RE: RE: [PATCH] xenoprof (xen)

RE: RE: [PATCH] xenoprof (xen)

[Santos, Jose Renato G]

[-- Attachment #1: Type: text/plain, Size: 1845 bytes --]

  Ian, Keir

  Here are the xenoprof patches for the linux-2.6-xen.hg tree.
  John Levon has reviewed these patches and I have included all his
suggestions on this version.
  According to John, it is unlikely to have these patches included into
main stream linux until Xen is also included.
  I am not sure how would you like to distribute it for now. I suggest
that we include the architecture specific driver for xen
(xenoprof-arch-linux-2.6.14.patch) into the linux sparse-tree for 2.6.14
(all code in this patch is under arch/i386/xen); and include the generic
oprofile driver modifications (xenoprof-generic-linux-2.6.14) in the
patches automatically applied when builting the linux trees for xen (all
code in this patch are under drivers/oprofile and include/linux). Please
let me know if this is acceptable or if you prefer an alternative
approach?

  Thanks

  Renato

>> -----Original Message-----
>> From: Ian Pratt [mailto:m+Ian.Pratt@cl.cam.ac.uk] 
>> Sent: Sunday, November 13, 2005 11:24 AM
>> To: Santos, Jose Renato G; Keir Fraser
>> Cc: Turner, Yoshio; xen-devel@lists.xensource.com; Jose 
>> Renato Santos; G John Janakiraman; ian.pratt@cl.cam.ac.uk
>> Subject: RE: [Xen-devel] RE: [PATCH] xenoprof (xen)
>> 
>> 
>> > Is there anything I can do to make your job easier?
>> > Would you like to receive multiple smaller patches (one patch
>> > per file)?
>> 
>> I think we should hold-off applying the linux patches to the 
>> sparse tree for the moment for a couple of reasons: John 
>> Levon would like to see some cleanups, and I think we'll be 
>> better off applying once the sparse tree has moved to 2.6.14 
>> anyhow (generated from the merge tree).
>> 
>> Jose: please could you have a go generating the patch 
>> against the linux-2.6-xen.hg tree.
>> 
>> Thanks,
>> Ian
>> 

[-- Attachment #2: xenoprof-generic-linux-2.6.14.patch --]
[-- Type: application/octet-stream, Size: 11433 bytes --]

diff -r e52af3942d55 drivers/oprofile/buffer_sync.c
--- a/drivers/oprofile/buffer_sync.c	Sat Nov 12 04:30:38 2005
+++ b/drivers/oprofile/buffer_sync.c	Wed Nov 23 14:44:03 2005
@@ -5,6 +5,10 @@
  * @remark Read the file COPYING
  *
  * @author John Levon <levon@movementarian.org>
+ *
+ * Modified by Aravind Menon for Xen
+ * These modifications are:
+ * Copyright (C) 2005 Hewlett-Packard Co.
  *
  * This is the core of the buffer management. Each
  * CPU buffer is processed and entered into the
@@ -272,15 +276,24 @@
 	last_cookie = INVALID_COOKIE;
 }
 
-static void add_kernel_ctx_switch(unsigned int in_kernel)
-{
-	add_event_entry(ESCAPE_CODE);
-	if (in_kernel)
-		add_event_entry(KERNEL_ENTER_SWITCH_CODE); 
-	else
-		add_event_entry(KERNEL_EXIT_SWITCH_CODE); 
-}
- 
+static void add_cpu_mode_switch(unsigned int cpu_mode)
+{
+	add_event_entry(ESCAPE_CODE);
+	switch (cpu_mode) {
+	case CPU_MODE_USER:
+		add_event_entry(USER_ENTER_SWITCH_CODE);
+		break;
+	case CPU_MODE_KERNEL:
+		add_event_entry(KERNEL_ENTER_SWITCH_CODE);
+		break;
+	case CPU_MODE_XEN:
+		add_event_entry(XEN_ENTER_SWITCH_CODE);
+		break;
+	default:
+		break;
+	}
+}
+
 static void
 add_user_ctx_switch(struct task_struct const * task, unsigned long cookie)
 {
@@ -345,9 +358,9 @@
  * for later lookup from userspace.
  */
 static int
-add_sample(struct mm_struct * mm, struct op_sample * s, int in_kernel)
-{
-	if (in_kernel) {
+add_sample(struct mm_struct * mm, struct op_sample * s, int cpu_mode)
+{
+	if (cpu_mode >= CPU_MODE_KERNEL) {
 		add_sample_entry(s->eip, s->event);
 		return 1;
 	} else if (mm) {
@@ -496,7 +509,7 @@
 	struct mm_struct *mm = NULL;
 	struct task_struct * new;
 	unsigned long cookie = 0;
-	int in_kernel = 1;
+	int cpu_mode = 1;
 	unsigned int i;
 	sync_buffer_state state = sb_buffer_start;
 	unsigned long available;
@@ -513,12 +526,12 @@
 		struct op_sample * s = &cpu_buf->buffer[cpu_buf->tail_pos];
  
 		if (is_code(s->eip)) {
-			if (s->event <= CPU_IS_KERNEL) {
+			if (s->event <= CPU_MODE_XEN) {
 				/* kernel/userspace switch */
-				in_kernel = s->event;
+				cpu_mode = s->event;
 				if (state == sb_buffer_start)
 					state = sb_sample_start;
-				add_kernel_ctx_switch(s->event);
+				add_cpu_mode_switch(s->event);
 			} else if (s->event == CPU_TRACE_BEGIN) {
 				state = sb_bt_start;
 				add_trace_begin();
@@ -536,7 +549,7 @@
 			}
 		} else {
 			if (state >= sb_bt_start &&
-			    !add_sample(mm, s, in_kernel)) {
+			    !add_sample(mm, s, cpu_mode)) {
 				if (state == sb_bt_start) {
 					state = sb_bt_ignore;
 					atomic_inc(&oprofile_stats.bt_lost_no_mapping);
diff -r e52af3942d55 drivers/oprofile/cpu_buffer.c
--- a/drivers/oprofile/cpu_buffer.c	Sat Nov 12 04:30:38 2005
+++ b/drivers/oprofile/cpu_buffer.c	Wed Nov 23 14:44:03 2005
@@ -5,6 +5,10 @@
  * @remark Read the file COPYING
  *
  * @author John Levon <levon@movementarian.org>
+ *
+ * Modified by Aravind Menon for Xen
+ * These modifications are:
+ * Copyright (C) 2005 Hewlett-Packard Co.
  *
  * Each CPU has a local buffer that stores PC value/event
  * pairs. We also log context switches when we notice them.
@@ -57,7 +61,7 @@
 			goto fail;
  
 		b->last_task = NULL;
-		b->last_is_kernel = -1;
+		b->last_cpu_mode = -1;
 		b->tracing = 0;
 		b->buffer_size = buffer_size;
 		b->tail_pos = 0;
@@ -113,7 +117,7 @@
 	 * collected will populate the buffer with proper
 	 * values to initialize the buffer
 	 */
-	cpu_buf->last_is_kernel = -1;
+	cpu_buf->last_cpu_mode = -1;
 	cpu_buf->last_task = NULL;
 }
 
@@ -163,13 +167,13 @@
  * because of the head/tail separation of the writer and reader
  * of the CPU buffer.
  *
- * is_kernel is needed because on some architectures you cannot
+ * cpu_mode is needed because on some architectures you cannot
  * tell if you are in kernel or user space simply by looking at
- * pc. We tag this in the buffer by generating kernel enter/exit
- * events whenever is_kernel changes
+ * pc. We tag this in the buffer by generating kernel/user (and xen)
+ *  enter events whenever cpu_mode changes
  */
 static int log_sample(struct oprofile_cpu_buffer * cpu_buf, unsigned long pc,
-		      int is_kernel, unsigned long event)
+		      int cpu_mode, unsigned long event)
 {
 	struct task_struct * task;
 
@@ -180,16 +184,16 @@
 		return 0;
 	}
 
-	is_kernel = !!is_kernel;
+	WARN_ON(cpu_mode > CPU_MODE_XEN);
 
 	task = current;
 
 	/* notice a switch from user->kernel or vice versa */
-	if (cpu_buf->last_is_kernel != is_kernel) {
-		cpu_buf->last_is_kernel = is_kernel;
-		add_code(cpu_buf, is_kernel);
-	}
-
+	if (cpu_buf->last_cpu_mode != cpu_mode) {
+		cpu_buf->last_cpu_mode = cpu_mode;
+		add_code(cpu_buf, cpu_mode);
+	}
+	
 	/* notice a task switch */
 	if (cpu_buf->last_task != task) {
 		cpu_buf->last_task = task;
diff -r e52af3942d55 drivers/oprofile/cpu_buffer.h
--- a/drivers/oprofile/cpu_buffer.h	Sat Nov 12 04:30:38 2005
+++ b/drivers/oprofile/cpu_buffer.h	Wed Nov 23 14:44:03 2005
@@ -36,7 +36,7 @@
 	volatile unsigned long tail_pos;
 	unsigned long buffer_size;
 	struct task_struct * last_task;
-	int last_is_kernel;
+	int last_cpu_mode;
 	int tracing;
 	struct op_sample * buffer;
 	unsigned long sample_received;
@@ -51,7 +51,9 @@
 void cpu_buffer_reset(struct oprofile_cpu_buffer * cpu_buf);
 
 /* transient events for the CPU buffer -> event buffer */
-#define CPU_IS_KERNEL 1
-#define CPU_TRACE_BEGIN 2
+#define CPU_MODE_USER    0
+#define CPU_MODE_KERNEL  1
+#define CPU_MODE_XEN     2
+#define CPU_TRACE_BEGIN  3
 
 #endif /* OPROFILE_CPU_BUFFER_H */
diff -r e52af3942d55 drivers/oprofile/event_buffer.c
--- a/drivers/oprofile/event_buffer.c	Sat Nov 12 04:30:38 2005
+++ b/drivers/oprofile/event_buffer.c	Wed Nov 23 14:44:03 2005
@@ -56,6 +56,7 @@
 /* Wake up the waiting process if any. This happens
  * on "echo 0 >/dev/oprofile/enable" so the daemon
  * processes the data remaining in the event buffer.
+ * also called on echo 1 > /dev/oprofile/dump
  */
 void wake_up_buffer_waiter(void)
 {
diff -r e52af3942d55 drivers/oprofile/event_buffer.h
--- a/drivers/oprofile/event_buffer.h	Sat Nov 12 04:30:38 2005
+++ b/drivers/oprofile/event_buffer.h	Wed Nov 23 14:44:03 2005
@@ -5,6 +5,10 @@
  * @remark Read the file COPYING
  *
  * @author John Levon <levon@movementarian.org>
+ *
+ * Modified by Aravind Menon for Xen
+ * These modifications are:
+ * Copyright (C) 2005 Hewlett-Packard Co.
  */
 
 #ifndef EVENT_BUFFER_H
@@ -29,11 +33,12 @@
 #define CPU_SWITCH_CODE 		2
 #define COOKIE_SWITCH_CODE 		3
 #define KERNEL_ENTER_SWITCH_CODE	4
-#define KERNEL_EXIT_SWITCH_CODE		5
+#define USER_ENTER_SWITCH_CODE		5
 #define MODULE_LOADED_CODE		6
 #define CTX_TGID_CODE			7
 #define TRACE_BEGIN_CODE		8
 #define TRACE_END_CODE			9
+#define XEN_ENTER_SWITCH_CODE		10
  
 #define INVALID_COOKIE ~0UL
 #define NO_COOKIE 0UL
diff -r e52af3942d55 drivers/oprofile/oprof.c
--- a/drivers/oprofile/oprof.c	Sat Nov 12 04:30:38 2005
+++ b/drivers/oprofile/oprof.c	Wed Nov 23 14:44:03 2005
@@ -5,6 +5,10 @@
  * @remark Read the file COPYING
  *
  * @author John Levon <levon@movementarian.org>
+ *
+ * Modified by Aravind Menon for Xen
+ * These modifications are:
+ * Copyright (C) 2005 Hewlett-Packard Co.
  */
 
 #include <linux/kernel.h>
@@ -32,6 +36,17 @@
    1 - use the timer int mechanism regardless
  */
 static int timer = 0;
+
+extern unsigned int adomains;
+extern int active_domains[MAX_OPROF_DOMAINS];
+
+int oprofile_set_active(void)
+{
+	if (oprofile_ops.set_active)
+		return oprofile_ops.set_active(active_domains, adomains);
+
+	return -EINVAL;
+}
 
 int oprofile_setup(void)
 {
diff -r e52af3942d55 drivers/oprofile/oprof.h
--- a/drivers/oprofile/oprof.h	Sat Nov 12 04:30:38 2005
+++ b/drivers/oprofile/oprof.h	Wed Nov 23 14:44:03 2005
@@ -35,5 +35,7 @@
 void oprofile_timer_init(struct oprofile_operations * ops);
 
 int oprofile_set_backtrace(unsigned long depth);
+
+int oprofile_set_active(void);
  
 #endif /* OPROF_H */
diff -r e52af3942d55 drivers/oprofile/oprofile_files.c
--- a/drivers/oprofile/oprofile_files.c	Sat Nov 12 04:30:38 2005
+++ b/drivers/oprofile/oprofile_files.c	Wed Nov 23 14:44:03 2005
@@ -5,10 +5,16 @@
  * @remark Read the file COPYING
  *
  * @author John Levon <levon@movementarian.org>
+ *
+ * Modified by Aravind Menon for Xen
+ * These modifications are:
+ * Copyright (C) 2005 Hewlett-Packard Co.	
  */
 
 #include <linux/fs.h>
 #include <linux/oprofile.h>
+#include <asm/uaccess.h>
+#include <linux/ctype.h>
 
 #include "event_buffer.h"
 #include "oprofile_stats.h"
@@ -117,11 +123,79 @@
 static struct file_operations dump_fops = {
 	.write		= dump_write,
 };
- 
+
+#define TMPBUFSIZE 512
+
+unsigned int adomains = 0;
+long active_domains[MAX_OPROF_DOMAINS];
+
+static ssize_t adomain_write(struct file * file, char const __user * buf, 
+			     size_t count, loff_t * offset)
+{
+	char tmpbuf[TMPBUFSIZE];
+	char * startp = tmpbuf;
+	char * endp = tmpbuf;
+	int i;
+	unsigned long val;
+	
+	if (*offset)
+		return -EINVAL;	
+	if (!count)
+		return 0;
+	if (count > TMPBUFSIZE - 1)
+		return -EINVAL;
+
+	memset(tmpbuf, 0x0, TMPBUFSIZE);
+
+	if (copy_from_user(tmpbuf, buf, count))
+		return -EFAULT;
+	
+	for (i = 0; i < MAX_OPROF_DOMAINS; i++)
+		active_domains[i] = -1;
+	adomains = 0;
+
+	while (1) {
+		val = simple_strtol(startp, &endp, 0);
+		if (endp == startp)
+			break;
+		while (ispunct(*endp))
+			endp++;
+		active_domains[adomains++] = val;
+		if (adomains >= MAX_OPROF_DOMAINS)
+			break;
+		startp = endp;
+	}
+	if (oprofile_set_active())
+		return -EINVAL; 
+	return count;
+}
+
+static ssize_t adomain_read(struct file * file, char __user * buf, 
+			    size_t count, loff_t * offset)
+{
+	char tmpbuf[TMPBUFSIZE];
+	size_t len = 0;
+	int i;
+	/* This is all screwed up if we run out of space */
+	for (i = 0; i < adomains; i++) 
+		len += snprintf(tmpbuf + len, TMPBUFSIZE - len, 
+				"%u ", (unsigned int)active_domains[i]);
+	len += snprintf(tmpbuf + len, TMPBUFSIZE - len, "\n");
+	return simple_read_from_buffer((void __user *)buf, count, 
+				       offset, tmpbuf, len);
+}
+
+
+static struct file_operations active_domain_ops = {
+	.read		= adomain_read,
+	.write		= adomain_write,
+};
+
 void oprofile_create_files(struct super_block * sb, struct dentry * root)
 {
 	oprofilefs_create_file(sb, root, "enable", &enable_fops);
 	oprofilefs_create_file_perm(sb, root, "dump", &dump_fops, 0666);
+	oprofilefs_create_file(sb, root, "active_domains", &active_domain_ops);
 	oprofilefs_create_file(sb, root, "buffer", &event_buffer_fops);
 	oprofilefs_create_ulong(sb, root, "buffer_size", &fs_buffer_size);
 	oprofilefs_create_ulong(sb, root, "buffer_watershed", &fs_buffer_watershed);
diff -r e52af3942d55 include/linux/oprofile.h
--- a/include/linux/oprofile.h	Sat Nov 12 04:30:38 2005
+++ b/include/linux/oprofile.h	Wed Nov 23 14:44:03 2005
@@ -8,6 +8,10 @@
  * @remark Read the file COPYING
  *
  * @author John Levon <levon@movementarian.org>
+ *
+ * Modified by Aravind Menon for Xen
+ * These modifications are:
+ * Copyright (C) 2005 Hewlett-Packard Co.
  */
 
 #ifndef OPROFILE_H
@@ -27,6 +31,8 @@
 	/* create any necessary configuration files in the oprofile fs.
 	 * Optional. */
 	int (*create_files)(struct super_block * sb, struct dentry * root);
+	/* setup active domains with Xen */
+	int (*set_active)(int *active_domains, unsigned int adomains);
 	/* Do any necessary interrupt setup. Optional. */
 	int (*setup)(void);
 	/* Do any necessary interrupt shutdown. Optional. */

[-- Attachment #3: xenoprof-arch-linux-2.6.14.patch --]
[-- Type: application/octet-stream, Size: 15842 bytes --]

diff -r e52af3942d55 arch/i386/Kconfig
--- a/arch/i386/Kconfig	Sat Nov 12 04:30:38 2005
+++ b/arch/i386/Kconfig	Wed Nov 23 14:48:18 2005
@@ -1325,6 +1325,10 @@
 source "arch/i386/oprofile/Kconfig"
 endif
 
+if X86_XEN
+source "arch/i386/xen/oprofile/Kconfig"
+endif
+
 source "arch/i386/Kconfig.debug"
 
 source "security/Kconfig"
diff -r e52af3942d55 arch/i386/Makefile
--- a/arch/i386/Makefile	Sat Nov 12 04:30:38 2005
+++ b/arch/i386/Makefile	Wed Nov 23 14:48:18 2005
@@ -136,7 +136,12 @@
 drivers-$(CONFIG_MATH_EMULATION)	+= arch/i386/math-emu/
 drivers-$(CONFIG_PCI)			+= $(pci)
 # must be linked after kernel/
+ifdef CONFIG_X86_XEN
+drivers-$(CONFIG_OPROFILE)		+= arch/i386/xen/oprofile/
+else
 drivers-$(CONFIG_OPROFILE)		+= arch/i386/oprofile/
+endif
+
 drivers-$(CONFIG_PM)			+= arch/i386/power/
 
 CFLAGS += $(mflags-y)
diff -r e52af3942d55 arch/i386/xen/mm/ioremap.c
--- a/arch/i386/xen/mm/ioremap.c	Sat Nov 12 04:30:38 2005
+++ b/arch/i386/xen/mm/ioremap.c	Wed Nov 23 14:48:18 2005
@@ -171,6 +171,32 @@
 } 
 
 EXPORT_SYMBOL(touch_pte_range);
+
+void *vm_map_xen_pages (unsigned long maddr, int vm_size, pgprot_t prot)
+{
+	int error;
+       
+	struct vm_struct *vma;
+	vma = get_vm_area (vm_size, VM_IOREMAP);
+      
+	if (vma == NULL) {
+		printk ("ioremap.c,vm_map_xen_pages(): Failed to get VMA area\n");
+		return NULL;
+	}
+
+	error = direct_kernel_remap_pfn_range((unsigned long) vma->addr,
+					      maddr >> PAGE_SHIFT, vm_size,
+					      prot, DOMID_SELF );
+	if (error == 0){
+		return  (vma->addr);
+		
+	}
+	else {
+		printk ("ioremap.c,vm_map_xen_pages(): Failed to map xen shared pages into kernel space\n");
+		return NULL;
+	}
+}
+EXPORT_SYMBOL(vm_map_xen_pages);
 
 #ifdef CONFIG_XEN_PHYSDEV_ACCESS
 
diff -r e52af3942d55 include/asm-i386/mach-xen/asm/hypercall.h
--- a/include/asm-i386/mach-xen/asm/hypercall.h	Sat Nov 12 04:30:38 2005
+++ b/include/asm-i386/mach-xen/asm/hypercall.h	Wed Nov 23 14:48:18 2005
@@ -300,6 +300,23 @@
 			   SHUTDOWN_suspend, srec);
 }
 
+static inline int
+HYPERVISOR_pmc_op(
+	int op, unsigned int arg1, unsigned int arg2)
+{
+	int ret;
+	unsigned long ign1, ign2, ign3;
+
+	__asm__ __volatile__ (
+		TRAP_INSTR
+		: "=a"(ret), "=b"(ign1), "=c"(ign2), "=d"(ign3)
+		: "0"(__HYPERVISOR_pmc_op), "1"(op), "2"(arg1), "3"(arg2)
+		: "memory" );
+
+	return ret;
+}
+
+
 #endif /* __HYPERCALL_H__ */
 
 /*
diff -r e52af3942d55 include/asm-xen/xen-public/xen.h
--- a/include/asm-xen/xen-public/xen.h	Sat Nov 12 04:30:38 2005
+++ b/include/asm-xen/xen-public/xen.h	Wed Nov 23 14:48:18 2005
@@ -4,6 +4,10 @@
  * Guest OS interface to Xen.
  * 
  * Copyright (c) 2004, K A Fraser
+ *
+ * Modified by Aravind Menon for supporting oprofile
+ * These modifications are:
+ * Copyright (C) 2005 Hewlett-Packard Co.
  */
 
 #ifndef __XEN_PUBLIC_XEN_H__
@@ -59,6 +63,7 @@
 #define __HYPERVISOR_set_segment_base     25 /* x86/64 only */
 #define __HYPERVISOR_mmuext_op            26
 #define __HYPERVISOR_acm_op               27
+#define __HYPERVISOR_pmc_op               28 
 
 /* 
  * VIRTUAL INTERRUPTS
@@ -72,7 +77,8 @@
 #define VIRQ_PARITY_ERR 4  /* (DOM0) NMI parity error.                    */
 #define VIRQ_IO_ERR     5  /* (DOM0) NMI I/O error.                       */
 #define VIRQ_DEBUGGER   6  /* (DOM0) A domain has paused for debugging.   */
-#define NR_VIRQS        7
+#define VIRQ_PMC_OVF	7  /* PMC Overflow */
+#define NR_VIRQS        8
 
 /*
  * MMU-UPDATE REQUESTS
@@ -211,6 +217,21 @@
 #define VMASST_TYPE_writable_pagetables  2
 #define MAX_VMASST_TYPE 2
 
+/*
+ * Commands to HYPERVISOR_pmc_op().
+ */
+#define PMC_INIT		0
+#define PMC_SET_ACTIVE		1
+#define PMC_SET_PASSIVE		2
+#define PMC_RESERVE_COUNTERS	3
+#define PMC_SETUP_EVENTS	4
+#define PMC_ENABLE_VIRQ		5
+#define PMC_START		6
+#define PMC_STOP		7
+#define PMC_DISABLE_VIRQ	8
+#define PMC_RELEASE_COUNTERS	9
+#define PMC_SHUTDOWN		10
+
 #ifndef __ASSEMBLY__
 
 typedef uint16_t domid_t;
@@ -266,6 +287,8 @@
  */
 #define NR_EVENT_CHANNELS (sizeof(unsigned long) * sizeof(unsigned long) * 64)
 
+#define MAX_OPROF_EVENTS	32
+#define MAX_OPROF_DOMAINS	25	
 /*
  * Per-VCPU information goes here. This will be cleaned up more when Xen 
  * actually supports multi-VCPU guests.
@@ -433,6 +456,33 @@
 
 typedef uint8_t xen_domain_handle_t[16];
 
+struct event_log {
+	uint32_t eip;
+	uint8_t mode;
+	uint8_t event;
+};
+
+/* Xenoprof buffer shared between Xen and domain - 1 per VCPU */
+typedef struct xenoprof_buf {
+    uint32_t event_head;
+    uint32_t event_tail;
+    uint32_t event_size;
+    uint32_t vcpu_id;
+    uint64_t xen_samples;
+    uint64_t kernel_samples;
+    uint64_t user_samples;
+    uint64_t lost_samples;
+    struct event_log event_log[1];
+} xenoprof_buf_t;
+
+typedef struct pmc_init_result {
+    int num_events;
+    int is_primary;
+    int nbuf;
+    int bufsize;
+    uint32_t buf_maddr;
+} pmc_init_result_t;
+
 #endif /* !__ASSEMBLY__ */
 
 #endif /* __XEN_PUBLIC_XEN_H__ */
diff -r e52af3942d55 arch/i386/xen/oprofile/Kconfig
--- /dev/null	Sat Nov 12 04:30:38 2005
+++ b/arch/i386/xen/oprofile/Kconfig	Wed Nov 23 14:48:18 2005
@@ -0,0 +1,25 @@
+
+menu "Profiling support"
+	depends on EXPERIMENTAL
+
+config PROFILING
+	bool "Profiling support (EXPERIMENTAL)"
+	default n
+	help
+	  Say Y here to enable the extended profiling support mechanisms used
+	  by profilers such as OProfile.
+	  
+
+config OPROFILE
+	tristate "OProfile system profiling (EXPERIMENTAL)"
+	depends on PROFILING
+	default n
+	help
+	  OProfile is a profiling system capable of profiling the
+	  whole system, include Xen, the kernel, kernel modules, libraries,
+	  and applications.
+
+	  If unsure, say N.
+
+endmenu
+
diff -r e52af3942d55 arch/i386/xen/oprofile/Makefile
--- /dev/null	Sat Nov 12 04:30:38 2005
+++ b/arch/i386/xen/oprofile/Makefile	Wed Nov 23 14:48:18 2005
@@ -0,0 +1,9 @@
+obj-$(CONFIG_OPROFILE) += oprofile.o
+
+DRIVER_OBJS = $(addprefix ../../../../drivers/oprofile/, \
+		oprof.o cpu_buffer.o buffer_sync.o \
+		event_buffer.o oprofile_files.o \
+		oprofilefs.o oprofile_stats.o  \
+		timer_int.o )
+
+oprofile-y				:= $(DRIVER_OBJS) pmc.o
diff -r e52af3942d55 arch/i386/xen/oprofile/op_counter.h
--- /dev/null	Sat Nov 12 04:30:38 2005
+++ b/arch/i386/xen/oprofile/op_counter.h	Wed Nov 23 14:48:18 2005
@@ -0,0 +1,29 @@
+/**
+ * @file op_counter.h
+ *
+ * @remark Copyright 2002 OProfile authors
+ * @remark Read the file COPYING
+ *
+ * @author John Levon
+ */
+ 
+#ifndef OP_COUNTER_H
+#define OP_COUNTER_H
+
+#define OP_MAX_COUNTER 8
+ 
+/* Per-perfctr configuration as set via
+ * oprofilefs.
+ */
+struct op_counter_config {
+        unsigned long count;
+        unsigned long enabled;
+        unsigned long event;
+        unsigned long kernel;
+        unsigned long user;
+        unsigned long unit_mask;
+};
+
+extern struct op_counter_config counter_config[];
+
+#endif /* OP_COUNTER_H */
diff -r e52af3942d55 arch/i386/xen/oprofile/pmc.c
--- /dev/null	Sat Nov 12 04:30:38 2005
+++ b/arch/i386/xen/oprofile/pmc.c	Wed Nov 23 14:48:18 2005
@@ -0,0 +1,431 @@
+/**
+ * @file nmi_int.c
+ *
+ * @remark Copyright 2002 OProfile authors
+ * @remark Read the file COPYING
+ *
+ * @author John Levon <levon@movementarian.org>
+ *
+ * Modified by Aravind Menon and Jose Renato Santos for Xen
+ * These modifications are:
+ * Copyright (C) 2005 Hewlett-Packard Co.
+ */
+
+#include <linux/init.h>
+#include <linux/notifier.h>
+#include <linux/smp.h>
+#include <linux/oprofile.h>
+#include <linux/sysdev.h>
+#include <linux/slab.h>
+#include <linux/interrupt.h>
+#include <linux/vmalloc.h>
+#include <asm/nmi.h>
+#include <asm/msr.h>
+#include <asm/apic.h>
+#include <asm/pgtable.h>
+#include <asm-xen/evtchn.h> 
+#include "op_counter.h"
+ 
+static int pmc_start(void);
+static void pmc_stop(void);
+
+extern void * vm_map_xen_pages (unsigned long maddr, int vm_size, pgprot_t prot);
+static int pmc_enabled = 0;
+static int num_events = 0;
+static int is_primary = 0;
+
+/* sample buffers shared with Xen */ 
+xenoprof_buf_t* xenoprof_buf[MAX_VIRT_CPUS];
+/* Shared buffer area */
+char * shared_buffer;
+/* Number of buffers in shared area (one per VCPU) */
+int nbuf;
+/* Mappings of VIRQ_PMC_OVF to irq number (per cpu) */
+int ovf_irq[NR_CPUS];
+
+#ifdef CONFIG_PM
+
+static int pmc_suspend(struct sys_device * dev, pm_message_t state)
+{
+	if (pmc_enabled == 1)
+		pmc_stop();
+	return 0;
+}
+
+
+static int pmc_resume(struct sys_device * dev)
+{
+	if (pmc_enabled == 1)
+		pmc_start();
+	return 0;
+}
+
+
+static struct sysdev_class oprofile_sysclass = {
+	set_kset_name("oprofile"),
+	.resume		= pmc_resume,
+	.suspend	= pmc_suspend
+};
+
+
+static struct sys_device device_oprofile = {
+	.id	= 0,
+	.cls	= &oprofile_sysclass,
+};
+
+
+static int __init init_driverfs(void)
+{
+	int error;
+	if (!(error = sysdev_class_register(&oprofile_sysclass)))
+		error = sysdev_register(&device_oprofile);
+	return error;
+}
+
+
+static void __exit exit_driverfs(void)
+{
+	sysdev_unregister(&device_oprofile);
+	sysdev_class_unregister(&oprofile_sysclass);
+}
+
+#else
+#define init_driverfs() do { } while (0)
+#define exit_driverfs() do { } while (0)
+#endif /* CONFIG_PM */
+
+unsigned long long oprofile_samples = 0;
+
+static irqreturn_t pmc_ovf_interrupt (int irq, void * dev_id, struct pt_regs * regs)
+{
+	int head, tail, size;
+	xenoprof_buf_t * buf;
+	int cpu;
+
+	cpu = smp_processor_id();
+	buf = xenoprof_buf[cpu];
+
+	head = buf->event_head;
+	tail = buf->event_tail;
+	size = buf->event_size;
+
+	if (tail > head) {
+		while (tail < size) {
+			oprofile_add_pc(buf->event_log[tail].eip, 
+					buf->event_log[tail].mode, 
+					buf->event_log[tail].event);
+			oprofile_samples++;
+			tail++;
+		}
+		tail = 0;
+	}
+	while (tail < head) {
+		oprofile_add_pc(buf->event_log[tail].eip, 
+				buf->event_log[tail].mode, 
+				buf->event_log[tail].event);
+		oprofile_samples++;
+		tail++;
+	}
+
+	buf->event_tail = tail;
+
+	return IRQ_HANDLED;
+}
+
+
+void unbind_pmc_ovf_cpu  (void * info)
+{
+	int cpu = smp_processor_id();
+	if (ovf_irq[cpu] >= 0) {
+		unbind_from_irqhandler(ovf_irq[cpu], NULL);
+		ovf_irq[cpu] = -1;
+	}
+}
+
+
+void unbind_pmc_ovf (void)
+{
+	on_each_cpu (unbind_pmc_ovf_cpu, NULL, 0, 1);
+}
+
+
+int bind_pmc_error;
+
+void bind_pmc_ovf_cpu  (void * info)
+{
+	int result;
+	int cpu = smp_processor_id();
+
+	result = bind_virq_to_irqhandler(VIRQ_PMC_OVF,
+					 cpu,
+					 pmc_ovf_interrupt, 
+					 SA_INTERRUPT, 
+					 "pmc_ovf", 
+					 NULL);
+
+	if (result<0) {
+		bind_pmc_error = result;
+		printk ("pmc.c: binding VIRQ_PMC_OVF to IRQ failed on CPU %d\n", cpu);
+	}
+	else
+		ovf_irq[cpu] = result;
+
+}
+
+
+int bind_pmc_ovf (void)
+{
+	bind_pmc_error = 0;
+	on_each_cpu (bind_pmc_ovf_cpu, NULL, 0, 1);
+	if (bind_pmc_error) {
+		unbind_pmc_ovf();
+		return bind_pmc_error;
+	}
+	else
+		return 0;
+}
+
+
+static int pmc_setup(void)
+{
+	int ret;
+
+	ret = bind_pmc_ovf();
+	if (ret)
+		return ret;
+
+	if (is_primary) {
+		ret = HYPERVISOR_pmc_op(PMC_RESERVE_COUNTERS, 
+					(unsigned int)NULL, 
+					(unsigned int)NULL);
+		if (ret)
+			goto err;
+
+		ret = HYPERVISOR_pmc_op(PMC_SETUP_EVENTS, 
+					(unsigned int)&counter_config, 
+					(unsigned int)num_events);
+		if (ret)
+			goto err;
+	}
+
+	ret = HYPERVISOR_pmc_op(PMC_ENABLE_VIRQ, 
+				(unsigned int)NULL, 
+				(unsigned int)NULL);
+	if (ret)
+		goto err;
+	
+	pmc_enabled = 1;
+	return 0;
+ err:
+	unbind_pmc_ovf();
+	return ret;
+
+}
+
+
+static void pmc_shutdown(void)
+{
+
+	pmc_enabled = 0;
+
+	HYPERVISOR_pmc_op(PMC_DISABLE_VIRQ, 
+			  (unsigned int)NULL, 
+			  (unsigned int)NULL);
+
+	if (is_primary) {
+		HYPERVISOR_pmc_op(PMC_RELEASE_COUNTERS, 
+				  (unsigned int)NULL, 
+				  (unsigned int)NULL);
+	}
+
+	unbind_pmc_ovf ();
+
+}
+
+
+static int pmc_start(void)
+{
+	int ret = 0;
+
+	if (is_primary)
+		ret = HYPERVISOR_pmc_op(PMC_START, 
+					(unsigned int)NULL, 
+					(unsigned int)NULL);
+	return ret;
+}
+ 
+
+static void pmc_stop(void)
+{
+	if (is_primary)
+		HYPERVISOR_pmc_op(PMC_STOP, 
+				  (unsigned int)NULL, 
+				  (unsigned int)NULL);
+}
+
+
+static int pmc_set_active(int * active_domains, 
+			  unsigned int adomains)
+{
+	int ret = 0;
+	if (is_primary) 
+		ret = HYPERVISOR_pmc_op(PMC_SET_ACTIVE, 
+					(unsigned int)active_domains, 
+					(unsigned int)adomains); 
+	return ret;
+}
+
+
+struct op_counter_config counter_config[OP_MAX_COUNTER];
+
+static int pmc_create_files(struct super_block * sb, struct dentry * root)
+{
+	unsigned int i;
+
+	for (i = 0; i < num_events; ++i) {
+		struct dentry * dir;
+		char buf[2];
+ 
+		snprintf(buf, 2, "%d", i);
+		dir = oprofilefs_mkdir(sb, root, buf);
+		oprofilefs_create_ulong(sb, dir, "enabled", &counter_config[i].enabled); 
+		oprofilefs_create_ulong(sb, dir, "event", &counter_config[i].event); 
+		oprofilefs_create_ulong(sb, dir, "count", &counter_config[i].count); 
+		oprofilefs_create_ulong(sb, dir, "unit_mask", &counter_config[i].unit_mask); 
+		oprofilefs_create_ulong(sb, dir, "kernel", &counter_config[i].kernel); 
+		oprofilefs_create_ulong(sb, dir, "user", &counter_config[i].user); 
+	}
+
+	return 0;
+}
+ 
+ 
+struct oprofile_operations pmc_ops = {
+	.create_files 	= pmc_create_files,
+	.set_active	= pmc_set_active,
+	.setup 		= pmc_setup,
+	.shutdown	= pmc_shutdown,
+	.start		= pmc_start,
+	.stop		= pmc_stop
+};
+ 
+static void __init p4_init(void)
+{
+	__u8 cpu_model = current_cpu_data.x86_model;
+
+	if (cpu_model > 3)
+		pmc_ops.cpu_type = "type_unknown";
+
+	/* We always use a non-HT system because that gives us more events */
+	pmc_ops.cpu_type = "i386/p4";
+}
+
+
+static void __init ppro_init(void)
+{
+	__u8 cpu_model = current_cpu_data.x86_model;
+
+	if (cpu_model > 0xd)
+		pmc_ops.cpu_type = "type_unknown";
+
+	if (cpu_model == 9) {
+		pmc_ops.cpu_type = "i386/p6_mobile";
+	} else if (cpu_model > 5) {
+		pmc_ops.cpu_type = "i386/piii";
+	} else if (cpu_model > 2) {
+		pmc_ops.cpu_type = "i386/pii";
+	} else {
+		pmc_ops.cpu_type = "i386/ppro";
+	}
+}
+
+
+/* in order to get driverfs right */
+static int using_pmc;
+
+int __init oprofile_arch_init(struct oprofile_operations * ops)
+{
+
+	pmc_init_result_t result;
+	xenoprof_buf_t * buf;
+	int max_samples = 16;
+	int vm_size;
+	int npages;
+	int i;
+
+	int ret = HYPERVISOR_pmc_op(PMC_INIT, 
+				    (unsigned int)max_samples, 
+				    (unsigned int)&result);
+	
+	if (!ret) {
+		__u8 vendor = current_cpu_data.x86_vendor;
+		__u8 family = current_cpu_data.x86;
+		pgprot_t prot = __pgprot(_KERNPG_TABLE);
+		
+		num_events = result.num_events;
+		is_primary = result.is_primary;
+		nbuf = result.nbuf;
+		
+		npages = (result.bufsize * nbuf - 1) / PAGE_SIZE + 1;
+		vm_size = npages * PAGE_SIZE;
+		
+
+		shared_buffer = (char *) vm_map_xen_pages(result.buf_maddr, 
+							  vm_size, prot);
+    
+		if (!shared_buffer)
+			return -ENOMEM;
+		
+		for (i=0; i< nbuf; i++) {
+			buf = (xenoprof_buf_t*) &shared_buffer[i * result.bufsize];
+			BUG_ON (buf->vcpu_id >= MAX_VIRT_CPUS);
+			xenoprof_buf[buf->vcpu_id] = buf;
+		}
+		
+		if (vendor == X86_VENDOR_INTEL) {
+			switch (family) {
+				/* Pentium IV */
+			case 0xf:
+				p4_init();
+				break;
+				/* A P6-class processor */
+			case 6:
+				ppro_init();
+				break;
+			default:
+				pmc_ops.cpu_type = "type_unknown";
+			}
+		} else pmc_ops.cpu_type = "type_unknown";
+		
+		init_driverfs();
+		using_pmc = 1;
+		*ops = pmc_ops;
+		
+		for (i=0; i<NR_CPUS; i++)
+			ovf_irq[i] = -1;
+
+    
+		printk (KERN_INFO "oprofile_arch_init: ret %d, events %d, is_primary %d\n",
+			ret, num_events, is_primary);
+		return ret;
+	}
+
+	return ret;
+}
+
+
+void __exit oprofile_arch_exit(void)
+{
+	if (using_pmc)
+		exit_driverfs();
+
+	if (shared_buffer) {
+		vunmap (shared_buffer);
+		shared_buffer = NULL;
+	}
+	if (is_primary)
+		HYPERVISOR_pmc_op(PMC_SHUTDOWN, 
+				  (unsigned int)NULL, 
+				  (unsigned int)NULL);
+
+}

[-- Attachment #4: Type: text/plain, Size: 138 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

Re: RE: [PATCH] xenoprof (xen)

[John Levon]

On Wed, Nov 23, 2005 at 05:44:16PM -0800, Santos, Jose Renato G wrote:

>   John Levon has reviewed these patches and I have included all his
> suggestions on this version.

To clarify, I've only reviewed the first one so far
(xenoprof-generic-linux-2.6.14.patch). I still have outstanding issues
with the arch one.

regards,
john

RE: RE: [PATCH] xenoprof (xen)

[Santos, Jose Renato G]

> -----Original Message-----
> From: John Levon [mailto:levon@movementarian.org
> Sent: Wednesday, November 23, 2005 6:52 PM
> To: Santos, Jose Renato G
> Cc: Ian Pratt; Keir Fraser; Jose Renato Santos; 
> xen-devel@lists.xensource.com; G John Janakiraman; Turner, Yoshio
> Subject: Re: [Xen-devel] RE: [PATCH] xenoprof (xen)
> 
> On Wed, Nov 23, 2005 at 05:44:16PM -0800, Santos, Jose Renato G wrote:
> 
> >   John Levon has reviewed these patches and I have included all his 
> > suggestions on this version.
> 
> To clarify, I've only reviewed the first one so far 
> (xenoprof-generic-linux-2.6.14.patch). I still have 
> outstanding issues with the arch one.

Not quite correct. You reviewed both patches and I thought I have
addressed all your concerns but apparently there are still some pending
issues on the arch patch. Will work on these after the Thanksgiving
holiday.
Regards
Renato

> 
> regards,
> john
> 

RE: RE: [PATCH] xenoprof (xen)

[Ian Pratt]

>   Oh, I see. I did not know about this reorganization of the 
> xen/linux tree.
>   Do you prefer that I port the xenoprof patch into the 
> "linux-2.6-xen.hg" tree right now or wait for when the tree 
> goes into the xen repository?

Yes, please.

Ian

RE: RE: [PATCH] xenoprof (xen)

[Santos, Jose Renato G]

  Ian,
  Oh, I see. I did not know about this reorganization of the xen/linux
tree.
  Do you prefer that I port the xenoprof patch into the
"linux-2.6-xen.hg" tree right now or wait for when the tree goes into
the xen repository?
  Thanks

  Renato

> -----Original Message-----
> From: Ian Pratt [mailto:m+Ian.Pratt@cl.cam.ac.uk] 
> Sent: Sunday, November 13, 2005 4:18 PM
> To: Santos, Jose Renato G; Keir Fraser; John Levon
> Cc: Turner, Yoshio; xen-devel@lists.xensource.com; Jose 
> Renato Santos; G John Janakiraman; ian.pratt@cl.cam.ac.uk
> Subject: RE: [Xen-devel] RE: [PATCH] xenoprof (xen)
> 
> >   From your message, I am assuming that you are agreeing to include 
> > the
> >   "arch/xen/oprofile" modifications into the sparse tree when
> >   it moves to the 2.6.14 (please, correct it if I am wrong).
> >   Any expectation of when this should happen? 
> 
> Well, arch/xen won't exist as xen is a subarch of i386/x86_64 in our
> 2.6.14 tree.
> 
> See http://xenbits.xensource.com/linux-2.6-xen.hg
> 
> 
> This will replace the 2.6.12 in the xen repository at some 
> point, after more testing.
> 
> Ian
> 

RE: RE: [PATCH] xenoprof (xen)

[Ian Pratt]

>   From your message, I am assuming that you are agreeing to 
> include the
>   "arch/xen/oprofile" modifications into the sparse tree when
>   it moves to the 2.6.14 (please, correct it if I am wrong).
>   Any expectation of when this should happen? 

Well, arch/xen won't exist as xen is a subarch of i386/x86_64 in our
2.6.14 tree.

See http://xenbits.xensource.com/linux-2.6-xen.hg


This will replace the 2.6.12 in the xen repository at some point, after
more testing.

Ian

RE: RE: [PATCH] xenoprof (xen)

[Santos, Jose Renato G]

 

> -----Original Message-----
> From: Ian Pratt [mailto:m+Ian.Pratt@cl.cam.ac.uk] 
> Sent: Sunday, November 13, 2005 11:24 AM
> To: Santos, Jose Renato G; Keir Fraser
> Cc: Turner, Yoshio; xen-devel@lists.xensource.com; Jose 
> Renato Santos; G John Janakiraman; ian.pratt@cl.cam.ac.uk
> Subject: RE: [Xen-devel] RE: [PATCH] xenoprof (xen)
> 
> > Is there anything I can do to make your job easier?
> > Would you like to receive multiple smaller patches (one patch per 
> > file)?
> 
> I think we should hold-off applying the linux patches to the 
> sparse tree for the moment for a couple of reasons: John 
> Levon would like to see some cleanups, and I think we'll be 
> better off applying once the sparse tree has moved to 2.6.14 
> anyhow (generated from the merge tree).
> 

  Ok, Ian
  Just a clarification. The cleanups suggested by John Levon,
  are for the user level tools and for the modifications on the
  generic code of Oprofile in "drivers/oprofile". Both of
  these patches are not expected to go into the sparse tree,
  anyhow. I was expecting that only the changes to "arch/xen/oprofile"
  would go to the sparse tree. There is no cleanups suggested
  by John Levon for this patch. 

  From your message, I am assuming that you are agreeing to include the
  "arch/xen/oprofile" modifications into the sparse tree when
  it moves to the 2.6.14 (please, correct it if I am wrong).
  Any expectation of when this should happen? 
 
  In the mean time, I will be working on the cleanups suggested
  by John for the user level tools and for generic Oprofile code
  modifications.
 
> Jose: please could you have a go generating the patch against 
> the linux-2.6-xen.hg tree.
> 
  Forgive me, but I am not sure I understand what you mean in this
  sentence. Could you please clarify ...

  Thanks

  Renato

> Thanks,
> Ian
> 

RE: RE: [PATCH] xenoprof (xen)

[Ian Pratt]

> Is there anything I can do to make your job easier?
> Would you like to receive multiple smaller patches (one patch 
> per file)?

I think we should hold-off applying the linux patches to the sparse tree
for the moment for a couple of reasons: John Levon would like to see
some cleanups, and I think we'll be better off applying once the sparse
tree has moved to 2.6.14 anyhow (generated from the merge tree).

Jose: please could you have a go generating the patch against the
linux-2.6-xen.hg tree.

Thanks,
Ian