Freescale Semiconductor i.MX51 EVK Reference Manual Download

Page: 190 / 198

OProfile

i.MX51 EVK Linux Reference Manual

33-2

Freescale Semiconductor

file. Each PC value is thus converted into a tuple (group or set) of binary-image offset. The userspace tools
can use this data to reconstruct where the code came from, including the particular assembly instructions,
symbol, and source line (through the binary debug information if present).

Regularly sampling the PC value like this approximates what actually was executed and how often and
more often than not, this statistical approximation is good enough to reflect reality. In common operation,
the time between each sample interrupt is regulated by a fixed number of clock cycles. This implies that
the results reflect where the CPU is spending the most time. This is a very useful information source for
performance analysis.

The ARM CPU provides hardware performance counters capable of measuring these events at the
hardware level. Typically, these counters increment once per each event and generate an interrupt on
reaching some pre-defined number of events. OProfile can use these interrupts to generate samples and the
profile results are a statistical approximation of which code caused how many instances of the given event.

33.4

Software Operation

33.4.1

Architecture Specific Components

If OProfile supports the hardware performance counters available on a particular architecture. Code for
managing the details of setting up and managing these counters can be located in the kernel source tree in
the relevant

<ltib_dir>/rpm/BUILD/linux

arch/arm/oprofile

directory. The architecture-specific

implementation operates through filling in the

oprofile_operations

structure at initialization. This

provides a set of operations, such as setup(), start(), stop(), and so on, that manage the hardware-specific
details the performance counter registers.

The other important facility available to the architecture code is

oprofile_add_sample

(). This is where a

particular sample taken at interrupt time is fed into the generic OProfile driver code.

33.4.2

oprofilefs Pseudo Filesystem

OProfile implements a pseudo-filesystem known as oprofilefs, which is mounted from userspace at

/dev/oprofile

. This consists of small files for reporting and receiving configuration from userspace, as

well as the actual character device that the OProfile userspace receives samples from. At setup() time, the
architecture-specific code may add further configuration files related to the details of the performance
counters. The filesystem also contains a

stats

directory with a number of useful counters for various

OProfile events.

33.4.3

Generic Kernel Driver

The generic kernel driver resides in

<ltib_dir>/rpm/BUILD/linux/drivers/oprofile/

, and forms the core

of how OProfile operates in the kernel. The generic kernel driver takes samples delivered from the
architecture-specific code (through

oprofile_add_sample

()), and buffers this data (in a transformed

configuration) until releasing the data to the userspace daemon through the