FASM lacks support for 64bit conventions (i think there were some macros, but i think they didn't work right), and also "proc" can't build exception data in PE header, which are replacement for dynamic SEH in FS segment.

MazeGen is creating such macros right now for 64bit MASM (even it doesn't have proper support for 64bit programming), so we could take them from him and adapt to FASM. I believe no assembler has support for generating these data right now.

Exception data? I recall there always was a "Exception" directory in the PE format (you couuld create it with "data 3"), but I never really knew what is it for.

yes, it's something like statical SEH info, instead of filling one on-the-run. It's "standard" for PE64, and for WinCE PE files, and maybe others.

search for "In the x86 world, exception handling is stack-based." here:
http://msdn.microsoft.com/msdnmag/issues/06/05/x64/default.aspx

ugly specs here:
http://msdn2.microsoft.com/en-us/library/0kd71y96.aspx

limited support for this in MASM described here:
http://msdn2.microsoft.com/en-us/library/ms235231.aspx

something here:
http://www.milw0rm.com/papers/68

this should also go to public, i think

oh, and how about macros for calling in 64bit mode? i remember you described them somewhere, but can't find it now.

are you aligning stack to 16bytes as required?

There's PROC64.INC in the fasmw package (for quite a long time, the history here: http://board.flatassembler.net/topic.php?t=4209). Yes, it does the 16-byte alignments.

think about adding ability to generate exception data to proc macros. Talk with mazegen for some more help (he is very helpful kind btw.)

But why is it needed in "proc" macro itself?

where elsewhere can you collect info about PROCEDURES?

But which procedures?
I'm against the label-racism that makes some labels more important than others.
Well, just kidding. But I wanted to say: actually I'm completely ignorant and have no idea why something related to exceptions would need information about all my procedures. And is it related to procedure entry points (where the "label discrimination" would happen ) or to diving the code into area of different procedures (the thing I don't like the most about the procs, BTW, but - you know - I'm one of these strange guys that refuse to understand what is so nifty about the structural programming).

maybe you could add OPTION to proc macro (like "c" to define __ccall procedure), that will add exception data for this procedure, maybe in style of "idata" macros. Not entire code area has to be covered by these structures, so you use it only when you wish.

Right now i am afraid these data are required in some cases, see end of FASMARM thread. btw, there is also example (for ARM processor) how to build this section.