This is the mail archive of the
binutils@sourceware.cygnus.com
mailing list for the binutils project.
Re: Patch to allow targets to prevent inlining
- To: nickc at cygnus dot com
- Subject: Re: Patch to allow targets to prevent inlining
- From: Mark Mitchell <mark at codesourcery dot com>
- Date: Tue, 15 Feb 2000 09:23:38 -0800
- Cc: dje at transmeta dot com, binutils at sourceware dot cygnus dot com
- Organization: CodeSourcery, LLC
- References: <200002150015.QAA02865@elmo.cygnus.com>
>>>>> "Nick" == Nick Clifton <nickc@cygnus.com> writes:
Nick> OK, you win. But since adding a naked attribute the to
Nick> generic part of gcc would increase the overall complexity of
Nick> the compiler, I doubt it I would be able to persuade the
Nick> steering committee to accept it unless several more ports
Nick> wanted the feature. Hmm, mnaybe we could start a campaign
Nick> :-)
I'm not on the steering committee. But, I think this would be a
useful feature in general -- if I understand `naked' correctly to mean
a function without a prologue or epilogue. For example, thunks in C++
could then be represented as ordinary functions, not with assembly
code.
However, I'm not convinced that such things should never be inlined,
in the abstract. What I'm convinced of is that GCC can't handle
inlining them at the moment. So, the documentation for `naked'
functions should explicitly say that they might be inlined in a future
release, etc., etc.
--
Mark Mitchell mark@codesourcery.com
CodeSourcery, LLC http://www.codesourcery.com