Import Traceline, a Windows performance trace event logger.

Import traceline, a Windows performance trace event logger.

This imports both the hook / event logger (traceline), and the SVG based UI for displaying trace results as a timeline (svgui).

[Dean McNamee, 2009-02-19 14:35:53]

About time I committed this.  This should be pretty straight forward, I made
sure everything has copyrights, etc.  This makes a copy of some sidestep files
from the sandbox/ src, but it's the easiest thing to do to keep the build
simple.

[M-A Ruel, 2009-02-19 20:48:16]

There's way too much code to do a complete review. You can check-in with most of
the changes. I have 2 question but you don't need to block on that, just reply
"why" :)

http://codereview.chromium.org/20494/diff/3021/3025
File tools/traceline/svgui/traceline.js (right):

http://codereview.chromium.org/20494/diff/3021/3025#newcode42
Line 42: // a TLThread respresents information about a thread in the traceline
data.
represents

http://codereview.chromium.org/20494/diff/3021/3025#newcode300
Line 300: // scrolled we rerending the scene.  This means that the SVG element
is never
rerender

http://codereview.chromium.org/20494/diff/3021/34
File tools/traceline/traceline/assembler.h (right):

http://codereview.chromium.org/20494/diff/3021/34#newcode544
Line 544: // put a unique cpu identifier into eax, uses sidt to fingerprint
cores.
Puts and other *s.

http://codereview.chromium.org/20494/diff/3021/42
File tools/traceline/traceline/main.cc (right):

http://codereview.chromium.org/20494/diff/3021/42#newcode108
Line 108: static void* kPlaygroundAddr = reinterpret_cast<void*>(0x66660000);
I don't understand why it is static.

http://codereview.chromium.org/20494/diff/3021/43
File tools/traceline/traceline/rdtsc.h (right):

http://codereview.chromium.org/20494/diff/3021/43#newcode17
Line 17: RDTSCNormalizer() { }
what for?

http://codereview.chromium.org/20494/diff/3021/49
File tools/traceline/traceline/sidestep/mini_disassembler.h (right):

http://codereview.chromium.org/20494/diff/3021/49#newcode6
Line 6: #ifndef SANDBOX_SRC_SIDESTEP_MINI_DISASSEMBLER_H__
update the define

http://codereview.chromium.org/20494/diff/3021/50
File tools/traceline/traceline/sidestep/mini_disassembler_types.h (right):

http://codereview.chromium.org/20494/diff/3021/50#newcode8
Line 8: #ifndef SANDBOX_SRC_SIDESTEP_MINI_DISASSEMBLER_TYPES_H__
update define

[Dean McNamee, 2009-02-20 11:24:19]

Hey MA, thanks for the good review.  I think I fixed all of the typos, include
guards, etc.

As for your two questions.  RDTSCNormalizer was from when I was trying to use
the cpu timestamp counters, and they needed to be normalized per-cpu since the
counters are not synchronized.  I couldn't get this approach to work well at
all, and eventually moved to the simpler but slower QueryPerformanceCounter
kernel call.  I'd like to revisit the idea again someday, which is why I left
the scaffolding in place.

The static on kPlaygroundAddr is sort of a note to myself that someday this
should be a variable so I can handle the case when this address is already used.

Committing, thanks!

On 2009/02/19 20:48:16, M-A wrote:
> There's way too much code to do a complete review. You can check-in with most
of
> the changes. I have 2 question but you don't need to block on that, just reply
> "why" :)
> 
> http://codereview.chromium.org/20494/diff/3021/3025
> File tools/traceline/svgui/traceline.js (right):
> 
> http://codereview.chromium.org/20494/diff/3021/3025#newcode42
> Line 42: // a TLThread respresents information about a thread in the traceline
> data.
> represents
> 
> http://codereview.chromium.org/20494/diff/3021/3025#newcode300
> Line 300: // scrolled we rerending the scene.  This means that the SVG element
> is never
> rerender
> 
> http://codereview.chromium.org/20494/diff/3021/34
> File tools/traceline/traceline/assembler.h (right):
> 
> http://codereview.chromium.org/20494/diff/3021/34#newcode544
> Line 544: // put a unique cpu identifier into eax, uses sidt to fingerprint
> cores.
> Puts and other *s.
> 
> http://codereview.chromium.org/20494/diff/3021/42
> File tools/traceline/traceline/main.cc (right):
> 
> http://codereview.chromium.org/20494/diff/3021/42#newcode108
> Line 108: static void* kPlaygroundAddr = reinterpret_cast<void*>(0x66660000);
> I don't understand why it is static.
> 
> http://codereview.chromium.org/20494/diff/3021/43
> File tools/traceline/traceline/rdtsc.h (right):
> 
> http://codereview.chromium.org/20494/diff/3021/43#newcode17
> Line 17: RDTSCNormalizer() { }
> what for?
> 
> http://codereview.chromium.org/20494/diff/3021/49
> File tools/traceline/traceline/sidestep/mini_disassembler.h (right):
> 
> http://codereview.chromium.org/20494/diff/3021/49#newcode6
> Line 6: #ifndef SANDBOX_SRC_SIDESTEP_MINI_DISASSEMBLER_H__
> update the define
> 
> http://codereview.chromium.org/20494/diff/3021/50
> File tools/traceline/traceline/sidestep/mini_disassembler_types.h (right):
> 
> http://codereview.chromium.org/20494/diff/3021/50#newcode8
> Line 8: #ifndef SANDBOX_SRC_SIDESTEP_MINI_DISASSEMBLER_TYPES_H__
> update define