[PATCH 0/7] Xen PV on HVM fixes and improvements

[PATCH 0/7] Xen PV on HVM fixes and improvements

[Stefano Stabellini]

Hi all,
this patch series is a collection of fixes and improvements for Linux
running as Xen PV on HVM guest.
Most of the patches have already been sent to the list at least once,
some have already been acked.


The list of patches with diffstat follows:

Stefano Stabellini (7):
      xen: no need to delay xen_setup_shutdown_event for hvm guests anymore
      xen: do not use xen_info on HVM, set pv_info name to "Xen HVM"
      xen-blkfront: handle Xen major numbers other than XENVBD
      xen: make the ballon driver work for hvm domains
      xen: PV on HVM: support PV spinlocks
      xen: enable event channels to send and receive IPIs for PV on HVM guests
      xen: fix compile issue if XEN is enabled but XEN_PVHVM is disabled

 arch/x86/xen/enlighten.c         |    6 ++-
 arch/x86/xen/smp.c               |   42 ++++++++++++++++++++
 arch/x86/xen/suspend.c           |    2 +
 arch/x86/xen/xen-ops.h           |    2 +
 drivers/block/xen-blkfront.c     |   79 +++++++++++++++++++++++++++++++++++--
 drivers/xen/balloon.c            |   14 ++++--
 drivers/xen/manage.c             |   17 ++------
 drivers/xen/platform-pci.c       |    3 -
 include/xen/interface/io/blkif.h |   21 ++++++++++
 9 files changed, 158 insertions(+), 28 deletions(-)


A branch with these patches on 2.6.38-rc5 is available here:

git://xenbits.xen.org/people/sstabellini/linux-pvhvm.git 2.6.38-rc5-pvhvm

Cheers,

Stefano

[PATCH 1/7] xen: no need to delay xen_setup_shutdown_event for hvm guests anymore

[stefano.stabellini]

From: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

Now that xenstore_ready is used correctly for PV on HVM guests too, we
don't need to delay the initialization of xen_setup_shutdown_event
anymore.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Jeremy Fitzhardinge <jeremy@goop.org>
---
 drivers/xen/manage.c       |   17 ++++-------------
 drivers/xen/platform-pci.c |    3 ---
 2 files changed, 4 insertions(+), 16 deletions(-)

diff --git a/drivers/xen/manage.c b/drivers/xen/manage.c
index db8c4c4..459cf8e 100644
--- a/drivers/xen/manage.c
+++ b/drivers/xen/manage.c
@@ -281,27 +281,18 @@ static int shutdown_event(struct notifier_block *notifier,
 	return NOTIFY_DONE;
 }
 
-static int __init __setup_shutdown_event(void)
-{
-	/* Delay initialization in the PV on HVM case */
-	if (xen_hvm_domain())
-		return 0;
-
-	if (!xen_pv_domain())
-		return -ENODEV;
-
-	return xen_setup_shutdown_event();
-}
-
 int xen_setup_shutdown_event(void)
 {
 	static struct notifier_block xenstore_notifier = {
 		.notifier_call = shutdown_event
 	};
+
+	if (!xen_domain())
+		return -ENODEV;
 	register_xenstore_notifier(&xenstore_notifier);
 
 	return 0;
 }
 EXPORT_SYMBOL_GPL(xen_setup_shutdown_event);
 
-subsys_initcall(__setup_shutdown_event);
+subsys_initcall(xen_setup_shutdown_event);
diff --git a/drivers/xen/platform-pci.c b/drivers/xen/platform-pci.c
index afbe041..319dd0a 100644
--- a/drivers/xen/platform-pci.c
+++ b/drivers/xen/platform-pci.c
@@ -156,9 +156,6 @@ static int __devinit platform_pci_init(struct pci_dev *pdev,
 	if (ret)
 		goto out;
 	xenbus_probe(NULL);
-	ret = xen_setup_shutdown_event();
-	if (ret)
-		goto out;
 	return 0;
 
 out:
-- 
1.5.6.5

[PATCH 2/7] xen: do not use xen_info on HVM, set pv_info name to "Xen HVM"

[stefano.stabellini]

From: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Jeremy Fitzhardinge <jeremy@goop.org>
---
 arch/x86/xen/enlighten.c |    3 +--
 1 files changed, 1 insertions(+), 2 deletions(-)

diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index 50542ef..2f67e2e 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -1284,8 +1284,7 @@ static int init_hvm_pv_info(int *major, int *minor)
 
 	xen_setup_features();
 
-	pv_info = xen_info;
-	pv_info.kernel_rpl = 0;
+	pv_info.name = "Xen HVM";
 
 	xen_domain_type = XEN_HVM_DOMAIN;
 
-- 
1.5.6.5

[PATCH 3/7] xen-blkfront: handle Xen major numbers other than XENVBD

[stefano.stabellini]

From: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

This patch makes sure blkfront handles correctly virtual device numbers
corresponding to Xen emulated IDE and SCSI disks: in those cases
blkfront translates the major number to XENVBD and the minor number to a
low xvd minor.

Note: this behaviour is different from what old xenlinux PV guests used
to do: they used to steal an IDE or SCSI major number and use it
instead.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Jeremy Fitzhardinge <jeremy@goop.org>
---
 drivers/block/xen-blkfront.c     |   79 +++++++++++++++++++++++++++++++++++--
 include/xen/interface/io/blkif.h |   21 ++++++++++
 2 files changed, 95 insertions(+), 5 deletions(-)

diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
index d7aa39e..64d9c6d 100644
--- a/drivers/block/xen-blkfront.c
+++ b/drivers/block/xen-blkfront.c
@@ -120,6 +120,10 @@ static DEFINE_SPINLOCK(minor_lock);
 #define EXTENDED (1<<EXT_SHIFT)
 #define VDEV_IS_EXTENDED(dev) ((dev)&(EXTENDED))
 #define BLKIF_MINOR_EXT(dev) ((dev)&(~EXTENDED))
+#define EMULATED_HD_DISK_MINOR_OFFSET (0)
+#define EMULATED_HD_DISK_NAME_OFFSET (EMULATED_HD_DISK_MINOR_OFFSET / 256)
+#define EMULATED_SD_DISK_MINOR_OFFSET (EMULATED_HD_DISK_MINOR_OFFSET + (4 * 16))
+#define EMULATED_SD_DISK_NAME_OFFSET (EMULATED_HD_DISK_NAME_OFFSET + 4)
 
 #define DEV_NAME	"xvd"	/* name in /dev */
 
@@ -434,6 +438,65 @@ static void xlvbd_flush(struct blkfront_info *info)
 	       info->feature_flush ? "enabled" : "disabled");
 }
 
+static int xen_translate_vdev(int vdevice, int *minor, unsigned int *offset)
+{
+	int major;
+	major = BLKIF_MAJOR(vdevice);
+	*minor = BLKIF_MINOR(vdevice);
+	switch (major) {
+		case XEN_IDE0_MAJOR:
+			*offset = (*minor / 64) + EMULATED_HD_DISK_NAME_OFFSET;
+			*minor = ((*minor / 64) * PARTS_PER_DISK) +
+				EMULATED_HD_DISK_MINOR_OFFSET;
+			break;
+		case XEN_IDE1_MAJOR:
+			*offset = (*minor / 64) + 2 + EMULATED_HD_DISK_NAME_OFFSET;
+			*minor = (((*minor / 64) + 2) * PARTS_PER_DISK) +
+				EMULATED_HD_DISK_MINOR_OFFSET;
+			break;
+		case XEN_SCSI_DISK0_MAJOR:
+			*offset = (*minor / PARTS_PER_DISK) + EMULATED_SD_DISK_NAME_OFFSET;
+			*minor = *minor + EMULATED_SD_DISK_MINOR_OFFSET;
+			break;
+		case XEN_SCSI_DISK1_MAJOR:
+		case XEN_SCSI_DISK2_MAJOR:
+		case XEN_SCSI_DISK3_MAJOR:
+		case XEN_SCSI_DISK4_MAJOR:
+		case XEN_SCSI_DISK5_MAJOR:
+		case XEN_SCSI_DISK6_MAJOR:
+		case XEN_SCSI_DISK7_MAJOR:
+			*offset = (*minor / PARTS_PER_DISK) + 
+				((major - XEN_SCSI_DISK1_MAJOR + 1) * 16) +
+				EMULATED_SD_DISK_NAME_OFFSET;
+			*minor = *minor +
+				((major - XEN_SCSI_DISK1_MAJOR + 1) * 16 * PARTS_PER_DISK) +
+				EMULATED_SD_DISK_MINOR_OFFSET;
+			break;
+		case XEN_SCSI_DISK8_MAJOR:
+		case XEN_SCSI_DISK9_MAJOR:
+		case XEN_SCSI_DISK10_MAJOR:
+		case XEN_SCSI_DISK11_MAJOR:
+		case XEN_SCSI_DISK12_MAJOR:
+		case XEN_SCSI_DISK13_MAJOR:
+		case XEN_SCSI_DISK14_MAJOR:
+		case XEN_SCSI_DISK15_MAJOR:
+			*offset = (*minor / PARTS_PER_DISK) + 
+				((major - XEN_SCSI_DISK8_MAJOR + 8) * 16) +
+				EMULATED_SD_DISK_NAME_OFFSET;
+			*minor = *minor +
+				((major - XEN_SCSI_DISK8_MAJOR + 8) * 16 * PARTS_PER_DISK) +
+				EMULATED_SD_DISK_MINOR_OFFSET;
+			break;
+		case XENVBD_MAJOR:
+			*offset = *minor / PARTS_PER_DISK;
+			break;
+		default:
+			printk(KERN_WARNING "blkfront: your disk configuration is "
+					"incorrect, please use an xvd device instead\n");
+			return -ENODEV;
+	}
+	return 0;
+}
 
 static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
 			       struct blkfront_info *info,
@@ -441,7 +504,7 @@ static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
 {
 	struct gendisk *gd;
 	int nr_minors = 1;
-	int err = -ENODEV;
+	int err;
 	unsigned int offset;
 	int minor;
 	int nr_parts;
@@ -456,12 +519,20 @@ static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
 	}
 
 	if (!VDEV_IS_EXTENDED(info->vdevice)) {
-		minor = BLKIF_MINOR(info->vdevice);
-		nr_parts = PARTS_PER_DISK;
+		err = xen_translate_vdev(info->vdevice, &minor, &offset);
+		if (err)
+			return err;		
+ 		nr_parts = PARTS_PER_DISK;
 	} else {
 		minor = BLKIF_MINOR_EXT(info->vdevice);
 		nr_parts = PARTS_PER_EXT_DISK;
+		offset = minor / nr_parts;
+		if (xen_hvm_domain() && offset <= EMULATED_HD_DISK_NAME_OFFSET + 4)
+			printk(KERN_WARNING "blkfront: vdevice 0x%x might conflict with "
+					"emulated IDE disks,\n\t choose an xvd device name"
+					"from xvde on\n", info->vdevice);
 	}
+	err = -ENODEV;
 
 	if ((minor % nr_parts) == 0)
 		nr_minors = nr_parts;
@@ -475,8 +546,6 @@ static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
 	if (gd == NULL)
 		goto release;
 
-	offset = minor / nr_parts;
-
 	if (nr_minors > 1) {
 		if (offset < 26)
 			sprintf(gd->disk_name, "%s%c", DEV_NAME, 'a' + offset);
diff --git a/include/xen/interface/io/blkif.h b/include/xen/interface/io/blkif.h
index c2d1fa4..68dd2b4 100644
--- a/include/xen/interface/io/blkif.h
+++ b/include/xen/interface/io/blkif.h
@@ -91,4 +91,25 @@ DEFINE_RING_TYPES(blkif, struct blkif_request, struct blkif_response);
 #define VDISK_REMOVABLE    0x2
 #define VDISK_READONLY     0x4
 
+/* Xen-defined major numbers for virtual disks, they look strangely
+ * familiar */
+#define XEN_IDE0_MAJOR	3
+#define XEN_IDE1_MAJOR	22
+#define XEN_SCSI_DISK0_MAJOR	8
+#define XEN_SCSI_DISK1_MAJOR	65
+#define XEN_SCSI_DISK2_MAJOR	66
+#define XEN_SCSI_DISK3_MAJOR	67
+#define XEN_SCSI_DISK4_MAJOR	68
+#define XEN_SCSI_DISK5_MAJOR	69
+#define XEN_SCSI_DISK6_MAJOR	70
+#define XEN_SCSI_DISK7_MAJOR	71
+#define XEN_SCSI_DISK8_MAJOR	128
+#define XEN_SCSI_DISK9_MAJOR	129
+#define XEN_SCSI_DISK10_MAJOR	130
+#define XEN_SCSI_DISK11_MAJOR	131
+#define XEN_SCSI_DISK12_MAJOR	132
+#define XEN_SCSI_DISK13_MAJOR	133
+#define XEN_SCSI_DISK14_MAJOR	134
+#define XEN_SCSI_DISK15_MAJOR	135
+
 #endif /* __XEN_PUBLIC_IO_BLKIF_H__ */
-- 
1.5.6.5

[PATCH 4/7] xen: make the ballon driver work for hvm domains

[stefano.stabellini]

From: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
---
 drivers/xen/balloon.c |   14 +++++++++-----
 1 files changed, 9 insertions(+), 5 deletions(-)

diff --git a/drivers/xen/balloon.c b/drivers/xen/balloon.c
index 43f9f02..9294f25 100644
--- a/drivers/xen/balloon.c
+++ b/drivers/xen/balloon.c
@@ -232,7 +232,7 @@ static int increase_reservation(unsigned long nr_pages)
 		set_phys_to_machine(pfn, frame_list[i]);
 
 		/* Link back into the page tables if not highmem. */
-		if (pfn < max_low_pfn) {
+		if (!xen_hvm_domain() && pfn < max_low_pfn) {
 			int ret;
 			ret = HYPERVISOR_update_va_mapping(
 				(unsigned long)__va(pfn << PAGE_SHIFT),
@@ -280,7 +280,7 @@ static int decrease_reservation(unsigned long nr_pages)
 
 		scrub_page(page);
 
-		if (!PageHighMem(page)) {
+		if (!xen_hvm_domain() && !PageHighMem(page)) {
 			ret = HYPERVISOR_update_va_mapping(
 				(unsigned long)__va(pfn << PAGE_SHIFT),
 				__pte_ma(0), 0);
@@ -392,15 +392,19 @@ static struct notifier_block xenstore_notifier;
 
 static int __init balloon_init(void)
 {
-	unsigned long pfn, extra_pfn_end;
+ 	unsigned long pfn, nr_pages, extra_pfn_end;
 	struct page *page;
 
-	if (!xen_pv_domain())
+	if (!xen_domain())
 		return -ENODEV;
 
 	pr_info("xen_balloon: Initialising balloon driver.\n");
 
-	balloon_stats.current_pages = min(xen_start_info->nr_pages, max_pfn);
+ 	if (xen_pv_domain())
+ 		nr_pages = xen_start_info->nr_pages;
+ 	else
+ 		nr_pages = max_pfn;
+ 	balloon_stats.current_pages = min(nr_pages, max_pfn);
 	balloon_stats.target_pages  = balloon_stats.current_pages;
 	balloon_stats.balloon_low   = 0;
 	balloon_stats.balloon_high  = 0;
-- 
1.5.6.5

[PATCH 5/7] xen: PV on HVM: support PV spinlocks

[stefano.stabellini]

From: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

Initialize PV spinlocks on boot CPU right after native_smp_prepare_cpus
(that switch to APIC mode and initialize APIC routing); on secondary
CPUs on CPU_UP_PREPARE.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
---
 arch/x86/xen/enlighten.c |   17 +++++++++++++++++
 1 files changed, 17 insertions(+), 0 deletions(-)

diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index 2f67e2e..9c1628b 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -1330,6 +1330,8 @@ static int __cpuinit xen_hvm_cpu_notify(struct notifier_block *self,
 	switch (action) {
 	case CPU_UP_PREPARE:
 		per_cpu(xen_vcpu, cpu) = &HYPERVISOR_shared_info->vcpu_info[cpu];
+		if (xen_have_vector_callback)
+			xen_init_lock_cpu(cpu);
 		break;
 	default:
 		break;
@@ -1341,6 +1343,20 @@ static struct notifier_block __cpuinitdata xen_hvm_cpu_notifier = {
 	.notifier_call	= xen_hvm_cpu_notify,
 };
 
+static void xen_hvm_spinlock_init(void)
+{
+	if (!xen_have_vector_callback)
+		return;
+	xen_init_lock_cpu(0);
+	xen_init_spinlocks();
+}
+
+static void __init xen_hvm_smp_prepare_cpus(unsigned int max_cpus)
+{
+	native_smp_prepare_cpus(max_cpus);
+	xen_hvm_spinlock_init();
+}
+
 static void __init xen_hvm_guest_init(void)
 {
 	int r;
@@ -1360,6 +1376,7 @@ static void __init xen_hvm_guest_init(void)
 	x86_init.irqs.intr_init = xen_init_IRQ;
 	xen_hvm_init_time_ops();
 	xen_hvm_init_mmu_ops();
+	smp_ops.smp_prepare_cpus = xen_hvm_smp_prepare_cpus;
 }
 
 static bool __init xen_hvm_platform(void)
-- 
1.5.6.5

[PATCH 6/7] xen: enable event channels to send and receive IPIs for PV on HVM guests

[stefano.stabellini]

From: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

Enable the usage of event channels to send and receive IPIs when
running as a PV on HVM guest.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
---
 arch/x86/xen/enlighten.c |   16 +---------------
 arch/x86/xen/smp.c       |   42 ++++++++++++++++++++++++++++++++++++++++++
 arch/x86/xen/xen-ops.h   |    2 ++
 3 files changed, 45 insertions(+), 15 deletions(-)

diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index 9c1628b..fe02574 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -1343,20 +1343,6 @@ static struct notifier_block __cpuinitdata xen_hvm_cpu_notifier = {
 	.notifier_call	= xen_hvm_cpu_notify,
 };
 
-static void xen_hvm_spinlock_init(void)
-{
-	if (!xen_have_vector_callback)
-		return;
-	xen_init_lock_cpu(0);
-	xen_init_spinlocks();
-}
-
-static void __init xen_hvm_smp_prepare_cpus(unsigned int max_cpus)
-{
-	native_smp_prepare_cpus(max_cpus);
-	xen_hvm_spinlock_init();
-}
-
 static void __init xen_hvm_guest_init(void)
 {
 	int r;
@@ -1370,13 +1356,13 @@ static void __init xen_hvm_guest_init(void)
 
 	if (xen_feature(XENFEAT_hvm_callback_vector))
 		xen_have_vector_callback = 1;
+	xen_hvm_smp_init();
 	register_cpu_notifier(&xen_hvm_cpu_notifier);
 	xen_unplug_emulated_devices();
 	have_vcpu_info_placement = 0;
 	x86_init.irqs.intr_init = xen_init_IRQ;
 	xen_hvm_init_time_ops();
 	xen_hvm_init_mmu_ops();
-	smp_ops.smp_prepare_cpus = xen_hvm_smp_prepare_cpus;
 }
 
 static bool __init xen_hvm_platform(void)
diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
index 72a4c79..2300d4b 100644
--- a/arch/x86/xen/smp.c
+++ b/arch/x86/xen/smp.c
@@ -509,3 +509,45 @@ void __init xen_smp_init(void)
 	xen_fill_possible_map();
 	xen_init_spinlocks();
 }
+
+static void xen_hvm_spinlock_init(void)
+{
+	if (!xen_have_vector_callback)
+		return;
+	xen_init_lock_cpu(0);
+	xen_init_spinlocks();
+}
+
+static void __init xen_hvm_smp_prepare_cpus(unsigned int max_cpus)
+{
+	native_smp_prepare_cpus(max_cpus);
+	WARN_ON(xen_smp_intr_init(0));
+	xen_hvm_spinlock_init();
+}
+
+static int __cpuinit xen_hvm_cpu_up(unsigned int cpu)
+{
+	int rc;
+	rc = native_cpu_up(cpu);
+	WARN_ON (xen_smp_intr_init(cpu));
+	return rc;
+}
+
+static void xen_hvm_cpu_die(unsigned int cpu)
+{
+	unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
+	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
+	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
+	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
+	native_cpu_die(cpu);
+}
+
+void __init xen_hvm_smp_init(void)
+{
+	smp_ops.smp_prepare_cpus = xen_hvm_smp_prepare_cpus;
+	smp_ops.smp_send_reschedule = xen_smp_send_reschedule;
+	smp_ops.cpu_up = xen_hvm_cpu_up;
+	smp_ops.cpu_die = xen_hvm_cpu_die;
+	smp_ops.send_call_func_ipi = xen_smp_send_call_function_ipi;
+	smp_ops.send_call_func_single_ipi = xen_smp_send_call_function_single_ipi;
+}
diff --git a/arch/x86/xen/xen-ops.h b/arch/x86/xen/xen-ops.h
index 9d41bf9..3112f55 100644
--- a/arch/x86/xen/xen-ops.h
+++ b/arch/x86/xen/xen-ops.h
@@ -64,10 +64,12 @@ void xen_setup_vcpu_info_placement(void);
 
 #ifdef CONFIG_SMP
 void xen_smp_init(void);
+void __init xen_hvm_smp_init(void);
 
 extern cpumask_var_t xen_cpu_initialized_map;
 #else
 static inline void xen_smp_init(void) {}
+static inline void xen_hvm_smp_init(void) {}
 #endif
 
 #ifdef CONFIG_PARAVIRT_SPINLOCKS
-- 
1.5.6.5

[PATCH 7/7] xen: fix compile issue if XEN is enabled but XEN_PVHVM is disabled

[stefano.stabellini]

From: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
---
 arch/x86/xen/suspend.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/arch/x86/xen/suspend.c b/arch/x86/xen/suspend.c
index 9bbd63a..4a3d3dd 100644
--- a/arch/x86/xen/suspend.c
+++ b/arch/x86/xen/suspend.c
@@ -28,6 +28,7 @@ void xen_pre_suspend(void)
 
 void xen_hvm_post_suspend(int suspend_cancelled)
 {
+#ifdef CONFIG_XEN_PVHVM
 	int cpu;
 	xen_hvm_init_shared_info();
 	xen_callback_vector();
@@ -37,6 +38,7 @@ void xen_hvm_post_suspend(int suspend_cancelled)
 			xen_setup_runstate_info(cpu);
 		}
 	}
+#endif
 }
 
 void xen_post_suspend(int suspend_cancelled)
-- 
1.5.6.5

Re: [PATCH 3/7] xen-blkfront: handle Xen major numbers other than XENVBD

[Konrad Rzeszutek Wilk]

On Wed, Feb 16, 2011 at 05:53:03PM +0000, stefano.stabellini@eu.citrix.com wrote:
> From: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
> 
> This patch makes sure blkfront handles correctly virtual device numbers
> corresponding to Xen emulated IDE and SCSI disks: in those cases
> blkfront translates the major number to XENVBD and the minor number to a
> low xvd minor.
> 
> Note: this behaviour is different from what old xenlinux PV guests used
> to do: they used to steal an IDE or SCSI major number and use it
> instead.
> 
> Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
> Acked-by: Jeremy Fitzhardinge <jeremy@goop.org>
> ---
>  drivers/block/xen-blkfront.c     |   79 +++++++++++++++++++++++++++++++++++--
>  include/xen/interface/io/blkif.h |   21 ++++++++++
>  2 files changed, 95 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
> index d7aa39e..64d9c6d 100644
> --- a/drivers/block/xen-blkfront.c
> +++ b/drivers/block/xen-blkfront.c
> @@ -120,6 +120,10 @@ static DEFINE_SPINLOCK(minor_lock);
>  #define EXTENDED (1<<EXT_SHIFT)
>  #define VDEV_IS_EXTENDED(dev) ((dev)&(EXTENDED))
>  #define BLKIF_MINOR_EXT(dev) ((dev)&(~EXTENDED))
> +#define EMULATED_HD_DISK_MINOR_OFFSET (0)
> +#define EMULATED_HD_DISK_NAME_OFFSET (EMULATED_HD_DISK_MINOR_OFFSET / 256)

0 / 256 ? Why not just 0?

> +#define EMULATED_SD_DISK_MINOR_OFFSET (EMULATED_HD_DISK_MINOR_OFFSET + (4 * 16))
> +#define EMULATED_SD_DISK_NAME_OFFSET (EMULATED_HD_DISK_NAME_OFFSET + 4)
>  
>  #define DEV_NAME	"xvd"	/* name in /dev */
>  
> @@ -434,6 +438,65 @@ static void xlvbd_flush(struct blkfront_info *info)
>  	       info->feature_flush ? "enabled" : "disabled");
>  }
>  
> +static int xen_translate_vdev(int vdevice, int *minor, unsigned int *offset)
> +{
> +	int major;
> +	major = BLKIF_MAJOR(vdevice);
> +	*minor = BLKIF_MINOR(vdevice);
> +	switch (major) {
> +		case XEN_IDE0_MAJOR:
> +			*offset = (*minor / 64) + EMULATED_HD_DISK_NAME_OFFSET;
> +			*minor = ((*minor / 64) * PARTS_PER_DISK) +
> +				EMULATED_HD_DISK_MINOR_OFFSET;
> +			break;
> +		case XEN_IDE1_MAJOR:
> +			*offset = (*minor / 64) + 2 + EMULATED_HD_DISK_NAME_OFFSET;
> +			*minor = (((*minor / 64) + 2) * PARTS_PER_DISK) +
> +				EMULATED_HD_DISK_MINOR_OFFSET;
> +			break;
> +		case XEN_SCSI_DISK0_MAJOR:
> +			*offset = (*minor / PARTS_PER_DISK) + EMULATED_SD_DISK_NAME_OFFSET;
> +			*minor = *minor + EMULATED_SD_DISK_MINOR_OFFSET;
> +			break;
> +		case XEN_SCSI_DISK1_MAJOR:
> +		case XEN_SCSI_DISK2_MAJOR:
> +		case XEN_SCSI_DISK3_MAJOR:
> +		case XEN_SCSI_DISK4_MAJOR:
> +		case XEN_SCSI_DISK5_MAJOR:
> +		case XEN_SCSI_DISK6_MAJOR:
> +		case XEN_SCSI_DISK7_MAJOR:
> +			*offset = (*minor / PARTS_PER_DISK) + 
> +				((major - XEN_SCSI_DISK1_MAJOR + 1) * 16) +
> +				EMULATED_SD_DISK_NAME_OFFSET;
> +			*minor = *minor +
> +				((major - XEN_SCSI_DISK1_MAJOR + 1) * 16 * PARTS_PER_DISK) +
> +				EMULATED_SD_DISK_MINOR_OFFSET;
> +			break;
> +		case XEN_SCSI_DISK8_MAJOR:
> +		case XEN_SCSI_DISK9_MAJOR:
> +		case XEN_SCSI_DISK10_MAJOR:
> +		case XEN_SCSI_DISK11_MAJOR:
> +		case XEN_SCSI_DISK12_MAJOR:
> +		case XEN_SCSI_DISK13_MAJOR:
> +		case XEN_SCSI_DISK14_MAJOR:
> +		case XEN_SCSI_DISK15_MAJOR:
> +			*offset = (*minor / PARTS_PER_DISK) + 
> +				((major - XEN_SCSI_DISK8_MAJOR + 8) * 16) +
> +				EMULATED_SD_DISK_NAME_OFFSET;
> +			*minor = *minor +
> +				((major - XEN_SCSI_DISK8_MAJOR + 8) * 16 * PARTS_PER_DISK) +
> +				EMULATED_SD_DISK_MINOR_OFFSET;
> +			break;
> +		case XENVBD_MAJOR:
> +			*offset = *minor / PARTS_PER_DISK;
> +			break;
> +		default:
> +			printk(KERN_WARNING "blkfront: your disk configuration is "
> +					"incorrect, please use an xvd device instead\n");
> +			return -ENODEV;
> +	}
> +	return 0;
> +}
>  
>  static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
>  			       struct blkfront_info *info,
> @@ -441,7 +504,7 @@ static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
>  {
>  	struct gendisk *gd;
>  	int nr_minors = 1;
> -	int err = -ENODEV;
> +	int err;
>  	unsigned int offset;
>  	int minor;
>  	int nr_parts;
> @@ -456,12 +519,20 @@ static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
>  	}
>  
>  	if (!VDEV_IS_EXTENDED(info->vdevice)) {
> -		minor = BLKIF_MINOR(info->vdevice);
> -		nr_parts = PARTS_PER_DISK;
> +		err = xen_translate_vdev(info->vdevice, &minor, &offset);
> +		if (err)
> +			return err;		
> + 		nr_parts = PARTS_PER_DISK;
>  	} else {
>  		minor = BLKIF_MINOR_EXT(info->vdevice);
>  		nr_parts = PARTS_PER_EXT_DISK;
> +		offset = minor / nr_parts;
> +		if (xen_hvm_domain() && offset <= EMULATED_HD_DISK_NAME_OFFSET + 4)
> +			printk(KERN_WARNING "blkfront: vdevice 0x%x might conflict with "
> +					"emulated IDE disks,\n\t choose an xvd device name"
> +					"from xvde on\n", info->vdevice);
>  	}
> +	err = -ENODEV;
>  
>  	if ((minor % nr_parts) == 0)
>  		nr_minors = nr_parts;
> @@ -475,8 +546,6 @@ static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
>  	if (gd == NULL)
>  		goto release;
>  
> -	offset = minor / nr_parts;
> -
>  	if (nr_minors > 1) {
>  		if (offset < 26)
>  			sprintf(gd->disk_name, "%s%c", DEV_NAME, 'a' + offset);
> diff --git a/include/xen/interface/io/blkif.h b/include/xen/interface/io/blkif.h
> index c2d1fa4..68dd2b4 100644
> --- a/include/xen/interface/io/blkif.h
> +++ b/include/xen/interface/io/blkif.h
> @@ -91,4 +91,25 @@ DEFINE_RING_TYPES(blkif, struct blkif_request, struct blkif_response);
>  #define VDISK_REMOVABLE    0x2
>  #define VDISK_READONLY     0x4
>  
> +/* Xen-defined major numbers for virtual disks, they look strangely
> + * familiar */
> +#define XEN_IDE0_MAJOR	3
> +#define XEN_IDE1_MAJOR	22
> +#define XEN_SCSI_DISK0_MAJOR	8
> +#define XEN_SCSI_DISK1_MAJOR	65
> +#define XEN_SCSI_DISK2_MAJOR	66
> +#define XEN_SCSI_DISK3_MAJOR	67
> +#define XEN_SCSI_DISK4_MAJOR	68
> +#define XEN_SCSI_DISK5_MAJOR	69
> +#define XEN_SCSI_DISK6_MAJOR	70
> +#define XEN_SCSI_DISK7_MAJOR	71
> +#define XEN_SCSI_DISK8_MAJOR	128
> +#define XEN_SCSI_DISK9_MAJOR	129
> +#define XEN_SCSI_DISK10_MAJOR	130
> +#define XEN_SCSI_DISK11_MAJOR	131
> +#define XEN_SCSI_DISK12_MAJOR	132
> +#define XEN_SCSI_DISK13_MAJOR	133
> +#define XEN_SCSI_DISK14_MAJOR	134
> +#define XEN_SCSI_DISK15_MAJOR	135
> +
>  #endif /* __XEN_PUBLIC_IO_BLKIF_H__ */
> -- 
> 1.5.6.5

Re: [PATCH 6/7] xen: enable event channels to send and receive IPIs for PV on HVM guests

[Konrad Rzeszutek Wilk]

On Wed, Feb 16, 2011 at 05:53:06PM +0000, stefano.stabellini@eu.citrix.com wrote:
> From: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
> 
> Enable the usage of event channels to send and receive IPIs when
> running as a PV on HVM guest.

Why not squash this with the other patch? After this, the only thing
that the other patch had done was to add in the 'xen_hvm_cpu_notify' function
a call to setup the spinlocks..

> 
> Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
> ---
>  arch/x86/xen/enlighten.c |   16 +---------------
>  arch/x86/xen/smp.c       |   42 ++++++++++++++++++++++++++++++++++++++++++
>  arch/x86/xen/xen-ops.h   |    2 ++
>  3 files changed, 45 insertions(+), 15 deletions(-)
> 
> diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
> index 9c1628b..fe02574 100644
> --- a/arch/x86/xen/enlighten.c
> +++ b/arch/x86/xen/enlighten.c
> @@ -1343,20 +1343,6 @@ static struct notifier_block __cpuinitdata xen_hvm_cpu_notifier = {
>  	.notifier_call	= xen_hvm_cpu_notify,
>  };
>  
> -static void xen_hvm_spinlock_init(void)
> -{
> -	if (!xen_have_vector_callback)
> -		return;
> -	xen_init_lock_cpu(0);
> -	xen_init_spinlocks();
> -}
> -
> -static void __init xen_hvm_smp_prepare_cpus(unsigned int max_cpus)
> -{
> -	native_smp_prepare_cpus(max_cpus);
> -	xen_hvm_spinlock_init();
> -}
> -
>  static void __init xen_hvm_guest_init(void)
>  {
>  	int r;
> @@ -1370,13 +1356,13 @@ static void __init xen_hvm_guest_init(void)
>  
>  	if (xen_feature(XENFEAT_hvm_callback_vector))
>  		xen_have_vector_callback = 1;
> +	xen_hvm_smp_init();
>  	register_cpu_notifier(&xen_hvm_cpu_notifier);
>  	xen_unplug_emulated_devices();
>  	have_vcpu_info_placement = 0;
>  	x86_init.irqs.intr_init = xen_init_IRQ;
>  	xen_hvm_init_time_ops();
>  	xen_hvm_init_mmu_ops();
> -	smp_ops.smp_prepare_cpus = xen_hvm_smp_prepare_cpus;
>  }
>  
>  static bool __init xen_hvm_platform(void)
> diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
> index 72a4c79..2300d4b 100644
> --- a/arch/x86/xen/smp.c
> +++ b/arch/x86/xen/smp.c
> @@ -509,3 +509,45 @@ void __init xen_smp_init(void)
>  	xen_fill_possible_map();
>  	xen_init_spinlocks();
>  }
> +
> +static void xen_hvm_spinlock_init(void)
> +{
> +	if (!xen_have_vector_callback)
> +		return;
> +	xen_init_lock_cpu(0);
> +	xen_init_spinlocks();
> +}
> +
> +static void __init xen_hvm_smp_prepare_cpus(unsigned int max_cpus)
> +{
> +	native_smp_prepare_cpus(max_cpus);
> +	WARN_ON(xen_smp_intr_init(0));
> +	xen_hvm_spinlock_init();

Why not merge 'xen_hvm_spinlock_init' in here?

> +}
> +
> +static int __cpuinit xen_hvm_cpu_up(unsigned int cpu)
> +{
> +	int rc;
> +	rc = native_cpu_up(cpu);
> +	WARN_ON (xen_smp_intr_init(cpu));
> +	return rc;
> +}
> +
> +static void xen_hvm_cpu_die(unsigned int cpu)
> +{
> +	unbind_from_irqhandler(per_cpu(xen_resched_irq, cpu), NULL);
> +	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
> +	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
> +	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
> +	native_cpu_die(cpu);
> +}
> +
> +void __init xen_hvm_smp_init(void)
> +{
> +	smp_ops.smp_prepare_cpus = xen_hvm_smp_prepare_cpus;
> +	smp_ops.smp_send_reschedule = xen_smp_send_reschedule;
> +	smp_ops.cpu_up = xen_hvm_cpu_up;
> +	smp_ops.cpu_die = xen_hvm_cpu_die;
> +	smp_ops.send_call_func_ipi = xen_smp_send_call_function_ipi;
> +	smp_ops.send_call_func_single_ipi = xen_smp_send_call_function_single_ipi;
> +}
> diff --git a/arch/x86/xen/xen-ops.h b/arch/x86/xen/xen-ops.h
> index 9d41bf9..3112f55 100644
> --- a/arch/x86/xen/xen-ops.h
> +++ b/arch/x86/xen/xen-ops.h
> @@ -64,10 +64,12 @@ void xen_setup_vcpu_info_placement(void);
>  
>  #ifdef CONFIG_SMP
>  void xen_smp_init(void);
> +void __init xen_hvm_smp_init(void);
>  
>  extern cpumask_var_t xen_cpu_initialized_map;
>  #else
>  static inline void xen_smp_init(void) {}
> +static inline void xen_hvm_smp_init(void) {}
>  #endif
>  
>  #ifdef CONFIG_PARAVIRT_SPINLOCKS
> -- 
> 1.5.6.5

Re: [PATCH 5/7] xen: PV on HVM: support PV spinlocks

[Konrad Rzeszutek Wilk]

> +static void xen_hvm_spinlock_init(void)
> +{
> +	if (!xen_have_vector_callback)
> +		return;
> +	xen_init_lock_cpu(0);
> +	xen_init_spinlocks();
> +}
> +
> +static void __init xen_hvm_smp_prepare_cpus(unsigned int max_cpus)
> +{
> +	native_smp_prepare_cpus(max_cpus);
> +	xen_hvm_spinlock_init();
> +}

You could squash those two together. Ah, next patch does something interesting.

Re: [PATCH 3/7] xen-blkfront: handle Xen major numbers other than XENVBD

[Stefano Stabellini]

On Tue, 22 Feb 2011, Konrad Rzeszutek Wilk wrote:
> > @@ -120,6 +120,10 @@ static DEFINE_SPINLOCK(minor_lock);
> >  #define EXTENDED (1<<EXT_SHIFT)
> >  #define VDEV_IS_EXTENDED(dev) ((dev)&(EXTENDED))
> >  #define BLKIF_MINOR_EXT(dev) ((dev)&(~EXTENDED))
> > +#define EMULATED_HD_DISK_MINOR_OFFSET (0)
> > +#define EMULATED_HD_DISK_NAME_OFFSET (EMULATED_HD_DISK_MINOR_OFFSET / 256)
> 
> 0 / 256 ? Why not just 0?

Because if we set EMULATED_HD_DISK_MINOR_OFFSET to something different
from 0, then the divison by 256 would matter.
If you look at the code below you'll see that:

offset = minor / nr_parts;

where nr_parts is 256 and offset is the the device name offset.
EMULATED_HD_DISK_MINOR_OFFSET is the offset of the variable "minor" and
EMULATED_HD_DISK_NAME_OFFSET is the offset of the variable "offset".

Re: [PATCH 6/7] xen: enable event channels to send and receive IPIs for PV on HVM guests

[Stefano Stabellini]

On Tue, 22 Feb 2011, Konrad Rzeszutek Wilk wrote:
> On Wed, Feb 16, 2011 at 05:53:06PM +0000, stefano.stabellini@eu.citrix.com wrote:
> > From: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
> > 
> > Enable the usage of event channels to send and receive IPIs when
> > running as a PV on HVM guest.
> 
> Why not squash this with the other patch? After this, the only thing
> that the other patch had done was to add in the 'xen_hvm_cpu_notify' function
> a call to setup the spinlocks..

Yeah, I'll do that.

> > Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
> > ---
> >  arch/x86/xen/enlighten.c |   16 +---------------
> >  arch/x86/xen/smp.c       |   42 ++++++++++++++++++++++++++++++++++++++++++
> >  arch/x86/xen/xen-ops.h   |    2 ++
> >  3 files changed, 45 insertions(+), 15 deletions(-)
> > 
> > diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
> > index 9c1628b..fe02574 100644
> > --- a/arch/x86/xen/enlighten.c
> > +++ b/arch/x86/xen/enlighten.c
> > @@ -1343,20 +1343,6 @@ static struct notifier_block __cpuinitdata xen_hvm_cpu_notifier = {
> >  	.notifier_call	= xen_hvm_cpu_notify,
> >  };
> >  
> > -static void xen_hvm_spinlock_init(void)
> > -{
> > -	if (!xen_have_vector_callback)
> > -		return;
> > -	xen_init_lock_cpu(0);
> > -	xen_init_spinlocks();
> > -}
> > -
> > -static void __init xen_hvm_smp_prepare_cpus(unsigned int max_cpus)
> > -{
> > -	native_smp_prepare_cpus(max_cpus);
> > -	xen_hvm_spinlock_init();
> > -}
> > -
> >  static void __init xen_hvm_guest_init(void)
> >  {
> >  	int r;
> > @@ -1370,13 +1356,13 @@ static void __init xen_hvm_guest_init(void)
> >  
> >  	if (xen_feature(XENFEAT_hvm_callback_vector))
> >  		xen_have_vector_callback = 1;
> > +	xen_hvm_smp_init();
> >  	register_cpu_notifier(&xen_hvm_cpu_notifier);
> >  	xen_unplug_emulated_devices();
> >  	have_vcpu_info_placement = 0;
> >  	x86_init.irqs.intr_init = xen_init_IRQ;
> >  	xen_hvm_init_time_ops();
> >  	xen_hvm_init_mmu_ops();
> > -	smp_ops.smp_prepare_cpus = xen_hvm_smp_prepare_cpus;
> >  }
> >  
> >  static bool __init xen_hvm_platform(void)
> > diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
> > index 72a4c79..2300d4b 100644
> > --- a/arch/x86/xen/smp.c
> > +++ b/arch/x86/xen/smp.c
> > @@ -509,3 +509,45 @@ void __init xen_smp_init(void)
> >  	xen_fill_possible_map();
> >  	xen_init_spinlocks();
> >  }
> > +
> > +static void xen_hvm_spinlock_init(void)
> > +{
> > +	if (!xen_have_vector_callback)
> > +		return;
> > +	xen_init_lock_cpu(0);
> > +	xen_init_spinlocks();
> > +}
> > +
> > +static void __init xen_hvm_smp_prepare_cpus(unsigned int max_cpus)
> > +{
> > +	native_smp_prepare_cpus(max_cpus);
> > +	WARN_ON(xen_smp_intr_init(0));
> > +	xen_hvm_spinlock_init();
> 
> Why not merge 'xen_hvm_spinlock_init' in here?
> 

sure