This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa] Computed gotos
- From: law at redhat dot com
- To: Richard Henderson <rth at redhat dot com>
- Cc: Paul Brook <paul at nowt dot org>, "gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>
- Date: Mon, 16 Jun 2003 13:46:13 -0600
- Subject: Re: [tree-ssa] Computed gotos
- Reply-to: law at redhat dot com
In message <20030616193024.GA4045@redhat.com>, Richard Henderson writes:
>On Mon, Jun 16, 2003 at 09:00:29AM -0600, law@redhat.com wrote:
>> >Apparently you set FORCED_LABEL.
>>
>> Or more correctly, it should be set during the gimplification process.
>> If the gimplifier isn't automatically setting FORCED_LABEL for the
>> labels in question, then we'd likely consider that a bug.
>
>Maybe. Thing is, in rtl-land, the forced_label list is those
>labels whose address is taken in an initializer for static storage.
>
>Labels whose address is taken in code are put on a separate list.
>This is important because the code that takes the address can be
>deleted, at which point it isn't a forced label anymore.
>
>Seems like we should do something similar in gimple-land.
Yes. At some point we'll probably make the distinction between those
which appear in static storage and those which appear as operands to
real code for tree-ssa. But it's not that high on the list of
priorities yet.
jeff