW)
+			diag308(DIAG308_LOAD_NORMAL_DUMP, NULL);
+		else
+			diag308(DIAG308_LOAD_NORMAL, NULL);
 	}
 	return count;
 }
···
  */
 static int __init zcore_reipl_init(void)
 {
+	struct os_info_entry *entry;
 	struct ipib_info ipib_info;
+	unsigned long os_info_addr;
+	struct os_info *os_info;
 	int rc;
 
 	rc = memcpy_hsa_kernel(&ipib_info, __LC_DUMP_REIPL, sizeof(ipib_info));
···
 		free_page((unsigned long) zcore_ipl_block);
 		zcore_ipl_block = NULL;
 	}
+	/*
+	 * Read the bit-flags field from os_info flags entry.
+	 * Return zero even for os_info read or entry checksum errors in order
+	 * to continue dump processing, considering that os_info could be
+	 * corrupted on the panicked system.
+	 */
+	os_info = (void *)__get_free_page(GFP_KERNEL);
+	if (!os_info)
+		return -ENOMEM;
+	rc = memcpy_hsa_kernel(&os_info_addr, __LC_OS_INFO, sizeof(os_info_addr));
+	if (rc)
+		goto out;
+	if (os_info_addr < sclp.hsa_size)
+		rc = memcpy_hsa_kernel(os_info, os_info_addr, PAGE_SIZE);
+	else
+		rc = memcpy_real(os_info, os_info_addr, PAGE_SIZE);
+	if (rc || os_info_csum(os_info) != os_info->csum)
+		goto out;
+	entry = &os_info->entry[OS_INFO_FLAGS_ENTRY];
+	if (entry->addr && entry->size) {
+		if (entry->addr < sclp.hsa_size)
+			rc = memcpy_hsa_kernel(&os_info_flags, entry->addr, sizeof(os_info_flags));
+		else
+			rc = memcpy_real(&os_info_flags, entry->addr, sizeof(os_info_flags));
+		if (rc || (__force u32)csum_partial(&os_info_flags, entry->size, 0) != entry->csum)
+			os_info_flags = 0;
+	}
+out:
+	free_page((unsigned long)os_info);
 	return 0;
 }
 

drivers/s390/cio/vfio_ccw_drv.c

···
 		return -ENODEV;
 	}
 
-	parent = kzalloc(sizeof(*parent), GFP_KERNEL);
+	parent = kzalloc(struct_size(parent, mdev_types, 1), GFP_KERNEL);
 	if (!parent)
 		return -ENOMEM;
 

drivers/s390/cio/vfio_ccw_private.h

···
 
 	struct mdev_parent	parent;
 	struct mdev_type	mdev_type;
-	struct mdev_type	*mdev_types[1];
+	struct mdev_type	*mdev_types[];
 };
 
 /**

drivers/s390/crypto/pkey_api.c

···
 /*
  *  pkey device driver
  *
- *  Copyright IBM Corp. 2017,2019
+ *  Copyright IBM Corp. 2017, 2023
+ *
  *  Author(s): Harald Freudenberger
  */
 
···
 MODULE_DESCRIPTION("s390 protected key interface");
 
 #define KEYBLOBBUFSIZE 8192	/* key buffer size used for internal processing */
+#define MINKEYBLOBBUFSIZE (sizeof(struct keytoken_header))
 #define PROTKEYBLOBBUFSIZE 256	/* protected key buffer size used internal */
 #define MAXAPQNSINLIST 64	/* max 64 apqns within a apqn list */
+#define AES_WK_VP_SIZE 32	/* Size of WK VP block appended to a prot key */
 
 /*
  * debug feature data and functions
···
 } __packed;
 
 /* inside view of a clear key token (type 0x00 version 0x02) */
-struct clearaeskeytoken {
-	u8  type;	 /* 0x00 for PAES specific key tokens */
+struct clearkeytoken {
+	u8  type;	/* 0x00 for PAES specific key tokens */
 	u8  res0[3];
-	u8  version;	 /* 0x02 for clear AES key token */
+	u8  version;	/* 0x02 for clear key token */
 	u8  res1[3];
-	u32 keytype;	 /* key type, one of the PKEY_KEYTYPE values */
-	u32 len;	 /* bytes actually stored in clearkey[] */
+	u32 keytype;	/* key type, one of the PKEY_KEYTYPE_* values */
+	u32 len;	/* bytes actually stored in clearkey[] */
 	u8  clearkey[]; /* clear key value */
 } __packed;
 
+/* helper function which translates the PKEY_KEYTYPE_AES_* to their keysize */
+static inline u32 pkey_keytype_aes_to_size(u32 keytype)
+{
+	switch (keytype) {
+	case PKEY_KEYTYPE_AES_128:
+		return 16;
+	case PKEY_KEYTYPE_AES_192:
+		return 24;
+	case PKEY_KEYTYPE_AES_256:
+		return 32;
+	default:
+		return 0;
+	}
+}
+
 /*
- * Create a protected key from a clear key value.
+ * Create a protected key from a clear key value via PCKMO instruction.
  */
-static int pkey_clr2protkey(u32 keytype,
-			    const struct pkey_clrkey *clrkey,
-			    struct pkey_protkey *protkey)
+static int pkey_clr2protkey(u32 keytype, const u8 *clrkey,
+			    u8 *protkey, u32 *protkeylen, u32 *protkeytype)
 {
 	/* mask of available pckmo subfunctions */
 	static cpacf_mask_t pckmo_functions;
 
-	long fc;
+	u8 paramblock[112];
+	u32 pkeytype;
 	int keysize;
-	u8 paramblock[64];
+	long fc;
 
 	switch (keytype) {
 	case PKEY_KEYTYPE_AES_128:
+		/* 16 byte key, 32 byte aes wkvp, total 48 bytes */
 		keysize = 16;
+		pkeytype = keytype;
 		fc = CPACF_PCKMO_ENC_AES_128_KEY;
 		break;
 	case PKEY_KEYTYPE_AES_192:
+		/* 24 byte key, 32 byte aes wkvp, total 56 bytes */
 		keysize = 24;
+		pkeytype = keytype;
 		fc = CPACF_PCKMO_ENC_AES_192_KEY;
 		break;
 	case PKEY_KEYTYPE_AES_256:
+		/* 32 byte key, 32 byte aes wkvp, total 64 bytes */
 		keysize = 32;
+		pkeytype = keytype;
 		fc = CPACF_PCKMO_ENC_AES_256_KEY;
 		break;
+	case PKEY_KEYTYPE_ECC_P256:
+		/* 32 byte key, 32 byte aes wkvp, total 64 bytes */
+		keysize = 32;
+		pkeytype = PKEY_KEYTYPE_ECC;
+		fc = CPACF_PCKMO_ENC_ECC_P256_KEY;
+		break;
+	case PKEY_KEYTYPE_ECC_P384:
+		/* 48 byte key, 32 byte aes wkvp, total 80 bytes */
+		keysize = 48;
+		pkeytype = PKEY_KEYTYPE_ECC;
+		fc = CPACF_PCKMO_ENC_ECC_P384_KEY;
+		break;
+	case PKEY_KEYTYPE_ECC_P521:
+		/* 80 byte key, 32 byte aes wkvp, total 112 bytes */
+		keysize = 80;
+		pkeytype = PKEY_KEYTYPE_ECC;
+		fc = CPACF_PCKMO_ENC_ECC_P521_KEY;
+		break;
+	case PKEY_KEYTYPE_ECC_ED25519:
+		/* 32 byte key, 32 byte aes wkvp, total 64 bytes */
+		keysize = 32;
+		pkeytype = PKEY_KEYTYPE_ECC;
+		fc = CPACF_PCKMO_ENC_ECC_ED25519_KEY;
+		break;
+	case PKEY_KEYTYPE_ECC_ED448:
+		/* 64 byte key, 32 byte aes wkvp, total 96 bytes */
+		keysize = 64;
+		pkeytype = PKEY_KEYTYPE_ECC;
+		fc = CPACF_PCKMO_ENC_ECC_ED448_KEY;
+		break;
 	default:
-		DEBUG_ERR("%s unknown/unsupported keytype %d\n",
+		DEBUG_ERR("%s unknown/unsupported keytype %u\n",
 			  __func__, keytype);
+		return -EINVAL;
+	}
+
+	if (*protkeylen < keysize + AES_WK_VP_SIZE) {
+		DEBUG_ERR("%s prot key buffer size too small: %u < %d\n",
+			  __func__, *protkeylen, keysize + AES_WK_VP_SIZE);
 		return -EINVAL;
 	}
 
···
 
 	/* prepare param block */
 	memset(paramblock, 0, sizeof(paramblock));
-	memcpy(paramblock, clrkey->clrkey, keysize);
+	memcpy(paramblock, clrkey, keysize);
 
 	/* call the pckmo instruction */
 	cpacf_pckmo(fc, paramblock);
 
-	/* copy created protected key */
-	protkey->type = keytype;
-	protkey->len = keysize + 32;
-	memcpy(protkey->protkey, paramblock, keysize + 32);
+	/* copy created protected key to key buffer including the wkvp block */
+	*protkeylen = keysize + AES_WK_VP_SIZE;
+	memcpy(protkey, paramblock, *protkeylen);
+	*protkeytype = pkeytype;
 
 	return 0;
 }
···
 /*
  * Find card and transform secure key into protected key.
  */
-static int pkey_skey2pkey(const u8 *key, struct pkey_protkey *pkey)
+static int pkey_skey2pkey(const u8 *key, u8 *protkey,
+			  u32 *protkeylen, u32 *protkeytype)
 {
-	int rc, verify;
-	u16 cardnr, domain;
 	struct keytoken_header *hdr = (struct keytoken_header *)key;
+	u16 cardnr, domain;
+	int rc, verify;
 
 	zcrypt_wait_api_operational();
 
···
 			continue;
 		switch (hdr->version) {
 		case TOKVER_CCA_AES:
-			rc = cca_sec2protkey(cardnr, domain,
-					     key, pkey->protkey,
-					     &pkey->len, &pkey->type);
+			rc = cca_sec2protkey(cardnr, domain, key,
+					     protkey, protkeylen, protkeytype);
 			break;
 		case TOKVER_CCA_VLSC:
-			rc = cca_cipher2protkey(cardnr, domain,
-						key, pkey->protkey,
-						&pkey->len, &pkey->type);
+			rc = cca_cipher2protkey(cardnr, domain, key,
+						protkey, protkeylen,
+						protkeytype);
 			break;
 		default:
 			return -EINVAL;
···
 static int pkey_clr2ep11key(const u8 *clrkey, size_t clrkeylen,
 			    u8 *keybuf, size_t *keybuflen)
 {
-	int i, rc;
-	u16 card, dom;
 	u32 nr_apqns, *apqns = NULL;
+	u16 card, dom;
+	int i, rc;
 
 	zcrypt_wait_api_operational();
 
···
 /*
  * Find card and transform EP11 secure key into protected key.
  */
-static int pkey_ep11key2pkey(const u8 *key, struct pkey_protkey *pkey)
+static int pkey_ep11key2pkey(const u8 *key, u8 *protkey,
+			     u32 *protkeylen, u32 *protkeytype)
 {
-	int i, rc;
-	u16 card, dom;
-	u32 nr_apqns, *apqns = NULL;
 	struct ep11keyblob *kb = (struct ep11keyblob *)key;
+	u32 nr_apqns, *apqns = NULL;
+	u16 card, dom;
+	int i, rc;
 
 	zcrypt_wait_api_operational();
 
···
 	for (rc = -ENODEV, i = 0; i < nr_apqns; i++) {
 		card = apqns[i] >> 16;
 		dom = apqns[i] & 0xFFFF;
-		pkey->len = sizeof(pkey->protkey);
 		rc = ep11_kblob2protkey(card, dom, key, kb->head.len,
-					pkey->protkey, &pkey->len, &pkey->type);
+					protkey, protkeylen, protkeytype);
 		if (rc == 0)
 			break;
 	}
···
 /*
  * Generate a random protected key
  */
-static int pkey_genprotkey(u32 keytype, struct pkey_protkey *protkey)
+static int pkey_genprotkey(u32 keytype, u8 *protkey,
+			   u32 *protkeylen, u32 *protkeytype)
 {
-	struct pkey_clrkey clrkey;
+	u8 clrkey[32];
 	int keysize;
 	int rc;
 
-	switch (keytype) {
-	case PKEY_KEYTYPE_AES_128:
-		keysize = 16;
-		break;
-	case PKEY_KEYTYPE_AES_192:
-		keysize = 24;
-		break;
-	case PKEY_KEYTYPE_AES_256:
-		keysize = 32;
-		break;
-	default:
+	keysize = pkey_keytype_aes_to_size(keytype);
+	if (!keysize) {
 		DEBUG_ERR("%s unknown/unsupported keytype %d\n", __func__,
 			  keytype);
 		return -EINVAL;
 	}
 
 	/* generate a dummy random clear key */
-	get_random_bytes(clrkey.clrkey, keysize);
+	get_random_bytes(clrkey, keysize);
 
 	/* convert it to a dummy protected key */
-	rc = pkey_clr2protkey(keytype, &clrkey, protkey);
+	rc = pkey_clr2protkey(keytype, clrkey,
+			      protkey, protkeylen, protkeytype);
 	if (rc)
 		return rc;
 
 	/* replace the key part of the protected key with random bytes */
-	get_random_bytes(protkey->protkey, keysize);
+	get_random_bytes(protkey, keysize);
 
 	return 0;
 }
···
 /*
  * Verify if a protected key is still valid
  */
-static int pkey_verifyprotkey(const struct pkey_protkey *protkey)
+static int pkey_verifyprotkey(const u8 *protkey, u32 protkeylen,
+			      u32 protkeytype)
 {
-	unsigned long fc;
 	struct {
 		u8 iv[AES_BLOCK_SIZE];
 		u8 key[MAXPROTKEYSIZE];
 	} param;
 	u8 null_msg[AES_BLOCK_SIZE];
 	u8 dest_buf[AES_BLOCK_SIZE];
-	unsigned int k;
+	unsigned int k, pkeylen;
+	unsigned long fc;
 
-	switch (protkey->type) {
+	switch (protkeytype) {
 	case PKEY_KEYTYPE_AES_128:
+		pkeylen = 16 + AES_WK_VP_SIZE;
 		fc = CPACF_KMC_PAES_128;
 		break;
 	case PKEY_KEYTYPE_AES_192:
+		pkeylen = 24 + AES_WK_VP_SIZE;
 		fc = CPACF_KMC_PAES_192;
 		break;
 	case PKEY_KEYTYPE_AES_256:
+		pkeylen = 32 + AES_WK_VP_SIZE;
 		fc = CPACF_KMC_PAES_256;
 		break;
 	default:
-		DEBUG_ERR("%s unknown/unsupported keytype %d\n", __func__,
-			  protkey->type);
+		DEBUG_ERR("%s unknown/unsupported keytype %u\n", __func__,
+			  protkeytype);
+		return -EINVAL;
+	}
+	if (protkeylen != pkeylen) {
+		DEBUG_ERR("%s invalid protected key size %u for keytype %u\n",
+			  __func__, protkeylen, protkeytype);
 		return -EINVAL;
 	}
 
 	memset(null_msg, 0, sizeof(null_msg));
 
 	memset(param.iv, 0, sizeof(param.iv));
-	memcpy(param.key, protkey->protkey, sizeof(param.key));
+	memcpy(param.key, protkey, protkeylen);
 
 	k = cpacf_kmc(fc | CPACF_ENCRYPT, &param, null_msg, dest_buf,
 		      sizeof(null_msg));
···
 	return 0;
 }
 
+/* Helper for pkey_nonccatok2pkey, handles aes clear key token */
+static int nonccatokaes2pkey(const struct clearkeytoken *t,
+			     u8 *protkey, u32 *protkeylen, u32 *protkeytype)
+{
+	size_t tmpbuflen = max_t(size_t, SECKEYBLOBSIZE, MAXEP11AESKEYBLOBSIZE);
+	u8 *tmpbuf = NULL;
+	u32 keysize;
+	int rc;
+
+	keysize = pkey_keytype_aes_to_size(t->keytype);
+	if (!keysize) {
+		DEBUG_ERR("%s unknown/unsupported keytype %u\n",
+			  __func__, t->keytype);
+		return -EINVAL;
+	}
+	if (t->len != keysize) {
+		DEBUG_ERR("%s non clear key aes token: invalid key len %u\n",
+			  __func__, t->len);
+		return -EINVAL;
+	}
+
+	/* try direct way with the PCKMO instruction */
+	rc = pkey_clr2protkey(t->keytype, t->clearkey,
+			      protkey, protkeylen, protkeytype);
+	if (!rc)
+		goto out;
+
+	/* PCKMO failed, so try the CCA secure key way */
+	tmpbuf = kmalloc(tmpbuflen, GFP_ATOMIC);
+	if (!tmpbuf)
+		return -ENOMEM;
+	zcrypt_wait_api_operational();
+	rc = cca_clr2seckey(0xFFFF, 0xFFFF, t->keytype, t->clearkey, tmpbuf);
+	if (rc)
+		goto try_via_ep11;
+	rc = pkey_skey2pkey(tmpbuf,
+			    protkey, protkeylen, protkeytype);
+	if (!rc)
+		goto out;
+
+try_via_ep11:
+	/* if the CCA way also failed, let's try via EP11 */
+	rc = pkey_clr2ep11key(t->clearkey, t->len,
+			      tmpbuf, &tmpbuflen);
+	if (rc)
+		goto failure;
+	rc = pkey_ep11key2pkey(tmpbuf,
+			       protkey, protkeylen, protkeytype);
+	if (!rc)
+		goto out;
+
+failure:
+	DEBUG_ERR("%s unable to build protected key from clear", __func__);
+
+out:
+	kfree(tmpbuf);
+	return rc;
+}
+
+/* Helper for pkey_nonccatok2pkey, handles ecc clear key token */
+static int nonccatokecc2pkey(const struct clearkeytoken *t,
+			     u8 *protkey, u32 *protkeylen, u32 *protkeytype)
+{
+	u32 keylen;
+	int rc;
+
+	switch (t->keytype) {
+	case PKEY_KEYTYPE_ECC_P256:
+		keylen = 32;
+		break;
+	case PKEY_KEYTYPE_ECC_P384:
+		keylen = 48;
+		break;
+	case PKEY_KEYTYPE_ECC_P521:
+		keylen = 80;
+		break;
+	case PKEY_KEYTYPE_ECC_ED25519:
+		keylen = 32;
+		break;
+	case PKEY_KEYTYPE_ECC_ED448:
+		keylen = 64;
+		break;
+	default:
+		DEBUG_ERR("%s unknown/unsupported keytype %u\n",
+			  __func__, t->keytype);
+		return -EINVAL;
+	}
+
+	if (t->len != keylen) {
+		DEBUG_ERR("%s non clear key ecc token: invalid key len %u\n",
+			  __func__, t->len);
+		return -EINVAL;
+	}
+
+	/* only one path possible: via PCKMO instruction */
+	rc = pkey_clr2protkey(t->keytype, t->clearkey,
+			      protkey, protkeylen, protkeytype);
+	if (rc) {
+		DEBUG_ERR("%s unable to build protected key from clear",
+			  __func__);
+	}
+
+	return rc;
+}
+
 /*
  * Transform a non-CCA key token into a protected key
  */
 static int pkey_nonccatok2pkey(const u8 *key, u32 keylen,
-			       struct pkey_protkey *protkey)
+			       u8 *protkey, u32 *protkeylen, u32 *protkeytype)
 {
-	int rc = -EINVAL;
-	u8 *tmpbuf = NULL;
 	struct keytoken_header *hdr = (struct keytoken_header *)key;
+	int rc = -EINVAL;
 
 	switch (hdr->version) {
 	case TOKVER_PROTECTED_KEY: {
···
 		if (keylen != sizeof(struct protaeskeytoken))
 			goto out;
 		t = (struct protaeskeytoken *)key;
-		protkey->len = t->len;
-		protkey->type = t->keytype;
-		memcpy(protkey->protkey, t->protkey,
-		       sizeof(protkey->protkey));
-		rc = pkey_verifyprotkey(protkey);
+		rc = pkey_verifyprotkey(t->protkey, t->len, t->keytype);
+		if (rc)
+			goto out;
+		memcpy(protkey, t->protkey, t->len);
+		*protkeylen = t->len;
+		*protkeytype = t->keytype;
 		break;
 	}
 	case TOKVER_CLEAR_KEY: {
-		struct clearaeskeytoken *t;
-		struct pkey_clrkey ckey;
-		union u_tmpbuf {
-			u8 skey[SECKEYBLOBSIZE];
-			u8 ep11key[MAXEP11AESKEYBLOBSIZE];
-		};
-		size_t tmpbuflen = sizeof(union u_tmpbuf);
+		struct clearkeytoken *t = (struct clearkeytoken *)key;
 
-		if (keylen < sizeof(struct clearaeskeytoken))
+		if (keylen < sizeof(struct clearkeytoken) ||
+		    keylen != sizeof(*t) + t->len)
 			goto out;
-		t = (struct clearaeskeytoken *)key;
-		if (keylen != sizeof(*t) + t->len)
-			goto out;
-		if ((t->keytype == PKEY_KEYTYPE_AES_128 && t->len == 16) ||
-		    (t->keytype == PKEY_KEYTYPE_AES_192 && t->len == 24) ||
-		    (t->keytype == PKEY_KEYTYPE_AES_256 && t->len == 32))
-			memcpy(ckey.clrkey, t->clearkey, t->len);
-		else
-			goto out;
-		/* alloc temp key buffer space */
-		tmpbuf = kmalloc(tmpbuflen, GFP_ATOMIC);
-		if (!tmpbuf) {
-			rc = -ENOMEM;
-			goto out;
+		switch (t->keytype) {
+		case PKEY_KEYTYPE_AES_128:
+		case PKEY_KEYTYPE_AES_192:
+		case PKEY_KEYTYPE_AES_256:
+			rc = nonccatokaes2pkey(t, protkey,
+					       protkeylen, protkeytype);
+			break;
+		case PKEY_KEYTYPE_ECC_P256:
+		case PKEY_KEYTYPE_ECC_P384:
+		case PKEY_KEYTYPE_ECC_P521:
+		case PKEY_KEYTYPE_ECC_ED25519:
+		case PKEY_KEYTYPE_ECC_ED448:
+			rc = nonccatokecc2pkey(t, protkey,
+					       protkeylen, protkeytype);
+			break;
+		default:
+			DEBUG_ERR("%s unknown/unsupported non cca clear key type %u\n",
+				  __func__, t->keytype);
+			return -EINVAL;
 		}
-		/* try direct way with the PCKMO instruction */
-		rc = pkey_clr2protkey(t->keytype, &ckey, protkey);
-		if (rc == 0)
-			break;
-		/* PCKMO failed, so try the CCA secure key way */
-		zcrypt_wait_api_operational();
-		rc = cca_clr2seckey(0xFFFF, 0xFFFF, t->keytype,
-				    ckey.clrkey, tmpbuf);
-		if (rc == 0)
-			rc = pkey_skey2pkey(tmpbuf, protkey);
-		if (rc == 0)
-			break;
-		/* if the CCA way also failed, let's try via EP11 */
-		rc = pkey_clr2ep11key(ckey.clrkey, t->len,
-				      tmpbuf, &tmpbuflen);
-		if (rc == 0)
-			rc = pkey_ep11key2pkey(tmpbuf, protkey);
-		/* now we should really have an protected key */
-		DEBUG_ERR("%s unable to build protected key from clear",
-			  __func__);
 		break;
 	}
 	case TOKVER_EP11_AES: {
···
 		rc = ep11_check_aes_key(debug_info, 3, key, keylen, 1);
 		if (rc)
 			goto out;
-		rc = pkey_ep11key2pkey(key, protkey);
+		rc = pkey_ep11key2pkey(key,
+				       protkey, protkeylen, protkeytype);
 		break;
 	}
 	case TOKVER_EP11_AES_WITH_HEADER:
···
 		if (rc)
 			goto out;
 		rc = pkey_ep11key2pkey(key + sizeof(struct ep11kblob_header),
-				       protkey);
+				       protkey, protkeylen, protkeytype);
 		break;
 	default:
 		DEBUG_ERR("%s unknown/unsupported non-CCA token version %d\n",
 			  __func__, hdr->version);
-		rc = -EINVAL;
 	}
 
 out:
-	kfree(tmpbuf);
 	return rc;
 }
 
···
  * Transform a CCA internal key token into a protected key
  */
 static int pkey_ccainttok2pkey(const u8 *key, u32 keylen,
-			       struct pkey_protkey *protkey)
+			       u8 *protkey, u32 *protkeylen, u32 *protkeytype)
 {
 	struct keytoken_header *hdr = (struct keytoken_header *)key;
 
···
 		return -EINVAL;
 	}
 
-	return pkey_skey2pkey(key, protkey);
+	return pkey_skey2pkey(key, protkey, protkeylen, protkeytype);
 }
 
 /*
  * Transform a key blob (of any type) into a protected key
  */
 int pkey_keyblob2pkey(const u8 *key, u32 keylen,
-		      struct pkey_protkey *protkey)
+		      u8 *protkey, u32 *protkeylen, u32 *protkeytype)
 {
-	int rc;
 	struct keytoken_header *hdr = (struct keytoken_header *)key;
+	int rc;
 
 	if (keylen < sizeof(struct keytoken_header)) {
 		DEBUG_ERR("%s invalid keylen %d\n", __func__, keylen);
···
 
 	switch (hdr->type) {
 	case TOKTYPE_NON_CCA:
-		rc = pkey_nonccatok2pkey(key, keylen, protkey);
+		rc = pkey_nonccatok2pkey(key, keylen,
+					 protkey, protkeylen, protkeytype);
 		break;
 	case TOKTYPE_CCA_INTERNAL:
-		rc = pkey_ccainttok2pkey(key, keylen, protkey);
+		rc = pkey_ccainttok2pkey(key, keylen,
+					 protkey, protkeylen, protkeytype);
 		break;
 	default:
 		DEBUG_ERR("%s unknown/unsupported blob type %d\n",
···
 			   enum pkey_key_type *ktype,
 			   enum pkey_key_size *ksize, u32 *flags)
 {
-	int rc;
-	u32 _nr_apqns, *_apqns = NULL;
 	struct keytoken_header *hdr = (struct keytoken_header *)key;
+	u32 _nr_apqns, *_apqns = NULL;
+	int rc;
 
 	if (keylen < sizeof(struct keytoken_header))
 		return -EINVAL;
···
 
 static int pkey_keyblob2pkey2(const struct pkey_apqn *apqns, size_t nr_apqns,
 			      const u8 *key, size_t keylen,
-			      struct pkey_protkey *pkey)
+			      u8 *protkey, u32 *protkeylen, u32 *protkeytype)
 {
-	int i, card, dom, rc;
 	struct keytoken_header *hdr = (struct keytoken_header *)key;
+	int i, card, dom, rc;
 
 	/* check for at least one apqn given */
 	if (!apqns || !nr_apqns)
···
 			if (ep11_check_aes_key(debug_info, 3, key, keylen, 1))
 				return -EINVAL;
 		} else {
-			return pkey_nonccatok2pkey(key, keylen, pkey);
+			return pkey_nonccatok2pkey(key, keylen,
+						   protkey, protkeylen,
+						   protkeytype);
 		}
 	} else {
 		DEBUG_ERR("%s unknown/unsupported blob type %d\n",
···
 		dom = apqns[i].domain;
 		if (hdr->type == TOKTYPE_CCA_INTERNAL &&
 		    hdr->version == TOKVER_CCA_AES) {
-			rc = cca_sec2protkey(card, dom, key, pkey->protkey,
-					     &pkey->len, &pkey->type);
+			rc = cca_sec2protkey(card, dom, key,
+					     protkey, protkeylen, protkeytype);
 		} else if (hdr->type == TOKTYPE_CCA_INTERNAL &&
 			   hdr->version == TOKVER_CCA_VLSC) {
-			rc = cca_cipher2protkey(card, dom, key, pkey->protkey,
-						&pkey->len, &pkey->type);
+			rc = cca_cipher2protkey(card, dom, key,
+						protkey, protkeylen,
+						protkeytype);
 		} else {
 			/* EP11 AES secure key blob */
 			struct ep11keyblob *kb = (struct ep11keyblob *)key;
 
-			pkey->len = sizeof(pkey->protkey);
 			rc = ep11_kblob2protkey(card, dom, key, kb->head.len,
-						pkey->protkey, &pkey->len,
-						&pkey->type);
+						protkey, protkeylen,
+						protkeytype);
 		}
 		if (rc == 0)
 			break;
···
 static int pkey_apqns4key(const u8 *key, size_t keylen, u32 flags,
 			  struct pkey_apqn *apqns, size_t *nr_apqns)
 {
-	int rc;
-	u32 _nr_apqns, *_apqns = NULL;
 	struct keytoken_header *hdr = (struct keytoken_header *)key;
+	u32 _nr_apqns, *_apqns = NULL;
+	int rc;
 
 	if (keylen < sizeof(struct keytoken_header) || flags == 0)
 		return -EINVAL;
···
 	    (hdr->version == TOKVER_EP11_AES_WITH_HEADER ||
 	     hdr->version == TOKVER_EP11_ECC_WITH_HEADER) &&
 	    is_ep11_keyblob(key + sizeof(struct ep11kblob_header))) {
-		int minhwtype = 0, api = 0;
 		struct ep11keyblob *kb = (struct ep11keyblob *)
 			(key + sizeof(struct ep11kblob_header));
+		int minhwtype = 0, api = 0;
 
 		if (flags != PKEY_FLAGS_MATCH_CUR_MKVP)
 			return -EINVAL;
···
 	} else if (hdr->type == TOKTYPE_NON_CCA &&
 		   hdr->version == TOKVER_EP11_AES &&
 		   is_ep11_keyblob(key)) {
-		int minhwtype = 0, api = 0;
 		struct ep11keyblob *kb = (struct ep11keyblob *)key;
+		int minhwtype = 0, api = 0;
 
 		if (flags != PKEY_FLAGS_MATCH_CUR_MKVP)
 			return -EINVAL;
···
 		if (rc)
 			goto out;
 	} else if (hdr->type == TOKTYPE_CCA_INTERNAL) {
-		int minhwtype = ZCRYPT_CEX3C;
 		u64 cur_mkvp = 0, old_mkvp = 0;
+		int minhwtype = ZCRYPT_CEX3C;
 
 		if (hdr->version == TOKVER_CCA_AES) {
 			struct secaeskeytoken *t = (struct secaeskeytoken *)key;
···
 		if (rc)
 			goto out;
 	} else if (hdr->type == TOKTYPE_CCA_INTERNAL_PKA) {
-		u64 cur_mkvp = 0, old_mkvp = 0;
 		struct eccprivkeytoken *t = (struct eccprivkeytoken *)key;
+		u64 cur_mkvp = 0, old_mkvp = 0;
 
 		if (t->secid == 0x20) {
 			if (flags & PKEY_FLAGS_MATCH_CUR_MKVP)
···
 			      u8 cur_mkvp[32], u8 alt_mkvp[32], u32 flags,
 			      struct pkey_apqn *apqns, size_t *nr_apqns)
 {
-	int rc;
 	u32 _nr_apqns, *_apqns = NULL;
+	int rc;
 
 	zcrypt_wait_api_operational();
 
···
 }
 
 static int pkey_keyblob2pkey3(const struct pkey_apqn *apqns, size_t nr_apqns,
-			      const u8 *key, size_t keylen, u32 *protkeytype,
-			      u8 *protkey, u32 *protkeylen)
+			      const u8 *key, size_t keylen,
+			      u8 *protkey, u32 *protkeylen, u32 *protkeytype)
 {
-	int i, card, dom, rc;
 	struct keytoken_header *hdr = (struct keytoken_header *)key;
+	int i, card, dom, rc;
 
 	/* check for at least one apqn given */
 	if (!apqns || !nr_apqns)
···
 		if (cca_check_sececckeytoken(debug_info, 3, key, keylen, 1))
 			return -EINVAL;
 	} else if (hdr->type == TOKTYPE_NON_CCA) {
-		struct pkey_protkey pkey;
-
-		rc = pkey_nonccatok2pkey(key, keylen, &pkey);
-		if (rc)
-			return rc;
-		memcpy(protkey, pkey.protkey, pkey.len);
-		*protkeylen = pkey.len;
-		*protkeytype = pkey.type;
-		return 0;
+		return pkey_nonccatok2pkey(key, keylen,
+					   protkey, protkeylen, protkeytype);
 	} else {
 		DEBUG_ERR("%s unknown/unsupported blob type %d\n",
 			  __func__, hdr->type);
···
 
 static void *_copy_key_from_user(void __user *ukey, size_t keylen)
 {
-	if (!ukey || keylen < MINKEYBLOBSIZE || keylen > KEYBLOBBUFSIZE)
+	if (!ukey || keylen < MINKEYBLOBBUFSIZE || keylen > KEYBLOBBUFSIZE)
 		return ERR_PTR(-EINVAL);
 
 	return memdup_user(ukey, keylen);
···
 
 		if (copy_from_user(&ksp, usp, sizeof(ksp)))
 			return -EFAULT;
+		ksp.protkey.len = sizeof(ksp.protkey.protkey);
 		rc = cca_sec2protkey(ksp.cardnr, ksp.domain,
 				     ksp.seckey.seckey, ksp.protkey.protkey,
 				     &ksp.protkey.len, &ksp.protkey.type);
···
 
 		if (copy_from_user(&kcp, ucp, sizeof(kcp)))
 			return -EFAULT;
-		rc = pkey_clr2protkey(kcp.keytype,
-				      &kcp.clrkey, &kcp.protkey);
+		kcp.protkey.len = sizeof(kcp.protkey.protkey);
+		rc = pkey_clr2protkey(kcp.keytype, kcp.clrkey.clrkey,
+				      kcp.protkey.protkey,
+				      &kcp.protkey.len, &kcp.protkey.type);
 		DEBUG_DBG("%s pkey_clr2protkey()=%d\n", __func__, rc);
 		if (rc)
 			break;
···
 
 		if (copy_from_user(&ksp, usp, sizeof(ksp)))
 			return -EFAULT;
-		rc = pkey_skey2pkey(ksp.seckey.seckey, &ksp.protkey);
+		ksp.protkey.len = sizeof(ksp.protkey.protkey);
+		rc = pkey_skey2pkey(ksp.seckey.seckey, ksp.protkey.protkey,
+				    &ksp.protkey.len, &ksp.protkey.type);
 		DEBUG_DBG("%s pkey_skey2pkey()=%d\n", __func__, rc);
 		if (rc)
 			break;
···
 
 		if (copy_from_user(&kgp, ugp, sizeof(kgp)))
 			return -EFAULT;
-		rc = pkey_genprotkey(kgp.keytype, &kgp.protkey);
+		kgp.protkey.len = sizeof(kgp.protkey.protkey);
+		rc = pkey_genprotkey(kgp.keytype, kgp.protkey.protkey,
+				     &kgp.protkey.len, &kgp.protkey.type);
 		DEBUG_DBG("%s pkey_genprotkey()=%d\n", __func__, rc);
 		if (rc)
 			break;
···
 
 		if (copy_from_user(&kvp, uvp, sizeof(kvp)))
 			return -EFAULT;
-		rc = pkey_verifyprotkey(&kvp.protkey);
+		rc = pkey_verifyprotkey(kvp.protkey.protkey,
+					kvp.protkey.len, kvp.protkey.type);
 		DEBUG_DBG("%s pkey_verifyprotkey()=%d\n", __func__, rc);
 		break;
 	}
···
 		kkey = _copy_key_from_user(ktp.key, ktp.keylen);
 		if (IS_ERR(kkey))
 			return PTR_ERR(kkey);
-		rc = pkey_keyblob2pkey(kkey, ktp.keylen, &ktp.protkey);
+		ktp.protkey.len = sizeof(ktp.protkey.protkey);
+		rc = pkey_keyblob2pkey(kkey, ktp.keylen, ktp.protkey.protkey,
+				       &ktp.protkey.len, &ktp.protkey.type);
 		DEBUG_DBG("%s pkey_keyblob2pkey()=%d\n", __func__, rc);
 		memzero_explicit(kkey, ktp.keylen);
 		kfree(kkey);
···
 	}
 	case PKEY_GENSECK2: {
 		struct pkey_genseck2 __user *ugs = (void __user *)arg;
+		size_t klen = KEYBLOBBUFSIZE;
 		struct pkey_genseck2 kgs;
 		struct pkey_apqn *apqns;
-		size_t klen = KEYBLOBBUFSIZE;
 		u8 *kkey;
 
 		if (copy_from_user(&kgs, ugs, sizeof(kgs)))
···
 	}
 	case PKEY_CLR2SECK2: {
 		struct pkey_clr2seck2 __user *ucs = (void __user *)arg;
+		size_t klen = KEYBLOBBUFSIZE;
 		struct pkey_clr2seck2 kcs;
 		struct pkey_apqn *apqns;
-		size_t klen = KEYBLOBBUFSIZE;
 		u8 *kkey;
 
 		if (copy_from_user(&kcs, ucs, sizeof(kcs)))
···
 	}
 	case PKEY_KBLOB2PROTK2: {
 		struct pkey_kblob2pkey2 __user *utp = (void __user *)arg;
-		struct pkey_kblob2pkey2 ktp;
 		struct pkey_apqn *apqns = NULL;
+		struct pkey_kblob2pkey2 ktp;
 		u8 *kkey;
 
 		if (copy_from_user(&ktp, utp, sizeof(ktp)))
···
 			kfree(apqns);
 			return PTR_ERR(kkey);
 		}
+		ktp.protkey.len = sizeof(ktp.protkey.protkey);
 		rc = pkey_keyblob2pkey2(apqns, ktp.apqn_entries,
-					kkey, ktp.keylen, &ktp.protkey);
+					kkey, ktp.keylen,
+					ktp.protkey.protkey, &ktp.protkey.len,
+					&ktp.protkey.type);
 		DEBUG_DBG("%s pkey_keyblob2pkey2()=%d\n", __func__, rc);
 		kfree(apqns);
 		memzero_explicit(kkey, ktp.keylen);
···
 	}
 	case PKEY_APQNS4K: {
 		struct pkey_apqns4key __user *uak = (void __user *)arg;
-		struct pkey_apqns4key kak;
 		struct pkey_apqn *apqns = NULL;
+		struct pkey_apqns4key kak;
 		size_t nr_apqns, len;
 		u8 *kkey;
 
···
 	}
 	case PKEY_APQNS4KT: {
 		struct pkey_apqns4keytype __user *uat = (void __user *)arg;
-		struct pkey_apqns4keytype kat;
 		struct pkey_apqn *apqns = NULL;
+		struct pkey_apqns4keytype kat;
 		size_t nr_apqns, len;
 
 		if (copy_from_user(&kat, uat, sizeof(kat)))
···
 	}
 	case PKEY_KBLOB2PROTK3: {
 		struct pkey_kblob2pkey3 __user *utp = (void __user *)arg;
-		struct pkey_kblob2pkey3 ktp;
-		struct pkey_apqn *apqns = NULL;
 		u32 protkeylen = PROTKEYBLOBBUFSIZE;
+		struct pkey_apqn *apqns = NULL;
+		struct pkey_kblob2pkey3 ktp;
 		u8 *kkey, *protkey;
 
 		if (copy_from_user(&ktp, utp, sizeof(ktp)))
···
 			kfree(kkey);
 			return -ENOMEM;
 		}
-		rc = pkey_keyblob2pkey3(apqns, ktp.apqn_entries, kkey,
-					ktp.keylen, &ktp.pkeytype,
-					protkey, &protkeylen);
+		rc = pkey_keyblob2pkey3(apqns, ktp.apqn_entries,
+					kkey, ktp.keylen,
+					protkey, &protkeylen, &ktp.pkeytype);
 		DEBUG_DBG("%s pkey_keyblob2pkey3()=%d\n", __func__, rc);
 		kfree(apqns);
 		memzero_explicit(kkey, ktp.keylen);
···
 	protkeytoken.version = TOKVER_PROTECTED_KEY;
 	protkeytoken.keytype = keytype;
 
-	rc = pkey_genprotkey(protkeytoken.keytype, &protkey);
+	protkey.len = sizeof(protkey.protkey);
+	rc = pkey_genprotkey(protkeytoken.keytype,
+			     protkey.protkey, &protkey.len, &protkey.type);
 	if (rc)
 		return rc;
 
···
 	memcpy(buf, &protkeytoken, sizeof(protkeytoken));
 
 	if (is_xts) {
-		rc = pkey_genprotkey(protkeytoken.keytype, &protkey);
+		/* xts needs a second protected key, reuse protkey struct */
+		protkey.len = sizeof(protkey.protkey);
+		rc = pkey_genprotkey(protkeytoken.keytype,
+				     protkey.protkey, &protkey.len, &protkey.type);
 		if (rc)
 			return rc;
 
···
 static ssize_t pkey_ccadata_aes_attr_read(u32 keytype, bool is_xts, char *buf,
 					  loff_t off, size_t count)
 {
-	int rc;
 	struct pkey_seckey *seckey = (struct pkey_seckey *)buf;
+	int rc;
 
 	if (off != 0 || count < sizeof(struct secaeskeytoken))
 		return -EINVAL;
···
 					    bool is_xts, char *buf, loff_t off,
 					    size_t count)
 {
-	int i, rc, card, dom;
-	u32 nr_apqns, *apqns = NULL;
 	size_t keysize = CCACIPHERTOKENSIZE;
+	u32 nr_apqns, *apqns = NULL;
+	int i, rc, card, dom;
 
 	if (off != 0 || count < CCACIPHERTOKENSIZE)
 		return -EINVAL;
···
 				       bool is_xts, char *buf, loff_t off,
 				       size_t count)
 {
-	int i, rc, card, dom;
-	u32 nr_apqns, *apqns = NULL;
 	size_t keysize = MAXEP11AESKEYBLOBSIZE;
+	u32 nr_apqns, *apqns = NULL;
+	int i, rc, card, dom;
 
 	if (off != 0 || count < MAXEP11AESKEYBLOBSIZE)
 		return -EINVAL;

drivers/s390/crypto/vfio_ap_ops.c

···
 	ret = vfio_register_emulated_iommu_dev(&matrix_mdev->vdev);
 	if (ret)
 		goto err_put_vdev;
+	matrix_mdev->req_trigger = NULL;
 	dev_set_drvdata(&mdev->dev, matrix_mdev);
 	mutex_lock(&matrix_dev->mdevs_lock);
 	list_add(&matrix_mdev->node, &matrix_dev->mdev_list);
···
 	vfio_ap_mdev_unset_kvm(matrix_mdev);
 }
 
+static void vfio_ap_mdev_request(struct vfio_device *vdev, unsigned int count)
+{
+	struct device *dev = vdev->dev;
+	struct ap_matrix_mdev *matrix_mdev;
+
+	matrix_mdev = container_of(vdev, struct ap_matrix_mdev, vdev);
+
+	if (matrix_mdev->req_trigger) {
+		if (!(count % 10))
+			dev_notice_ratelimited(dev,
+					       "Relaying device request to user (#%u)\n",
+					       count);
+
+		eventfd_signal(matrix_mdev->req_trigger, 1);
+	} else if (count == 0) {
+		dev_notice(dev,
+			   "No device request registered, blocked until released by user\n");
+	}
+}
+
 static int vfio_ap_mdev_get_device_info(unsigned long arg)
 {
 	unsigned long minsz;
···
 
 	info.flags = VFIO_DEVICE_FLAGS_AP | VFIO_DEVICE_FLAGS_RESET;
 	info.num_regions = 0;
-	info.num_irqs = 0;
+	info.num_irqs = VFIO_AP_NUM_IRQS;
 
 	return copy_to_user((void __user *)arg, &info, minsz) ? -EFAULT : 0;
+}
+
+static ssize_t vfio_ap_get_irq_info(unsigned long arg)
+{
+	unsigned long minsz;
+	struct vfio_irq_info info;
+
+	minsz = offsetofend(struct vfio_irq_info, count);
+
+	if (copy_from_user(&info, (void __user *)arg, minsz))
+		return -EFAULT;
+
+	if (info.argsz < minsz || info.index >= VFIO_AP_NUM_IRQS)
+		return -EINVAL;
+
+	switch (info.index) {
+	case VFIO_AP_REQ_IRQ_INDEX:
+		info.count = 1;
+		info.flags = VFIO_IRQ_INFO_EVENTFD;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return copy_to_user((void __user *)arg, &info, minsz) ? -EFAULT : 0;
+}
+
+static int vfio_ap_irq_set_init(struct vfio_irq_set *irq_set, unsigned long arg)
+{
+	int ret;
+	size_t data_size;
+	unsigned long minsz;
+
+	minsz = offsetofend(struct vfio_irq_set, count);
+
+	if (copy_from_user(irq_set, (void __user *)arg, minsz))
+		return -EFAULT;
+
+	ret = vfio_set_irqs_validate_and_prepare(irq_set, 1, VFIO_AP_NUM_IRQS,
+						 &data_size);
+	if (ret)
+		return ret;
+
+	if (!(irq_set->flags & VFIO_IRQ_SET_ACTION_TRIGGER))
+		return -EINVAL;
+
+	return 0;
+}
+
+static int vfio_ap_set_request_irq(struct ap_matrix_mdev *matrix_mdev,
+				   unsigned long arg)
+{
+	s32 fd;
+	void __user *data;
+	unsigned long minsz;
+	struct eventfd_ctx *req_trigger;
+
+	minsz = offsetofend(struct vfio_irq_set, count);
+	data = (void __user *)(arg + minsz);
+
+	if (get_user(fd, (s32 __user *)data))
+		return -EFAULT;
+
+	if (fd == -1) {
+		if (matrix_mdev->req_trigger)
+			eventfd_ctx_put(matrix_mdev->req_trigger);
+		matrix_mdev->req_trigger = NULL;
+	} else if (fd >= 0) {
+		req_trigger = eventfd_ctx_fdget(fd);
+		if (IS_ERR(req_trigger))
+			return PTR_ERR(req_trigger);
+
+		if (matrix_mdev->req_trigger)
+			eventfd_ctx_put(matrix_mdev->req_trigger);
+
+		matrix_mdev->req_trigger = req_trigger;
+	} else {
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int vfio_ap_set_irqs(struct ap_matrix_mdev *matrix_mdev,
+			    unsigned long arg)
+{
+	int ret;
+	struct vfio_irq_set irq_set;
+
+	ret = vfio_ap_irq_set_init(&irq_set, arg);
+	if (ret)
+		return ret;
+
+	switch (irq_set.flags & VFIO_IRQ_SET_DATA_TYPE_MASK) {
+	case VFIO_IRQ_SET_DATA_EVENTFD:
+		switch (irq_set.index) {
+		case VFIO_AP_REQ_IRQ_INDEX:
+			return vfio_ap_set_request_irq(matrix_mdev, arg);
+		default:
+			return -EINVAL;
+		}
+	default:
+		return -EINVAL;
+	}
 }
 
 static ssize_t vfio_ap_mdev_ioctl(struct vfio_device *vdev,
···
 		break;
 	case VFIO_DEVICE_RESET:
 		ret = vfio_ap_mdev_reset_queues(&matrix_mdev->qtable);
+		break;
+	case VFIO_DEVICE_GET_IRQ_INFO:
+			ret = vfio_ap_get_irq_info(arg);
+			break;
+	case VFIO_DEVICE_SET_IRQS:
+		ret = vfio_ap_set_irqs(matrix_mdev, arg);
 		break;
 	default:
 		ret = -EOPNOTSUPP;
···
 	.bind_iommufd = vfio_iommufd_emulated_bind,
 	.unbind_iommufd = vfio_iommufd_emulated_unbind,
 	.attach_ioas = vfio_iommufd_emulated_attach_ioas,
+	.request = vfio_ap_mdev_request
 };
 
 static struct mdev_driver vfio_ap_matrix_driver = {

drivers/s390/crypto/vfio_ap_private.h

···
 #include <linux/types.h>
 #include <linux/mdev.h>
 #include <linux/delay.h>
+#include <linux/eventfd.h>
 #include <linux/mutex.h>
 #include <linux/kvm_host.h>
 #include <linux/vfio.h>
···
  *		PQAP(AQIC) instruction.
  * @mdev:	the mediated device
  * @qtable:	table of queues (struct vfio_ap_queue) assigned to the mdev
+ * @req_trigger eventfd ctx for signaling userspace to return a device
  * @apm_add:	bitmap of APIDs added to the host's AP configuration
  * @aqm_add:	bitmap of APQIs added to the host's AP configuration
  * @adm_add:	bitmap of control domain numbers added to the host's AP
···
 	crypto_hook pqap_hook;
 	struct mdev_device *mdev;
 	struct ap_queue_table qtable;
+	struct eventfd_ctx *req_trigger;
 	DECLARE_BITMAP(apm_add, AP_DEVICES);
 	DECLARE_BITMAP(aqm_add, AP_DOMAINS);
 	DECLARE_BITMAP(adm_add, AP_DOMAINS);

include/uapi/linux/vfio.h

···
 	VFIO_CCW_NUM_IRQS
 };
 
+/*
+ * The vfio-ap bus driver makes use of the following IRQ index mapping.
+ * Unimplemented IRQ types return a count of zero.
+ */
+enum {
+	VFIO_AP_REQ_IRQ_INDEX,
+	VFIO_AP_NUM_IRQS
+};
+
 /**
  * VFIO_DEVICE_GET_PCI_HOT_RESET_INFO - _IOWR(VFIO_TYPE, VFIO_BASE + 12,
  *					      struct vfio_pci_hot_reset_info)