When I build CVS HEAD with RTS debugging I sometimes get, at exit time:
ghc-5.03: fatal error: Found non-frame during stack squeezing at 0x504797a0 (prev
frame was 0x504797a0)
and sometimes segmentation faults:
#0 0x87da233 in threadSqueezeStack (tso=0x500ba000) at GC.c:3820
3820 switch (get_itbl(frame)->type) {
(gdb) bt
#0 0x87da233 in threadSqueezeStack (tso=0x500ba000) at GC.c:3820
#1 0x87da4ff in threadPaused (tso=0x500ba000) at GC.c:4054
#2 0x87e2777 in suspendThread (reg=0x8896770, concCall=rtsFalse) at Schedule.c:1520
#3 0x87a5661 in s5rx_fast1 ()
#4 0x87e6f60 in stg_catch_frame_info ()
#5 0x4c58308 in ?? ()
Cannot access memory at address 0x887fe0be.
(gdb) print frame
$1 = (StgUpdateFrame *) 0x1
and generally behaviour which suggests there isn't an update frame where
it is expecting one - and indeed tso->su doesn't look like it's pointing
at an update frame.
Here s5rx_fast1 is a bit of TopHandler that calls shutdownHaskellAndExit.
Anyone know what's up with it?
Matt
_______________________________________________
Glasgow-haskell-bugs mailing list
[EMAIL PROTECTED]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
