Re: cancellation with HAVE_FORCED_UNWIND

Good day, 
 
> > You'll have to look through the call chain and determine which compiler
> > generated the code for all the open call frames.
> 
> As a first guess, this may be a binutils bug.  The linker was recently
> (last month or so) fixed to not add invalid padding to unwind
> information.

Apparently it was an ld bug. Surprisingly (at least to me), in some 
cases, the kernel participates in stack unwinding. For example when 
thread cancellation is delivered by an async-signal. After re-linking
the kernel with the new ld everything seem to work fine.

thanks,
-boris

-- 
Boris Kolpackov  [email protected] http://kolpackov.net
3072R/AE084F1D - F608 942F 312E D82E 5B84  0407 C880 290B AE08 4F1D

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)
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=HG/C
-----END PGP SIGNATURE-----