Richard Kenner - Re: [patch] Fix pr23046 (original) (raw)

This is the mail archive of the gcc-patches@gcc.gnu.orgmailing list for the GCC project.

• From: kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)
• To: rth at redhat dot com
• Cc: gcc-patches at gcc dot gnu dot org
• Date: Mon, 19 Sep 05 14:37:44 EDT
• Subject: Re: [patch] Fix pr23046

 You'd V_C_E to a type with the same precision but with bounds not
 narrowed beyond what the bits can hold.  If that's the base type,
 fine, but something else if not.

Suppose the base type is signed SImode with precision of 32. The subtype has a range of 2-7, so is QImode, unsigned, and precision of 3.

So you are suggesting making a new type with what? Unsigned QImode, precision 8 and range 0-255?

How local would that type be? Do we make one of such for each subtype (meaning we'd likely have to find a place to stick it) or do we make a new such type every time we need to do the operation?

• Follow-Ups:
  • Re: [patch] Fix pr23046
    * From: Richard Henderson