KVM: s390: hardware support for guest debugging

This patch adds support to debug the guest using the PER facility on s390.
Single-stepping, hardware breakpoints and hardware watchpoints are supported. In
order to use the PER facility of the guest without it noticing it, the control
registers of the guest have to be patched and access to them has to be
intercepted(stctl, stctg, lctl, lctlg).

All PER program interrupts have to be intercepted and only the relevant PER
interrupts for the guest have to be given back. Special care has to be taken
about repeated exits on the same hardware breakpoint. The intervention of the
host in the guests PER configuration is not fully transparent. PER instruction
nullification can not be used by the guest and too many storage alteration
events may be reported to the guest (if it is activated for special address
ranges only) when the host concurrently debugging it.

Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>

1 files changed, 479 insertions, 0 deletions

diff --git a/arch/s390/kvm/guestdbg.c b/arch/s390/kvm/guestdbg.c
new file mode 100644
index 000000000000..100e99d1030d
--- /dev/null
+++ b/arch/s390/kvm/guestdbg.c
@@ -0,0 +1,479 @@
+/*
+ * kvm guest debug support
+ *
+ * Copyright IBM Corp. 2014
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License (version 2 only)
+ * as published by the Free Software Foundation.
+ *
+ *    Author(s): David Hildenbrand <dahi@linux.vnet.ibm.com>
+ */
+#include <linux/kvm_host.h>
+#include <linux/errno.h>
+#include "kvm-s390.h"
+#include "gaccess.h"
+
+/*
+ * Extends the address range given by *start and *stop to include the address
+ * range starting with estart and the length len. Takes care of overflowing
+ * intervals and tries to minimize the overall intervall size.
+ */
+static void extend_address_range(u64 *start, u64 *stop, u64 estart, int len)
+{
+	u64 estop;
+
+	if (len > 0)
+		len--;
+	else
+		len = 0;
+
+	estop = estart + len;
+
+	/* 0-0 range represents "not set" */
+	if ((*start == 0) && (*stop == 0)) {
+		*start = estart;
+		*stop = estop;
+	} else if (*start <= *stop) {
+		/* increase the existing range */
+		if (estart < *start)
+			*start = estart;
+		if (estop > *stop)
+			*stop = estop;
+	} else {
+		/* "overflowing" interval, whereby *stop > *start */
+		if (estart <= *stop) {
+			if (estop > *stop)
+				*stop = estop;
+		} else if (estop > *start) {
+			if (estart < *start)
+				*start = estart;
+		}
+		/* minimize the range */
+		else if ((estop - *stop) < (*start - estart))
+			*stop = estop;
+		else
+			*start = estart;
+	}
+}
+
+#define MAX_INST_SIZE 6
+
+static void enable_all_hw_bp(struct kvm_vcpu *vcpu)
+{
+	unsigned long start, len;
+	u64 *cr9 = &vcpu->arch.sie_block->gcr[9];
+	u64 *cr10 = &vcpu->arch.sie_block->gcr[10];
+	u64 *cr11 = &vcpu->arch.sie_block->gcr[11];
+	int i;
+
+	if (vcpu->arch.guestdbg.nr_hw_bp <= 0 ||
+	    vcpu->arch.guestdbg.hw_bp_info == NULL)
+		return;
+
+	/*
+	 * If the guest is not interrested in branching events, we can savely
+	 * limit them to the PER address range.
+	 */
+	if (!(*cr9 & PER_EVENT_BRANCH))
+		*cr9 |= PER_CONTROL_BRANCH_ADDRESS;
+	*cr9 |= PER_EVENT_IFETCH | PER_EVENT_BRANCH;
+
+	for (i = 0; i < vcpu->arch.guestdbg.nr_hw_bp; i++) {
+		start = vcpu->arch.guestdbg.hw_bp_info[i].addr;
+		len = vcpu->arch.guestdbg.hw_bp_info[i].len;
+
+		/*
+		 * The instruction in front of the desired bp has to
+		 * report instruction-fetching events
+		 */
+		if (start < MAX_INST_SIZE) {
+			len += start;
+			start = 0;
+		} else {
+			start -= MAX_INST_SIZE;
+			len += MAX_INST_SIZE;
+		}
+
+		extend_address_range(cr10, cr11, start, len);
+	}
+}
+
+static void enable_all_hw_wp(struct kvm_vcpu *vcpu)
+{
+	unsigned long start, len;
+	u64 *cr9 = &vcpu->arch.sie_block->gcr[9];
+	u64 *cr10 = &vcpu->arch.sie_block->gcr[10];
+	u64 *cr11 = &vcpu->arch.sie_block->gcr[11];
+	int i;
+
+	if (vcpu->arch.guestdbg.nr_hw_wp <= 0 ||
+	    vcpu->arch.guestdbg.hw_wp_info == NULL)
+		return;
+
+	/* if host uses storage alternation for special address
+	 * spaces, enable all events and give all to the guest */
+	if (*cr9 & PER_EVENT_STORE && *cr9 & PER_CONTROL_ALTERATION) {
+		*cr9 &= ~PER_CONTROL_ALTERATION;
+		*cr10 = 0;
+		*cr11 = PSW_ADDR_INSN;
+	} else {
+		*cr9 &= ~PER_CONTROL_ALTERATION;
+		*cr9 |= PER_EVENT_STORE;
+
+		for (i = 0; i < vcpu->arch.guestdbg.nr_hw_wp; i++) {
+			start = vcpu->arch.guestdbg.hw_wp_info[i].addr;
+			len = vcpu->arch.guestdbg.hw_wp_info[i].len;
+
+			extend_address_range(cr10, cr11, start, len);
+		}
+	}
+}
+
+void kvm_s390_backup_guest_per_regs(struct kvm_vcpu *vcpu)
+{
+	vcpu->arch.guestdbg.cr0 = vcpu->arch.sie_block->gcr[0];
+	vcpu->arch.guestdbg.cr9 = vcpu->arch.sie_block->gcr[9];
+	vcpu->arch.guestdbg.cr10 = vcpu->arch.sie_block->gcr[10];
+	vcpu->arch.guestdbg.cr11 = vcpu->arch.sie_block->gcr[11];
+}
+
+void kvm_s390_restore_guest_per_regs(struct kvm_vcpu *vcpu)
+{
+	vcpu->arch.sie_block->gcr[0] = vcpu->arch.guestdbg.cr0;
+	vcpu->arch.sie_block->gcr[9] = vcpu->arch.guestdbg.cr9;
+	vcpu->arch.sie_block->gcr[10] = vcpu->arch.guestdbg.cr10;
+	vcpu->arch.sie_block->gcr[11] = vcpu->arch.guestdbg.cr11;
+}
+
+void kvm_s390_patch_guest_per_regs(struct kvm_vcpu *vcpu)
+{
+	/*
+	 * TODO: if guest psw has per enabled, otherwise 0s!
+	 * This reduces the amount of reported events.
+	 * Need to intercept all psw changes!
+	 */
+
+	if (guestdbg_sstep_enabled(vcpu)) {
+		vcpu->arch.sie_block->gcr[9] |= PER_EVENT_IFETCH;
+		vcpu->arch.sie_block->gcr[10] = 0;
+		vcpu->arch.sie_block->gcr[11] = PSW_ADDR_INSN;
+	}
+
+	if (guestdbg_hw_bp_enabled(vcpu)) {
+		enable_all_hw_bp(vcpu);
+		enable_all_hw_wp(vcpu);
+	}
+
+	/* TODO: Instruction-fetching-nullification not allowed for now */
+	if (vcpu->arch.sie_block->gcr[9] & PER_EVENT_NULLIFICATION)
+		vcpu->arch.sie_block->gcr[9] &= ~PER_EVENT_NULLIFICATION;
+}
+
+#define MAX_WP_SIZE 100
+
+static int __import_wp_info(struct kvm_vcpu *vcpu,
+			    struct kvm_hw_breakpoint *bp_data,
+			    struct kvm_hw_wp_info_arch *wp_info)
+{
+	int ret = 0;
+	wp_info->len = bp_data->len;
+	wp_info->addr = bp_data->addr;
+	wp_info->phys_addr = bp_data->phys_addr;
+	wp_info->old_data = NULL;
+
+	if (wp_info->len < 0 || wp_info->len > MAX_WP_SIZE)
+		return -EINVAL;
+
+	wp_info->old_data = kmalloc(bp_data->len, GFP_KERNEL);
+	if (!wp_info->old_data)
+		return -ENOMEM;
+	/* try to backup the original value */
+	ret = read_guest(vcpu, wp_info->phys_addr, wp_info->old_data,
+			 wp_info->len);
+	if (ret) {
+		kfree(wp_info->old_data);
+		wp_info->old_data = NULL;
+	}
+
+	return ret;
+}
+
+#define MAX_BP_COUNT 50
+
+int kvm_s390_import_bp_data(struct kvm_vcpu *vcpu,
+			    struct kvm_guest_debug *dbg)
+{
+	int ret = 0, nr_wp = 0, nr_bp = 0, i, size;
+	struct kvm_hw_breakpoint *bp_data = NULL;
+	struct kvm_hw_wp_info_arch *wp_info = NULL;
+	struct kvm_hw_bp_info_arch *bp_info = NULL;
+
+	if (dbg->arch.nr_hw_bp <= 0 || !dbg->arch.hw_bp)
+		return 0;
+	else if (dbg->arch.nr_hw_bp > MAX_BP_COUNT)
+		return -EINVAL;
+
+	size = dbg->arch.nr_hw_bp * sizeof(struct kvm_hw_breakpoint);
+	bp_data = kmalloc(size, GFP_KERNEL);
+	if (!bp_data) {
+		ret = -ENOMEM;
+		goto error;
+	}
+
+	ret = copy_from_user(bp_data, dbg->arch.hw_bp, size);
+	if (ret)
+		goto error;
+
+	for (i = 0; i < dbg->arch.nr_hw_bp; i++) {
+		switch (bp_data[i].type) {
+		case KVM_HW_WP_WRITE:
+			nr_wp++;
+			break;
+		case KVM_HW_BP:
+			nr_bp++;
+			break;
+		default:
+			break;
+		}
+	}
+
+	size = nr_wp * sizeof(struct kvm_hw_wp_info_arch);
+	if (size > 0) {
+		wp_info = kmalloc(size, GFP_KERNEL);
+		if (!wp_info) {
+			ret = -ENOMEM;
+			goto error;
+		}
+	}
+	size = nr_bp * sizeof(struct kvm_hw_bp_info_arch);
+	if (size > 0) {
+		bp_info = kmalloc(size, GFP_KERNEL);
+		if (!bp_info) {
+			ret = -ENOMEM;
+			goto error;
+		}
+	}
+
+	for (nr_wp = 0, nr_bp = 0, i = 0; i < dbg->arch.nr_hw_bp; i++) {
+		switch (bp_data[i].type) {
+		case KVM_HW_WP_WRITE:
+			ret = __import_wp_info(vcpu, &bp_data[i],
+					       &wp_info[nr_wp]);
+			if (ret)
+				goto error;
+			nr_wp++;
+			break;
+		case KVM_HW_BP:
+			bp_info[nr_bp].len = bp_data[i].len;
+			bp_info[nr_bp].addr = bp_data[i].addr;
+			nr_bp++;
+			break;
+		}
+	}
+
+	vcpu->arch.guestdbg.nr_hw_bp = nr_bp;
+	vcpu->arch.guestdbg.hw_bp_info = bp_info;
+	vcpu->arch.guestdbg.nr_hw_wp = nr_wp;
+	vcpu->arch.guestdbg.hw_wp_info = wp_info;
+	return 0;
+error:
+	kfree(bp_data);
+	kfree(wp_info);
+	kfree(bp_info);
+	return ret;
+}
+
+void kvm_s390_clear_bp_data(struct kvm_vcpu *vcpu)
+{
+	int i;
+	struct kvm_hw_wp_info_arch *hw_wp_info = NULL;
+
+	for (i = 0; i < vcpu->arch.guestdbg.nr_hw_wp; i++) {
+		hw_wp_info = &vcpu->arch.guestdbg.hw_wp_info[i];
+		kfree(hw_wp_info->old_data);
+		hw_wp_info->old_data = NULL;
+	}
+	kfree(vcpu->arch.guestdbg.hw_wp_info);
+	vcpu->arch.guestdbg.hw_wp_info = NULL;
+
+	kfree(vcpu->arch.guestdbg.hw_bp_info);
+	vcpu->arch.guestdbg.hw_bp_info = NULL;
+
+	vcpu->arch.guestdbg.nr_hw_wp = 0;
+	vcpu->arch.guestdbg.nr_hw_bp = 0;
+}
+
+static inline int in_addr_range(u64 addr, u64 a, u64 b)
+{
+	if (a <= b)
+		return (addr >= a) && (addr <= b);
+	else
+		/* "overflowing" interval */
+		return (addr <= a) && (addr >= b);
+}
+
+#define end_of_range(bp_info) (bp_info->addr + bp_info->len - 1)
+
+static struct kvm_hw_bp_info_arch *find_hw_bp(struct kvm_vcpu *vcpu,
+					      unsigned long addr)
+{
+	struct kvm_hw_bp_info_arch *bp_info = vcpu->arch.guestdbg.hw_bp_info;
+	int i;
+
+	if (vcpu->arch.guestdbg.nr_hw_bp == 0)
+		return NULL;
+
+	for (i = 0; i < vcpu->arch.guestdbg.nr_hw_bp; i++) {
+		/* addr is directly the start or in the range of a bp */
+		if (addr == bp_info->addr)
+			goto found;
+		if (bp_info->len > 0 &&
+		    in_addr_range(addr, bp_info->addr, end_of_range(bp_info)))
+			goto found;
+
+		bp_info++;
+	}
+
+	return NULL;
+found:
+	return bp_info;
+}
+
+static struct kvm_hw_wp_info_arch *any_wp_changed(struct kvm_vcpu *vcpu)
+{
+	int i;
+	struct kvm_hw_wp_info_arch *wp_info = NULL;
+	void *temp = NULL;
+
+	if (vcpu->arch.guestdbg.nr_hw_wp == 0)
+		return NULL;
+
+	for (i = 0; i < vcpu->arch.guestdbg.nr_hw_wp; i++) {
+		wp_info = &vcpu->arch.guestdbg.hw_wp_info[i];
+		if (!wp_info || !wp_info->old_data || wp_info->len <= 0)
+			continue;
+
+		temp = kmalloc(wp_info->len, GFP_KERNEL);
+		if (!temp)
+			continue;
+
+		/* refetch the wp data and compare it to the old value */
+		if (!read_guest(vcpu, wp_info->phys_addr, temp,
+				wp_info->len)) {
+			if (memcmp(temp, wp_info->old_data, wp_info->len)) {
+				kfree(temp);
+				return wp_info;
+			}
+		}
+		kfree(temp);
+		temp = NULL;
+	}
+
+	return NULL;
+}
+
+void kvm_s390_prepare_debug_exit(struct kvm_vcpu *vcpu)
+{
+	vcpu->run->exit_reason = KVM_EXIT_DEBUG;
+	vcpu->guest_debug &= ~KVM_GUESTDBG_EXIT_PENDING;
+}
+
+#define per_bp_event(code) \
+			(code & (PER_EVENT_IFETCH | PER_EVENT_BRANCH))
+#define per_write_wp_event(code) \
+			(code & (PER_EVENT_STORE | PER_EVENT_STORE_REAL))
+
+static int debug_exit_required(struct kvm_vcpu *vcpu)
+{
+	u32 perc = (vcpu->arch.sie_block->perc << 24);
+	struct kvm_debug_exit_arch *debug_exit = &vcpu->run->debug.arch;
+	struct kvm_hw_wp_info_arch *wp_info = NULL;
+	struct kvm_hw_bp_info_arch *bp_info = NULL;
+	unsigned long addr = vcpu->arch.sie_block->gpsw.addr;
+	unsigned long peraddr = vcpu->arch.sie_block->peraddr;
+
+	if (guestdbg_hw_bp_enabled(vcpu)) {
+		if (per_write_wp_event(perc) &&
+		    vcpu->arch.guestdbg.nr_hw_wp > 0) {
+			wp_info = any_wp_changed(vcpu);
+			if (wp_info) {
+				debug_exit->addr = wp_info->addr;
+				debug_exit->type = KVM_HW_WP_WRITE;
+				goto exit_required;
+			}
+		}
+		if (per_bp_event(perc) &&
+			 vcpu->arch.guestdbg.nr_hw_bp > 0) {
+			bp_info = find_hw_bp(vcpu, addr);
+			/* remove duplicate events if PC==PER address */
+			if (bp_info && (addr != peraddr)) {
+				debug_exit->addr = addr;
+				debug_exit->type = KVM_HW_BP;
+				vcpu->arch.guestdbg.last_bp = addr;
+				goto exit_required;
+			}
+			/* breakpoint missed */
+			bp_info = find_hw_bp(vcpu, peraddr);
+			if (bp_info && vcpu->arch.guestdbg.last_bp != peraddr) {
+				debug_exit->addr = peraddr;
+				debug_exit->type = KVM_HW_BP;
+				goto exit_required;
+			}
+		}
+	}
+	if (guestdbg_sstep_enabled(vcpu) && per_bp_event(perc)) {
+		debug_exit->addr = addr;
+		debug_exit->type = KVM_SINGLESTEP;
+		goto exit_required;
+	}
+
+	return 0;
+exit_required:
+	return 1;
+}
+
+#define guest_per_enabled(vcpu) \
+			     (vcpu->arch.sie_block->gpsw.mask & PSW_MASK_PER)
+
+static void filter_guest_per_event(struct kvm_vcpu *vcpu)
+{
+	u32 perc = vcpu->arch.sie_block->perc << 24;
+	u64 peraddr = vcpu->arch.sie_block->peraddr;
+	u64 addr = vcpu->arch.sie_block->gpsw.addr;
+	u64 cr9 = vcpu->arch.sie_block->gcr[9];
+	u64 cr10 = vcpu->arch.sie_block->gcr[10];
+	u64 cr11 = vcpu->arch.sie_block->gcr[11];
+	/* filter all events, demanded by the guest */
+	u32 guest_perc = perc & cr9 & PER_EVENT_MASK;
+
+	if (!guest_per_enabled(vcpu))
+		guest_perc = 0;
+
+	/* filter "successful-branching" events */
+	if (guest_perc & PER_EVENT_BRANCH &&
+	    cr9 & PER_CONTROL_BRANCH_ADDRESS &&
+	    !in_addr_range(addr, cr10, cr11))
+		guest_perc &= ~PER_EVENT_BRANCH;
+
+	/* filter "instruction-fetching" events */
+	if (guest_perc & PER_EVENT_IFETCH &&
+	    !in_addr_range(peraddr, cr10, cr11))
+		guest_perc &= ~PER_EVENT_IFETCH;
+
+	/* All other PER events will be given to the guest */
+	/* TODO: Check alterated address/address space */
+
+	vcpu->arch.sie_block->perc = guest_perc >> 24;
+
+	if (!guest_perc)
+		vcpu->arch.sie_block->iprcc &= ~PGM_PER;
+}
+
+void kvm_s390_handle_per_event(struct kvm_vcpu *vcpu)
+{
+	if (debug_exit_required(vcpu))
+		vcpu->guest_debug |= KVM_GUESTDBG_EXIT_PENDING;
+
+	filter_guest_per_event(vcpu);
+}