public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Richard Biener <rguenther@suse.de>
To: Jakub Jelinek <jakub@redhat.com>
Cc: Jason Merrill <jason@redhat.com>, gcc-patches@gcc.gnu.org
Subject: Re: [PATCH] dwarf2out: Fix up CONST_WIDE_INT handling once more [PR103046]
Date: Wed, 3 Nov 2021 14:38:03 +0100 (CET)	[thread overview]
Message-ID: <83s743q4-orr4-o0r1-8oo1-38q014p183n7@fhfr.qr> (raw)
In-Reply-To: <20211103131456.GN304296@tucnak>

On Wed, 3 Nov 2021, Jakub Jelinek wrote:

> On Wed, Nov 03, 2021 at 01:59:27PM +0100, Richard Biener wrote:
> > >      case CONST_WIDE_INT:
> > > -      {
> > > -	wide_int w1 = rtx_mode_t (rtl, MAX_MODE_INT);
> > > -	unsigned int prec = MIN (wi::min_precision (w1, UNSIGNED),
> > > -				 (unsigned int) CONST_WIDE_INT_NUNITS (rtl)
> > > -				 * HOST_BITS_PER_WIDE_INT);
> > > -	wide_int w = wide_int::from (w1, prec, UNSIGNED);
> > > -	add_AT_wide (die, DW_AT_const_value, w);
> > > -      }
> > > +      if (is_int_mode (mode, &int_mode)
> > > +	  && (GET_MODE_PRECISION (int_mode)
> > > +	      & (HOST_BITS_PER_WIDE_INT - 1)) == 0)
> > > +	{
> > > +	  wide_int w = rtx_mode_t (rtl, int_mode);
> > > +	  add_AT_wide (die, DW_AT_const_value, w);
> > > +	}
> > 
> > ... this at least deserves a comment why we return true doing
> > nothing when the mode precision isn't a multiple of 64.  The other
> > patch simply extended things.
> 
> Oops, that is obviously a bug.  For non-integral mode or mode with
> precision not a multiple of 64 I meant to punt.  The reason for the latter
> is that if precision is something weird like 123 bits, then the printing
> code will print the whole 64-bit elt which would be wrong for at least
> unsigned numbers with MSB set.  I guess such modes are still very rare to
> non-existing.

Agreed.  The patch looks good to me now.

Thanks,
Richard.

> 2021-11-03  Jakub Jelinek  <jakub@redhat.com>
> 
> 	PR debug/103046
> 	* dwarf2out.c (add_const_value_attribute): Add MODE argument, use it
> 	in CONST_WIDE_INT handling.  Adjust recursive calls.
> 	(add_location_or_const_value_attribute): Pass DECL_MODE (decl) to
> 	new add_const_value_attribute argument.
> 	(tree_add_const_value_attribute): Pass TYPE_MODE (type) to new
> 	add_const_value_attribute argument.
> 
> --- gcc/dwarf2out.c.jj	2021-10-08 10:52:47.086531820 +0200
> +++ gcc/dwarf2out.c	2021-11-03 14:06:22.750387774 +0100
> @@ -3854,7 +3854,7 @@ static void add_AT_location_description
>  					 dw_loc_list_ref);
>  static void add_data_member_location_attribute (dw_die_ref, tree,
>  						struct vlr_context *);
> -static bool add_const_value_attribute (dw_die_ref, rtx);
> +static bool add_const_value_attribute (dw_die_ref, machine_mode, rtx);
>  static void insert_int (HOST_WIDE_INT, unsigned, unsigned char *);
>  static void insert_wide_int (const wide_int &, unsigned char *, int);
>  static unsigned insert_float (const_rtx, unsigned char *);
> @@ -20081,8 +20081,10 @@ insert_float (const_rtx rtl, unsigned ch
>     constants do not necessarily get memory "homes".  */
>  
>  static bool
> -add_const_value_attribute (dw_die_ref die, rtx rtl)
> +add_const_value_attribute (dw_die_ref die, machine_mode mode, rtx rtl)
>  {
> +  scalar_mode int_mode;
> +
>    switch (GET_CODE (rtl))
>      {
>      case CONST_INT:
> @@ -20097,15 +20099,15 @@ add_const_value_attribute (dw_die_ref di
>        return true;
>  
>      case CONST_WIDE_INT:
> -      {
> -	wide_int w1 = rtx_mode_t (rtl, MAX_MODE_INT);
> -	unsigned int prec = MIN (wi::min_precision (w1, UNSIGNED),
> -				 (unsigned int) CONST_WIDE_INT_NUNITS (rtl)
> -				 * HOST_BITS_PER_WIDE_INT);
> -	wide_int w = wide_int::from (w1, prec, UNSIGNED);
> -	add_AT_wide (die, DW_AT_const_value, w);
> -      }
> -      return true;
> +      if (is_int_mode (mode, &int_mode)
> +	  && (GET_MODE_PRECISION (int_mode)
> +	      & (HOST_BITS_PER_WIDE_INT - 1)) == 0)
> +	{
> +	  wide_int w = rtx_mode_t (rtl, int_mode);
> +	  add_AT_wide (die, DW_AT_const_value, w);
> +	  return true;
> +	}
> +      return false;
>  
>      case CONST_DOUBLE:
>        /* Note that a CONST_DOUBLE rtx could represent either an integer or a
> @@ -20190,7 +20192,7 @@ add_const_value_attribute (dw_die_ref di
>  
>      case CONST:
>        if (CONSTANT_P (XEXP (rtl, 0)))
> -	return add_const_value_attribute (die, XEXP (rtl, 0));
> +	return add_const_value_attribute (die, mode, XEXP (rtl, 0));
>        /* FALLTHROUGH */
>      case SYMBOL_REF:
>        if (!const_ok_for_output (rtl))
> @@ -20703,7 +20705,7 @@ add_location_or_const_value_attribute (d
>  
>    rtl = rtl_for_decl_location (decl);
>    if (rtl && (CONSTANT_P (rtl) || GET_CODE (rtl) == CONST_STRING)
> -      && add_const_value_attribute (die, rtl))
> +      && add_const_value_attribute (die, DECL_MODE (decl), rtl))
>      return true;
>  
>    /* See if we have single element location list that is equivalent to
> @@ -20724,7 +20726,7 @@ add_location_or_const_value_attribute (d
>        if (GET_CODE (rtl) == EXPR_LIST)
>  	rtl = XEXP (rtl, 0);
>        if ((CONSTANT_P (rtl) || GET_CODE (rtl) == CONST_STRING)
> -	  && add_const_value_attribute (die, rtl))
> +	  && add_const_value_attribute (die, DECL_MODE (decl), rtl))
>  	 return true;
>      }
>    /* If this decl is from BLOCK_NONLOCALIZED_VARS, we might need its
> @@ -20799,7 +20801,7 @@ tree_add_const_value_attribute (dw_die_r
>       symbols.  */
>    rtl = rtl_for_decl_init (init, type);
>    if (rtl && !early_dwarf)
> -    return add_const_value_attribute (die, rtl);
> +    return add_const_value_attribute (die, TYPE_MODE (type), rtl);
>    /* If the host and target are sane, try harder.  */
>    if (CHAR_BIT == 8 && BITS_PER_UNIT == 8
>        && initializer_constant_valid_p (init, type))
> 
> 
> 	Jakub
> 
> 

-- 
Richard Biener <rguenther@suse.de>
SUSE Software Solutions Germany GmbH, Maxfeldstrasse 5, 90409 Nuernberg,
Germany; GF: Ivo Totev; HRB 36809 (AG Nuernberg)

      reply	other threads:[~2021-11-03 13:38 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-11-03  8:35 Jakub Jelinek
2021-11-03  9:26 ` Jakub Jelinek
2021-11-03 12:59   ` Richard Biener
2021-11-03 13:14     ` Jakub Jelinek
2021-11-03 13:38       ` Richard Biener [this message]

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=83s743q4-orr4-o0r1-8oo1-38q014p183n7@fhfr.qr \
    --to=rguenther@suse.de \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=jakub@redhat.com \
    --cc=jason@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).