This is my patch:

[ 1005891 ] support --with-tsc on PPC

plus a trivial change to settscdump's docstring and a Misc/NEWS entry.
This commit is contained in:
Michael W. Hudson 2004-08-12 18:19:17 +00:00
parent d459f536c5
commit 800ba2375a
4 changed files with 62 additions and 2 deletions

View file

@ -57,6 +57,9 @@ Tools/Demos
Build
-----
- The --with-tsc flag to configure to enable VM profiling with the
processor's timestamp counter now works on PPC platforms.
C API
-----

View file

@ -227,3 +227,28 @@ When this symbol is defined, the ceval mainloop and helper functions
count the number of function calls made. It keeps detailed statistics
about what kind of object was called and whether the call hit any of
the special fast paths in the code.
---------------------------------------------------------------------------
WITH_TSC introduced for Python 2.4
Super-lowlevel profiling of the interpreter. When enabled, the sys
module grows a new function:
settscdump(bool)
If true, tell the Python interpreter to dump VM measurements to
stderr. If false, turn off dump. The measurements are based on the
processor's time-stamp counter.
This build option requires a small amount of platform specific code.
Currently this code is present for linux/x86 and any PowerPC platform
that uses GCC (i.e. OS X and linux/ppc).
On the PowerPC the rate at which the time base register is incremented
is not defined by the architecture specification, so you'll need to
find the manual for your specific processor. For the 750CX, 750CXe,
750FX (all sold as the G3) we find:
The time base counter is clocked at a frequency that is
one-fourth that of the bus clock.
This build is enabled by the --with-tsc flag to configure.

View file

@ -17,10 +17,38 @@
#include <ctype.h>
#ifdef WITH_TSC
#include <asm/msr.h>
typedef unsigned long long uint64;
#if defined(__ppc__) /* <- Don't know if this is the correct symbol; this
section should work for GCC on any PowerPC platform,
irrespective of OS. POWER? Who knows :-) */
#define rdtscll(var) ppc_getcounter(&var)
static void
ppc_getcounter(uint64 *v)
{
register unsigned long tbu, tb, tbu2;
loop:
asm volatile ("mftbu %0" : "=r" (tbu) );
asm volatile ("mftb %0" : "=r" (tb) );
asm volatile ("mftbu %0" : "=r" (tbu2));
if (__builtin_expect(tbu != tbu2, 0)) goto loop;
/* The slightly peculiar way of writing the next lines is
compiled better by GCC than any other way I tried. */
((long*)(v))[0] = tbu;
((long*)(v))[1] = tb;
}
#else /* this section is for linux/x86 */
#include <asm/msr.h>
#endif
void dump_tsc(int opcode, int ticked, uint64 inst0, uint64 inst1,
uint64 loop0, uint64 loop1, uint64 intr0, uint64 intr1)
{
@ -34,6 +62,7 @@ void dump_tsc(int opcode, int ticked, uint64 inst0, uint64 inst1,
fprintf(stderr, "opcode=%03d t=%d inst=%06lld loop=%06lld\n",
opcode, ticked, inst, loop);
}
#endif
/* Turn this on if your compiler chokes on the big switch: */
@ -545,6 +574,9 @@ PyEval_EvalFrame(PyFrameObject *f)
rdtscll(inst1);
rdtscll(loop0);
rdtscll(loop1);
/* shut up the compiler */
opcode = 0;
#endif
/* Code access macros */

View file

@ -465,7 +465,7 @@ PyDoc_STRVAR(settscdump_doc,
\n\
If true, tell the Python interpreter to dump VM measurements to\n\
stderr. If false, turn off dump. The measurements are based on the\n\
Pentium time-stamp counter."
processor's time-stamp counter."
);
#endif /* TSC */