On Thu, 5 Oct 2023, Jakub Jelinek wrote:
> Hi!
>
> The following patch removes ipa_bits struct pointer/vector from ipa
> jump functions and ipa cp transformations.
>
> The reason is because the struct uses widest_int to represent
> mask/value pair, which in the RFC patches to allow larger precisions
> for wide_int/widest_int is GC unfriendly because those types become
> non-trivially default constructible/copyable/destructible.
> One option would be to use trailing_wide_int for that instead, but
> as pointed out by Aldy, irange_storage which we already use under
> the hood for ipa_vr when type of parameter is integral or pointer
> already stores the mask/value pair because VRP now does the bit cp
> as well.
> So, this patch just uses m_vr to store both the value range and
> the bitmask. There is still separate propagation of the
> ipcp_bits_lattice from propagation of the ipcp_vr_lattice, but
> when storing we merge the two into the same container.
>
> I've bootstrapped/regtested a slightly older version of this
> patch on x86_64-linux and i686-linux and that version regressed
> +FAIL: gcc.dg/ipa/propalign-3.c scan-ipa-dump-not cp "align:"
> +FAIL: gcc.dg/ipa/propalign-3.c scan-tree-dump optimized "fail_the_test"
> +FAIL: gcc.dg/ipa/propbits-1.c scan-ipa-dump cp "Adjusting mask for param 0
> to 0x7"
> +FAIL: gcc.dg/ipa/propbits-2.c scan-ipa-dump cp "Adjusting mask for param 0
> to 0xf"
> The last 2 were solely about the earlier patch not actually copying
> the if (dump_file) dumping of message that we set some mask for some
> parameter (since then added in the @@ -5985,6 +5741,77 @@ hunk).
> The first testcase is a test for -fno-ipa-bit-cp disabling bit cp
> for alignments. For integral types I'm afraid it is a lost case
> when -fno-ipa-bit-cp -fipa-vrp is on when value ranges track bit cp
> as well, but for pointer alignments I've added
> && opt_for_fn (cs->caller->decl, flag_ipa_bit_cp)
> and
> && opt_for_fn (node->decl, flag_ipa_bit_cp)
> guards such that even just -fno-ipa-bit-cp disables it (alternatively
> we could just add -fno-ipa-vrp to propalign-3.c dg-options).
>
> Ok for trunk if this passes another bootstrap/regtest?
OK.
Richard.
> Or defer until it is really needed (when the wide_int/widest_int
> changes are about to be committed)?
>
> 2023-10-05 Jakub Jelinek <ja...@redhat.com>
>
> * ipa-prop.h (ipa_bits): Remove.
> (struct ipa_jump_func): Remove bits member.
> (struct ipcp_transformation): Remove bits member, adjust
> ctor and dtor.
> (ipa_get_ipa_bits_for_value): Remove.
> * ipa-prop.cc (struct ipa_bit_ggc_hash_traits): Remove.
> (ipa_bits_hash_table): Remove.
> (ipa_print_node_jump_functions_for_edge): Don't print bits.
> (ipa_get_ipa_bits_for_value): Remove.
> (ipa_set_jfunc_bits): Remove.
> (ipa_compute_jump_functions_for_edge): For pointers query
> pointer alignment before ipa_set_jfunc_vr and update_bitmask
> in there. For integral types, just rely on bitmask already
> being handled in value ranges.
> (ipa_check_create_edge_args): Don't create ipa_bits_hash_table.
> (ipcp_transformation_initialize): Neither here.
> (ipcp_transformation_t::duplicate): Don't copy bits vector.
> (ipa_write_jump_function): Don't stream bits here.
> (ipa_read_jump_function): Neither here.
> (useful_ipcp_transformation_info_p): Don't test bits vec.
> (write_ipcp_transformation_info): Don't stream bits here.
> (read_ipcp_transformation_info): Neither here.
> (ipcp_get_parm_bits): Get mask and value from m_vr rather
> than bits.
> (ipcp_update_bits): Remove.
> (ipcp_update_vr): For pointers, set_ptr_info_alignment from
> bitmask stored in value range.
> (ipcp_transform_function): Don't test bits vector, don't call
> ipcp_update_bits.
> * ipa-cp.cc (propagate_bits_across_jump_function): Don't use
> jfunc->bits, instead get mask and value from jfunc->m_vr.
> (ipcp_store_bits_results): Remove.
> (ipcp_store_vr_results): Incorporate parts of
> ipcp_store_bits_results here, merge the bitmasks with value
> range if both are supplied.
> (ipcp_driver): Don't call ipcp_store_bits_results.
> * ipa-sra.cc (zap_useless_ipcp_results): Remove *ts->bits
> clearing.
>
> --- gcc/ipa-prop.h.jj 2023-10-05 11:32:40.172739988 +0200
> +++ gcc/ipa-prop.h 2023-10-05 11:36:45.405378086 +0200
> @@ -292,18 +292,6 @@ public:
> array_slice<const ipa_argagg_value> m_elts;
> };
>
> -/* Information about zero/non-zero bits. */
> -class GTY(()) ipa_bits
> -{
> -public:
> - /* The propagated value. */
> - widest_int value;
> - /* Mask corresponding to the value.
> - Similar to ccp_lattice_t, if xth bit of mask is 0,
> - implies xth bit of value is constant. */
> - widest_int mask;
> -};
> -
> /* Info about value ranges. */
>
> class GTY(()) ipa_vr
> @@ -342,11 +330,6 @@ struct GTY (()) ipa_jump_func
> and its description. */
> struct ipa_agg_jump_function agg;
>
> - /* Information about zero/non-zero bits. The pointed to structure is
> shared
> - betweed different jump functions. Use ipa_set_jfunc_bits to set this
> - field. */
> - class ipa_bits *bits;
> -
> /* Information about value range, containing valid data only when vr_known
> is
> true. The pointed to structure is shared betweed different jump
> functions. Use ipa_set_jfunc_vr to set this field. */
> @@ -940,15 +923,13 @@ struct GTY(()) ipcp_transformation
> {
> /* Default constructor. */
> ipcp_transformation ()
> - : m_agg_values (nullptr), bits (nullptr), m_vr (nullptr),
> - m_uid_to_idx (nullptr)
> + : m_agg_values (nullptr), m_vr (nullptr), m_uid_to_idx (nullptr)
> { }
>
> /* Default destructor. */
> ~ipcp_transformation ()
> {
> vec_free (m_agg_values);
> - vec_free (bits);
> vec_free (m_vr);
> }
>
> @@ -968,8 +949,6 @@ struct GTY(()) ipcp_transformation
>
> /* Known aggregate values. */
> vec<ipa_argagg_value, va_gc> *m_agg_values;
> - /* Known bits information. */
> - vec<ipa_bits *, va_gc> *bits;
> /* Value range information. */
> vec<ipa_vr, va_gc> *m_vr;
> /* If there are many parameters, this is a vector sorted by their DECL_UIDs
> @@ -1172,8 +1151,6 @@ tree ipa_get_indirect_edge_target (struc
> struct cgraph_edge *ipa_make_edge_direct_to_target (struct cgraph_edge *,
> tree,
> bool speculative = false);
> tree ipa_impossible_devirt_target (struct cgraph_edge *, tree);
> -ipa_bits *ipa_get_ipa_bits_for_value (const widest_int &value,
> - const widest_int &mask);
>
>
> /* Functions related to both. */
> --- gcc/ipa-prop.cc.jj 2023-10-05 11:32:40.154740234 +0200
> +++ gcc/ipa-prop.cc 2023-10-05 14:06:13.742885839 +0200
> @@ -66,49 +66,6 @@ function_summary <ipcp_transformation *>
> /* Edge summary for IPA-CP edge information. */
> ipa_edge_args_sum_t *ipa_edge_args_sum;
>
> -/* Traits for a hash table for reusing already existing ipa_bits. */
> -
> -struct ipa_bit_ggc_hash_traits : public ggc_cache_remove <ipa_bits *>
> -{
> - typedef ipa_bits *value_type;
> - typedef ipa_bits *compare_type;
> - static hashval_t
> - hash (const ipa_bits *p)
> - {
> - hashval_t t = (hashval_t) p->value.to_shwi ();
> - return iterative_hash_host_wide_int (p->mask.to_shwi (), t);
> - }
> - static bool
> - equal (const ipa_bits *a, const ipa_bits *b)
> - {
> - return a->value == b->value && a->mask == b->mask;
> - }
> - static const bool empty_zero_p = true;
> - static void
> - mark_empty (ipa_bits *&p)
> - {
> - p = NULL;
> - }
> - static bool
> - is_empty (const ipa_bits *p)
> - {
> - return p == NULL;
> - }
> - static bool
> - is_deleted (const ipa_bits *p)
> - {
> - return p == reinterpret_cast<const ipa_bits *> (1);
> - }
> - static void
> - mark_deleted (ipa_bits *&p)
> - {
> - p = reinterpret_cast<ipa_bits *> (1);
> - }
> -};
> -
> -/* Hash table for avoid repeated allocations of equal ipa_bits. */
> -static GTY ((cache)) hash_table<ipa_bit_ggc_hash_traits>
> *ipa_bits_hash_table;
> -
> /* Traits for a hash table for reusing ranges. */
>
> struct ipa_vr_ggc_hash_traits : public ggc_cache_remove <ipa_vr *>
> @@ -528,17 +485,6 @@ ipa_print_node_jump_functions_for_edge (
> ctx->dump (dump_file);
> }
>
> - if (jump_func->bits)
> - {
> - fprintf (f, " value: ");
> - print_hex (jump_func->bits->value, f);
> - fprintf (f, ", mask: ");
> - print_hex (jump_func->bits->mask, f);
> - fprintf (f, "\n");
> - }
> - else
> - fprintf (f, " Unknown bits\n");
> -
> if (jump_func->m_vr)
> {
> jump_func->m_vr->dump (f);
> @@ -2267,39 +2213,6 @@ ipa_get_callee_param_type (struct cgraph
> return NULL;
> }
>
> -/* Return ipa_bits with VALUE and MASK values, which can be either a newly
> - allocated structure or a previously existing one shared with other jump
> - functions and/or transformation summaries. */
> -
> -ipa_bits *
> -ipa_get_ipa_bits_for_value (const widest_int &value, const widest_int &mask)
> -{
> - ipa_bits tmp;
> - tmp.value = value;
> - tmp.mask = mask;
> -
> - ipa_bits **slot = ipa_bits_hash_table->find_slot (&tmp, INSERT);
> - if (*slot)
> - return *slot;
> -
> - ipa_bits *res = ggc_alloc<ipa_bits> ();
> - res->value = value;
> - res->mask = mask;
> - *slot = res;
> -
> - return res;
> -}
> -
> -/* Assign to JF a pointer to ipa_bits structure with VALUE and MASK. Use
> hash
> - table in order to avoid creating multiple same ipa_bits structures. */
> -
> -static void
> -ipa_set_jfunc_bits (ipa_jump_func *jf, const widest_int &value,
> - const widest_int &mask)
> -{
> - jf->bits = ipa_get_ipa_bits_for_value (value, mask);
> -}
> -
> /* Return a pointer to an ipa_vr just like TMP, but either find it in
> ipa_vr_hash_table or allocate it in GC memory. */
>
> @@ -2393,10 +2306,31 @@ ipa_compute_jump_functions_for_edge (str
> addr_nonzero = true;
>
> if (addr_nonzero)
> + vr.set_nonzero (TREE_TYPE (arg));
> +
> + unsigned HOST_WIDE_INT bitpos;
> + unsigned align, prec = TYPE_PRECISION (TREE_TYPE (arg));
> +
> + get_pointer_alignment_1 (arg, &align, &bitpos);
> +
> + if (align > BITS_PER_UNIT
> + && opt_for_fn (cs->caller->decl, flag_ipa_bit_cp))
> {
> - vr.set_nonzero (TREE_TYPE (arg));
> + wide_int mask
> + = wi::bit_and_not (wi::mask (prec, false, prec),
> + wide_int::from (align / BITS_PER_UNIT - 1,
> + prec, UNSIGNED));
> + wide_int value = wide_int::from (bitpos / BITS_PER_UNIT, prec,
> + UNSIGNED);
> + irange_bitmask bm (value, mask);
> + if (!addr_nonzero)
> + vr.set_varying (TREE_TYPE (arg));
> + irange &r = as_a <irange> (vr);
> + r.update_bitmask (bm);
> ipa_set_jfunc_vr (jfunc, vr);
> }
> + else if (addr_nonzero)
> + ipa_set_jfunc_vr (jfunc, vr);
> else
> gcc_assert (!jfunc->m_vr);
> }
> @@ -2421,30 +2355,6 @@ ipa_compute_jump_functions_for_edge (str
> gcc_assert (!jfunc->m_vr);
> }
>
> - if (INTEGRAL_TYPE_P (TREE_TYPE (arg)) && !vr.undefined_p ())
> - {
> - irange &r = as_a <irange> (vr);
> - irange_bitmask bm = r.get_bitmask ();
> - signop sign = TYPE_SIGN (TREE_TYPE (arg));
> - ipa_set_jfunc_bits (jfunc,
> - widest_int::from (bm.value (), sign),
> - widest_int::from (bm.mask (), sign));
> - }
> - else if (POINTER_TYPE_P (TREE_TYPE (arg)))
> - {
> - unsigned HOST_WIDE_INT bitpos;
> - unsigned align;
> -
> - get_pointer_alignment_1 (arg, &align, &bitpos);
> - widest_int mask = wi::bit_and_not
> - (wi::mask<widest_int> (TYPE_PRECISION (TREE_TYPE (arg)), false),
> - align / BITS_PER_UNIT - 1);
> - widest_int value = bitpos / BITS_PER_UNIT;
> - ipa_set_jfunc_bits (jfunc, value, mask);
> - }
> - else
> - gcc_assert (!jfunc->bits);
> -
> if (is_gimple_ip_invariant (arg)
> || (VAR_P (arg)
> && is_global_var (arg)
> @@ -4398,8 +4308,6 @@ ipa_check_create_edge_args (void)
> ipa_edge_args_sum
> = (new (ggc_alloc_no_dtor<ipa_edge_args_sum_t> ())
> ipa_edge_args_sum_t (symtab, true));
> - if (!ipa_bits_hash_table)
> - ipa_bits_hash_table = hash_table<ipa_bit_ggc_hash_traits>::create_ggc
> (37);
> if (!ipa_vr_hash_table)
> ipa_vr_hash_table = hash_table<ipa_vr_ggc_hash_traits>::create_ggc (37);
> }
> @@ -4432,8 +4340,6 @@ ipa_free_all_node_params (void)
> void
> ipcp_transformation_initialize (void)
> {
> - if (!ipa_bits_hash_table)
> - ipa_bits_hash_table = hash_table<ipa_bit_ggc_hash_traits>::create_ggc
> (37);
> if (!ipa_vr_hash_table)
> ipa_vr_hash_table = hash_table<ipa_vr_ggc_hash_traits>::create_ggc (37);
> if (ipcp_transformation_sum == NULL)
> @@ -4636,7 +4542,6 @@ ipcp_transformation_t::duplicate(cgraph_
> if (dst->inlined_to)
> return;
> dst_trans->m_agg_values = vec_safe_copy (src_trans->m_agg_values);
> - dst_trans->bits = vec_safe_copy (src_trans->bits);
> dst_trans->m_vr = vec_safe_copy (src_trans->m_vr);
> }
>
> @@ -4859,13 +4764,6 @@ ipa_write_jump_function (struct output_b
> }
>
> bp = bitpack_create (ob->main_stream);
> - bp_pack_value (&bp, !!jump_func->bits, 1);
> - streamer_write_bitpack (&bp);
> - if (jump_func->bits)
> - {
> - streamer_write_widest_int (ob, jump_func->bits->value);
> - streamer_write_widest_int (ob, jump_func->bits->mask);
> - }
> if (jump_func->m_vr)
> jump_func->m_vr->streamer_write (ob);
> else
> @@ -4992,18 +4890,6 @@ ipa_read_jump_function (class lto_input_
> jump_func->agg.items->quick_push (item);
> }
>
> - struct bitpack_d bp = streamer_read_bitpack (ib);
> - bool bits_known = bp_unpack_value (&bp, 1);
> - if (bits_known)
> - {
> - widest_int value = streamer_read_widest_int (ib);
> - widest_int mask = streamer_read_widest_int (ib);
> - if (prevails)
> - ipa_set_jfunc_bits (jump_func, value, mask);
> - }
> - else
> - jump_func->bits = NULL;
> -
> ipa_vr vr;
> vr.streamer_read (ib, data_in);
> if (vr.known_p ())
> @@ -5387,7 +5273,6 @@ useful_ipcp_transformation_info_p (ipcp_
> if (!ts)
> return false;
> if (!vec_safe_is_empty (ts->m_agg_values)
> - || !vec_safe_is_empty (ts->bits)
> || !vec_safe_is_empty (ts->m_vr))
> return true;
> return false;
> @@ -5420,19 +5305,6 @@ write_ipcp_transformation_info (output_b
> streamer_write_uhwi (ob, vec_safe_length (ts->m_vr));
> for (const ipa_vr &parm_vr : ts->m_vr)
> parm_vr.streamer_write (ob);
> -
> - streamer_write_uhwi (ob, vec_safe_length (ts->bits));
> - for (const ipa_bits *bits_jfunc : ts->bits)
> - {
> - struct bitpack_d bp = bitpack_create (ob->main_stream);
> - bp_pack_value (&bp, !!bits_jfunc, 1);
> - streamer_write_bitpack (&bp);
> - if (bits_jfunc)
> - {
> - streamer_write_widest_int (ob, bits_jfunc->value);
> - streamer_write_widest_int (ob, bits_jfunc->mask);
> - }
> - }
> }
>
> /* Stream in the aggregate value replacement chain for NODE from IB. */
> @@ -5473,24 +5345,6 @@ read_ipcp_transformation_info (lto_input
> parm_vr->streamer_read (ib, data_in);
> }
> }
> - count = streamer_read_uhwi (ib);
> - if (count > 0)
> - {
> - vec_safe_grow_cleared (ts->bits, count, true);
> - for (i = 0; i < count; i++)
> - {
> - struct bitpack_d bp = streamer_read_bitpack (ib);
> - bool known = bp_unpack_value (&bp, 1);
> - if (known)
> - {
> - const widest_int value = streamer_read_widest_int (ib);
> - const widest_int mask = streamer_read_widest_int (ib);
> - ipa_bits *bits
> - = ipa_get_ipa_bits_for_value (value, mask);
> - (*ts->bits)[i] = bits;
> - }
> - }
> - }
> }
>
> /* Write all aggregate replacement for nodes in set. */
> @@ -5796,7 +5650,9 @@ ipcp_get_parm_bits (tree parm, tree *val
> {
> cgraph_node *cnode = cgraph_node::get (current_function_decl);
> ipcp_transformation *ts = ipcp_get_transformation_summary (cnode);
> - if (!ts || vec_safe_length (ts->bits) == 0)
> + if (!ts
> + || vec_safe_length (ts->m_vr) == 0
> + || !irange::supports_p (TREE_TYPE (parm)))
> return false;
>
> int i = ts->get_param_index (current_function_decl, parm);
> @@ -5810,120 +5666,20 @@ ipcp_get_parm_bits (tree parm, tree *val
> return false;
> }
>
> - vec<ipa_bits *, va_gc> &bits = *ts->bits;
> - if (!bits[i])
> + vec<ipa_vr, va_gc> &vr = *ts->m_vr;
> + if (!vr[i].known_p ())
> return false;
> - *mask = bits[i]->mask;
> - *value = wide_int_to_tree (TREE_TYPE (parm), bits[i]->value);
> + Value_Range tmp;
> + vr[i].get_vrange (tmp);
> + if (tmp.undefined_p () || tmp.varying_p ())
> + return false;
> + irange &r = as_a <irange> (tmp);
> + irange_bitmask bm = r.get_bitmask ();
> + *mask = widest_int::from (bm.mask (), TYPE_SIGN (TREE_TYPE (parm)));
> + *value = wide_int_to_tree (TREE_TYPE (parm), bm.value ());
> return true;
> }
>
> -/* Update bits info of formal parameters of NODE as described in TS. */
> -
> -static void
> -ipcp_update_bits (struct cgraph_node *node, ipcp_transformation *ts)
> -{
> - if (vec_safe_is_empty (ts->bits))
> - return;
> - vec<ipa_bits *, va_gc> &bits = *ts->bits;
> - unsigned count = bits.length ();
> - if (!count)
> - return;
> -
> - auto_vec<int, 16> new_indices;
> - bool need_remapping = false;
> - clone_info *cinfo = clone_info::get (node);
> - if (cinfo && cinfo->param_adjustments)
> - {
> - cinfo->param_adjustments->get_updated_indices (&new_indices);
> - need_remapping = true;
> - }
> - auto_vec <tree, 16> parm_decls;
> - push_function_arg_decls (&parm_decls, node->decl);
> -
> - for (unsigned i = 0; i < count; ++i)
> - {
> - tree parm;
> - if (need_remapping)
> - {
> - if (i >= new_indices.length ())
> - continue;
> - int idx = new_indices[i];
> - if (idx < 0)
> - continue;
> - parm = parm_decls[idx];
> - }
> - else
> - parm = parm_decls[i];
> - gcc_checking_assert (parm);
> -
> -
> - if (!bits[i]
> - || !(INTEGRAL_TYPE_P (TREE_TYPE (parm))
> - || POINTER_TYPE_P (TREE_TYPE (parm)))
> - || !is_gimple_reg (parm))
> - continue;
> -
> - tree ddef = ssa_default_def (DECL_STRUCT_FUNCTION (node->decl), parm);
> - if (!ddef)
> - continue;
> -
> - if (dump_file)
> - {
> - fprintf (dump_file, "Adjusting mask for param %u to ", i);
> - print_hex (bits[i]->mask, dump_file);
> - fprintf (dump_file, "\n");
> - }
> -
> - if (INTEGRAL_TYPE_P (TREE_TYPE (ddef)))
> - {
> - unsigned prec = TYPE_PRECISION (TREE_TYPE (ddef));
> - signop sgn = TYPE_SIGN (TREE_TYPE (ddef));
> - wide_int mask = wide_int::from (bits[i]->mask, prec, UNSIGNED);
> - wide_int value = wide_int::from (bits[i]->value, prec, sgn);
> - set_bitmask (ddef, value, mask);
> - }
> - else
> - {
> - unsigned tem = bits[i]->mask.to_uhwi ();
> - unsigned HOST_WIDE_INT bitpos = bits[i]->value.to_uhwi ();
> - unsigned align = tem & -tem;
> - unsigned misalign = bitpos & (align - 1);
> -
> - if (align > 1)
> - {
> - if (dump_file)
> - fprintf (dump_file, "Adjusting align: %u, misalign: %u\n",
> align, misalign);
> -
> - unsigned old_align, old_misalign;
> - struct ptr_info_def *pi = get_ptr_info (ddef);
> - bool old_known = get_ptr_info_alignment (pi, &old_align,
> &old_misalign);
> -
> - if (old_known
> - && old_align > align)
> - {
> - if (dump_file)
> - {
> - fprintf (dump_file, "But alignment was already %u.\n",
> old_align);
> - if ((old_misalign & (align - 1)) != misalign)
> - fprintf (dump_file, "old_misalign (%u) and misalign
> (%u) mismatch\n",
> - old_misalign, misalign);
> - }
> - continue;
> - }
> -
> - if (old_known
> - && ((misalign & (old_align - 1)) != old_misalign)
> - && dump_file)
> - fprintf (dump_file, "old_misalign (%u) and misalign (%u)
> mismatch\n",
> - old_misalign, misalign);
> -
> - set_ptr_info_alignment (pi, align, misalign);
> - }
> - }
> - }
> -}
> -
> /* Update value range of formal parameters of NODE as described in TS. */
>
> static void
> @@ -5985,6 +5741,77 @@ ipcp_update_vr (struct cgraph_node *node
> fprintf (dump_file, "]\n");
> }
> set_range_info (ddef, tmp);
> +
> + if (POINTER_TYPE_P (TREE_TYPE (parm))
> + && opt_for_fn (node->decl, flag_ipa_bit_cp))
> + {
> + irange &r = as_a<irange> (tmp);
> + irange_bitmask bm = r.get_bitmask ();
> + unsigned tem = bm.mask ().to_uhwi ();
> + unsigned HOST_WIDE_INT bitpos = bm.value ().to_uhwi ();
> + unsigned align = tem & -tem;
> + unsigned misalign = bitpos & (align - 1);
> +
> + if (align > 1)
> + {
> + if (dump_file)
> + {
> + fprintf (dump_file,
> + "Adjusting mask for param %u to ", i);
> + print_hex (bm.mask (), dump_file);
> + fprintf (dump_file, "\n");
> + }
> +
> + if (dump_file)
> + fprintf (dump_file,
> + "Adjusting align: %u, misalign: %u\n",
> + align, misalign);
> +
> + unsigned old_align, old_misalign;
> + struct ptr_info_def *pi = get_ptr_info (ddef);
> + bool old_known = get_ptr_info_alignment (pi, &old_align,
> + &old_misalign);
> +
> + if (old_known && old_align > align)
> + {
> + if (dump_file)
> + {
> + fprintf (dump_file,
> + "But alignment was already %u.\n",
> + old_align);
> + if ((old_misalign & (align - 1)) != misalign)
> + fprintf (dump_file,
> + "old_misalign (%u) and misalign "
> + "(%u) mismatch\n",
> + old_misalign, misalign);
> + }
> + continue;
> + }
> +
> + if (dump_file
> + && old_known
> + && ((misalign & (old_align - 1)) != old_misalign))
> + fprintf (dump_file,
> + "old_misalign (%u) and misalign (%u) "
> + "mismatch\n",
> + old_misalign, misalign);
> +
> + set_ptr_info_alignment (pi, align, misalign);
> + }
> + }
> + else if (dump_file && INTEGRAL_TYPE_P (TREE_TYPE (parm)))
> + {
> + irange &r = as_a<irange> (tmp);
> + irange_bitmask bm = r.get_bitmask ();
> + unsigned prec = TYPE_PRECISION (TREE_TYPE (parm));
> + if (wi::ne_p (bm.mask (), wi::shwi (-1, prec)))
> + {
> + fprintf (dump_file,
> + "Adjusting mask for param %u to ", i);
> + print_hex (bm.mask (), dump_file);
> + fprintf (dump_file, "\n");
> + }
> + }
> }
> }
> }
> @@ -6008,12 +5835,10 @@ ipcp_transform_function (struct cgraph_n
> ipcp_transformation *ts = ipcp_get_transformation_summary (node);
> if (!ts
> || (vec_safe_is_empty (ts->m_agg_values)
> - && vec_safe_is_empty (ts->bits)
> && vec_safe_is_empty (ts->m_vr)))
> return 0;
>
> ts->maybe_create_parm_idx_map (cfun->decl);
> - ipcp_update_bits (node, ts);
> ipcp_update_vr (node, ts);
> if (vec_safe_is_empty (ts->m_agg_values))
> return 0;
> --- gcc/ipa-cp.cc.jj 2023-10-05 11:32:39.828744703 +0200
> +++ gcc/ipa-cp.cc 2023-10-05 11:36:45.408378045 +0200
> @@ -2749,11 +2749,22 @@ propagate_bits_across_jump_function (cgr
> }
> }
>
> - if (jfunc->bits)
> - return dest_lattice->meet_with (jfunc->bits->value, jfunc->bits->mask,
> - precision);
> - else
> - return dest_lattice->set_to_bottom ();
> + Value_Range vr (parm_type);
> + if (jfunc->m_vr)
> + {
> + jfunc->m_vr->get_vrange (vr);
> + if (!vr.undefined_p () && !vr.varying_p ())
> + {
> + irange &r = as_a <irange> (vr);
> + irange_bitmask bm = r.get_bitmask ();
> + widest_int mask
> + = widest_int::from (bm.mask (), TYPE_SIGN (parm_type));
> + widest_int value
> + = widest_int::from (bm.value (), TYPE_SIGN (parm_type));
> + return dest_lattice->meet_with (value, mask, precision);
> + }
> + }
> + return dest_lattice->set_to_bottom ();
> }
>
> /* Propagate value range across jump function JFUNC that is associated with
> @@ -6521,89 +6532,8 @@ ipcp_decision_stage (class ipa_topo_info
> }
> }
>
> -/* Look up all the bits information that we have discovered and copy it over
> - to the transformation summary. */
> -
> -static void
> -ipcp_store_bits_results (void)
> -{
> - cgraph_node *node;
> -
> - FOR_EACH_FUNCTION_WITH_GIMPLE_BODY (node)
> - {
> - ipa_node_params *info = ipa_node_params_sum->get (node);
> - bool dumped_sth = false;
> - bool found_useful_result = false;
> -
> - if (!opt_for_fn (node->decl, flag_ipa_bit_cp) || !info)
> - {
> - if (dump_file)
> - fprintf (dump_file, "Not considering %s for ipa bitwise propagation
> "
> - "; -fipa-bit-cp: disabled.\n",
> - node->dump_name ());
> - continue;
> - }
> -
> - if (info->ipcp_orig_node)
> - info = ipa_node_params_sum->get (info->ipcp_orig_node);
> - if (!info->lattices)
> - /* Newly expanded artificial thunks do not have lattices. */
> - continue;
> -
> - unsigned count = ipa_get_param_count (info);
> - for (unsigned i = 0; i < count; i++)
> - {
> - ipcp_param_lattices *plats = ipa_get_parm_lattices (info, i);
> - if (plats->bits_lattice.constant_p ())
> - {
> - found_useful_result = true;
> - break;
> - }
> - }
> -
> - if (!found_useful_result)
> - continue;
> -
> - ipcp_transformation_initialize ();
> - ipcp_transformation *ts = ipcp_transformation_sum->get_create (node);
> - vec_safe_reserve_exact (ts->bits, count);
> -
> - for (unsigned i = 0; i < count; i++)
> - {
> - ipcp_param_lattices *plats = ipa_get_parm_lattices (info, i);
> - ipa_bits *jfbits;
> -
> - if (plats->bits_lattice.constant_p ())
> - {
> - jfbits
> - = ipa_get_ipa_bits_for_value (plats->bits_lattice.get_value (),
> - plats->bits_lattice.get_mask ());
> - if (!dbg_cnt (ipa_cp_bits))
> - jfbits = NULL;
> - }
> - else
> - jfbits = NULL;
> -
> - ts->bits->quick_push (jfbits);
> - if (!dump_file || !jfbits)
> - continue;
> - if (!dumped_sth)
> - {
> - fprintf (dump_file, "Propagated bits info for function %s:\n",
> - node->dump_name ());
> - dumped_sth = true;
> - }
> - fprintf (dump_file, " param %i: value = ", i);
> - print_hex (jfbits->value, dump_file);
> - fprintf (dump_file, ", mask = ");
> - print_hex (jfbits->mask, dump_file);
> - fprintf (dump_file, "\n");
> - }
> - }
> -}
> -
> -/* Look up all VR information that we have discovered and copy it over
> - to the transformation summary. */
> +/* Look up all VR and bits information that we have discovered and copy it
> + over to the transformation summary. */
>
> static void
> ipcp_store_vr_results (void)
> @@ -6613,7 +6543,10 @@ ipcp_store_vr_results (void)
> FOR_EACH_FUNCTION_WITH_GIMPLE_BODY (node)
> {
> ipa_node_params *info = ipa_node_params_sum->get (node);
> + bool dumped_sth = false;
> bool found_useful_result = false;
> + bool do_vr = true;
> + bool do_bits = true;
>
> if (!info || !opt_for_fn (node->decl, flag_ipa_vrp))
> {
> @@ -6621,8 +6554,18 @@ ipcp_store_vr_results (void)
> fprintf (dump_file, "Not considering %s for VR discovery "
> "and propagate; -fipa-ipa-vrp: disabled.\n",
> node->dump_name ());
> - continue;
> + do_vr = false;
> + }
> + if (!info || !opt_for_fn (node->decl, flag_ipa_bit_cp))
> + {
> + if (dump_file)
> + fprintf (dump_file, "Not considering %s for ipa bitwise "
> + "propagation ; -fipa-bit-cp: disabled.\n",
> + node->dump_name ());
> + do_bits = false;
> }
> + if (!do_bits && !do_vr)
> + continue;
>
> if (info->ipcp_orig_node)
> info = ipa_node_params_sum->get (info->ipcp_orig_node);
> @@ -6634,12 +6577,18 @@ ipcp_store_vr_results (void)
> for (unsigned i = 0; i < count; i++)
> {
> ipcp_param_lattices *plats = ipa_get_parm_lattices (info, i);
> - if (!plats->m_value_range.bottom_p ()
> + if (do_vr
> + && !plats->m_value_range.bottom_p ()
> && !plats->m_value_range.top_p ())
> {
> found_useful_result = true;
> break;
> }
> + if (do_bits && plats->bits_lattice.constant_p ())
> + {
> + found_useful_result = true;
> + break;
> + }
> }
> if (!found_useful_result)
> continue;
> @@ -6651,12 +6600,53 @@ ipcp_store_vr_results (void)
> for (unsigned i = 0; i < count; i++)
> {
> ipcp_param_lattices *plats = ipa_get_parm_lattices (info, i);
> + ipcp_bits_lattice *bits = NULL;
> +
> + if (do_bits
> + && plats->bits_lattice.constant_p ()
> + && dbg_cnt (ipa_cp_bits))
> + bits = &plats->bits_lattice;
>
> - if (!plats->m_value_range.bottom_p ()
> + if (do_vr
> + && !plats->m_value_range.bottom_p ()
> && !plats->m_value_range.top_p ()
> && dbg_cnt (ipa_cp_vr))
> {
> - ipa_vr vr (plats->m_value_range.m_vr);
> + if (bits)
> + {
> + Value_Range tmp = plats->m_value_range.m_vr;
> + tree type = ipa_get_type (info, i);
> + irange &r = as_a<irange> (tmp);
> + irange_bitmask bm (wide_int::from (bits->get_value (),
> + TYPE_PRECISION (type),
> + TYPE_SIGN (type)),
> + wide_int::from (bits->get_mask (),
> + TYPE_PRECISION (type),
> + TYPE_SIGN (type)));
> + r.update_bitmask (bm);
> + ipa_vr vr (tmp);
> + ts->m_vr->quick_push (vr);
> + }
> + else
> + {
> + ipa_vr vr (plats->m_value_range.m_vr);
> + ts->m_vr->quick_push (vr);
> + }
> + }
> + else if (bits)
> + {
> + tree type = ipa_get_type (info, i);
> + Value_Range tmp;
> + tmp.set_varying (type);
> + irange &r = as_a<irange> (tmp);
> + irange_bitmask bm (wide_int::from (bits->get_value (),
> + TYPE_PRECISION (type),
> + TYPE_SIGN (type)),
> + wide_int::from (bits->get_mask (),
> + TYPE_PRECISION (type),
> + TYPE_SIGN (type)));
> + r.update_bitmask (bm);
> + ipa_vr vr (tmp);
> ts->m_vr->quick_push (vr);
> }
> else
> @@ -6664,6 +6654,21 @@ ipcp_store_vr_results (void)
> ipa_vr vr;
> ts->m_vr->quick_push (vr);
> }
> +
> + if (!dump_file || !bits)
> + continue;
> +
> + if (!dumped_sth)
> + {
> + fprintf (dump_file, "Propagated bits info for function %s:\n",
> + node->dump_name ());
> + dumped_sth = true;
> + }
> + fprintf (dump_file, " param %i: value = ", i);
> + print_hex (bits->get_value (), dump_file);
> + fprintf (dump_file, ", mask = ");
> + print_hex (bits->get_mask (), dump_file);
> + fprintf (dump_file, "\n");
> }
> }
> }
> @@ -6696,9 +6701,7 @@ ipcp_driver (void)
> ipcp_propagate_stage (&topo);
> /* Decide what constant propagation and cloning should be performed. */
> ipcp_decision_stage (&topo);
> - /* Store results of bits propagation. */
> - ipcp_store_bits_results ();
> - /* Store results of value range propagation. */
> + /* Store results of value range and bits propagation. */
> ipcp_store_vr_results ();
>
> /* Free all IPCP structures. */
> --- gcc/ipa-sra.cc.jj 2023-10-05 11:32:40.233739151 +0200
> +++ gcc/ipa-sra.cc 2023-10-05 11:36:45.408378045 +0200
> @@ -4134,22 +4134,8 @@ zap_useless_ipcp_results (const isra_fun
> else if (removed_item)
> ts->m_agg_values->truncate (dst_index);
>
> - bool useful_bits = false;
> - unsigned count = vec_safe_length (ts->bits);
> - for (unsigned i = 0; i < count; i++)
> - if ((*ts->bits)[i])
> - {
> - const isra_param_desc *desc = &(*ifs->m_parameters)[i];
> - if (desc->locally_unused)
> - (*ts->bits)[i] = NULL;
> - else
> - useful_bits = true;
> - }
> - if (!useful_bits)
> - ts->bits = NULL;
> -
> bool useful_vr = false;
> - count = vec_safe_length (ts->m_vr);
> + unsigned count = vec_safe_length (ts->m_vr);
> for (unsigned i = 0; i < count; i++)
> if ((*ts->m_vr)[i].known_p ())
> {
>
> Jakub
>
>
--
Richard Biener <rguent...@suse.de>
SUSE Software Solutions Germany GmbH,
Frankenstrasse 146, 90461 Nuernberg, Germany;
GF: Ivo Totev, Andrew McDonald, Werner Knoblich; (HRB 36809, AG Nuernberg)
