sigframe inlining

Chris Faylor
Sat Sep 2 21:52:00 GMT 2000

On Sat, Sep 02, 2000 at 07:51:59PM +0400, Egor Duda wrote:
>Saturday, 02 September, 2000 Chris Faylor wrote:
>>>I've duplicated the problem. A test case is below.
>>>I'll look into this.
>CF> I've checked in a fix.
>today i've got another bug, when i've compiled cygwin without -O2 flag
>since   gcc   stops   inining   class   sigframe   member   functions,
>__builtin_frame_address   got   wrong  stack  offset.  this  leads  to
>api_fatal in signal handler. how about this patch?
>Egor. ICQ 5165414 FidoNet 2:5020/496.19

Yeah, I knew about this problem and have studiously avoided compiling
sigframe functions without optimization.  I wish there was some way to
force inlining no matter what.

I guess we'll probably have to go with something like this eventually
but I'd like to ruminate on it for a while first.


More information about the Cygwin-developers mailing list