From: Paul G. <pga...@at...> - 2002-11-01 01:14:01
|
Not being very intimate with eithe sjlj or dw2, I did notice some things w.r.t. observations of Ranjit below: At any rate, this is not intended as a solution at all, it is just an observation based on experience with earlier bugs re: gcc that I did force re: -fomit-frame-pointer. It is most likely a gcc related thing (code not fully functional, but minimally functional sort of thing). On 31 Oct 2002 at 16:04, Ranjit Mathew wrote: > ------------------------- 8< ---------------------------------- > Some observations: > > 1. Happens only if -fomit-frame-pointer is specified.] > 2. Does not happen if only -momit-leaf-frame-pointer is specified. 3. Does not happen if the called function does not take any args. 4. The generated ASM code shows sane procedure call frame > setup/cleanup w.r.t. the ESP and EBP registers. > > So what in the heaven's *could* be wrong here? I wish I > knew more about DW2 EH - I can only guess at the moment > that the FDE is somehow being wrongly generated when a > function calls an stdcall function with at least one > arg under -fomit-frame-pointer. I think I can confirm, based on past experience with gcc-2.95.2 stress testing, similar behaviors were exhibited. I assume we are discussing gcc 3.x, which leads me to think that maybe a portion the -fomit-frame-pointer code has not been completely repaired &/or revised. Disclaimer: I must state that the above is mostly what comes when I brainstorm on stuff. In other words, I am simply offering a possiblity, as far out and as unlikely as it may seem (or is), in terms of what may be further complicating matters when it comes to enabling DW2 EH for gcc. Paul G. |