On Thu, May 26, 2011 at 12:45 PM, Jan Hubicka <hubi...@ucw.cz> wrote: >> >> This looks all very hackish with no immediate benefit mostly because >> of the use of lto_output_string. I think what you should do instead >> is split up lto_output_string_with_length into the piece that streams >> the string itself to the string-stream and returns an index into it >> and the piece streaming the index to the specified stream. Then you >> can simply bitpack that index and the two int line / column fields. > > Hmm, I plan to optimize string streaming (since we always stream one uleb to > set it is non-NULL that can be easilly handled by assigining NULL string index > 0). How precisely you however suggest to bitpack line/column and string > offset > together?
Similar to how you suggested, stream bits for a changed flag but instead of finishing the bitpack simply stream HOST_BITS_PER_INT bits for line (if changed), colunn (if changed) and file string index (if changed and the index is 'int'). I mostly want to avoid the split between the changed bits and the data output, esp. breaking the bitpack. > The point is to make location info occupy not even whole byte most of time. > Adding a simple stats claims that for tramp3d 30% of time location is > undefined, 15% of time file changes, 39% of time line changes and 44% of time > column changes (on tramp3d). So assuming one byte for each uleb (that is > optimistic, of course) one need 4 bits for the changed flags + less than a > byte > for the data. Situation is similar for combine.c where unknown id 10%, file > change is 0.5% and line change is 30%. > > If we want to get fancy, we could probably mix index/line/column as you > suggest > (so we will have only undefined bit and location change bit) and record only > changes to reduce uleb's range. > > All location streaming I know of somehow takes advantage of fact that location > do not change completely all the time from one place to another. > > Honza >
