This fixes the regression reported at
<http://sourceware.org/ml/gdb-patches/2013-06/msg00185.html>.
GDBserver was reaching:
static int
regsets_fetch_inferior_registers (struct regsets_info *regsets_info,
struct regcache *regcache)
{
struct regset_info *regset;
int saw_general_regs = 0;
int pid;
struct iovec iov;
regset = regsets_info->regsets;
pid = lwpid_of (get_thread_lwp (current_inferior));
while (regset->size >= 0)
{
void *buf, *data;
int nt_type, res;
if (regset->size == 0
|| regsets_info->disabled_regsets[regset - regsets_info->regsets])
{
>>>>>>> regset ++; <<<<<<< HERE
continue;
}
Because info->disabled_regsets[] was not being initialized, and that
causes all sorts of wrong.
gdb/gdbserver/
2013-06-11 Pedro Alves <palves@redhat.com>
* linux-low.c (initialize_regsets_info): Use xcalloc instead of
xmalloc.
2013-06-11 Pedro Alves <palves@redhat.com>
+ * linux-low.c (initialize_regsets_info): Use xcalloc instead of
+ xmalloc.
+
+2013-06-11 Pedro Alves <palves@redhat.com>
+
* linux-x86-low.c (initialize_low_arch): Call
init_registers_x32_avx_linux.
info->regsets[info->num_regsets].size >= 0;
info->num_regsets++)
;
- info->disabled_regsets = xmalloc (info->num_regsets);
+ info->disabled_regsets = xcalloc (1, info->num_regsets);
}
#endif