* [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers
2006-06-14 3:47 [PATCH 0/3] powerpc: Instrument Hypervisor Calls Mike Kravetz
@ 2006-06-14 3:50 ` Mike Kravetz
0 siblings, 0 replies; 22+ messages in thread
From: Mike Kravetz @ 2006-06-14 3:50 UTC (permalink / raw)
To: linuxppc-dev; +Cc: Bryan Rosenburg, Christopher Yeoh
Move all the Hypervisor call definitions to to a single header file.
--
Signed-off-by: Mike Kravetz <kravetz@us.ibm.com>
diff -Naupr linux-2.6.17-rc6/drivers/net/ibmveth.h linux-2.6.17-rc6.work/drivers/net/ibmveth.h
--- linux-2.6.17-rc6/drivers/net/ibmveth.h 2006-06-06 00:57:02.000000000 +0000
+++ linux-2.6.17-rc6.work/drivers/net/ibmveth.h 2006-06-13 21:35:10.000000000 +0000
@@ -41,16 +41,6 @@
#define IbmVethMcastRemoveFilter 0x2UL
#define IbmVethMcastClearFilterTable 0x3UL
-/* hcall numbers */
-#define H_VIO_SIGNAL 0x104
-#define H_REGISTER_LOGICAL_LAN 0x114
-#define H_FREE_LOGICAL_LAN 0x118
-#define H_ADD_LOGICAL_LAN_BUFFER 0x11C
-#define H_SEND_LOGICAL_LAN 0x120
-#define H_MULTICAST_CTRL 0x130
-#define H_CHANGE_LOGICAL_LAN_MAC 0x14C
-#define H_FREE_LOGICAL_LAN_BUFFER 0x1D4
-
/* hcall macros */
#define h_register_logical_lan(ua, buflst, rxq, fltlst, mac) \
plpar_hcall_norets(H_REGISTER_LOGICAL_LAN, ua, buflst, rxq, fltlst, mac)
diff -Naupr linux-2.6.17-rc6/include/asm-powerpc/hvcall.h linux-2.6.17-rc6.work/include/asm-powerpc/hvcall.h
--- linux-2.6.17-rc6/include/asm-powerpc/hvcall.h 2006-06-06 00:57:02.000000000 +0000
+++ linux-2.6.17-rc6.work/include/asm-powerpc/hvcall.h 2006-06-13 21:35:10.000000000 +0000
@@ -155,9 +155,15 @@
#define H_VIO_SIGNAL 0x104
#define H_SEND_CRQ 0x108
#define H_COPY_RDMA 0x110
+#define H_REGISTER_LOGICAL_LAN 0x114
+#define H_FREE_LOGICAL_LAN 0x118
+#define H_ADD_LOGICAL_LAN_BUFFER 0x11C
+#define H_SEND_LOGICAL_LAN 0x120
+#define H_MULTICAST_CTRL 0x130
#define H_SET_XDABR 0x134
#define H_STUFF_TCE 0x138
#define H_PUT_TCE_INDIRECT 0x13C
+#define H_CHANGE_LOGICAL_LAN_MAC 0x14C
#define H_VTERM_PARTNER_INFO 0x150
#define H_REGISTER_VTERM 0x154
#define H_FREE_VTERM 0x158
@@ -187,11 +193,14 @@
#define H_GET_HCA_INFO 0x1B8
#define H_GET_PERF_COUNT 0x1BC
#define H_MANAGE_TRACE 0x1C0
+#define H_FREE_LOGICAL_LAN_BUFFER 0x1D4
#define H_QUERY_INT_STATE 0x1E4
#define H_POLL_PENDING 0x1D8
#define H_JOIN 0x298
#define H_ENABLE_CRQ 0x2B0
+#define MAX_HCALL_OPCODES (H_ENABLE_CRQ >> 2)
+
#ifndef __ASSEMBLY__
/* plpar_hcall() -- Generic call interface using above opcodes
^ permalink raw reply [flat|nested] 22+ messages in thread
* [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers
2006-06-22 22:56 [PATCH 0/3] powerpc: Instrument Hypervisor Calls Mike Kravetz
@ 2006-06-22 22:57 ` Mike Kravetz
0 siblings, 0 replies; 22+ messages in thread
From: Mike Kravetz @ 2006-06-22 22:57 UTC (permalink / raw)
To: linuxppc-dev
Cc: Bryan Rosenburg, Christopher Yeoh, Nathan Lynch, Arnd Bergmann
Move all the Hypervisor call definitions to to a single header file.
--
Signed-off-by: Mike Kravetz <kravetz@us.ibm.com>
diff -Naupr linux-2.6.17.1/drivers/net/ibmveth.h linux-2.6.17.1.work/drivers/net/ibmveth.h
--- linux-2.6.17.1/drivers/net/ibmveth.h 2006-06-20 09:31:55.000000000 +0000
+++ linux-2.6.17.1.work/drivers/net/ibmveth.h 2006-06-22 22:54:28.000000000 +0000
@@ -41,16 +41,6 @@
#define IbmVethMcastRemoveFilter 0x2UL
#define IbmVethMcastClearFilterTable 0x3UL
-/* hcall numbers */
-#define H_VIO_SIGNAL 0x104
-#define H_REGISTER_LOGICAL_LAN 0x114
-#define H_FREE_LOGICAL_LAN 0x118
-#define H_ADD_LOGICAL_LAN_BUFFER 0x11C
-#define H_SEND_LOGICAL_LAN 0x120
-#define H_MULTICAST_CTRL 0x130
-#define H_CHANGE_LOGICAL_LAN_MAC 0x14C
-#define H_FREE_LOGICAL_LAN_BUFFER 0x1D4
-
/* hcall macros */
#define h_register_logical_lan(ua, buflst, rxq, fltlst, mac) \
plpar_hcall_norets(H_REGISTER_LOGICAL_LAN, ua, buflst, rxq, fltlst, mac)
diff -Naupr linux-2.6.17.1/include/asm-powerpc/hvcall.h linux-2.6.17.1.work/include/asm-powerpc/hvcall.h
--- linux-2.6.17.1/include/asm-powerpc/hvcall.h 2006-06-20 09:31:55.000000000 +0000
+++ linux-2.6.17.1.work/include/asm-powerpc/hvcall.h 2006-06-22 22:54:28.000000000 +0000
@@ -155,9 +155,15 @@
#define H_VIO_SIGNAL 0x104
#define H_SEND_CRQ 0x108
#define H_COPY_RDMA 0x110
+#define H_REGISTER_LOGICAL_LAN 0x114
+#define H_FREE_LOGICAL_LAN 0x118
+#define H_ADD_LOGICAL_LAN_BUFFER 0x11C
+#define H_SEND_LOGICAL_LAN 0x120
+#define H_MULTICAST_CTRL 0x130
#define H_SET_XDABR 0x134
#define H_STUFF_TCE 0x138
#define H_PUT_TCE_INDIRECT 0x13C
+#define H_CHANGE_LOGICAL_LAN_MAC 0x14C
#define H_VTERM_PARTNER_INFO 0x150
#define H_REGISTER_VTERM 0x154
#define H_FREE_VTERM 0x158
@@ -187,11 +193,14 @@
#define H_GET_HCA_INFO 0x1B8
#define H_GET_PERF_COUNT 0x1BC
#define H_MANAGE_TRACE 0x1C0
+#define H_FREE_LOGICAL_LAN_BUFFER 0x1D4
#define H_QUERY_INT_STATE 0x1E4
#define H_POLL_PENDING 0x1D8
#define H_JOIN 0x298
#define H_ENABLE_CRQ 0x2B0
+#define MAX_HCALL_OPCODES (H_ENABLE_CRQ >> 2)
+
#ifndef __ASSEMBLY__
/* plpar_hcall() -- Generic call interface using above opcodes
^ permalink raw reply [flat|nested] 22+ messages in thread
* [PATCH 0/3] powerpc: Instrument Hypervisor Calls
@ 2006-07-14 23:37 Mike Kravetz
2006-07-14 23:39 ` [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers Mike Kravetz
` (3 more replies)
0 siblings, 4 replies; 22+ messages in thread
From: Mike Kravetz @ 2006-07-14 23:37 UTC (permalink / raw)
To: Paul Mackerras
Cc: Arnd Bergmann, Bryan Rosenburg, linuxppc-dev, Nathan Lynch,
Christopher Yeoh
Hi Paul,
Here is an updated version of the patch(es) to instrument hcalls. All
issues from the previous versions have been addressed. Although, I
haven't really discovered an elegant solution to the assembly routine
name juggling.
In addition, there has been some discussion of these patches on IRC.
Some remaining issues/questions are:
- Exactly how much overhead does the statistic gathering introduce?
- What would be the cost of disabling preemption for more accurate statistics?
- What would be the cost of disabling interrupts for more accurate statistics?
- Should we extend this statistic gathering to RTAS calls?
I would like to get this basic code integrated and then start to address
some of these questions. The IBM performance group will help in
benchmarking to determine overhead.
Please note that all code is behind a config option (off by default) with
zero impact unless enabled.
--
Mike
^ permalink raw reply [flat|nested] 22+ messages in thread
* [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers
2006-07-14 23:37 [PATCH 0/3] powerpc: Instrument Hypervisor Calls Mike Kravetz
@ 2006-07-14 23:39 ` Mike Kravetz
2006-07-14 23:40 ` [PATCH 2/3] powerpc: Instrument Hypervisor Calls: add wrappers Mike Kravetz
` (2 subsequent siblings)
3 siblings, 0 replies; 22+ messages in thread
From: Mike Kravetz @ 2006-07-14 23:39 UTC (permalink / raw)
To: Paul Mackerras
Cc: Arnd Bergmann, Bryan Rosenburg, linuxppc-dev, Nathan Lynch,
Christopher Yeoh
Move all the Hypervisor call definitions to to a single header file.
--
Signed-off-by: Mike Kravetz <kravetz@us.ibm.com>
diff -Naupr linux-2.6.17.4/drivers/net/ibmveth.h linux-2.6.17.4.work/drivers/net/ibmveth.h
--- linux-2.6.17.4/drivers/net/ibmveth.h 2006-07-06 20:02:28.000000000 +0000
+++ linux-2.6.17.4.work/drivers/net/ibmveth.h 2006-07-14 23:05:37.000000000 +0000
@@ -41,16 +41,6 @@
#define IbmVethMcastRemoveFilter 0x2UL
#define IbmVethMcastClearFilterTable 0x3UL
-/* hcall numbers */
-#define H_VIO_SIGNAL 0x104
-#define H_REGISTER_LOGICAL_LAN 0x114
-#define H_FREE_LOGICAL_LAN 0x118
-#define H_ADD_LOGICAL_LAN_BUFFER 0x11C
-#define H_SEND_LOGICAL_LAN 0x120
-#define H_MULTICAST_CTRL 0x130
-#define H_CHANGE_LOGICAL_LAN_MAC 0x14C
-#define H_FREE_LOGICAL_LAN_BUFFER 0x1D4
-
/* hcall macros */
#define h_register_logical_lan(ua, buflst, rxq, fltlst, mac) \
plpar_hcall_norets(H_REGISTER_LOGICAL_LAN, ua, buflst, rxq, fltlst, mac)
diff -Naupr linux-2.6.17.4/include/asm-powerpc/hvcall.h linux-2.6.17.4.work/include/asm-powerpc/hvcall.h
--- linux-2.6.17.4/include/asm-powerpc/hvcall.h 2006-07-06 20:02:28.000000000 +0000
+++ linux-2.6.17.4.work/include/asm-powerpc/hvcall.h 2006-07-14 23:05:37.000000000 +0000
@@ -155,9 +155,15 @@
#define H_VIO_SIGNAL 0x104
#define H_SEND_CRQ 0x108
#define H_COPY_RDMA 0x110
+#define H_REGISTER_LOGICAL_LAN 0x114
+#define H_FREE_LOGICAL_LAN 0x118
+#define H_ADD_LOGICAL_LAN_BUFFER 0x11C
+#define H_SEND_LOGICAL_LAN 0x120
+#define H_MULTICAST_CTRL 0x130
#define H_SET_XDABR 0x134
#define H_STUFF_TCE 0x138
#define H_PUT_TCE_INDIRECT 0x13C
+#define H_CHANGE_LOGICAL_LAN_MAC 0x14C
#define H_VTERM_PARTNER_INFO 0x150
#define H_REGISTER_VTERM 0x154
#define H_FREE_VTERM 0x158
@@ -187,11 +193,14 @@
#define H_GET_HCA_INFO 0x1B8
#define H_GET_PERF_COUNT 0x1BC
#define H_MANAGE_TRACE 0x1C0
+#define H_FREE_LOGICAL_LAN_BUFFER 0x1D4
#define H_QUERY_INT_STATE 0x1E4
#define H_POLL_PENDING 0x1D8
#define H_JOIN 0x298
#define H_ENABLE_CRQ 0x2B0
+#define MAX_HCALL_OPCODES (H_ENABLE_CRQ >> 2)
+
#ifndef __ASSEMBLY__
/* plpar_hcall() -- Generic call interface using above opcodes
^ permalink raw reply [flat|nested] 22+ messages in thread
* [PATCH 2/3] powerpc: Instrument Hypervisor Calls: add wrappers
2006-07-14 23:37 [PATCH 0/3] powerpc: Instrument Hypervisor Calls Mike Kravetz
2006-07-14 23:39 ` [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers Mike Kravetz
@ 2006-07-14 23:40 ` Mike Kravetz
2006-07-15 0:15 ` Nathan Lynch
2006-07-14 23:41 ` [PATCH 3/3] powerpc: Instrument Hypervisor Calls: add debugfs files Mike Kravetz
2006-07-15 0:00 ` [PATCH 0/3] powerpc: Instrument Hypervisor Calls Arnd Bergmann
3 siblings, 1 reply; 22+ messages in thread
From: Mike Kravetz @ 2006-07-14 23:40 UTC (permalink / raw)
To: Paul Mackerras
Cc: Arnd Bergmann, Bryan Rosenburg, linuxppc-dev, Nathan Lynch,
Christopher Yeoh
Add wrappers which perform the actual hypervisor call instrumentation.
--
Signed-off-by: Mike Kravetz <kravetz@us.ibm.com>
diff -Naupr linux-2.6.17.4/arch/powerpc/Kconfig.debug linux-2.6.17.4.work/arch/powerpc/Kconfig.debug
--- linux-2.6.17.4/arch/powerpc/Kconfig.debug 2006-07-06 20:02:28.000000000 +0000
+++ linux-2.6.17.4.work/arch/powerpc/Kconfig.debug 2006-07-14 23:28:20.000000000 +0000
@@ -18,6 +18,19 @@ config DEBUG_STACK_USAGE
This option will slow down process creation somewhat.
+config HCALL_STATS
+ bool "Hypervisor call instrumentation"
+ depends on PPC_PSERIES && DEBUG_FS
+ help
+ Adds code to keep track of the number of hypervisor calls made and
+ the amount of time spent in hypervisor calls. A directory named
+ hcall_inst is added at the root of the debugfs filesystem. Within
+ the hcall_inst directory are files that contain CPU specific call
+ statistics.
+
+ This option will add a small amount of overhead to all hypervisor
+ calls.
+
config DEBUGGER
bool "Enable debugger hooks"
depends on DEBUG_KERNEL
diff -Naupr linux-2.6.17.4/arch/powerpc/platforms/pseries/Makefile linux-2.6.17.4.work/arch/powerpc/platforms/pseries/Makefile
--- linux-2.6.17.4/arch/powerpc/platforms/pseries/Makefile 2006-07-06 20:02:28.000000000 +0000
+++ linux-2.6.17.4.work/arch/powerpc/platforms/pseries/Makefile 2006-07-14 23:28:20.000000000 +0000
@@ -9,3 +9,4 @@ obj-$(CONFIG_EEH) += eeh.o eeh_cache.o e
obj-$(CONFIG_HVC_CONSOLE) += hvconsole.o
obj-$(CONFIG_HVCS) += hvcserver.o
+obj-$(CONFIG_HCALL_STATS) += hvCall_inst.o
diff -Naupr linux-2.6.17.4/arch/powerpc/platforms/pseries/hvCall.S linux-2.6.17.4.work/arch/powerpc/platforms/pseries/hvCall.S
--- linux-2.6.17.4/arch/powerpc/platforms/pseries/hvCall.S 2006-07-06 20:02:28.000000000 +0000
+++ linux-2.6.17.4.work/arch/powerpc/platforms/pseries/hvCall.S 2006-07-14 23:28:20.000000000 +0000
@@ -11,7 +11,35 @@
#include <asm/hvcall.h>
#include <asm/processor.h>
#include <asm/ppc_asm.h>
-
+
+/*
+ * If hcall statistics are desired, all routines are wrapped with code
+ * that does the statistic gathering.
+ */
+#ifndef CONFIG_HCALL_STATS
+#define PLPAR_HCALL plpar_hcall
+#define PLPAR_HCALL_NORETS plpar_hcall_norets
+#define PLPAR_HCALL_8ARG_2RET plpar_hcall_8arg_2ret
+#define PLPAR_HCALL_4OUT plpar_hcall_4out
+#define PLPAR_HCALL_7ARG_7RET plpar_hcall_7arg_7ret
+#define PLPAR_HCALL_9ARG_9RET plpar_hcall_9arg_9ret
+#else
+#define PLPAR_HCALL plpar_hcall_base
+#define PLPAR_HCALL_NORETS plpar_hcall_norets_base
+#define PLPAR_HCALL_8ARG_2RET plpar_hcall_8arg_2ret_base
+#define PLPAR_HCALL_4OUT plpar_hcall_4out_base
+#define PLPAR_HCALL_7ARG_7RET plpar_hcall_7arg_7ret_base
+#define PLPAR_HCALL_9ARG_9RET plpar_hcall_9arg_9ret_base
+
+/*
+ * A special 'indirect' call to a C based wrapper if statistics are desired.
+ * See plpar_hcall_norets_C function header for more details.
+ */
+_GLOBAL(plpar_hcall_norets)
+ b plpar_hcall_norets_C
+
+#endif
+
#define STK_PARM(i) (48 + ((i)-3)*8)
.text
@@ -25,7 +53,7 @@
unsigned long *out2, R9
unsigned long *out3); R10
*/
-_GLOBAL(plpar_hcall)
+_GLOBAL(PLPAR_HCALL)
HMT_MEDIUM
mfcr r0
@@ -52,7 +80,7 @@ _GLOBAL(plpar_hcall)
/* Simple interface with no output values (other than status) */
-_GLOBAL(plpar_hcall_norets)
+_GLOBAL(PLPAR_HCALL_NORETS)
HMT_MEDIUM
mfcr r0
@@ -76,7 +104,7 @@ _GLOBAL(plpar_hcall_norets)
unsigned long arg8, 112(R1)
unsigned long *out1); 120(R1)
*/
-_GLOBAL(plpar_hcall_8arg_2ret)
+_GLOBAL(PLPAR_HCALL_8ARG_2RET)
HMT_MEDIUM
mfcr r0
@@ -102,7 +130,7 @@ _GLOBAL(plpar_hcall_8arg_2ret)
unsigned long *out3, R10
unsigned long *out4); 112(R1)
*/
-_GLOBAL(plpar_hcall_4out)
+_GLOBAL(PLPAR_HCALL_4OUT)
HMT_MEDIUM
mfcr r0
@@ -144,7 +172,7 @@ _GLOBAL(plpar_hcall_4out)
unsigned long *out6, 102(R1)
unsigned long *out7); 100(R1)
*/
-_GLOBAL(plpar_hcall_7arg_7ret)
+_GLOBAL(PLPAR_HCALL_7ARG_7RET)
HMT_MEDIUM
mfcr r0
@@ -193,7 +221,7 @@ _GLOBAL(plpar_hcall_7arg_7ret)
unsigned long *out8, 94(R1)
unsigned long *out9, 92(R1)
*/
-_GLOBAL(plpar_hcall_9arg_9ret)
+_GLOBAL(PLPAR_HCALL_9ARG_9RET)
HMT_MEDIUM
mfcr r0
diff -Naupr linux-2.6.17.4/arch/powerpc/platforms/pseries/hvCall_inst.c linux-2.6.17.4.work/arch/powerpc/platforms/pseries/hvCall_inst.c
--- linux-2.6.17.4/arch/powerpc/platforms/pseries/hvCall_inst.c 1970-01-01 00:00:00.000000000 +0000
+++ linux-2.6.17.4.work/arch/powerpc/platforms/pseries/hvCall_inst.c 2006-07-14 23:28:44.000000000 +0000
@@ -0,0 +1,207 @@
+/*
+ * Copyright (C) 2006 Mike Kravetz IBM Corporation
+ *
+ * Hypervisor Call Instrumentation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/kernel.h>
+#include <linux/percpu.h>
+#include <linux/debugfs.h>
+#include <linux/seq_file.h>
+#include <linux/cpumask.h>
+#include <asm/hvcall.h>
+#include <asm/firmware.h>
+
+DEFINE_PER_CPU(struct hcall_stats[MAX_HCALL_OPCODES+1], hcall_stats);
+
+/*
+ * Common update of the per-CPU/per-hcall statistics
+ */
+static inline void update_stats(unsigned long opcode, unsigned long t_before)
+{
+ unsigned long op_index = opcode >> 2;
+ struct hcall_stats *hs = &__get_cpu_var(hcall_stats[op_index]);
+
+ hs->total_time += (mftb() - t_before);
+ hs->num_calls++;
+}
+
+/*
+ * plpar_hcall wrapper
+ */
+long plpar_hcall(unsigned long opcode,
+ unsigned long arg1,
+ unsigned long arg2,
+ unsigned long arg3,
+ unsigned long arg4,
+ unsigned long *out1,
+ unsigned long *out2,
+ unsigned long *out3)
+{
+ long rc;
+ unsigned long t_before;
+
+ t_before = mftb();
+ rc = plpar_hcall_base(opcode, arg1, arg2, arg3, arg4, out1, out2, out3);
+
+ update_stats(opcode, t_before);
+ return rc;
+}
+
+/*
+ * A C based wrapper for plpar_hcall_norets
+ * The wrapper for plpar_hcall_norets is a special case because the function
+ * takes a variable number of arguments. It is almost impossible to write a
+ * wrapper for a function that takes a variable number of arguments in C.
+ * Therefore, there is an assembly routine in hvCall.S that simply branches
+ * to this C wrapper. This 'indirection' takes care of the variable arguments
+ * issue. This C wrapper has a fixed maximum number of arguments.
+ */
+long plpar_hcall_norets_C(unsigned long opcode,
+ unsigned long arg1,
+ unsigned long arg2,
+ unsigned long arg3,
+ unsigned long arg4,
+ unsigned long arg5,
+ unsigned long arg6)
+{
+ long rc;
+ unsigned long t_before;
+
+ t_before = mftb();
+ rc = plpar_hcall_norets_base(opcode, arg1, arg2, arg3, arg4, arg5,
+ arg6);
+
+ update_stats(opcode, t_before);
+ return rc;
+}
+
+/*
+ * plpar_hcall_8arg_2ret wrapper
+ */
+long plpar_hcall_8arg_2ret(unsigned long opcode,
+ unsigned long arg1,
+ unsigned long arg2,
+ unsigned long arg3,
+ unsigned long arg4,
+ unsigned long arg5,
+ unsigned long arg6,
+ unsigned long arg7,
+ unsigned long arg8,
+ unsigned long *out1)
+{
+ long rc;
+ unsigned long t_before;
+
+ t_before = mftb();
+ rc = plpar_hcall_8arg_2ret_base(opcode, arg1, arg2, arg3, arg4, arg5,
+ arg6, arg7, arg8, out1);
+
+ update_stats(opcode, t_before);
+ return rc;
+}
+
+/*
+ * plpar_hcall_4out wrapper
+ */
+long plpar_hcall_4out(unsigned long opcode,
+ unsigned long arg1,
+ unsigned long arg2,
+ unsigned long arg3,
+ unsigned long arg4,
+ unsigned long *out1,
+ unsigned long *out2,
+ unsigned long *out3,
+ unsigned long *out4)
+{
+ long rc;
+ unsigned long t_before;
+
+ t_before = mftb();
+ rc = plpar_hcall_4out_base(opcode, arg1, arg2, arg3, arg4, out1,
+ out2, out3, out4);
+
+ update_stats(opcode, t_before);
+ return rc;
+}
+
+/*
+ * plpar_hcall_7arg_7ret wrapper
+ */
+long plpar_hcall_7arg_7ret(unsigned long opcode,
+ unsigned long arg1,
+ unsigned long arg2,
+ unsigned long arg3,
+ unsigned long arg4,
+ unsigned long arg5,
+ unsigned long arg6,
+ unsigned long arg7,
+ unsigned long *out1,
+ unsigned long *out2,
+ unsigned long *out3,
+ unsigned long *out4,
+ unsigned long *out5,
+ unsigned long *out6,
+ unsigned long *out7)
+{
+ long rc;
+ unsigned long t_before;
+
+ t_before = mftb();
+ rc = plpar_hcall_7arg_7ret_base(opcode, arg1, arg2, arg3, arg4, arg5,
+ arg6, arg7, out1, out2, out3, out4,
+ out5, out6, out7);
+
+ update_stats(opcode, t_before);
+ return rc;
+}
+
+/*
+ * plpar_hcall_9arg_9ret wrapper
+ */
+long plpar_hcall_9arg_9ret(unsigned long opcode,
+ unsigned long arg1,
+ unsigned long arg2,
+ unsigned long arg3,
+ unsigned long arg4,
+ unsigned long arg5,
+ unsigned long arg6,
+ unsigned long arg7,
+ unsigned long arg8,
+ unsigned long arg9,
+ unsigned long *out1,
+ unsigned long *out2,
+ unsigned long *out3,
+ unsigned long *out4,
+ unsigned long *out5,
+ unsigned long *out6,
+ unsigned long *out7,
+ unsigned long *out8,
+ unsigned long *out9)
+{
+ long rc;
+ unsigned long t_before;
+
+ t_before = mftb();
+ rc = plpar_hcall_9arg_9ret_base(opcode, arg1, arg2, arg3, arg4, arg5,
+ arg6, arg7, arg8, arg9, out1, out2,
+ out3, out4, out5, out6, out7, out8,
+ out9);
+
+ update_stats(opcode, t_before);
+ return rc;
+}
diff -Naupr linux-2.6.17.4/include/asm-powerpc/hvcall.h linux-2.6.17.4.work/include/asm-powerpc/hvcall.h
--- linux-2.6.17.4/include/asm-powerpc/hvcall.h 2006-07-14 23:27:52.000000000 +0000
+++ linux-2.6.17.4.work/include/asm-powerpc/hvcall.h 2006-07-14 23:28:20.000000000 +0000
@@ -292,6 +292,86 @@ long plpar_hcall_9arg_9ret(unsigned long
unsigned long *out8,
unsigned long *out9);
+
+/* For hcall instrumentation. One structure per-hcall, per-CPU */
+struct hcall_stats {
+ unsigned long num_calls; /* number of calls (on this CPU) */
+ unsigned long total_time; /* total cputime (PURR) of calls. */
+};
+
+/* If Hypervisor call instrumentation is enabled, the assembly routine
+ * names are changed from 'plpar_hcall*' to 'plpar_hcall*_base' and
+ * 'plpar_hcall*' routines become instrumented wrappers. The following
+ * are declarations for the renamed 'plpar_hcall*_base' routines.
+ */
+long plpar_hcall_base (unsigned long opcode,
+ unsigned long arg1,
+ unsigned long arg2,
+ unsigned long arg3,
+ unsigned long arg4,
+ unsigned long *out1,
+ unsigned long *out2,
+ unsigned long *out3);
+
+long plpar_hcall_norets_base(unsigned long opcode, ...);
+
+long plpar_hcall_8arg_2ret_base(unsigned long opcode,
+ unsigned long arg1,
+ unsigned long arg2,
+ unsigned long arg3,
+ unsigned long arg4,
+ unsigned long arg5,
+ unsigned long arg6,
+ unsigned long arg7,
+ unsigned long arg8,
+ unsigned long *out1);
+
+long plpar_hcall_4out_base(unsigned long opcode,
+ unsigned long arg1,
+ unsigned long arg2,
+ unsigned long arg3,
+ unsigned long arg4,
+ unsigned long *out1,
+ unsigned long *out2,
+ unsigned long *out3,
+ unsigned long *out4);
+
+long plpar_hcall_7arg_7ret_base(unsigned long opcode,
+ unsigned long arg1,
+ unsigned long arg2,
+ unsigned long arg3,
+ unsigned long arg4,
+ unsigned long arg5,
+ unsigned long arg6,
+ unsigned long arg7,
+ unsigned long *out1,
+ unsigned long *out2,
+ unsigned long *out3,
+ unsigned long *out4,
+ unsigned long *out5,
+ unsigned long *out6,
+ unsigned long *out7);
+
+long plpar_hcall_9arg_9ret_base(unsigned long opcode,
+ unsigned long arg1,
+ unsigned long arg2,
+ unsigned long arg3,
+ unsigned long arg4,
+ unsigned long arg5,
+ unsigned long arg6,
+ unsigned long arg7,
+ unsigned long arg8,
+ unsigned long arg9,
+ unsigned long *out1,
+ unsigned long *out2,
+ unsigned long *out3,
+ unsigned long *out4,
+ unsigned long *out5,
+ unsigned long *out6,
+ unsigned long *out7,
+ unsigned long *out8,
+ unsigned long *out9);
+
#endif /* __ASSEMBLY__ */
#endif /* __KERNEL__ */
#endif /* _ASM_POWERPC_HVCALL_H */
^ permalink raw reply [flat|nested] 22+ messages in thread
* [PATCH 3/3] powerpc: Instrument Hypervisor Calls: add debugfs files
2006-07-14 23:37 [PATCH 0/3] powerpc: Instrument Hypervisor Calls Mike Kravetz
2006-07-14 23:39 ` [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers Mike Kravetz
2006-07-14 23:40 ` [PATCH 2/3] powerpc: Instrument Hypervisor Calls: add wrappers Mike Kravetz
@ 2006-07-14 23:41 ` Mike Kravetz
2006-07-15 0:00 ` [PATCH 0/3] powerpc: Instrument Hypervisor Calls Arnd Bergmann
3 siblings, 0 replies; 22+ messages in thread
From: Mike Kravetz @ 2006-07-14 23:41 UTC (permalink / raw)
To: Paul Mackerras
Cc: Arnd Bergmann, Bryan Rosenburg, linuxppc-dev, Nathan Lynch,
Christopher Yeoh
Make statistics available via files in debugfs.
--
Signed-off-by: Mike Kravetz <kravetz@us.ibm.com>
diff -Naupr linux-2.6.17.4/arch/powerpc/platforms/pseries/hvCall_inst.c linux-2.6.17.4.work/arch/powerpc/platforms/pseries/hvCall_inst.c
--- linux-2.6.17.4/arch/powerpc/platforms/pseries/hvCall_inst.c 2006-07-14 23:32:01.000000000 +0000
+++ linux-2.6.17.4.work/arch/powerpc/platforms/pseries/hvCall_inst.c 2006-07-14 23:32:54.000000000 +0000
@@ -205,3 +205,94 @@ long plpar_hcall_9arg_9ret(unsigned long
update_stats(opcode, t_before);
return rc;
}
+
+/*
+ * Routines for displaying the statistics in debugfs
+ */
+static void *hc_start(struct seq_file *m, loff_t *pos)
+{
+ if ((int)*pos < MAX_HCALL_OPCODES)
+ return (void *)(unsigned long)(*pos + 1);
+
+ return NULL;
+}
+
+static void *hc_next(struct seq_file *m, void *p, loff_t * pos)
+{
+ ++*pos;
+
+ return hc_start(m, pos);
+}
+
+static void hc_stop(struct seq_file *m, void *p)
+{
+}
+
+static int hc_show(struct seq_file *m, void *p)
+{
+ unsigned long h_num = (unsigned long)p;
+ struct hcall_stats *hs = (struct hcall_stats *)m->private;
+
+ if (hs[h_num].num_calls)
+ seq_printf(m, "%lu %lu %lu\n", h_num<<2, hs[h_num].num_calls,
+ hs[h_num].total_time);
+
+ return 0;
+}
+
+static struct seq_operations hcall_inst_seq_ops = {
+ .start = hc_start,
+ .next = hc_next,
+ .stop = hc_stop,
+ .show = hc_show
+};
+
+static int hcall_inst_seq_open(struct inode *inode, struct file *file)
+{
+ int rc;
+ struct seq_file *seq;
+
+ rc = seq_open(file, &hcall_inst_seq_ops);
+ seq = file->private_data;
+ seq->private = file->f_dentry->d_inode->u.generic_ip;
+
+ return rc;
+}
+
+static struct file_operations hcall_inst_seq_fops = {
+ .open = hcall_inst_seq_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = seq_release,
+};
+
+#define HCALL_ROOT_DIR "hcall_inst"
+#define CPU_NAME_BUF_SIZE 32
+
+static int __init hcall_inst_init(void)
+{
+ struct dentry *hcall_root;
+ struct dentry *hcall_file;
+ char cpu_name_buf[CPU_NAME_BUF_SIZE];
+ int cpu;
+
+ if (!firmware_has_feature(FW_FEATURE_LPAR))
+ return 0;
+
+ hcall_root = debugfs_create_dir(HCALL_ROOT_DIR, NULL);
+ if (!hcall_root)
+ return -ENOMEM;
+
+ for_each_cpu(cpu) {
+ snprintf(cpu_name_buf, CPU_NAME_BUF_SIZE, "cpu%d", cpu);
+ hcall_file = debugfs_create_file(cpu_name_buf, S_IRUGO,
+ hcall_root,
+ per_cpu(hcall_stats, cpu),
+ &hcall_inst_seq_fops);
+ if (!hcall_file)
+ return -ENOMEM;
+ }
+
+ return 0;
+}
+__initcall(hcall_inst_init);
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 0/3] powerpc: Instrument Hypervisor Calls
2006-07-14 23:37 [PATCH 0/3] powerpc: Instrument Hypervisor Calls Mike Kravetz
` (2 preceding siblings ...)
2006-07-14 23:41 ` [PATCH 3/3] powerpc: Instrument Hypervisor Calls: add debugfs files Mike Kravetz
@ 2006-07-15 0:00 ` Arnd Bergmann
2006-07-15 0:06 ` Mike Kravetz
` (2 more replies)
3 siblings, 3 replies; 22+ messages in thread
From: Arnd Bergmann @ 2006-07-15 0:00 UTC (permalink / raw)
To: linuxppc-dev
Cc: Bryan Rosenburg, Nathan Lynch, Paul Mackerras, Christopher Yeoh
On Saturday 15 July 2006 01:37, Mike Kravetz wrote:
> In addition, there has been some discussion of these patches on IRC.
> Some remaining issues/questions are:
> - Exactly how much overhead does the statistic gathering introduce?
> - What would be the cost of disabling preemption for more accurate statistics?
> - What would be the cost of disabling interrupts for more accurate statistics?
> - Should we extend this statistic gathering to RTAS calls?
What happened to the question whether to use PURR values for also measuring
cycles spent executing the hcall as opposed to cycles that passed before
the hcall returns. Did that turn out not giving extra information after all
or was there a different reason to drop that idea?
Arnd <><
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 0/3] powerpc: Instrument Hypervisor Calls
2006-07-15 0:00 ` [PATCH 0/3] powerpc: Instrument Hypervisor Calls Arnd Bergmann
@ 2006-07-15 0:06 ` Mike Kravetz
2006-07-15 15:30 ` Anton Blanchard
2006-07-16 3:53 ` Olof Johansson
2 siblings, 0 replies; 22+ messages in thread
From: Mike Kravetz @ 2006-07-15 0:06 UTC (permalink / raw)
To: Arnd Bergmann
Cc: Bryan Rosenburg, linuxppc-dev, Nathan Lynch, Paul Mackerras,
Christopher Yeoh
On Sat, Jul 15, 2006 at 02:00:02AM +0200, Arnd Bergmann wrote:
> What happened to the question whether to use PURR values for also measuring
> cycles spent executing the hcall as opposed to cycles that passed before
> the hcall returns. Did that turn out not giving extra information after all
> or was there a different reason to drop that idea?
Oops, forgot that as an additional issue/question. In this patch, I went
back to mftb() as 'wall time' made more sense for the group wanting this
functionality. It is easy to switch, or collect both (I think). Since I
started with mftb went to PURR and then back to mftb, it certainly does
look like an agument to try and collect both. :)
--
Mike
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 2/3] powerpc: Instrument Hypervisor Calls: add wrappers
2006-07-14 23:40 ` [PATCH 2/3] powerpc: Instrument Hypervisor Calls: add wrappers Mike Kravetz
@ 2006-07-15 0:15 ` Nathan Lynch
2006-07-15 0:41 ` Mike Kravetz
0 siblings, 1 reply; 22+ messages in thread
From: Nathan Lynch @ 2006-07-15 0:15 UTC (permalink / raw)
To: Mike Kravetz
Cc: Bryan Rosenburg, linuxppc-dev, Paul Mackerras, Arnd Bergmann,
Christopher Yeoh
Mike Kravetz wrote:
> +static inline void update_stats(unsigned long opcode, unsigned long t_before)
> +{
> + unsigned long op_index = opcode >> 2;
> + struct hcall_stats *hs = &__get_cpu_var(hcall_stats[op_index]);
> +
> + hs->total_time += (mftb() - t_before);
> + hs->num_calls++;
> +}
> +
> +/*
> + * plpar_hcall wrapper
> + */
> +long plpar_hcall(unsigned long opcode,
> + unsigned long arg1,
> + unsigned long arg2,
> + unsigned long arg3,
> + unsigned long arg4,
> + unsigned long *out1,
> + unsigned long *out2,
> + unsigned long *out3)
> +{
> + long rc;
> + unsigned long t_before;
> +
> + t_before = mftb();
> + rc = plpar_hcall_base(opcode, arg1, arg2, arg3, arg4, out1, out2, out3);
> +
> + update_stats(opcode, t_before);
> + return rc;
> +}
Hmm, isn't it possible that the stats could be corrupted if we process
an interrupt/softirq which does an hcall while the stats are being
updated? Maybe it's not a show-stopper, but it seems to me that
accuracy could suffer under some workloads.
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 2/3] powerpc: Instrument Hypervisor Calls: add wrappers
2006-07-15 0:15 ` Nathan Lynch
@ 2006-07-15 0:41 ` Mike Kravetz
2006-07-15 8:03 ` Nathan Lynch
0 siblings, 1 reply; 22+ messages in thread
From: Mike Kravetz @ 2006-07-15 0:41 UTC (permalink / raw)
To: Nathan Lynch
Cc: Bryan Rosenburg, linuxppc-dev, Paul Mackerras, Arnd Bergmann,
Christopher Yeoh
On Fri, Jul 14, 2006 at 07:15:45PM -0500, Nathan Lynch wrote:
> Hmm, isn't it possible that the stats could be corrupted if we process
> an interrupt/softirq which does an hcall while the stats are being
> updated? Maybe it's not a show-stopper, but it seems to me that
> accuracy could suffer under some workloads.
Yes, they can become inaccurate. Is this what you mean by 'corrupted' or
you thinking of something worse?
There is going to be a trade off between accuracy and the overhead of the
data gathering. What really needs to happen is some good benchmarking to
measure the costs of different approaches. I'm pretty sure I can get
someone to provide such data.
In this patch, I have minimized the overhead in data collection at the
cost of accuracy.
--
Mike
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 2/3] powerpc: Instrument Hypervisor Calls: add wrappers
2006-07-15 0:41 ` Mike Kravetz
@ 2006-07-15 8:03 ` Nathan Lynch
0 siblings, 0 replies; 22+ messages in thread
From: Nathan Lynch @ 2006-07-15 8:03 UTC (permalink / raw)
To: Mike Kravetz
Cc: Bryan Rosenburg, linuxppc-dev, Paul Mackerras, Arnd Bergmann,
Christopher Yeoh
Mike Kravetz wrote:
> On Fri, Jul 14, 2006 at 07:15:45PM -0500, Nathan Lynch wrote:
> > Hmm, isn't it possible that the stats could be corrupted if we process
> > an interrupt/softirq which does an hcall while the stats are being
> > updated? Maybe it's not a show-stopper, but it seems to me that
> > accuracy could suffer under some workloads.
>
> Yes, they can become inaccurate. Is this what you mean by 'corrupted' or
> you thinking of something worse?
Yes, "corrupted" was perhaps too strong, but we could lose the
accounting for hcalls from interrupts that we take at exactly the
wrong time. I guess it's a pretty tiny risk.
I just wanted to make sure this had been considered, but if you're
comfortable with the overhead/accuracy tradeoff, then okay.
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 0/3] powerpc: Instrument Hypervisor Calls
2006-07-15 0:00 ` [PATCH 0/3] powerpc: Instrument Hypervisor Calls Arnd Bergmann
2006-07-15 0:06 ` Mike Kravetz
@ 2006-07-15 15:30 ` Anton Blanchard
2006-07-15 16:42 ` Arnd Bergmann
2006-07-17 2:02 ` Luke Browning
2006-07-16 3:53 ` Olof Johansson
2 siblings, 2 replies; 22+ messages in thread
From: Anton Blanchard @ 2006-07-15 15:30 UTC (permalink / raw)
To: Arnd Bergmann
Cc: Bryan Rosenburg, linuxppc-dev, Nathan Lynch, Paul Mackerras,
Christopher Yeoh
> What happened to the question whether to use PURR values for also measuring
> cycles spent executing the hcall as opposed to cycles that passed before
> the hcall returns. Did that turn out not giving extra information after all
> or was there a different reason to drop that idea?
You have to be careful with PURR since it may be context switched
between partitions.
Anton
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 0/3] powerpc: Instrument Hypervisor Calls
2006-07-15 15:30 ` Anton Blanchard
@ 2006-07-15 16:42 ` Arnd Bergmann
2006-07-15 22:07 ` Anton Blanchard
2006-07-17 2:02 ` Luke Browning
1 sibling, 1 reply; 22+ messages in thread
From: Arnd Bergmann @ 2006-07-15 16:42 UTC (permalink / raw)
To: Anton Blanchard
Cc: Bryan Rosenburg, linuxppc-dev, Nathan Lynch, Paul Mackerras,
Christopher Yeoh
On Saturday 15 July 2006 17:30, Anton Blanchard wrote:
> > What happened to the question whether to use PURR values for also measuring
> > cycles spent executing the hcall as opposed to cycles that passed before
> > the hcall returns. Did that turn out not giving extra information after all
> > or was there a different reason to drop that idea?
>
> You have to be careful with PURR since it may be context switched
> between partitions.
>
Not sure I follow you. I would expect the PURR value to be restored after
a context switch, even if we continue on a different physical CPU.
The idea behind monitoring both PURR and timebase is that the difference
between the two tells you how long the partition was suspended during
the hcall.
Arnd <><
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 0/3] powerpc: Instrument Hypervisor Calls
2006-07-15 16:42 ` Arnd Bergmann
@ 2006-07-15 22:07 ` Anton Blanchard
2006-07-16 23:02 ` Luke Browning
0 siblings, 1 reply; 22+ messages in thread
From: Anton Blanchard @ 2006-07-15 22:07 UTC (permalink / raw)
To: Arnd Bergmann
Cc: Bryan Rosenburg, linuxppc-dev, Nathan Lynch, Paul Mackerras,
Christopher Yeoh
> Not sure I follow you. I would expect the PURR value to be restored after
> a context switch, even if we continue on a different physical CPU.
>
> The idea behind monitoring both PURR and timebase is that the difference
> between the two tells you how long the partition was suspended during
> the hcall.
Sounds good, last time I looked at the patch I thought it was gathering
the PURR only. That on its own would make for some confusing results.
Anton
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 0/3] powerpc: Instrument Hypervisor Calls
2006-07-15 0:00 ` [PATCH 0/3] powerpc: Instrument Hypervisor Calls Arnd Bergmann
2006-07-15 0:06 ` Mike Kravetz
2006-07-15 15:30 ` Anton Blanchard
@ 2006-07-16 3:53 ` Olof Johansson
2006-07-16 22:53 ` Luke Browning
2 siblings, 1 reply; 22+ messages in thread
From: Olof Johansson @ 2006-07-16 3:53 UTC (permalink / raw)
To: Arnd Bergmann
Cc: Bryan Rosenburg, linuxppc-dev, Nathan Lynch, Paul Mackerras,
Christopher Yeoh
On Sat, Jul 15, 2006 at 02:00:02AM +0200, Arnd Bergmann wrote:
> What happened to the question whether to use PURR values for also measuring
> cycles spent executing the hcall as opposed to cycles that passed before
> the hcall returns. Did that turn out not giving extra information after all
> or was there a different reason to drop that idea?
Just so people don't forget: this can't be done on all processors. For
example, PPC970 and POWER4 don't implement the PURR SPR. And it doesn't
make sense to use H_PURR to get the software emulated ones there. Not
really an issue on POWER4 since they don't do shared processor LPAR, but
on JS21 I think they might do?
-Olof
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 0/3] powerpc: Instrument Hypervisor Calls
2006-07-16 3:53 ` Olof Johansson
@ 2006-07-16 22:53 ` Luke Browning
2006-07-16 23:09 ` Olof Johansson
0 siblings, 1 reply; 22+ messages in thread
From: Luke Browning @ 2006-07-16 22:53 UTC (permalink / raw)
To: Olof Johansson
Cc: linuxppc-dev-bounces+lukebrowning=us.ibm.com, Arnd Bergmann,
Bryan S Rosenburg, linuxppc-dev, Nathan Lynch, Christopher Yeoh,
Paul Mackerras
[-- Attachment #1: Type: text/plain, Size: 910 bytes --]
linuxppc-dev-bounces+lukebrowning=us.ibm.com@ozlabs.org wrote on
07/16/2006 12:53:54 AM:
> On Sat, Jul 15, 2006 at 02:00:02AM +0200, Arnd Bergmann wrote:
>
> > What happened to the question whether to use PURR values for also
measuring
> > cycles spent executing the hcall as opposed to cycles that passed
before
> > the hcall returns. Did that turn out not giving extra information
after all
> > or was there a different reason to drop that idea?
>
> Just so people don't forget: this can't be done on all processors. For
> example, PPC970 and POWER4 don't implement the PURR SPR. And it doesn't
> make sense to use H_PURR to get the software emulated ones there. Not
> really an issue on POWER4 since they don't do shared processor LPAR, but
> on JS21 I think they might do?
>
Isn't there someway to do a platform specific processor overlay, where you
get PURR if it exists and TB otherwise.
Luke
[-- Attachment #2: Type: text/html, Size: 1129 bytes --]
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 0/3] powerpc: Instrument Hypervisor Calls
2006-07-15 22:07 ` Anton Blanchard
@ 2006-07-16 23:02 ` Luke Browning
0 siblings, 0 replies; 22+ messages in thread
From: Luke Browning @ 2006-07-16 23:02 UTC (permalink / raw)
To: Anton Blanchard
Cc: linuxppc-dev-bounces+lukebrowning=us.ibm.com, Arnd Bergmann,
Bryan S Rosenburg, linuxppc-dev, Nathan Lynch, Christopher Yeoh,
Paul Mackerras
[-- Attachment #1: Type: text/plain, Size: 986 bytes --]
linuxppc-dev-bounces+lukebrowning=us.ibm.com@ozlabs.org wrote on
07/15/2006 07:07:58 PM:
>
> > Not sure I follow you. I would expect the PURR value to be restored
after
> > a context switch, even if we continue on a different physical CPU.
> >
> > The idea behind monitoring both PURR and timebase is that the
difference
> > between the two tells you how long the partition was suspended during
> > the hcall.
>
> Sounds good, last time I looked at the patch I thought it was gathering
> the PURR only. That on its own would make for some confusing results.
>
It would be more efficient to have a separate trace log for PHYP
dispatches as
there are many more hypervisor calls than PHYP dispatches. I believe PHYP
provides
a trace records for dispatches, which could be seperately written. Perf
tools can
assemble the information from multiple records. It would be nice to have
processor
specific time function that reads either PURR or TB based on the platform.
Luke
[-- Attachment #2: Type: text/html, Size: 1376 bytes --]
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 0/3] powerpc: Instrument Hypervisor Calls
2006-07-16 22:53 ` Luke Browning
@ 2006-07-16 23:09 ` Olof Johansson
0 siblings, 0 replies; 22+ messages in thread
From: Olof Johansson @ 2006-07-16 23:09 UTC (permalink / raw)
To: Luke Browning
Cc: Arnd Bergmann, Bryan S Rosenburg, linuxppc-dev, Paul Mackerras,
Christopher Yeoh, Nathan Lynch
On Sun, Jul 16, 2006 at 08:53:21PM -0200, Luke Browning wrote:
> > Just so people don't forget: this can't be done on all processors. For
> > example, PPC970 and POWER4 don't implement the PURR SPR. And it doesn't
> > make sense to use H_PURR to get the software emulated ones there. Not
> > really an issue on POWER4 since they don't do shared processor LPAR, but
> > on JS21 I think they might do?
> >
> Isn't there someway to do a platform specific processor overlay, where you
> get PURR if it exists and TB otherwise.
Sure, it's what we have cpu feature bits for. It's not a difficult
problem to solve, I was just reminding of the lack of SPRN_PURR on some
processors that can run hypervisors.
-Olof
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 0/3] powerpc: Instrument Hypervisor Calls
2006-07-15 15:30 ` Anton Blanchard
2006-07-15 16:42 ` Arnd Bergmann
@ 2006-07-17 2:02 ` Luke Browning
1 sibling, 0 replies; 22+ messages in thread
From: Luke Browning @ 2006-07-17 2:02 UTC (permalink / raw)
To: Anton Blanchard
Cc: linuxppc-dev-bounces+lukebrowning=us.ibm.com, Arnd Bergmann,
Bryan S Rosenburg, linuxppc-dev, Nathan Lynch, Christopher Yeoh,
Paul Mackerras
[-- Attachment #1: Type: text/plain, Size: 602 bytes --]
linuxppc-dev-bounces+lukebrowning=us.ibm.com@ozlabs.org wrote on
07/15/2006 12:30:54 PM:
>
> > What happened to the question whether to use PURR values for also
measuring
> > cycles spent executing the hcall as opposed to cycles that passed
before
> > the hcall returns. Did that turn out not giving extra information
after all
> > or was there a different reason to drop that idea?
>
> You have to be careful with PURR since it may be context switched
> between partitions.
>
That shouldn't be an issue as the PURR is supposed to be virtualized by
PHYP at
the virtual processor level.
Luke
[-- Attachment #2: Type: text/html, Size: 828 bytes --]
^ permalink raw reply [flat|nested] 22+ messages in thread
* [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers
2006-07-18 20:47 [PATCH 0/3] powerpc: Instrument Hypervisor Calls Mike Kravetz
@ 2006-07-18 20:48 ` Mike Kravetz
0 siblings, 0 replies; 22+ messages in thread
From: Mike Kravetz @ 2006-07-18 20:48 UTC (permalink / raw)
To: Paul Mackerras; +Cc: linuxppc-dev
Move all the Hypervisor call definitions to to a single header file.
--
Signed-off-by: Mike Kravetz <kravetz@us.ibm.com>
diff -Naupr linux-2.6.17.6/drivers/net/ibmveth.h linux-2.6.17.6.work/drivers/net/ibmveth.h
--- linux-2.6.17.6/drivers/net/ibmveth.h 2006-07-15 19:00:43.000000000 +0000
+++ linux-2.6.17.6.work/drivers/net/ibmveth.h 2006-07-18 19:33:47.000000000 +0000
@@ -41,16 +41,6 @@
#define IbmVethMcastRemoveFilter 0x2UL
#define IbmVethMcastClearFilterTable 0x3UL
-/* hcall numbers */
-#define H_VIO_SIGNAL 0x104
-#define H_REGISTER_LOGICAL_LAN 0x114
-#define H_FREE_LOGICAL_LAN 0x118
-#define H_ADD_LOGICAL_LAN_BUFFER 0x11C
-#define H_SEND_LOGICAL_LAN 0x120
-#define H_MULTICAST_CTRL 0x130
-#define H_CHANGE_LOGICAL_LAN_MAC 0x14C
-#define H_FREE_LOGICAL_LAN_BUFFER 0x1D4
-
/* hcall macros */
#define h_register_logical_lan(ua, buflst, rxq, fltlst, mac) \
plpar_hcall_norets(H_REGISTER_LOGICAL_LAN, ua, buflst, rxq, fltlst, mac)
diff -Naupr linux-2.6.17.6/include/asm-powerpc/hvcall.h linux-2.6.17.6.work/include/asm-powerpc/hvcall.h
--- linux-2.6.17.6/include/asm-powerpc/hvcall.h 2006-07-15 19:00:43.000000000 +0000
+++ linux-2.6.17.6.work/include/asm-powerpc/hvcall.h 2006-07-18 19:33:47.000000000 +0000
@@ -155,9 +155,15 @@
#define H_VIO_SIGNAL 0x104
#define H_SEND_CRQ 0x108
#define H_COPY_RDMA 0x110
+#define H_REGISTER_LOGICAL_LAN 0x114
+#define H_FREE_LOGICAL_LAN 0x118
+#define H_ADD_LOGICAL_LAN_BUFFER 0x11C
+#define H_SEND_LOGICAL_LAN 0x120
+#define H_MULTICAST_CTRL 0x130
#define H_SET_XDABR 0x134
#define H_STUFF_TCE 0x138
#define H_PUT_TCE_INDIRECT 0x13C
+#define H_CHANGE_LOGICAL_LAN_MAC 0x14C
#define H_VTERM_PARTNER_INFO 0x150
#define H_REGISTER_VTERM 0x154
#define H_FREE_VTERM 0x158
@@ -187,11 +193,14 @@
#define H_GET_HCA_INFO 0x1B8
#define H_GET_PERF_COUNT 0x1BC
#define H_MANAGE_TRACE 0x1C0
+#define H_FREE_LOGICAL_LAN_BUFFER 0x1D4
#define H_QUERY_INT_STATE 0x1E4
#define H_POLL_PENDING 0x1D8
#define H_JOIN 0x298
#define H_ENABLE_CRQ 0x2B0
+#define MAX_HCALL_OPCODES (H_ENABLE_CRQ >> 2)
+
#ifndef __ASSEMBLY__
/* plpar_hcall() -- Generic call interface using above opcodes
^ permalink raw reply [flat|nested] 22+ messages in thread
* [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers
2006-07-21 6:38 Mike Kravetz
@ 2006-07-21 6:39 ` Mike Kravetz
2006-07-21 12:14 ` Stephen Rothwell
0 siblings, 1 reply; 22+ messages in thread
From: Mike Kravetz @ 2006-07-21 6:39 UTC (permalink / raw)
To: Paul Mackerras; +Cc: linuxppc-dev
Move all the Hypervisor call definitions to to a single header file.
--
Signed-off-by: Mike Kravetz <kravetz@us.ibm.com>
diff -Naupr powerpc/drivers/net/ibmveth.h powerpc.work/drivers/net/ibmveth.h
--- powerpc/drivers/net/ibmveth.h 2006-07-19 18:58:18.000000000 +0000
+++ powerpc.work/drivers/net/ibmveth.h 2006-07-21 07:03:44.000000000 +0000
@@ -41,16 +41,6 @@
#define IbmVethMcastRemoveFilter 0x2UL
#define IbmVethMcastClearFilterTable 0x3UL
-/* hcall numbers */
-#define H_VIO_SIGNAL 0x104
-#define H_REGISTER_LOGICAL_LAN 0x114
-#define H_FREE_LOGICAL_LAN 0x118
-#define H_ADD_LOGICAL_LAN_BUFFER 0x11C
-#define H_SEND_LOGICAL_LAN 0x120
-#define H_MULTICAST_CTRL 0x130
-#define H_CHANGE_LOGICAL_LAN_MAC 0x14C
-#define H_FREE_LOGICAL_LAN_BUFFER 0x1D4
-
/* hcall macros */
#define h_register_logical_lan(ua, buflst, rxq, fltlst, mac) \
plpar_hcall_norets(H_REGISTER_LOGICAL_LAN, ua, buflst, rxq, fltlst, mac)
diff -Naupr powerpc/include/asm-powerpc/hvcall.h powerpc.work/include/asm-powerpc/hvcall.h
--- powerpc/include/asm-powerpc/hvcall.h 2006-07-19 18:58:18.000000000 +0000
+++ powerpc.work/include/asm-powerpc/hvcall.h 2006-07-21 07:03:44.000000000 +0000
@@ -164,9 +164,15 @@
#define H_VIO_SIGNAL 0x104
#define H_SEND_CRQ 0x108
#define H_COPY_RDMA 0x110
+#define H_REGISTER_LOGICAL_LAN 0x114
+#define H_FREE_LOGICAL_LAN 0x118
+#define H_ADD_LOGICAL_LAN_BUFFER 0x11C
+#define H_SEND_LOGICAL_LAN 0x120
+#define H_MULTICAST_CTRL 0x130
#define H_SET_XDABR 0x134
#define H_STUFF_TCE 0x138
#define H_PUT_TCE_INDIRECT 0x13C
+#define H_CHANGE_LOGICAL_LAN_MAC 0x14C
#define H_VTERM_PARTNER_INFO 0x150
#define H_REGISTER_VTERM 0x154
#define H_FREE_VTERM 0x158
@@ -196,11 +202,13 @@
#define H_GET_HCA_INFO 0x1B8
#define H_GET_PERF_COUNT 0x1BC
#define H_MANAGE_TRACE 0x1C0
+#define H_FREE_LOGICAL_LAN_BUFFER 0x1D4
#define H_QUERY_INT_STATE 0x1E4
#define H_POLL_PENDING 0x1D8
#define H_JOIN 0x298
#define H_VASI_STATE 0x2A4
#define H_ENABLE_CRQ 0x2B0
+#define MAX_HCALL_OPCODES (H_ENABLE_CRQ >> 2)
#ifndef __ASSEMBLY__
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers
2006-07-21 6:39 ` [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers Mike Kravetz
@ 2006-07-21 12:14 ` Stephen Rothwell
0 siblings, 0 replies; 22+ messages in thread
From: Stephen Rothwell @ 2006-07-21 12:14 UTC (permalink / raw)
To: paulus; +Cc: linuxppc-dev
[-- Attachment #1: Type: text/plain, Size: 414 bytes --]
Hi Paulus,
On Thu, 20 Jul 2006 23:39:51 -0700 Mike Kravetz <kravetz@us.ibm.com> wrote:
>
> Move all the Hypervisor call definitions to to a single header file.
> --
> Signed-off-by: Mike Kravetz <kravetz@us.ibm.com>
Can you put this one in immediately just to save Mike having to repost it :-)
--
Cheers,
Stephen Rothwell sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/
[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]
^ permalink raw reply [flat|nested] 22+ messages in thread
end of thread, other threads:[~2006-07-21 12:15 UTC | newest]
Thread overview: 22+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-07-14 23:37 [PATCH 0/3] powerpc: Instrument Hypervisor Calls Mike Kravetz
2006-07-14 23:39 ` [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers Mike Kravetz
2006-07-14 23:40 ` [PATCH 2/3] powerpc: Instrument Hypervisor Calls: add wrappers Mike Kravetz
2006-07-15 0:15 ` Nathan Lynch
2006-07-15 0:41 ` Mike Kravetz
2006-07-15 8:03 ` Nathan Lynch
2006-07-14 23:41 ` [PATCH 3/3] powerpc: Instrument Hypervisor Calls: add debugfs files Mike Kravetz
2006-07-15 0:00 ` [PATCH 0/3] powerpc: Instrument Hypervisor Calls Arnd Bergmann
2006-07-15 0:06 ` Mike Kravetz
2006-07-15 15:30 ` Anton Blanchard
2006-07-15 16:42 ` Arnd Bergmann
2006-07-15 22:07 ` Anton Blanchard
2006-07-16 23:02 ` Luke Browning
2006-07-17 2:02 ` Luke Browning
2006-07-16 3:53 ` Olof Johansson
2006-07-16 22:53 ` Luke Browning
2006-07-16 23:09 ` Olof Johansson
-- strict thread matches above, loose matches on Subject: below --
2006-07-21 6:38 Mike Kravetz
2006-07-21 6:39 ` [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers Mike Kravetz
2006-07-21 12:14 ` Stephen Rothwell
2006-07-18 20:47 [PATCH 0/3] powerpc: Instrument Hypervisor Calls Mike Kravetz
2006-07-18 20:48 ` [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers Mike Kravetz
2006-06-22 22:56 [PATCH 0/3] powerpc: Instrument Hypervisor Calls Mike Kravetz
2006-06-22 22:57 ` [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers Mike Kravetz
2006-06-14 3:47 [PATCH 0/3] powerpc: Instrument Hypervisor Calls Mike Kravetz
2006-06-14 3:50 ` [PATCH 1/3] powerpc: Instrument Hypervisor Calls: merge headers Mike Kravetz
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).