Message ID | CAO2gOZUpyzR1-Uh7mZbNptGiQvTifqtw+GapUBM-gU_b1d5SAQ@mail.gmail.com |
---|---|
State | New |
Headers | show |
On 06/23/2015 04:18 PM, Dehao Chen wrote: > This patch disables reorder_blocks_and_partition for DWARF2 unwinding. > This is causing an internal application to segfault when built with > AutoFDO+EH. Unfortunately, I don't have a small reproducible. But from > the comment above the code, I do not see why UI_DWARF2 will survive > the issue. Any comments? > > Bootstrapped and test on-going. > > OK for trunk? > > Thanks, > Dehao > > ChangeLog: > > 2015-06-23 Dehao Chen <dehao@google.com> > > * opts.c(finish_options): Disable reorder_blocks_and_partition for DWARF2. This seems wrong to me -- shouldn't we be looking at what went wrong with the application and fixing whatever interaction between dwarf2 unwinding and reorder_and_partition? jeff
Index: opts.c =================================================================== --- opts.c (revision 220823) +++ opts.c (working copy) @@ -740,7 +740,7 @@ finish_options (struct gcc_options *opts, struct g if (opts->x_flag_exceptions && opts->x_flag_reorder_blocks_and_partition - && (ui_except == UI_SJLJ || ui_except >= UI_TARGET)) + && (ui_except >= UI_SJLJ)) { if (opts_set->x_flag_reorder_blocks_and_partition) inform (loc,