Re: gdb, cores & kg switchtoact
Re: gdb, cores & kg switchtoact
- Subject: Re: gdb, cores & kg switchtoact
- From: Terry Lambert <email@hidden>
- Date: Thu, 25 Aug 2005 10:58:22 -0700
On Aug 25, 2005, at 10:30 AM, Brian Bergstrand wrote:
I'm investigating a kernel core and trying to use kgmacros'
switchtoact. It keeps failing with a memory error though.
Try using "switchtocorethread".
This is just not my day:
activation thread pri state wait_queue wait_event
0x01d6d948 0x01d6d948 47 UW 0x003d6120 0x1149610
(gdb) switchtocorethread 0x01d6d948
Left operand of assignment is not an lvalue.
I even restarted gdb thinking I may have messed up some macro
internal state with my fumblings from before. Still no dice.
Did see this comment for the macro:
#Use of this macro requires the gdb submission from 3401283
I have the latest dev tools.
I assume you have the symbolic kernel that matches the mach_kernel
you are running, along with the kgmacros that match the kernel you
are running, and that you have NMI'ed the machine to get it into this
state.
You need to continue in the debugger, and NMI again; this may take
several tries, if your machine happens to be in a high-interrupt-with-
nothing-much-else-going-on state. Eventually, you will be able to
"switchtoact".
The problem is that if you hit the NMI at just the wrong time
(interrupt context), the debugger doesn't work so well.
Someone should do something about this, but since there's a
workaround, it's been lower priority than it might be.
-- Terry
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-kernel mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden