d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
3753c2c3f1
freebsd-dev/sys/i386/isa/prof_machdep.c

364 lines
10 KiB
C
Raw Normal View History

Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
/*-
* Copyright (c) 1996 Bruce D. Evans.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
Added a $Id$ keyword. Bruce still needs to put a copyright notice on this file.
1996-04-08 16:41:06 +00:00
*
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
Added a $Id$ keyword. Bruce still needs to put a copyright notice on this file.
1996-04-08 16:41:06 +00:00
*/
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
Use __FBSDID().
2003-06-02 16:32:55 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Fix low res profiling kernel build. Move two defines to collapse the #ifdef GUPROF case.
2005-05-19 05:22:52 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <machine/asmacros.h>
#include <machine/timerreg.h>
Fix LINT build, original breakage was rev 1.23. There are 2 definitions of MCOUNT to have a C version and an asm version with the same name and not have LOCORE ifdefs to distinguish them. <machine/profile.h> provides a C version and <machine/asmacros.h> provides an assembler version. Discussed with: bde
2005-05-20 17:16:24 +00:00
/*
* There are 2 definitions of MCOUNT to have a C version and an asm version
* with the same name and not have LOCORE #ifdefs to distinguish them.
* <machine/profile.h> provides a C version, and <machine/asmacros.h>
* provides an asm version. To avoid conflicts, #undef the asm version.
*/
#undef MCOUNT
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
#ifdef GUPROF
#include "opt_i586_guprof.h"
#include "opt_perfmon.h"
#include <sys/gmon.h>
Added a sysctl (machdep.cputime_clock) to select the clock used by "high resolution" profiling. The available clocks are: - the i8254 clock - on non-SMP i586's and i686's: the TSC - on systems with I586_PMC_GUPROF configured, and PERFMON configured and available: all the performance counters. This is unfinshed (there are problems with locking out the PERFMON device driver, and with losing calibration after switching the clock), but better than static configuration or writing to kmem. Changed ifdefs to avoid generating code for non-working option combinations.
1997-11-24 18:16:23 +00:00
#include <sys/kernel.h>
#include <sys/sysctl.h>
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
#include <machine/clock.h>
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
#include <machine/perfmon.h>
#include <machine/profile.h>
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
#define CPUTIME_CLOCK_UNINITIALIZED 0
#define CPUTIME_CLOCK_I8254 1
Rename "i586_ctr" to "tsc" (both upper and lower case instances). Fix a couple of printfs too. Warning: This changes the names of a couple of kernel options!
1997-12-26 20:42:37 +00:00
#define CPUTIME_CLOCK_TSC 2
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
#define CPUTIME_CLOCK_I586_PMC 3
#define CPUTIME_CLOCK_I8254_SHIFT 7
int cputime_bias = 1; /* initialize for locality of reference */
static int cputime_clock = CPUTIME_CLOCK_UNINITIALIZED;
#ifdef I586_PMC_GUPROF
static u_int cputime_clock_pmc_conf = I586_PMC_GUPROF;
static int cputime_clock_pmc_init;
static struct gmonparam saved_gmp;
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
#endif
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
#endif /* GUPROF */
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
netchild's mega-patch to isolate compiler dependencies into a central place. This moves the dependency on GCC's and other compiler's features into the central sys/cdefs.h file, while the individual source files can then refer to #ifdef __COMPILER_FEATURE_FOO where they by now used to refer to #if __GNUC__ > 3.1415 && __BARC__ <= 42. By now, GCC and ICC (the Intel compiler) have been actively tested on IA32 platforms by netchild. Extension to other compilers is supposed to be possible, of course. Submitted by: netchild Reviewed by: various developers on arch@, some time ago
2005-03-02 21:33:29 +00:00
#ifdef __GNUCLIKE_ASM
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
__asm(" \n\
GM_STATE = 0 \n\
GMON_PROF_OFF = 3 \n\
\n\
.text \n\
Use .p2align to ensure consistant a.out/elf alignment. I'd have used SUPERALIGN_TEXT, but this is inline assembler and after cpp has run. Inspired by bde's comments on linux_locore.s.
1999-08-25 23:50:03 +00:00
.p2align 4,0x90 \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
.globl __mcount \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
.type __mcount,@function \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
__mcount: \n\
# \n\
# Check that we are profiling. Do it early for speed. \n\
# \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
cmpl $GMON_PROF_OFF," __XSTRING(CNAME(_gmonparam)) "+GM_STATE \n\
je .mcount_exit \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
# \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
# __mcount is the same as [.]mcount except the caller \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
# hasn't changed the stack except to call here, so the \n\
# caller's raddr is above our raddr. \n\
# \n\
movl 4(%esp),%edx \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
jmp .got_frompc \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
\n\
Use .p2align to ensure consistant a.out/elf alignment. I'd have used SUPERALIGN_TEXT, but this is inline assembler and after cpp has run. Inspired by bde's comments on linux_locore.s.
1999-08-25 23:50:03 +00:00
.p2align 4,0x90 \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
.globl " __XSTRING(HIDENAME(mcount)) " \n\
" __XSTRING(HIDENAME(mcount)) ": \n\
Quick fix for high resolution kernel profiling on i386's. Use -finstrument-functions instead of -mprofiler-epilogue. The former works essentially the same as the latter but has a higher overhead (about 22 more bytes per function for passing unused args to the profiling functions). Removed all traces of the IDENT Makefile variable, which had been reduced to just a place for holding profiling's contribution to CFLAGS (the IDENT that gives the kernel identity was renamed to KERN_IDENT).
2002-07-13 22:28:34 +00:00
.globl __cyg_profile_func_enter \n\
__cyg_profile_func_enter: \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
cmpl $GMON_PROF_OFF," __XSTRING(CNAME(_gmonparam)) "+GM_STATE \n\
je .mcount_exit \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
# \n\
# The caller's stack frame has already been built, so \n\
# %ebp is the caller's frame pointer. The caller's \n\
# raddr is in the caller's frame following the caller's \n\
# caller's frame pointer. \n\
# \n\
movl 4(%ebp),%edx \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
.got_frompc: \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
# \n\
# Our raddr is the caller's pc. \n\
# \n\
movl (%esp),%eax \n\
\n\
pushfl \n\
pushl %eax \n\
pushl %edx \n\
cli \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
call " __XSTRING(CNAME(mcount)) " \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
addl $8,%esp \n\
popfl \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
.mcount_exit: \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
ret \n\
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
");
netchild's mega-patch to isolate compiler dependencies into a central place. This moves the dependency on GCC's and other compiler's features into the central sys/cdefs.h file, while the individual source files can then refer to #ifdef __COMPILER_FEATURE_FOO where they by now used to refer to #if __GNUC__ > 3.1415 && __BARC__ <= 42. By now, GCC and ICC (the Intel compiler) have been actively tested on IA32 platforms by netchild. Extension to other compilers is supposed to be possible, of course. Submitted by: netchild Reviewed by: various developers on arch@, some time ago
2005-03-02 21:33:29 +00:00
#else /* !__GNUCLIKE_ASM */
#error this file needs to be ported to your compiler
#endif /* __GNUCLIKE_ASM */
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
#ifdef GUPROF
/*
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
* [.]mexitcount saves the return register(s), loads selfpc and calls
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
* mexitcount(selfpc) to do the work. Someday it should be in a machine
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
* dependent file together with cputime(), __mcount and [.]mcount. cputime()
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
* can't just be put in machdep.c because it has to be compiled without -pg.
*/
netchild's mega-patch to isolate compiler dependencies into a central place. This moves the dependency on GCC's and other compiler's features into the central sys/cdefs.h file, while the individual source files can then refer to #ifdef __COMPILER_FEATURE_FOO where they by now used to refer to #if __GNUC__ > 3.1415 && __BARC__ <= 42. By now, GCC and ICC (the Intel compiler) have been actively tested on IA32 platforms by netchild. Extension to other compilers is supposed to be possible, of course. Submitted by: netchild Reviewed by: various developers on arch@, some time ago
2005-03-02 21:33:29 +00:00
#ifdef __GNUCLIKE_ASM
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
__asm(" \n\
.text \n\
# \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
# Dummy label to be seen when gprof -u hides [.]mexitcount. \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
# \n\
Use .p2align to ensure consistant a.out/elf alignment. I'd have used SUPERALIGN_TEXT, but this is inline assembler and after cpp has run. Inspired by bde's comments on linux_locore.s.
1999-08-25 23:50:03 +00:00
.p2align 4,0x90 \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
.globl __mexitcount \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
.type __mexitcount,@function \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
__mexitcount: \n\
nop \n\
\n\
GMON_PROF_HIRES = 4 \n\
\n\
Use .p2align to ensure consistant a.out/elf alignment. I'd have used SUPERALIGN_TEXT, but this is inline assembler and after cpp has run. Inspired by bde's comments on linux_locore.s.
1999-08-25 23:50:03 +00:00
.p2align 4,0x90 \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
.globl " __XSTRING(HIDENAME(mexitcount)) " \n\
" __XSTRING(HIDENAME(mexitcount)) ": \n\
Quick fix for high resolution kernel profiling on i386's. Use -finstrument-functions instead of -mprofiler-epilogue. The former works essentially the same as the latter but has a higher overhead (about 22 more bytes per function for passing unused args to the profiling functions). Removed all traces of the IDENT Makefile variable, which had been reduced to just a place for holding profiling's contribution to CFLAGS (the IDENT that gives the kernel identity was renamed to KERN_IDENT).
2002-07-13 22:28:34 +00:00
.globl __cyg_profile_func_exit \n\
__cyg_profile_func_exit: \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
cmpl $GMON_PROF_HIRES," __XSTRING(CNAME(_gmonparam)) "+GM_STATE \n\
jne .mexitcount_exit \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
pushl %edx \n\
pushl %eax \n\
movl 8(%esp),%eax \n\
pushfl \n\
pushl %eax \n\
cli \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
call " __XSTRING(CNAME(mexitcount)) " \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
addl $4,%esp \n\
popfl \n\
popl %eax \n\
popl %edx \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
.mexitcount_exit: \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
ret \n\
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
");
netchild's mega-patch to isolate compiler dependencies into a central place. This moves the dependency on GCC's and other compiler's features into the central sys/cdefs.h file, while the individual source files can then refer to #ifdef __COMPILER_FEATURE_FOO where they by now used to refer to #if __GNUC__ > 3.1415 && __BARC__ <= 42. By now, GCC and ICC (the Intel compiler) have been actively tested on IA32 platforms by netchild. Extension to other compilers is supposed to be possible, of course. Submitted by: netchild Reviewed by: various developers on arch@, some time ago
2005-03-02 21:33:29 +00:00
#endif /* __GNUCLIKE_ASM */
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
/*
* Return the time elapsed since the last call. The units are machine-
* dependent.
*/
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
int
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
cputime()
{
u_int count;
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
int delta;
Ifdefed the declarations of conditionally used variables.
1998-12-14 18:21:34 +00:00
#if (defined(I586_CPU) || defined(I686_CPU)) && !defined(SMP) && \
defined(PERFMON) && defined(I586_PMC_GUPROF)
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
u_quad_t event_count;
#endif
u_char high, low;
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
static u_int prev_count;
Added a sysctl (machdep.cputime_clock) to select the clock used by "high resolution" profiling. The available clocks are: - the i8254 clock - on non-SMP i586's and i686's: the TSC - on systems with I586_PMC_GUPROF configured, and PERFMON configured and available: all the performance counters. This is unfinshed (there are problems with locking out the PERFMON device driver, and with losing calibration after switching the clock), but better than static configuration or writing to kmem. Changed ifdefs to avoid generating code for non-working option combinations.
1997-11-24 18:16:23 +00:00
#if (defined(I586_CPU) || defined(I686_CPU)) && !defined(SMP)
Rename "i586_ctr" to "tsc" (both upper and lower case instances). Fix a couple of printfs too. Warning: This changes the names of a couple of kernel options!
1997-12-26 20:42:37 +00:00
if (cputime_clock == CPUTIME_CLOCK_TSC) {
Quick fix for overflow when tsc_freq >= 2^31. "int profrate" in struct gmon and struct gmonhdr was originally just to represent the kernel (profiling) clock frequency and it remains poorly suited to representing the frequencies of fast counters like the TSC. It broke a year or two ago. This quick fix keeps it working for another year or month or two until TSC frequencies can exceed 2^32, by dividing the frequency by 2. Dividing the frequency by 4 would work for a little longer but would lose a little too much precision.
2004-05-26 09:43:38 +00:00
/*
* Scale the TSC a little to make cputime()'s frequency
* fit in an int, assuming that the TSC frequency fits
* in a u_int. Use a fixed scale since dynamic scaling
* would be slower and we can't really use the low bit
* of precision.
*/
count = (u_int)rdtsc() & ~1u;
delta = (int)(count - prev_count) >> 1;
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
prev_count = count;
return (delta);
}
Added a sysctl (machdep.cputime_clock) to select the clock used by "high resolution" profiling. The available clocks are: - the i8254 clock - on non-SMP i586's and i686's: the TSC - on systems with I586_PMC_GUPROF configured, and PERFMON configured and available: all the performance counters. This is unfinshed (there are problems with locking out the PERFMON device driver, and with losing calibration after switching the clock), but better than static configuration or writing to kmem. Changed ifdefs to avoid generating code for non-working option combinations.
1997-11-24 18:16:23 +00:00
#if defined(PERFMON) && defined(I586_PMC_GUPROF)
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
if (cputime_clock == CPUTIME_CLOCK_I586_PMC) {
/*
* XXX permon_read() should be inlined so that the
* perfmon module doesn't need to be compiled with
* profiling disabled and so that it is fast.
*/
perfmon_read(0, &event_count);
count = (u_int)event_count;
delta = (int)(count - prev_count);
prev_count = count;
return (delta);
}
Added a sysctl (machdep.cputime_clock) to select the clock used by "high resolution" profiling. The available clocks are: - the i8254 clock - on non-SMP i586's and i686's: the TSC - on systems with I586_PMC_GUPROF configured, and PERFMON configured and available: all the performance counters. This is unfinshed (there are problems with locking out the PERFMON device driver, and with losing calibration after switching the clock), but better than static configuration or writing to kmem. Changed ifdefs to avoid generating code for non-working option combinations.
1997-11-24 18:16:23 +00:00
#endif /* PERFMON && I586_PMC_GUPROF */
#endif /* (I586_CPU || I686_CPU) && !SMP */
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
/*
* Read the current value of the 8254 timer counter 0.
*/
outb(TIMER_MODE, TIMER_SEL0 | TIMER_LATCH);
low = inb(TIMER_CNTR0);
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
high = inb(TIMER_CNTR0);
count = ((high << 8) | low) << CPUTIME_CLOCK_I8254_SHIFT;
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
/*
* The timer counts down from TIMER_CNTR0_MAX to 0 and then resets.
* While profiling is enabled, this routine is called at least twice
* per timer reset (for mcounting and mexitcounting hardclock()),
* so at most one reset has occurred since the last call, and one
* has occurred iff the current count is larger than the previous
* count. This allows counter underflow to be detected faster
* than in microtime().
*/
delta = prev_count - count;
prev_count = count;
if ((int) delta <= 0)
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
return (delta + (timer0_max_count << CPUTIME_CLOCK_I8254_SHIFT));
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
return (delta);
}
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
Added a sysctl (machdep.cputime_clock) to select the clock used by "high resolution" profiling. The available clocks are: - the i8254 clock - on non-SMP i586's and i686's: the TSC - on systems with I586_PMC_GUPROF configured, and PERFMON configured and available: all the performance counters. This is unfinshed (there are problems with locking out the PERFMON device driver, and with losing calibration after switching the clock), but better than static configuration or writing to kmem. Changed ifdefs to avoid generating code for non-working option combinations.
1997-11-24 18:16:23 +00:00
static int
Previous commit changing SYSCTL_HANDLER_ARGS violated KNF. Pointed out by: bde
2000-07-04 11:25:35 +00:00
sysctl_machdep_cputime_clock(SYSCTL_HANDLER_ARGS)
Added a sysctl (machdep.cputime_clock) to select the clock used by "high resolution" profiling. The available clocks are: - the i8254 clock - on non-SMP i586's and i686's: the TSC - on systems with I586_PMC_GUPROF configured, and PERFMON configured and available: all the performance counters. This is unfinshed (there are problems with locking out the PERFMON device driver, and with losing calibration after switching the clock), but better than static configuration or writing to kmem. Changed ifdefs to avoid generating code for non-working option combinations.
1997-11-24 18:16:23 +00:00
{
int clock;
int error;
Ifdefed the declarations of conditionally used variables.
1998-12-14 18:21:34 +00:00
#if defined(PERFMON) && defined(I586_PMC_GUPROF)
int event;
Added a sysctl (machdep.cputime_clock) to select the clock used by "high resolution" profiling. The available clocks are: - the i8254 clock - on non-SMP i586's and i686's: the TSC - on systems with I586_PMC_GUPROF configured, and PERFMON configured and available: all the performance counters. This is unfinshed (there are problems with locking out the PERFMON device driver, and with losing calibration after switching the clock), but better than static configuration or writing to kmem. Changed ifdefs to avoid generating code for non-working option combinations.
1997-11-24 18:16:23 +00:00
struct pmc pmc;
Ifdefed the declarations of conditionally used variables.
1998-12-14 18:21:34 +00:00
#endif
Added a sysctl (machdep.cputime_clock) to select the clock used by "high resolution" profiling. The available clocks are: - the i8254 clock - on non-SMP i586's and i686's: the TSC - on systems with I586_PMC_GUPROF configured, and PERFMON configured and available: all the performance counters. This is unfinshed (there are problems with locking out the PERFMON device driver, and with losing calibration after switching the clock), but better than static configuration or writing to kmem. Changed ifdefs to avoid generating code for non-working option combinations.
1997-11-24 18:16:23 +00:00
clock = cputime_clock;
#if defined(PERFMON) && defined(I586_PMC_GUPROF)
if (clock == CPUTIME_CLOCK_I586_PMC) {
pmc.pmc_val = cputime_clock_pmc_conf;
clock += pmc.pmc_event;
}
#endif
error = sysctl_handle_opaque(oidp, &clock, sizeof clock, req);
if (error == 0 && req->newptr != NULL) {
#if defined(PERFMON) && defined(I586_PMC_GUPROF)
if (clock >= CPUTIME_CLOCK_I586_PMC) {
event = clock - CPUTIME_CLOCK_I586_PMC;
if (event >= 256)
return (EINVAL);
pmc.pmc_num = 0;
pmc.pmc_event = event;
pmc.pmc_unit = 0;
pmc.pmc_flags = PMCF_E | PMCF_OS | PMCF_USR;
pmc.pmc_mask = 0;
cputime_clock_pmc_conf = pmc.pmc_val;
cputime_clock = CPUTIME_CLOCK_I586_PMC;
} else
#endif
{
if (clock < 0 || clock >= CPUTIME_CLOCK_I586_PMC)
return (EINVAL);
cputime_clock = clock;
}
}
return (error);
}
SYSCTL_PROC(_machdep, OID_AUTO, cputime_clock, CTLTYPE_INT | CTLFLAG_RW,
0, sizeof(u_int), sysctl_machdep_cputime_clock, "I", "");
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
/*
* The start and stop routines need not be here since we turn off profiling
* before calling them. They are here for convenience.
*/
void
startguprof(gp)
struct gmonparam *gp;
{
if (cputime_clock == CPUTIME_CLOCK_UNINITIALIZED) {
cputime_clock = CPUTIME_CLOCK_I8254;
Added a sysctl (machdep.cputime_clock) to select the clock used by "high resolution" profiling. The available clocks are: - the i8254 clock - on non-SMP i586's and i686's: the TSC - on systems with I586_PMC_GUPROF configured, and PERFMON configured and available: all the performance counters. This is unfinshed (there are problems with locking out the PERFMON device driver, and with losing calibration after switching the clock), but better than static configuration or writing to kmem. Changed ifdefs to avoid generating code for non-working option combinations.
1997-11-24 18:16:23 +00:00
#if (defined(I586_CPU) || defined(I686_CPU)) && !defined(SMP)
Rename "i586_ctr" to "tsc" (both upper and lower case instances). Fix a couple of printfs too. Warning: This changes the names of a couple of kernel options!
1997-12-26 20:42:37 +00:00
if (tsc_freq != 0)
cputime_clock = CPUTIME_CLOCK_TSC;
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
#endif
}
gp->profrate = timer_freq << CPUTIME_CLOCK_I8254_SHIFT;
Added a sysctl (machdep.cputime_clock) to select the clock used by "high resolution" profiling. The available clocks are: - the i8254 clock - on non-SMP i586's and i686's: the TSC - on systems with I586_PMC_GUPROF configured, and PERFMON configured and available: all the performance counters. This is unfinshed (there are problems with locking out the PERFMON device driver, and with losing calibration after switching the clock), but better than static configuration or writing to kmem. Changed ifdefs to avoid generating code for non-working option combinations.
1997-11-24 18:16:23 +00:00
#if (defined(I586_CPU) || defined(I686_CPU)) && !defined(SMP)
Rename "i586_ctr" to "tsc" (both upper and lower case instances). Fix a couple of printfs too. Warning: This changes the names of a couple of kernel options!
1997-12-26 20:42:37 +00:00
if (cputime_clock == CPUTIME_CLOCK_TSC)
Quick fix for overflow when tsc_freq >= 2^31. "int profrate" in struct gmon and struct gmonhdr was originally just to represent the kernel (profiling) clock frequency and it remains poorly suited to representing the frequencies of fast counters like the TSC. It broke a year or two ago. This quick fix keeps it working for another year or month or two until TSC frequencies can exceed 2^32, by dividing the frequency by 2. Dividing the frequency by 4 would work for a little longer but would lose a little too much precision.
2004-05-26 09:43:38 +00:00
gp->profrate = tsc_freq >> 1;
Added a sysctl (machdep.cputime_clock) to select the clock used by "high resolution" profiling. The available clocks are: - the i8254 clock - on non-SMP i586's and i686's: the TSC - on systems with I586_PMC_GUPROF configured, and PERFMON configured and available: all the performance counters. This is unfinshed (there are problems with locking out the PERFMON device driver, and with losing calibration after switching the clock), but better than static configuration or writing to kmem. Changed ifdefs to avoid generating code for non-working option combinations.
1997-11-24 18:16:23 +00:00
#if defined(PERFMON) && defined(I586_PMC_GUPROF)
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
else if (cputime_clock == CPUTIME_CLOCK_I586_PMC) {
if (perfmon_avail() &&
perfmon_setup(0, cputime_clock_pmc_conf) == 0) {
if (perfmon_start(0) != 0)
perfmon_fini(0);
else {
/* XXX 1 event == 1 us. */
gp->profrate = 1000000;
saved_gmp = *gp;
/* Zap overheads. They are invalid. */
gp->cputime_overhead = 0;
gp->mcount_overhead = 0;
gp->mcount_post_overhead = 0;
gp->mcount_pre_overhead = 0;
gp->mexitcount_overhead = 0;
gp->mexitcount_post_overhead = 0;
gp->mexitcount_pre_overhead = 0;
cputime_clock_pmc_init = TRUE;
}
}
}
Added a sysctl (machdep.cputime_clock) to select the clock used by "high resolution" profiling. The available clocks are: - the i8254 clock - on non-SMP i586's and i686's: the TSC - on systems with I586_PMC_GUPROF configured, and PERFMON configured and available: all the performance counters. This is unfinshed (there are problems with locking out the PERFMON device driver, and with losing calibration after switching the clock), but better than static configuration or writing to kmem. Changed ifdefs to avoid generating code for non-working option combinations.
1997-11-24 18:16:23 +00:00
#endif /* PERFMON && I586_PMC_GUPROF */
#endif /* (I586_CPU || I686_CPU) && !SMP */
Improved non-statistical (GUPROF) profiling: - use a more accurate and more efficient method of compensating for overheads. The old method counted too much time against leaf functions. - normally use the Pentium timestamp counter if available. On Pentiums, the times are now accurate to within a couple of cpu clock cycles per function call in the (unlikely) event that there are no cache misses in or caused by the profiling code. - optionally use an arbitrary Pentium event counter if available. - optionally regress to using the i8254 counter. - scaled the i8254 counter by a factor of 128. Now the i8254 counters overflow slightly faster than the TSC counters for a 150MHz Pentium :-) (after about 16 seconds). This is to avoid fractional overheads. files.i386: permon.c temporarily has to be classified as a profiling-routine because a couple of functions in it may be called from profiling code. options.i386: - I586_CTR_GUPROF is currently unused (oops). - I586_PMC_GUPROF should be something like 0x70000 to enable (but not use unless prof_machdep.c is changed) support for Pentium event counters. 7 is a control mode and the counter number 0 is somewhere in the 0000 bits (see perfmon.h for the encoding). profile.h: - added declarations. - cleaned up separation of user mode declarations. prof_machdep.c: Mostly clock-select changes. The default clock can be changed by editing kmem. There should be a sysctl for this. subr_prof.c: - added copyright. - calibrate overheads for the new method. - documented new method. - fixed races and and machine dependencies in start/stop code. mcount.c: Use the new overhead compensation method. gmon.h: - changed GPROF4 counter type from unsigned to int. Oops, this should be machine-dependent and/or int32_t. - reorganized overhead counters. Submitted by: Pentium event counter changes mostly by wollman
1996-10-17 19:32:31 +00:00
cputime_bias = 0;
cputime();
}
void
stopguprof(gp)
struct gmonparam *gp;
{
#if defined(PERFMON) && defined(I586_PMC_GUPROF)
if (cputime_clock_pmc_init) {
*gp = saved_gmp;
perfmon_fini(0);
cputime_clock_pmc_init = FALSE;
}
#endif
}
#else /* !GUPROF */
netchild's mega-patch to isolate compiler dependencies into a central place. This moves the dependency on GCC's and other compiler's features into the central sys/cdefs.h file, while the individual source files can then refer to #ifdef __COMPILER_FEATURE_FOO where they by now used to refer to #if __GNUC__ > 3.1415 && __BARC__ <= 42. By now, GCC and ICC (the Intel compiler) have been actively tested on IA32 platforms by netchild. Extension to other compilers is supposed to be possible, of course. Submitted by: netchild Reviewed by: various developers on arch@, some time ago
2005-03-02 21:33:29 +00:00
#ifdef __GNUCLIKE_ASM
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
__asm(" \n\
.text \n\
Use .p2align to ensure consistant a.out/elf alignment. I'd have used SUPERALIGN_TEXT, but this is inline assembler and after cpp has run. Inspired by bde's comments on linux_locore.s.
1999-08-25 23:50:03 +00:00
.p2align 4,0x90 \n\
Fixed profiling of elf kernels. Made high resolution profiling compile for elf kernels (it is broken for all kernels due to lack of egcs support). Renaming of many assembler labels is avoided by declaring by declaring the labels that need to be visible to gprof as having type "function" and depending on the elf version of gprof being zealous about discarding the others. A few type declarations are still missing, mainly for SMP. PR: 9413 Submitted by: Assar Westerlund <assar@sics.se> (initial parts)
1999-05-06 09:44:57 +00:00
.globl " __XSTRING(HIDENAME(mexitcount)) " \n\
" __XSTRING(HIDENAME(mexitcount)) ": \n\
Support compiling with `gcc -pedantic' (don't use hard newlines in (asm) string constants).
1998-04-19 15:41:06 +00:00
ret \n\
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
");
netchild's mega-patch to isolate compiler dependencies into a central place. This moves the dependency on GCC's and other compiler's features into the central sys/cdefs.h file, while the individual source files can then refer to #ifdef __COMPILER_FEATURE_FOO where they by now used to refer to #if __GNUC__ > 3.1415 && __BARC__ <= 42. By now, GCC and ICC (the Intel compiler) have been actively tested on IA32 platforms by netchild. Extension to other compilers is supposed to be possible, of course. Submitted by: netchild Reviewed by: various developers on arch@, some time ago
2005-03-02 21:33:29 +00:00
#endif /* __GNUCLIKE_ASM */
Implemented non-statistical kernel profiling. This is based on looking at a high resolution clock for each of the following events: function call, function return, interrupt entry, interrupt exit, and interesting branches. The differences between the times of these events are added at appropriate places in a ordinary histogram (as if very fast statistical profiling sampled the pc at those places) so that ordinary gprof can be used to analyze the times. gmon.h: Histogram counters need to be 4 bytes for microsecond resolutions. They will need to be larger for the 586 clock. The comments were vax-centric and wrong even on vaxes. Does anyone disagree? gprof4.c: The standard gprof should support counters of all integral sizes and the size of the counter should be in the gmon header. This hack will do until then. (Use gprof4 -u to examine the results of non-statistical profiling.) config/*: Non-statistical profiling is configured with `config -pp'. `config -p' still gives ordinary profiling. kgmon/*: Non-statistical profiling is enabled with `kgmon -B'. `kgmon -b' still enables ordinary profiling (and distables non-statistical profiling) if non-statistical profiling is configured.
1995-12-29 15:30:05 +00:00
#endif /* GUPROF */
Reference in New Issue Copy Permalink