Dan McDonald
2007-Mar-09 12:32 UTC
[dtrace-discuss] breakpoint(), ecb, and resumption of control.
I have a bug in my kernel that occurs under a specific set of circumstances (last packet of an FTP of a particular file - I can''t just send a singleton). I''m using a FBT predicate with a breakpoint() to catch this packet. Now I drop into kmdb with an ecb. I''d like to single-step through and beyond the function in question. The problem is, when I "::step out" it''s like typing ":c" and I hit my bug. And if I do single-stepping, I get stuck in all sorts of DTrace code. Finally, I can''t enable a breakpoint... because DTrace is loaded! :) So how do I get kmdb to actually stop in the function whose function boundary I intercepted? Thanks, Dan This message posted from opensolaris.org
Jonathan Adams
2007-Mar-13 21:41 UTC
[dtrace-discuss] breakpoint(), ecb, and resumption of control.
On 3/9/07, Dan McDonald <danmcd at sun.com> wrote:> I have a bug in my kernel that occurs under a specific set of > circumstances (last packet of an FTP of a particular file - I > can''t just send a singleton). I''m using a FBT predicate with > a breakpoint() to catch this packet.kdi_dtrace_state/W1 should allow breakpoints to then be installed. ::step out should (I believe) then work. Cheers, - jonathan