kdb
[Top] [All Lists]

Re: KDB_REASON_CALL

To: Rob Tarte <robt@xxxxxxxxxxxxxxxxxxxx>
Subject: Re: KDB_REASON_CALL
From: Keith Owens <kaos@xxxxxxx>
Date: Thu, 27 Feb 2003 11:22:21 +1100
Cc: kdb@xxxxxxxxxxx
In-reply-to: Your message of "Wed, 26 Feb 2003 10:45:47 -0800." <3E5D0B5B.7070705@xxxxxxxxxxxxxxxxxxxx>
Sender: kdb-bounce@xxxxxxxxxxx
On Wed, 26 Feb 2003 10:45:47 -0800, 
Rob Tarte <robt@xxxxxxxxxxxxxxxxxxxx> wrote:
>It seems that KDB_REASON_CALL doesn't give a stack backtrace anymore.
>I noticed kdb_local() used to build a register context with functions 
>like kdb_getcurrentframe()
>and kdba_setpc().  Now the code just says "kdb() called with no 
>registers, restricted function".
>Any particular reason?  Do I have the wrong patch?

For i386, kdb can sort of generate a stack frame, but it was always
flakey.  For ia64 it is not possible to generate a stack frame from
REASON_CALL.  Trying to support an environment without struct pt_regs
was causing lots of kdb problems and bugs, so I dropped the support.  I
strongly recommend that yoiu use KDB_ENTER() instead.  That generates
an interrupt and has registers.


<Prev in Thread] Current Thread [Next in Thread>