.PARISC.unwind has 32-bit addresses in both 32-bit ELF and 64-bit ELF.
Well, strictly speaking, the 32-bit "start" and "end" fields are
segment relative offsets. (The 64-bit ABI says so, while the 32-bit
ABI says they are addresses but it appears they are segment relative
offsets in practice. Likely the 32-bit ABI lacks an update.)
* readelf.c (hppa_process_unwind): Don't use eh_addr_size to
calculate number of entries.
(slurp_hppa_unwind_table): Don't use eh_addr_size here either.
+2018-10-10 Helge Deller <deller@gmx.de>
+ Alan Modra <amodra@gmail.com>
+
+ * readelf.c (hppa_process_unwind): Don't use eh_addr_size to
+ calculate number of entries.
+ (slurp_hppa_unwind_table): Don't use eh_addr_size here either.
+
2018-10-10 Alan Modra <amodra@gmail.com>
* objdump.c (dump_dwarf): Set s12z eh_addr_size to 4.
i = rp->r_offset / unw_ent_size;
- switch ((rp->r_offset % unw_ent_size) / eh_addr_size)
+ switch ((rp->r_offset % unw_ent_size) / 4)
{
case 0:
aux->table[i].start.section = sym->st_shndx;
{
if (streq (SECTION_NAME (sec), ".PARISC.unwind"))
{
- unsigned long num_unwind = sec->sh_size / (2 * eh_addr_size + 8);
+ unsigned long num_unwind = sec->sh_size / 16;
printf (ngettext ("\nUnwind section '%s' at offset 0x%lx "
"contains %lu entry:\n",