Re: Kernel debugging on a g4 iBook
Re: Kernel debugging on a g4 iBook
- Subject: Re: Kernel debugging on a g4 iBook
- From: Brian Bergstrand <email@hidden>
- Date: Wed, 10 Aug 2005 00:18:08 -0500
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
I have seen the exact same behavior on my PowerBook 15'' 1ghz. The
screen will go dark and the debug session is lost (can't connect).
This happens with 10.3.x and 10.4.x, however, with 10.2.x there is no
problem at all.
I've since switched my test machine to a 500Mhz G3 iMac and there are
no more problems. My personal thought is that something with the PB
power management in 10.3.x and 10.4.x interferes with kernel
debugging (and yes, I tried turning off all power management options).
On Aug 9, 2005, at 11:45 AM, Deker wrote:
So I've been trying to do some kernel debugging on a g4 iBook
recently and once I've connected with GDB from the debug host if I
pause for more than a few seconds the next stepi/nexti just causes
the target host to power off.
I'm working with 10.3.8 (and no, Tiger isn't an option), boot-args
set to 'debug=0x141` in OpenFirmware. If there's any more pertinent
data I can provide, please let me know.
Here's a private query I sent to someone at Apple -- they had no idea
what the problem could be:
I have a Powerbook 15"" Al 1ghz, 768 RAM, DVD/CD-RW that I use for my
test machine. The problem is, that the machine will disappear as soon
as I break into the debugger (either with a crash or bp) and the
debug session will die.
Here's what happens:
Setup normal two machine debugging, set a breakpoint and wait for it
to hit. When it hits, the debugger on my test machine (latest Tiger
tools) stops and then when I try 'n' or 's' the target machine's
screen goes dark and the debug session dies
Breakpoint 1, TcpCtlIndication (pSocket=0x1fc69d8, socketState=2,
stateParam=0x0) at /Users/bdb/src/prosoft/ParisCur/dd/ipctl/osx/
tcp_kpi.c:52
52 int tcpStatus, socketAccessType = 0,
sopt, soptlen;
(gdb) n
kdp_reply_wait: error from kdp_receive: receive timeout exceeded
kdp_transaction (kdp_xfer_memory): transaction timed out
kdp_transaction (kdp_xfer_memory): re-sending transaction
I had debug=0x0144 set and thinking ARP may have been the problem
tried 0x0004 (which requires a permanent arp conn). Didn't make a
difference. The machine just disappears and won't come back without a
hard reboot.
Now, here's the really weird part. Under Jaguar, there are no
problems whatsoever. But under any version of Panther or Tiger, the
problem occurs every time.
Brian Bergstrand
<http://www.bergstrand.org/brian/> PGP Key ID: 0xB6C7B6A2
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (Darwin)
iD8DBQFC+Y4kedHYW7bHtqIRAouIAJ0e2qBVvemgZwHpVuhgQZzSQkwcLgCg2wNp
sjNSdVVLN3GbdeDpYqbqo3Q=
=qIgS
-----END PGP SIGNATURE-----
_______________________________________________
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