Re: pointer address changes for no reason
Re: pointer address changes for no reason
- Subject: Re: pointer address changes for no reason
- From: Bill Bumgarner <email@hidden>
- Date: Tue, 10 Apr 2007 11:32:28 -0700
On Apr 10, 2007, at 11:22 AM, John Stiles wrote:
It is just very, very slow. I've clocked it at over an hour to run
one call to CreateNewWindow (in Carbon).
This is just a performance issue in the current debugger. I don't
know if they plan to address it, but the upshot is that watchpoints
are basically useless on current hardware until they improve the
speed. :( I have an open radar on the performance issue (as well as
Xcode's general uselessness with creating watchpoints) and have
heard back that it's known and they're working on it; don't know any
more than this.
There are two kinds of watchpoints in the world; in the CPU and via
the VM subsystem.
A CPU watchpoint is implemented in the CPU and is very very fast.
A VM based watchpoint is typically implemented by VM protecting the
pages containing the addresses to be watched and then conditionally
continuing or stopping based on what addresses were actually modified
in said pages of memory. And that is slow. Very very slow,
comparatively.
http://developer.apple.com/releasenotes/DeveloperTools/RN-GDB/index.html
http://developer.apple.com/documentation/DeveloperTools/gdb/gdb/gdb_6.html#SEC31
b.bum
_______________________________________________
Cocoa-dev mailing list (email@hidden)
Do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden