search for: lockstat

I''m looking for a suggestion on a good way to hunt down the source of high context switching and mutex contentions... Is dtrace the way to go now, or should I stick with something like lockstat? Russ This is a 5 second interval for mpstat: CPU minf mjf xcal intr ithr csw icsw migr smtx srw syscl usr sys wt idl 16 0 0 1115 1241 206 9095 912 2420 7393 0 12105 68 25 0 7 17 1 0 1054 1049 86 9290 921 2420 7355 0 12123 67 25 0 8 18 0 0 1004...

I just had a quick play with gzip compression on a filesystem and the result was the machine grinding to a halt while copying some large (.wav) files to it from another filesystem in the same pool. The system became very unresponsive, taking several seconds to echo keystrokes. The box is a maxed out AMD QuadFX, so it should have plenty of grunt for this. Comments? Ian

...eeded but these server do not create the devices found under /devices/pseudo/ as you can see via a truss on a dtrace. open("/devices/pseudo/fasttrap at 0:fasttrap", O_RDONLY) Err#2 ENOENT open("/devices/pseudo/fbt at 0:fbt", O_RDONLY) Err#2 ENOENT open("/devices/pseudo/lockstat at 0:lockstat", O_RDONLY) Err#2 ENOENT open("/devices/pseudo/profile at 0:profile", O_RDONLY) Err#2 ENOENT open("/devices/pseudo/sdt at 0:sdt", O_RDONLY) Err#2 ENOENT open("/devices/pseudo/systrace at 0:systrace", O_RDONLY) Err#2 ENOENT And i confirm these d...

I''m an Oracle DBA, a dtrace noob, and I''m attempting to resolve an Oracle write performance problem on Solaris 10, Oracle 9iR2. The Oracle DBWR processes have poor write throughput writing to VxFS files. I suspect that the problem may be with POSIX readers/writer file locking. Ideally I''d like to identify, for a given DBWR process, readers/writer file lock contention

...te description of what probes are offered ? Found some already under 7D category: dtrace dtrace (7d) - DTrace dynamic tracing facility fasttrap fasttrap (7d) - DTrace user instruction tracing provider fbt fbt (7d) - DTrace function boundary tracing provider lockstat lockstat (7d) - DTrace kernel lock instrumentation provider profile profile (7d) - DTrace profile interrupt provider sdt sdt (7d) - DTrace statically defined tracing provider systrace systrace (7d) - DTrace system call tracing provider Can we have as...

Is there a trick to listing available providers? I can''t find it.

Dtrace and Performance Teams, I have the following IO Performance Specific Questions (and I''m already savy with the lockstat and pre-dtrace utilities for performance analysis.. but in need of details regarding specifying IO bottlenecks @ the controller or IO bus..) : **Q.A*> Determining IO Saturation bottlenecks ( */.. beyond service times and kernel contention.. )/ I''m trying to quantify the /*% of resou...

...ve these directives (and this block comment) automatically, run * "dtrace -A" without additional arguments. See the "Anonymous Tracing" * chapter of the Solaris Dynamic Tracing Guide for details. * forceload: drv/systrace forceload: drv/sdt forceload: drv/profile forceload: drv/lockstat forceload: drv/fbt forceload: drv/fasttrap forceload: drv/systrace forceload: drv/sdt forceload: drv/profile forceload: drv/lockstat forceload: drv/fbt forceload: drv/fasttrap forceload: drv/dtrace * ^^^^ Added by DTrace It''s because when libdtrace searches for providers, it searches...

...166 36 3102 14 973 51300 221 68 0 61 0 38 5 0 0 15 206 25 3397 13 965 51468 238 62 0 61 0 39 6 0 0 10 205 29 3254 11 958 52947 236 107 0 62 0 38 7 0 0 19 124 30 3889 15 962 51948 230 44 0 62 0 38 We ran lockstat to see what was causing the high SMTX and found this: root at dcivolume01:~# lockstat -D 10 sleep 2 Adaptive mutex spin: 706601 events in 2.060 seconds (342956 events/sec) Count indv cuml rcnt nsec Lock Caller ------------------------------------------------------...

...0 symbol tables should not be discarded for UNDEAD processes Files: update: usr/src/cmd/dtrace/dtrace.c update: usr/src/cmd/fm/modules/sun4u/USII-io-diagnosis/iod_main.c update: usr/src/cmd/fm/modules/sun4u/cpumem-diagnosis/cmd_cpu.c update: usr/src/cmd/intrstat/intrstat.c update: usr/src/cmd/lockstat/lockstat.c update: usr/src/cmd/mdb/common/modules/dtrace/dtrace.c update: usr/src/cmd/plockstat/plockstat.c update: usr/src/lib/libdtrace/common/dt_aggregate.c update: usr/src/lib/libdtrace/common/dt_cc.c update: usr/src/lib/libdtrace/common/dt_consume.c update: usr/src/lib/libdtrace/common/d...

Hi, I would like to know why dtrace -l |grep plockstat show plockstat28099 lines. Process with pid 28099 is a java process. Why can I not examine other processes with the plockstat provider? Is it a java startup option? Regards Hans -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/piperma...

...PARAVIRT_SPINLOCKS) +#ifdef CONFIG_QUEUE_SPINLOCK + +static __always_inline void pv_kick_cpu(int cpu) +{ + PVOP_VCALLEE1(pv_lock_ops.kick_cpu, cpu); +} + +static __always_inline void pv_lockwait(u8 *lockbyte) +{ + PVOP_VCALLEE1(pv_lock_ops.lockwait, lockbyte); +} + +static __always_inline void pv_lockstat(enum pv_lock_stats type) +{ + PVOP_VCALLEE1(pv_lock_ops.lockstat, type); +} + +#else static __always_inline void __ticket_lock_spinning(struct arch_spinlock *lock, __ticket_t ticket) { @@ -723,6 +741,7 @@ static __always_inline void __ticket_unlock_kick(struct arch_spinlock *lock, { PV...

...PARAVIRT_SPINLOCKS) +#ifdef CONFIG_QUEUE_SPINLOCK + +static __always_inline void pv_kick_cpu(int cpu) +{ + PVOP_VCALLEE1(pv_lock_ops.kick_cpu, cpu); +} + +static __always_inline void pv_lockwait(u8 *lockbyte) +{ + PVOP_VCALLEE1(pv_lock_ops.lockwait, lockbyte); +} + +static __always_inline void pv_lockstat(enum pv_lock_stats type) +{ + PVOP_VCALLEE1(pv_lock_ops.lockstat, type); +} + +#else static __always_inline void __ticket_lock_spinning(struct arch_spinlock *lock, __ticket_t ticket) { @@ -723,6 +741,7 @@ static __always_inline void __ticket_unlock_kick(struct arch_spinlock *lock, { PV...

...code is identical for sparc and intel except swapping involved since the hardware is little endian . Any idea how to debug this issue ?? > The machine i tried to bench mark is T5440 and i have tried setting ip_soft_rings_count = 16 on T5440 but result is same . > You might want to run lockstat(1M) to see if the driver routines are implicated as a problem. If the driver is not a problem, then you might want to try CPU binding with dladm set-linkprop -p cpus=0,1,...,24 <link> There may be other things to try here. Also, which OpenSolaris build are you running? In gen...

Hi there, I am trying to diagnose a performance problem on a Sun T5120 running Solaris 10. I see general poor performance using Java and, in particular, loading classes. I have developed a simple java test that loads all of the classes in a Crystal Reports JAR file (around 7000 classes) which takes much longer on the T5120 (40 secs) than on a Core 2 Duo windows PC (5 secs). I recognise that this

...guest CPU utilization does not decrease when a lock bottleneck has been removed. Performance figures: Host: 2.6.34 upstream kernel, qemu-kvm-0.12.4 if=virtio,cache=none Guest: 2.6.35-rc3-kvm.git upstream kernel Storage: 12 disks as striped LVM volume Benchmark: 4 concurrent dd bs=4k iflag=direct Lockstat data for &vblk->lock: test con-bounces contentions waittime-min waittime-max waittime-total unmodified 7097 7108 0.31 956.09 161165.4 patched 11484 11550 0.30 411.80 50245.83 The maximum wait time went down by 544.29 us (-57...

...guest CPU utilization does not decrease when a lock bottleneck has been removed. Performance figures: Host: 2.6.34 upstream kernel, qemu-kvm-0.12.4 if=virtio,cache=none Guest: 2.6.35-rc3-kvm.git upstream kernel Storage: 12 disks as striped LVM volume Benchmark: 4 concurrent dd bs=4k iflag=direct Lockstat data for &vblk->lock: test con-bounces contentions waittime-min waittime-max waittime-total unmodified 7097 7108 0.31 956.09 161165.4 patched 11484 11550 0.30 411.80 50245.83 The maximum wait time went down by 544.29 us (-57...

...gt; +static __always_inline void pv_kick_cpu(int cpu) > +{ > + PVOP_VCALLEE1(pv_lock_ops.kick_cpu, cpu); > +} > + > +static __always_inline void pv_lockwait(u8 *lockbyte) > +{ > + PVOP_VCALLEE1(pv_lock_ops.lockwait, lockbyte); > +} > + > +static __always_inline void pv_lockstat(enum pv_lock_stats type) > +{ > + PVOP_VCALLEE1(pv_lock_ops.lockstat, type); > +} Why are any of these PV ops? they're only called from other pv_*() functions. What's the point of pv ops you only call from pv code? > +/* > + * Queue Spinlock Para-Virtualization (PV) Support...

...ARAVIRT_SPINLOCKS) +#ifdef CONFIG_QUEUE_SPINLOCK + +static __always_inline void pv_kick_cpu(int cpu) +{ + PVOP_VCALLEE1(pv_lock_ops.kick_cpu, cpu); +} + +static __always_inline void +pv_lockwait(u8 *lockbyte) +{ + PVOP_VCALLEE1(pv_lock_ops.lockwait, lockbyte); +} + +static __always_inline void pv_lockstat(enum pv_lock_stats type) +{ + PVOP_VCALLEE1(pv_lock_ops.lockstat, type); +} + +#else static __always_inline void __ticket_lock_spinning(struct arch_spinlock *lock, __ticket_t ticket) { @@ -723,6 +742,7 @@ static __always_inline void __ticket_unlock_kick(struct arch_spinlock *lock, { PV...

...ARAVIRT_SPINLOCKS) +#ifdef CONFIG_QUEUE_SPINLOCK + +static __always_inline void pv_kick_cpu(int cpu) +{ + PVOP_VCALLEE1(pv_lock_ops.kick_cpu, cpu); +} + +static __always_inline void +pv_lockwait(u8 *lockbyte) +{ + PVOP_VCALLEE1(pv_lock_ops.lockwait, lockbyte); +} + +static __always_inline void pv_lockstat(enum pv_lock_stats type) +{ + PVOP_VCALLEE1(pv_lock_ops.lockstat, type); +} + +#else static __always_inline void __ticket_lock_spinning(struct arch_spinlock *lock, __ticket_t ticket) { @@ -723,6 +742,7 @@ static __always_inline void __ticket_unlock_kick(struct arch_spinlock *lock, { PV...