The problem appeared due to a fact that stubs doesn't create a stack
frame, reusing the stack frame of the caller function. When building
stack traces, the current function is retrieved from PC, and its
callees are retrieved by traversing the stack backwards. Thus, for
stubs, the stub itself was discovered via PC, and then stub's caller's
caller was retrieved from stack.
To fix this problem, a pointer to JSFunction object is now captured
from the topmost stack frame, and is saved into stack trace log
record. Then a simple heuristics is applied whether a referred
function should be added to decoded stack, or not, to avoid reporting
the same function twice (from PC and from the pointer.)
BUG=553
TEST=added to mjsunit/tools/tickprocessor
Review URL: http://codereview.chromium.org/546089
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3673 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
This is an effort to reuse profiler data processing code both in
TickProcessor and Dev Tools Profiler. The old Python implementation
will be removed.
The new TickProcessor works almost identical to the previous one.
However, it has some differences:
1. Not very useful "Call profile" section is replaced with a new
WebKit-like "Bottom up (heavy) profile" which shows the most
expensive functions together with their callers. I used it
personally in order to find and remove bottlenecks in the
tickprocessor script itself, and found it quite helpful.
2. Code entries with duplicate names (they occur for RegExes, stubs
and sometimes for anonymous Function objects) are now distinguished
by adding an occurence number inside curly brackets.
3. (Address -> code entry) mapping is more precise in boundary cases.
4. Windows version no more requires specifying .map file location.
5. Works faster.
Review URL: http://codereview.chromium.org/99054
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1802 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Currently only two stack frames are sampled (current function and its caller).
Output of tick processor looks like this:
[Call profile]:
total call path
15.2% LazyCompile: am3 crypto.js:108 <- LazyCompile: montReduce crypto.js:583
6.5% LazyCompile: am3 crypto.js:108 <- LazyCompile: bnpSquareTo crypto.js:431
2.9% Builtin: KeyedStoreIC_Generic <- LazyCompile: montReduce crypto.js:583
2.3% LazyCompile: am3 crypto.js:108 <- LazyCompile: bnpMultiplyTo crypto.js:415
Tested under Windows, Linux and OS X.
Review URL: http://codereview.chromium.org/21403
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1292 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Thus, instead of the following profiler records:
1.5% 1.5% LazyCompile: <anonymous>
we'll now have these:
1.5% 1.5% LazyCompile: <anonymous> richards.js:309
Basically, I translated two functions from messages.js into C++.
In the next CL I will update messages.js to use added native functions.
Review URL: http://codereview.chromium.org/19537
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1216 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
of the generated code. These can be used by the profiler to
categorize the ticks that occur within generated code and thereby show
more detailed information about where time is spent in generated code.
For instance, this is what the profiler displayed for a simple regexp
benchmark with irregexp-native before:
[JavaScript]:
total nonlib name
87.2% 87.2% RegExp: (?:\w*\W+)*
This is what we can display now:
[JavaScript]:
total nonlib name
87.2% 87.2% RegExp: (?:\w*\W+)*
- 53.0% 56.7% BranchOrBacktrack
- 14.9% 59.8% CheckCharacterLT
- 13.7% 20.4% CheckStackLimit
- 6.7% 6.7% SafeCall
- 2.7% 7.0% CheckCharacterGT
- 2.4% 2.4% SafeReturn
- 2.1% 2.1% LoadCurrentCharacter
- 1.8% 1.8% PushRegister
- 0.9% 0.9% PopRegister
- 0.9% 0.9% AdvanceRegister
- 0.3% 0.3% PopCurrentPosition
- 0.3% 0.3% CheckGreedyLoop
- 0.0% 20.4% PushBacktrack
- 0.0% 22.3% CheckCharacter
- 0.0% 2.4% IfRegisterLT
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@1010 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
