home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
NeXTSTEP 3.3 (Developer)…68k, x86, SPARC, PA-RISC]
/
NeXTSTEP 3.3 Dev Intel.iso
/
NextDeveloper
/
Headers
/
mach-o
/
gmon.h
< prev
next >
Wrap
C/C++ Source or Header
|
1993-01-20
|
5KB
|
164 lines
/*
* Copyright (c) 1980 Regents of the University of California.
* All rights reserved. The Berkeley software License Agreement
* specifies the terms and conditions for redistribution.
*
* @(#)gmon.h 5.1 (Berkeley) 5/30/85
*/
/*
* Histogram counters are unsigned shorts (according to the kernel).
*/
#define HISTCOUNTER unsigned short
/*
* Fraction of text space to allocate for histogram counters here, 1/2
*/
#define HISTFRACTION 2
/*
* Fraction of text space to allocate for from hash buckets.
* The value of HASHFRACTION is based on the minimum number of bytes
* of separation between two subroutine call points in the object code.
* Given MIN_SUBR_SEPARATION bytes of separation the value of
* HASHFRACTION is calculated as:
*
* HASHFRACTION = MIN_SUBR_SEPARATION / (2 * sizeof(short) - 1);
*
* For the VAX,
* the shortest two call sequence is:
* calls $0,(r0)
* calls $0,(r0)
* which is separated by only three bytes, thus HASHFRACTION is
* calculated as:
* HASHFRACTION = 3 / (2 * 2 - 1) = 1
*
* For the m68k,
* the shortest two call sequence is:
* jsr a0
* jsr a0
* which is separated by only four bytes, thus HASHFRACTION is
* calculated as:
* HASHFRACTION = 4 / (2 * 2 - 1) = 1
*
* For all RISC machines
* the shortest two call sequence is 2 32-bit instructions,
* which is separated by only four bytes, thus HASHFRACTION is
* calculated as:
* HASHFRACTION = 4 / (2 * 2 - 1) = 1
*
* For the i386,
* the shortest two call sequence is:
* call %eax
* call %eax
* which is separated by only two bytes, thus HASHFRACTION is
* calculated as:
* HASHFRACTION = 2 / (2 * 2 - 1) = 0
* So on the i386 we use a HASHFRACTION of 1 instead and it can fail
* to determine that two call sites are different. But since all
* the call site address in gprof(1) is currently used for is
* to determine which routine was doing the calling it works for now.
*
* Note that the division above rounds down, thus if MIN_SUBR_FRACTION
* is less than three, this algorithm will not work!
*/
#define HASHFRACTION 1
/*
* percent of text space to allocate for tostructs with a minimum.
*/
#define ARCDENSITY 4
#define MINARCS 50
#ifndef ASSEMBLER
/*
* The tostruct is used internal to the monitor library routines to implement
* the recording of calls via mcount().
*/
struct tostruct {
char *selfpc;
long count;
unsigned short link;
unsigned short order;
};
/*
* The phdr (profile header) structure is what appears at the beginning of a
* mon.out (cc(1) -p) and gmon.out (cc(1) -pg) file and describes the histogram
* counters. The histogram counters are unsigned shorts which follow after the
* header for ncnt - sizeof(struct phdr) bytes.
*/
struct phdr {
char *lpc; /* low program counter */
char *hpc; /* high program counter */
int ncnt; /* number of bytes of histogram counters minius
sizeof(struct phdr) that follow */
};
/*
* In a gmon.out (cc(1) -pg) file what follows the above histogram counters are
* the raw arcs. A raw arc contains pointers to the calling site, the called
* site and a count. These repeat in the gmon.out file after the histogram
* counters to the end of the file.
*/
struct rawarc {
unsigned long raw_frompc;
unsigned long raw_selfpc;
unsigned long raw_count;
};
/*
* In order to support more information than in the original mon.out and
* gmon.out files there is an alternate gmon.out file format. The alternate
* gmon.out file format starts with a magic number then separates the
* information with gmon_data structs.
*/
#define GMON_MAGIC 0xbeefbabe
struct gmon_data {
unsigned long type; /* constant for type of data following this struct */
unsigned long size; /* size in bytes of the data following this struct */
};
/*
* The GMONTYPE_SAMPLES gmon_data.type is for the histogram counters described
* above and has a struct phdr followed by the counters.
*/
#define GMONTYPE_SAMPLES 1
/*
* The GMONTYPE_RAWARCS gmon_data.type is for the raw arcs described above.
*/
#define GMONTYPE_RAWARCS 2
/*
* The GMONTYPE_ARCS_ORDERS gmon_data.type is for the raw arcs with a call
* order field. The order is the order is a sequence number for the order each
* call site was executed. Raw_order values start at 1 not zero. Other than
* the raw_order field this is the same information as in the struct rawarc.
*/
#define GMONTYPE_ARCS_ORDERS 3
struct rawarc_order {
unsigned long raw_frompc;
unsigned long raw_selfpc;
unsigned long raw_count;
unsigned long raw_order;
};
/*
* The GMONTYPE_RLD_STATE gmon_data.type is for the rld_load()'ed state of the
* program.
* The informations starts with an unsigned long with the count of states:
* rld_nloaded_states
* Then each state follows in the file. The state is made up of
* header_addr (where rld loaded this set of objects)
* nobjectfiles (the number of objects in this set)
* offsets into the string table (one for each object in the set)
* nbytes of string table
* the file name strings null terminated.
*/
#define GMONTYPE_RLD_STATE 4
#endif !ASSEMBLER
/*
* general rounding functions.
*/
#define ROUNDDOWN(x,y) (((x)/(y))*(y))
#define ROUNDUP(x,y) ((((x)+(y)-1)/(y))*(y))