On 10/17/2016 05:45 PM, Jakub Jelinek wrote:
+/* First abbrev_id that can be optimized based on usage. */ +static unsigned int abbrev_opt_start; + +/* Vector of usage counts during build_abbrev_table. Indexed by + abbrev_id - abbrev_opt_start. */ +static vec<unsigned int> abbrev_usage_count; + +/* Vector of all DIEs added with die_abbrev >= abbrev_opt_start. */ +static vec<dw_die_ref> abbrev_dies;
I think these could use a comment that they are temporary data only live between build_abbrev_table and optimize_abbrev_table. And perhaps the abbrev_dies variable should be named something like sorted_abbrev_dies to avoid confusion with abbrev_die_table.
OK with that change. Jason