| commit | 01980517c348d4f35eb94f7a1baccd60ae50badd | |
| author | strange <kawk256@gmail.com> | |
| Sun, 13 Dec 2009 18:45:40 +0000 (13 11:45 -0700) | ||
| committer | strange <kawk256@gmail.com> | |
| Sun, 13 Dec 2009 18:45:40 +0000 (13 11:45 -0700) | ||
| tree | 3770699300ae0cdab2691137198fb436432374da | treesnapshot (tar.gz zip) |
| parent | 6296c7a148c1557d95f8fe2883085afb9718a937 | commitdiff |
Fixed a bug in MallocObserver::notify().
It was a bit of a re-entrant problem, combined with a gdb error . . .
Basically, when the gdb interpreter gets '-exec-finish', it finishes the
execution of the current frame. However, if there is a breakpoint on the same
place that the '-exec-finish' goes to, then the message sent out is not a
"function-finished", it is a "breakpoint-hit". That, and the observers were
added in an incorrect order . . .
Now, I have to do something similar for free(). Unfortunately, the difference
here is that I don't have a nice return value to watch for . . . so I don't
know how to watch for it . . .
It was a bit of a re-entrant problem, combined with a gdb error . . .
Basically, when the gdb interpreter gets '-exec-finish', it finishes the
execution of the current frame. However, if there is a breakpoint on the same
place that the '-exec-finish' goes to, then the message sent out is not a
"function-finished", it is a "breakpoint-hit". That, and the observers were
added in an incorrect order . . .
Now, I have to do something similar for free(). Unfortunately, the difference
here is that I don't have a nice return value to watch for . . . so I don't
know how to watch for it . . .