> > OK, I see. Then the only way out I can think of is to stop going up the > > chain of COMPONENT_REFs as soon as the offset becomes negative. > > Yeah, that sounds like a better solution.
PR ada/51775 shows that this pessimizes though because if you have an enclosing sub-component at offset 0 on the LHS and not on the RHS, and you're using the model of the LHS for the RHS, you'll build a convoluted (but correct) MEM on the RHS, and this is apparently enough to thwart FRE/PRE and the likes in some simple cases: t$F$i2_9 = MEM[(struct pack9__r2 *)y_1(D)].F.i2; D.2233_3 = y_1(D)->i2; The compiler doesn't see that D.2233_3 == t$F$i2_9. Additionally using the type trick would help, at least for PR ada/51775 and I guess in most cases. Thoughts? -- Eric Botcazou
