public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Richard Biener <richard.guenther@gmail.com>
To: Patrick Palka <ppalka@redhat.com>
Cc: gcc-patches@gcc.gnu.org, nathan@acm.org
Subject: Re: [PATCH 2/2] c++ modules: always stream TYPE_MIN/MAX_VALUE for enums [PR106848]
Date: Wed, 19 Oct 2022 09:31:32 +0200	[thread overview]
Message-ID: <CAFiYyc1aCEMoA2LN6LyZCA+EJx95dMpJyU1SSFZY+ojiJne6qg@mail.gmail.com> (raw)
In-Reply-To: <20221018181050.1629201-2-ppalka@redhat.com>

On Tue, Oct 18, 2022 at 8:11 PM Patrick Palka via Gcc-patches
<gcc-patches@gcc.gnu.org> wrote:
>
> We currently stream TYPE_MIN/MAX_VALUE of an enum only if the enum is
> defined rather than just declared.  But that seems inconsistent with
> what start_enum does, which always sets TYPE_MIN/MAX_VALUE on the
> ENUMERAL_TYPE (via copy_type_enum) even for an opaque enum that lacks a
> definition.
>
> This patch makes us stream TYPE_MIN/MAX_VALUE as part of the
> ENUMERAL_TYPE rather than as part of the defining TYPE_DECL so that
> TYPE_MIN/MAX_VALUE gets set even for opaque enums.
>
> Incidentally, this turns out to fix the below testcase in which during
> stream in we end up having created a type variant for the enum before we
> read the enum's definition, and thus the variant inherited stale
> TYPE_MIN/MAX_VALUE, which leads to an ICE from verify_type (with -g).
> (The stale variant got created from set_underlying_type during earlier
> stream in of the redundant typedef for the enum.)  By streaming these
> fields as part of the ENUMERAL_TYPE, we guarantee they won't be stale.
>
> Tested on x86_64-pc-linux-gnu, does this look OK for trunk?

Looks reasonable to me.

Thanks,
Richard.

>         PR c++/106848
>
> gcc/cp/ChangeLog:
>
>         * module.cc (trees_out::core_vals): Stream TYPE_MAX_VALUE and
>         TYPE_MIN_VALUE of ENUMERAL_TYPE.
>         (trees_in::core_vals): Likewise.
>         (trees_out::write_enum_def): Don't stream them here.
>         (trees_in::read_enum_def): Likewise.
>
> gcc/testsuite/ChangeLog:
>
>         * g++.dg/modules/enum-10_a.H: New test.
>         * g++.dg/modules/enum-10_b.C: New test.
> ---
>  gcc/cp/module.cc                         | 29 ++++++++++++------------
>  gcc/testsuite/g++.dg/modules/enum-10_a.H |  5 ++++
>  gcc/testsuite/g++.dg/modules/enum-10_b.C |  6 +++++
>  3 files changed, 25 insertions(+), 15 deletions(-)
>  create mode 100644 gcc/testsuite/g++.dg/modules/enum-10_a.H
>  create mode 100644 gcc/testsuite/g++.dg/modules/enum-10_b.C
>
> diff --git a/gcc/cp/module.cc b/gcc/cp/module.cc
> index cc704817718..bb406a5cf01 100644
> --- a/gcc/cp/module.cc
> +++ b/gcc/cp/module.cc
> @@ -6016,9 +6016,14 @@ trees_out::core_vals (tree t)
>
>    if (CODE_CONTAINS_STRUCT (code, TS_TYPE_NON_COMMON))
>      {
> +      if (code == ENUMERAL_TYPE)
> +       {
> +         WT (t->type_non_common.maxval);
> +         WT (t->type_non_common.minval);
> +       }
>        /* Records and unions hold FIELDS, VFIELD & BINFO on these
>          things.  */
> -      if (!RECORD_OR_UNION_CODE_P (code) && code != ENUMERAL_TYPE)
> +      else if (!RECORD_OR_UNION_CODE_P (code))
>         {
>           // FIXME: These are from tpl_parm_value's 'type' writing.
>           // Perhaps it should just be doing them directly?
> @@ -6529,9 +6534,14 @@ trees_in::core_vals (tree t)
>
>    if (CODE_CONTAINS_STRUCT (code, TS_TYPE_NON_COMMON))
>      {
> +      if (code == ENUMERAL_TYPE)
> +       {
> +         RT (t->type_non_common.maxval);
> +         RT (t->type_non_common.minval);
> +       }
>        /* Records and unions hold FIELDS, VFIELD & BINFO on these
>          things.  */
> -      if (!RECORD_OR_UNION_CODE_P (code) && code != ENUMERAL_TYPE)
> +      else if (!RECORD_OR_UNION_CODE_P (code))
>         {
>           /* This is not clobbering TYPE_CACHED_VALUES, because this
>              is a type that doesn't have any.  */
> @@ -12217,8 +12227,6 @@ trees_out::write_enum_def (tree decl)
>    tree type = TREE_TYPE (decl);
>
>    tree_node (TYPE_VALUES (type));
> -  tree_node (TYPE_MIN_VALUE (type));
> -  tree_node (TYPE_MAX_VALUE (type));
>  }
>
>  void
> @@ -12242,8 +12250,6 @@ trees_in::read_enum_def (tree defn, tree maybe_template)
>  {
>    tree type = TREE_TYPE (defn);
>    tree values = tree_node ();
> -  tree min = tree_node ();
> -  tree max = tree_node ();
>
>    if (get_overrun ())
>      return false;
> @@ -12254,8 +12260,6 @@ trees_in::read_enum_def (tree defn, tree maybe_template)
>    if (installing)
>      {
>        TYPE_VALUES (type) = values;
> -      TYPE_MIN_VALUE (type) = min;
> -      TYPE_MAX_VALUE (type) = max;
>
>        rest_of_type_compilation (type, DECL_NAMESPACE_SCOPE_P (defn));
>      }
> @@ -12269,22 +12273,17 @@ trees_in::read_enum_def (tree defn, tree maybe_template)
>           tree new_decl = TREE_VALUE (values);
>
>           if (DECL_NAME (known_decl) != DECL_NAME (new_decl))
> -           goto bad;
> +           break;
>
>           new_decl = maybe_duplicate (new_decl);
>
>           if (!cp_tree_equal (DECL_INITIAL (known_decl),
>                               DECL_INITIAL (new_decl)))
> -           goto bad;
> +           break;
>         }
>
>        if (known || values)
> -       goto bad;
> -
> -      if (!cp_tree_equal (TYPE_MIN_VALUE (type), min)
> -         || !cp_tree_equal (TYPE_MAX_VALUE (type), max))
>         {
> -       bad:;
>           error_at (DECL_SOURCE_LOCATION (maybe_dup),
>                     "definition of %qD does not match", maybe_dup);
>           inform (DECL_SOURCE_LOCATION (defn),
> diff --git a/gcc/testsuite/g++.dg/modules/enum-10_a.H b/gcc/testsuite/g++.dg/modules/enum-10_a.H
> new file mode 100644
> index 00000000000..fb7d10ad3b6
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/modules/enum-10_a.H
> @@ -0,0 +1,5 @@
> +// PR c++/106848
> +// { dg-additional-options -fmodule-header }
> +// { dg-module-cmi {} }
> +
> +typedef enum memory_order { memory_order_seq_cst } memory_order;
> diff --git a/gcc/testsuite/g++.dg/modules/enum-10_b.C b/gcc/testsuite/g++.dg/modules/enum-10_b.C
> new file mode 100644
> index 00000000000..76dc3152963
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/modules/enum-10_b.C
> @@ -0,0 +1,6 @@
> +// PR c++/106848
> +// { dg-additional-options "-fmodules-ts -g" }
> +
> +import "enum-10_a.H";
> +
> +memory_order x = memory_order_seq_cst;
> --
> 2.38.0.118.g4732897cf0
>

  reply	other threads:[~2022-10-19  7:31 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-10-18 18:10 [PATCH 1/2] c++ modules: streaming enum with no enumerators [PR102600] Patrick Palka
2022-10-18 18:10 ` [PATCH 2/2] c++ modules: always stream TYPE_MIN/MAX_VALUE for enums [PR106848] Patrick Palka
2022-10-19  7:31   ` Richard Biener [this message]
2022-10-18 18:54 ` [PATCH 1/2] c++ modules: streaming enum with no enumerators [PR102600] Patrick Palka

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=CAFiYyc1aCEMoA2LN6LyZCA+EJx95dMpJyU1SSFZY+ojiJne6qg@mail.gmail.com \
    --to=richard.guenther@gmail.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=nathan@acm.org \
    --cc=ppalka@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).