This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH]: Fix places where we expect non-null loops in the loop array


> >no, it is expected behavior.  As for why I decided to do it this way --
> >so that the numbers of the loops do not change in the middle of an
> >optimization (so you can for example store some information for loops
> >in an array indexed by loop numbers).  I do not think it is seriously
> >used anywhere (except perhaps for scev, that store loop numbers in
> >CHREC nodes).
> Okay, so, uh, guys, what do you want me to do?
> Make things ignore null loops (which is what everything does now)?

it seems to be the simplest choice to me.

> Invalidating the loop indexes by compression will screw over scev's cache, 
> which will cause a performance regression at the expense of a few if's in 
> the other code.
> I can of course, go through and figure out the new num_levels and depth, 
> etc, without compressing the array.

loop->depth is always updated, so you do not have to go through everything to
get it.  I do not know what num_levels is supposed to be (which very
likely means that it is not updated, and you would really have to go
over the loops and determine it).


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]