This is the mail archive of the
insight@sources.redhat.com
mailing list for the Insight project.
Re: 2 Segfaults in Insights (Latest CVS)
Steven,
This does not seem to be a GUI bug. Maybe the way things are being used by the GUI and
your script create the appropriate scenario for this to happen, but there is not much
that the GUI could do to influence the particular code that is executing.
As we can see from the stack, the set command was called with a valid argument
(frame #6, exp=0x89dd87c "$ictrl=0x7"). It is following it's normal course of execution,
calling evaluate_expression(), which goes through evaluate_subexp() to
evaluate_subexp_standard(). Noticing that it is an "assign" (because there are two
arguments separated by an "="), it calls value assign.
If I could I would do an "up 2" and print *toval and *fromval. But if everything is OK
toval must be an lval_register and value_assign will go through some register
reading/writting depending on the register type.
At the end, a value of the same type as the register is created, filled with the
new value and returned. The core dump happens when the copy of the register is
being filled.
memcpy (VALUE_CONTENTS_ALL_RAW (val), VALUE_CONTENTS_ALL_RAW (arg),
TYPE_LENGTH (VALUE_ENCLOSING_TYPE (arg)));
Either VALUE_CONTENTS_ALL_RAW (val) or VALUE_CONTENTS_ALL_RAW (arg) or
TYPE_LENGTH (VALUE_ENCLOSING_TYPE (arg)) is screwed or...
Your stack frame #1 looks strange. You should see value_copy () in there.
If we are facing some kind of stack corruption then we can just forget all
the above analysis.
Bottom line: unless I can reproduce it here it will be very hard for me to track
it down. I don't have a powerPC though.
P.S: Your PR says: "The segfault occurs at line 1060 in gdbtypes.c"
Is the core dump location constant? If not, the stack corruption hypothesis
is reinforced :-(
Steven Johnson wrote:
>
> Here tis.
>
> GNU gdb 19991116
> Copyright 1998 Free Software Foundation, Inc.
> GDB is free software, covered by the GNU General Public License, and you are
> welcome to change it and/or distribute copies of it under certain conditions.
> Type "show copying" to see the conditions.
> There is absolutely no warranty for GDB. Type "show warranty" for details.
> This GDB was configured as "i586-mandrake-linux".
>
> Program received signal SIGSEGV, Segmentation fault.
> 0x4018915c in memcpy () from /lib/libc.so.6
> #0 0x4018915c in memcpy () from /lib/libc.so.6
> #1 0x891b4f8 in ?? ()
> #2 0x80a5372 in value_assign (toval=0x891b4f8, fromval=0x89bdc50) at ../../src/gdb/valops.c:811
> #3 0x80a275a in evaluate_subexp_standard (expect_type=0x0, exp=0x89b8d90, pos=0xbfffc658,
> noside=EVAL_NORMAL) at ../../src/gdb/eval.c:1150
> #4 0x80a053f in evaluate_subexp (expect_type=0x0, exp=0x89b8d90, pos=0xbfffc658, noside=EVAL_NORMAL)
> at ../../src/gdb/eval.c:71
> #5 0x80a06c6 in evaluate_expression (exp=0x89b8d90) at ../../src/gdb/eval.c:145
> #6 0x80ae8dc in set_command (exp=0x89dd87c "$ictrl=0x7", from_tty=0)
> at ../../src/gdb/printcmd.c:1041
> #7 0x80e2c9a in gdbtk_call_command (cmdblk=0x84ce898, arg=0x89dd87c "$ictrl=0x7", from_tty=0)
> at ../../src/gdb/gdbtk/generic/gdbtk-hooks.c:519
> #8 0x80fec23 in execute_command (p=0x89dd885 "7", from_tty=0) at ../../src/gdb/top.c:1521
> #9 0x80fe3a5 in execute_control_command (cmd=0x8541188) at ../../src/gdb/top.c:1085
> #10 0x80fe9af in execute_user_command (c=0x84c9300, args=0x0) at ../../src/gdb/top.c:1437
> #11 0x80febcf in execute_command (p=0x89d5e0b "", from_tty=0) at ../../src/gdb/top.c:1515
> #12 0x80fe3a5 in execute_control_command (cmd=0x8694710) at ../../src/gdb/top.c:1085
> #13 0x80fe9af in execute_user_command (c=0x85cb398, args=0x0) at ../../src/gdb/top.c:1437
> #14 0x80feba0 in execute_command (p=0x89afc3a "e", from_tty=0) at ../../src/gdb/top.c:1507
> #15 0x80dd931 in gdb_cmd (clientData=0x80dd850, interp=0x84e06c0, objc=2, objv=0x84e136c)
> at ../../src/gdb/gdbtk/generic/gdbtk-cmds.c:830
> #16 0x80dd593 in wrapped_call (opaque_args=0xbfffc9c8)
> at ../../src/gdb/gdbtk/generic/gdbtk-cmds.c:536
> #17 0x80fddfc in catch_errors (func=0x80dd578 <wrapped_call>, args=0xbfffc9c8,
> errstring=0x83c34c3 "", mask=6) at ../../src/gdb/top.c:607
> #18 0x80dd4b6 in call_wrapper (clientData=0x80dd850, interp=0x84e06c0, objc=2, objv=0x84e136c)
> at ../../src/gdb/gdbtk/generic/gdbtk-cmds.c:474
> #19 0x837847a in TclExecuteByteCode (interp=0x84e06c0, codePtr=0x89afee0)
> at ../../../src/tcl/unix/../generic/tclExecute.c:955
> #20 0x835f2a0 in Tcl_EvalObj (interp=0x84e06c0, objPtr=0x85ed220)
> at ../../../src/tcl/unix/../generic/tclBasic.c:2645
> #21 0x82c8290 in Itcl_EvalMemberCode (interp=0x84e06c0, mfunc=0x89891a8, member=0x89891c0,
> contextObj=0x0, objc=1, objv=0x84e1368)
> at /home/steven/dev/tools/gdb/src/itcl/itcl/generic/itcl_methods.c:1029
> #22 0x82c8c6f in Itcl_ExecProc (clientData=0x89891a8, interp=0x84e06c0, objc=1, objv=0x84e1368)
> at /home/steven/dev/tools/gdb/src/itcl/itcl/generic/itcl_methods.c:1605
> #23 0x837847a in TclExecuteByteCode (interp=0x84e06c0, codePtr=0x89b54f8)
> at ../../../src/tcl/unix/../generic/tclExecute.c:955
> #24 0x835f2a0 in Tcl_EvalObj (interp=0x84e06c0, objPtr=0x898d850)
> at ../../../src/tcl/unix/../generic/tclBasic.c:2645
> #25 0x8395b1c in Tcl_UplevelObjCmd (dummy=0x0, interp=0x84e06c0, objc=3, objv=0x84e135c)
> at ../../../src/tcl/unix/../generic/tclProc.c:609
> #26 0x837847a in TclExecuteByteCode (interp=0x84e06c0, codePtr=0x8530d68)
> at ../../../src/tcl/unix/../generic/tclExecute.c:955
> #27 0x835f2a0 in Tcl_EvalObj (interp=0x84e06c0, objPtr=0x84f9de0)
> at ../../../src/tcl/unix/../generic/tclBasic.c:2645
> #28 0x839605a in TclObjInterpProc (clientData=0x85046d0, interp=0x84e06c0, objc=2, objv=0x84e1354)
> at ../../../src/tcl/unix/../generic/tclProc.c:996
> #29 0x837847a in TclExecuteByteCode (interp=0x84e06c0, codePtr=0x8991f18)
> at ../../../src/tcl/unix/../generic/tclExecute.c:955
> #30 0x835f2a0 in Tcl_EvalObj (interp=0x84e06c0, objPtr=0x868f2a8)
> at ../../../src/tcl/unix/../generic/tclBasic.c:2645
> #31 0x835f06b in Tcl_Eval (interp=0x84e06c0, string=0xbfffe024 "Download::download_it")
> at ../../../src/tcl/unix/../generic/tclBasic.c:2453
> #32 0x83607fa in Tcl_GlobalEval (interp=0x84e06c0, command=0xbfffe024 "Download::download_it")
> at ../../../src/tcl/unix/../generic/tclBasic.c:3983
> #33 0x8350700 in TkCopyAndGlobalEval (interp=0x84e06c0, script=0x8871ef8 "Download::download_it")
> at ../../../src/tk/unix/../generic/tkBind.c:4547
> #34 0x831c76f in TkInvokeMenu (interp=0x84e06c0, menuPtr=0x886aa68, index=1)
> at ../../../src/tk/unix/../generic/tkMenu.c:922
> #35 0x831c2b1 in MenuWidgetCmd (clientData=0x886aa68, interp=0x84e06c0, argc=3, argv=0xbfffe27c)
> at ../../../src/tk/unix/../generic/tkMenu.c:749
> #36 0x835e995 in TclInvokeStringCommand (clientData=0x886d868, interp=0x84e06c0, objc=3,
> objv=0x84e1348) at ../../../src/tcl/unix/../generic/tclBasic.c:1745
> #37 0x837847a in TclExecuteByteCode (interp=0x84e06c0, codePtr=0x89c2438)
> at ../../../src/tcl/unix/../generic/tclExecute.c:955
> #38 0x835f2a0 in Tcl_EvalObj (interp=0x84e06c0, objPtr=0x89e2040)
> at ../../../src/tcl/unix/../generic/tclBasic.c:2645
> #39 0x8395b1c in Tcl_UplevelObjCmd (dummy=0x0, interp=0x84e06c0, objc=3, objv=0x84e133c)
> at ../../../src/tcl/unix/../generic/tclProc.c:609
> #40 0x837847a in TclExecuteByteCode (interp=0x84e06c0, codePtr=0x89939f0)
> at ../../../src/tcl/unix/../generic/tclExecute.c:955
> #41 0x835f2a0 in Tcl_EvalObj (interp=0x84e06c0, objPtr=0x8506ec8)
> at ../../../src/tcl/unix/../generic/tclBasic.c:2645
> #42 0x839605a in TclObjInterpProc (clientData=0x851bd70, interp=0x84e06c0, objc=3, objv=0x84e1330)
> at ../../../src/tcl/unix/../generic/tclProc.c:996
> #43 0x837847a in TclExecuteByteCode (interp=0x84e06c0, codePtr=0x899b5d8)
> at ../../../src/tcl/unix/../generic/tclExecute.c:955
> #44 0x835f2a0 in Tcl_EvalObj (interp=0x84e06c0, objPtr=0x85c4858)
> at ../../../src/tcl/unix/../generic/tclBasic.c:2645
> #45 0x835f06b in Tcl_Eval (interp=0x84e06c0,
> string=0x8991fa8 "\n tkMenuInvoke .srcwin0.#srcwin0#srcwin#container#pane0#childsite#con#m.#srcwin0#srcwin#container#pane0#childsite#con#m#run 1\n")
> at ../../../src/tcl/unix/../generic/tclBasic.c:2453
> #46 0x83607fa in Tcl_GlobalEval (interp=0x84e06c0,
> command=0x8991fa8 "\n tkMenuInvoke .srcwin0.#srcwin0#srcwin#container#pane0#childsite#con#m.#srcwin0#srcwin#container#pane0#childsite#con#m#run 1\n")
> at ../../../src/tcl/unix/../generic/tclBasic.c:3983
> #47 0x834d730 in Tk_BindEvent (bindingTable=0x84f4a38, eventPtr=0x89c6a40, tkwin=0x886a898,
> numObjects=0, objectPtr=0xbffff49c) at ../../../src/tk/unix/../generic/tkBind.c:1731
> #48 0x835137a in TkBindEventProc (winPtr=0x886a898, eventPtr=0x89c6a40)
> at ../../../src/tk/unix/../generic/tkCmds.c:242
> #49 0x82fe5e0 in Tk_HandleEvent (eventPtr=0x89c6a40) at ../../../src/tk/unix/../generic/tkEvent.c:657
> #50 0x82fe8aa in WindowEventProc (evPtr=0x89c6a38, flags=-3)
> at ../../../src/tk/unix/../generic/tkEvent.c:983
> #51 0x838fdf7 in Tcl_ServiceEvent (flags=-3) at ../../../src/tcl/unix/../generic/tclNotify.c:444
> #52 0x83900d1 in Tcl_DoOneEvent (flags=0) at ../../../src/tcl/unix/../generic/tclNotify.c:683
> #53 0x82fe912 in Tk_MainLoop () at ../../../src/tk/unix/../generic/tkEvent.c:1041
> #54 0x80e2a66 in tk_command_loop () at ../../src/gdb/gdbtk/generic/gdbtk-hooks.c:383
> #55 0x809ba68 in captured_command_loop (data=0x0) at ../../src/gdb/main.c:104
> #56 0x80fddfc in catch_errors (func=0x809ba50 <captured_command_loop>, args=0x0,
> errstring=0x83a3c62 "", mask=6) at ../../src/gdb/top.c:607
> #57 0x809c523 in captured_main (data=0xbffff914) at ../../src/gdb/main.c:749
> #58 0x80fddfc in catch_errors (func=0x809ba9c <captured_main>, args=0xbffff914,
> errstring=0x83a3c62 "", mask=6) at ../../src/gdb/top.c:607
> #59 0x809c54f in main (argc=2, argv=0xbffff984) at ../../src/gdb/main.c:761
>
> > >
> > Your are being too optimistic :-) Insight is not that aware of things that are
> > done in the simulated console or even in the scripts (as they are still gdb scripts,
> > not insight scripts as we may have in the future).
> >
>
> I appreciate this, what I believed was happening is Insight itself would "set" when it wanted to update something. And that the hook was for the purpose of a general catch all of Insights "set"
> operations. (im sure i saw a comment to this effect somewhere) Insight definately calls the set_command operation, and is executing it "After" the set command has been successfully performed on the
> target. If I comment out the body of set_command, problem goes away. But im not sure what else i am breaking. The normal do_setshow_command operation does not appear to call set_command. What is
> set_command in printcmd.c actually providing??
>
> > But I would need to look at that stack and at the contents of the cmd_list_struct
> > that is passed back by gdb to the GUI in the hook callback.
> >
>
> How do I get this for you? (What commands.)
>
> Is there a way of getting GDB (Normal command line) to log its session? I can't find it if there is.
>
> Steven Johnson
--
Fernando Nasser
Red Hat Canada Ltd. E-Mail: fnasser@cygnus.com
2323 Yonge Street, Suite #300
Toronto, Ontario M4P 2C9