Article on RAW hazards

Feels like it's been a long while since I wrote up an article for OTN, so I'm pleased that I've finally got around to fixing that.

I've written about RAW hazards in the past. But I recently went through a patch of discovering them in a number of places, so I've written up a full article on them.

What is "nice" about RAW hazards is that once you recognise them for what they are (and that's the tricky bit), they are typically easy to avoid. So if you see 10 seconds of time attributable to RAW hazards in the profile, then you can often get the entire 10 seconds back by tweaking the code.

OpenMP, macros, and #define

Sometimes you need to include directives in macros. The classic example would be putting OpenMP directives into macros. The "obvious" way of doing this is:

#define BARRIER \
#pragma omp barrier

void foo()
{
  BARRIER
}

Which produces the following error:

"test.c", line 6: invalid source character: '#'
"test.c", line 6: undefined symbol: pragma
"test.c", line 6: syntax error before or at: omp

Fortunately C99 introduced the _Pragma mechanism to solve this problem. So the functioning code looks like:

#define BARRIER \
_Pragma("omp barrier")

void foo()
{
  BARRIER
}

Privileges

Before I start, this is not about security, it's probably the antithesis of security. So I'd recommend starting by reading about how using privileges can break the security of your system.

There are three tools that I regularly use that require escalated privileges: dtrace, cpustat, and busstat. You can read up on the way that Solaris manages privileges. But if you know what you want to do, the process to figure out how to get the necessary privileges is reasonable straightforward.

To find out what privileges you have you can use the ppriv -v $$ command. This will report all the privileges for the current shell.

To find out what privileges are stopping you from running a command, you should run it under ppriv -eD command. For example:

ppriv -eD cpustat -c instruction_counts 1 1
cpustat[13222]: missing privilege "sys_resource" (euid = 84945, syscall = 128) needed at rctl_rlimit_set+0x98
cpustat[13222]: missing privilege "cpc_cpu" (euid = 84945, syscall = 5) needed at kcpc_open+0x4
...

It is also possible to list all the privileges on the system using ppriv -l. This is helpful if the privilege is has a name that maps onto what you want to do. The privileges for dtrace are good examples of this:

$ ppriv -l|grep dtrace
dtrace_kernel
dtrace_proc
dtrace_user

You can then use usermod -K ... to assign the necessary privileges to a user. For example:

$ usermod -K defaultpriv=basic,sys_resource,cpc_cpu username

Information about privileges for users is recorded in /etc/user_attr, so it is possible to directly edit that file to add or remove privileges.

Using this approach you can determine that busstat needs sys_config, cpustat needs sys_resource and cpc_cpu, and dtrace needs dtrace_kernel, dtrace_proc, and dtrace_user.

JavaOne and Oracle Open World Presentations available

The slides from all the JavaOne and Oracle Open World presentations have been posted. You can find my slides at:

• Performance Tuning Where Java Meets the Hardware
• Mixed-Language Development: Leveraging Native Code from Java
• Developing Efficient Database Applications with C and C++

Timezone troubles when dual booting

I have a laptop that dual boots Solaris and Windows XP. When I switched between the two OSes I would have to reset the clock because the time would be eight hours out. This has been naggging at me for a while, so I dug into what was going on.

It seems that Windows assumes that the Real-Time Clock (RTC) in the bios is using local time. So it will read the clock and display whatever time is shown there.

Solaris on the other hand assumes that the clock is in Universal Time Format (UTC), so you have to apply a time zone transformation in order to get to the local time.

Obviously, if you adjust the clock to make one correct, then the other is wrong.

To me, it seems more natural to have the clock in a laptop work on UTC - because when you travel the local time changes. There is a registry setting in Windows that, when set to 1, tells the machine to use UTC:

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\TimeZoneInformation\RealTimeIsUniversal

However, it has some problems and is potentially not robust over sleep. So we have to work the other way, and get Solaris to use local time. Fortunately, this is a relatively simple change running the following as root (pick the appropriate timezone for your location):

rtc -z US/Pacific

My Oracle Open World and JavaOne schedule

I've got my schedule for Oracle Open World and JavaOne:

• Performance Tuning Where Java Meets the Hardware - CON3762
  9/23/13 (Monday) 1:00 PM - Hilton - Continental Ballroom 5
• Developing Efficient Database Applications with C and C++ - CON9355
  9/26/13 (Thursday) 12:30 PM - Marriott Marquis - Golden Gate C2
• Mixed-Language Development: Leveraging Native Code from Java - CON3408
  9/26/13 (Thursday) 2:00 PM - Hilton - Continental Ballroom 6

Note that on Thursday I have about 30 minutes between my two talks, so expect me to rush out of the database talk in order to get to the Java talk.

How to use a lot of threads ....

The SPARC M5-32 has 1,536 virtual CPUs. There is a lot that you can do with that much resource and a bunch of us sat down to write a white paper discussing the options(revised link).

There are a couple of key observations in there. First of all it is quite likely that such a large system will not end up running a single instance of a single application. Therefore it is useful to understand the various options for virtualising the system. The second observation is that there are a number of details to take care of when writing an application that is expected to scale to large numbers of threads.

Anyhow, I hope you find it a useful read.