John Millikin wrote: > I've not been able to reproduce this on my workstation, which is running > Jessie. Installed package versions: > > ghc: [7.6.3, 7.8.4, 7.10.1] > c2hs: [0.17.2] > haskell-ncurses: [0.2.7, 0.2.11] > libncursesw5-dev: [5.9+20140913-1+b1] > > That stack trace posted in the OP is suspicious. It looks like drawBorder is > being called somehow, but there is no path for the posted code snippet to > reach > drawBorder. This makes me believe the stack is getting corrupted. > > The only change between 0.2.7 and 0.2.11 that seems related is > 08137f194f66796a68bf38886d3944caf7b9d08e, which added support for c2hs>=0.18.2 > by adding ifdefs for some new pointer semantics. I believe this change is > semantically a no-op, but there might be something going on in the FFI code > that I'm not aware of. > > What version of c2hs are you using? I'll try building with that to see if it > triggers the issue for me.
c2hs is 0.25.2
I'll bet that if you debootstrap a chroot of debian unstable, you'll be
able to reproduce it in there.
Adding to the stack-smashing hypothesis, I got this backtrace just now:
#0 __memcpy_sse2_unaligned ()
at ../sysdeps/x86_64/multiarch/memcpy-sse2-unaligned.S:33
#1 0x0000000000415ede in ckTj_info ()
#2 0x0000000000000000 in ?? ()
--
see shy jo
signature.asc
Description: Digital signature
