Hi,

Here's another update to the ensure_size_t patch. The following changes
have been made to the patch:

* The function is now called ulongest_fits_host_or_error and is now
  in utils.c. I have named it so to give room for a future
  'longest_fits_host_orr_error' if needed. They have to be different
  because the error message will have to be printed differently
  (plongest vs pulongest). The name also looks neater in general.
* I have reverted the size_t checks from most tdep files in favour of
  promoting the value_contents call ahead. In fact, barring
  h8300h_return_value, all of them have an earlier value_contents call,
  making the check unnecessary.

This change applies on top of the bitpos patch and I have verified that
the change does not cause any regressions.

Regards,
Siddhesh

gdb/ChangeLog

	* alpha-tdep.c (alpha_push_dummy_call) Check for underflow in
	SP.
	* cp-valprint (cp_print_value): Ensure BASECLASS fits into
	size_t.
	* dwarf2loc.c (read_pieced_value): Ensure that THIS_SIZE fits
	into size_t.
	(write_pieced_value): Likewise.
	* h8300-tdep.c (h8300h_return_value): Ensure that TYPE fits
	into size_t.
	* p-valprint (pascal_object_print_value): Ensure BASECLASS fits
	into size_t.
	* utils.c (ulongest_fits_host_or_error): New function to find if
	a ULONGEST number fits into size_t.
	* utils.h: Declare ulongest_fits_host_or_error.
	* valops.c (search_struct_method): Ensure BASECLASS fits into
	size_t.
	* value.c (allocate_value_lazy): Ensure TYPE fits into size_t.
	(allocate_value_contents): Likewise.
	(set_value_enclosing_type): Ensure NEW_ENCL_TYPE fits into
	size_t.
	* xstormy16-tdep.c (xstormy16_push_dummy_call): Call
	value_contents earlier to ensure that the argument fits into
	host memory.