http://codereview.chromium.org/77014/diff/1/2 File test/mjsunit/tools/profile.js (right):
http://codereview.chromium.org/77014/diff/1/2#newcode87 Line 87: ProfileTestDriver.prototype.enter = function(funcName) { On 2009/04/17 16:27:25, Søren Gjesse wrote: > Please add a comment about TOS being in stack_[0] therefore unshith/shits is > used instead of push/pop. Done. Actually, we use 'unshift' here in order to produce a following stack layout: [pc, caller, ..., main]. This is because when unwinding the stack we are moving in this direction. http://codereview.chromium.org/77014/diff/1/4 File tools/profile.js (right): http://codereview.chromium.org/77014/diff/1/4#newcode79 Line 79: name, startAddr, endAddr) { On 2009/04/17 16:27:25, Søren Gjesse wrote: > Why does addStaticCode use startAddr and endAddr and addCode use addr and size > for specifying the code range? Because shared libs and funcs are reported from V8 using address intervals, while compiled code is reported using start address and length. http://codereview.chromium.org/77014/diff/1/4#newcode149 Line 149: devtools.profiler.Profile.prototype.processStack_ = function(stack) { On 2009/04/17 16:27:25, Søren Gjesse wrote: > procesStack seems a bit generic, how about resolveFunctionNames? Good idea! Renamed to resolveAndFilterFuncs_. http://codereview.chromium.org/77014/diff/1/4#newcode250 Line 250: devtools.profiler.Profile.DynamicCodeEntry.prototype.getName = function() { On 2009/04/17 16:27:25, Søren Gjesse wrote: > Why is type not added to anonymous functions? Why not added? Please revisit the 'return' statement. http://codereview.chromium.org/77014/diff/1/4#newcode254 Line 254: } else if (name.charAt(0) == ' ') { On 2009/04/17 16:27:25, Søren Gjesse wrote: > Please explain this ' ' name check. Done. http://codereview.chromium.org/77014 --~--~---------~--~----~------------~-------~--~----~ v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev -~----------~----~----~----~------~----~------~--~---
