This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Is this a bug in libffi testsuite on solaris2.8?
- From: "Zack Weinberg" <zack at codesourcery dot com>
- To: dewar at gnat dot com (Robert Dewar)
- Cc: aph at redhat dot com, gcc at gcc dot gnu dot org, lucier at math dot purdue dot edu
- Date: Sun, 02 Nov 2003 19:38:18 -0800
- Subject: Re: Is this a bug in libffi testsuite on solaris2.8?
- References: <20031103031415.A2990F2D5D@nile.gnat.com>
dewar@gnat.com (Robert Dewar) writes:
>> No we don't. It's up to the caller to provide the memory.
>
> I don't understand, we are talking about the code generated to build a
> trampoline. That call to build the trampoline is generated by the compiler,
> it does not result from direct user code. The user knows nothing about
> the trampoline (or the need for it).
I think there's confusion of different kinds of trampolines.
The trampolines that GCC generates when the address of a nested
function is taken, are built by the compiler and invisible to the
programmer.
The trampolines that libffi uses -- presumably for similar purposes --
are explicitly allocated by the programmer.
zw