This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [RFA] Add Windows x64 SEH unwinder


> 2012-06-15  Tristan Gingold  <gingold@adacore.com>
> 
> 	* amd64-windows-tdep.c (struct x64_frame_cache): Declare.
> 	(x64_w2gdb_regnum): New array.
> 	(x64_frame_decode_epilogue,	x64_frame_decode_insns)
> 	(x64_frame_cache, x64_frame_prev_register, x64_frame_this_id): New
> 	functions.
> 	(x64_frame_unwind): New variable.
> 	(amd64_windows_init_abi): Register this unwinder.
> 	* NEWS: Mention it.

Just for the record, Tristan and I found a weakness in this approach
which causes regressions for leaf functions when the compiler is not
configured to produce native SEH info (IIUC). My understanding is that,
the problem is that there is no entry for a function if the function
is a leaf function. So, for the case of such function, we don't know
if there is no info because it is a leaf function, or because it wasn't
going to be provided in the first place.

We discussed this situation, and Tristan proposed to make the DWARF
unwinder take priority over the native SEH info. I think that's
a good idea, in any case.

I also see that Tristan made some changes in our tree today, providing
a prologue-skipping routine, but I think that this is for a different
issue (I haven't had a chance to talk to him yet).

-- 
Joel


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]