public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* [PATCH][gcc][middle-end] PR104498: Fix comparing symbol reference
@ 2022-02-16 11:12 Andre Vieira (lists)
  2022-02-17  9:54 ` Richard Biener
  0 siblings, 1 reply; 2+ messages in thread
From: Andre Vieira (lists) @ 2022-02-16 11:12 UTC (permalink / raw)
  To: gcc-patches; +Cc: Richard Sandiford, Richard Biener

[-- Attachment #1: Type: text/plain, Size: 604 bytes --]

Hi,

As reported on PR104498, the issue here is that when 
compare_base_symbol_refs swaps x and y but doesn't take that into 
account when computing the distance.
This patch makes sure that if x and y are swapped, we correct the 
distance computation by multiplying it by -1 to end up with the correct 
expected result of the original Y_BASE - X_BASE.

Bootstrapped and regression tested on aarch64-none-linux.

OK for trunk?

gcc/ChangeLog:

         PR middle-end/104498
         * alias.cc (compare_base_symbol_refs): Correct distance 
computation when
         swapping x and y.

[-- Attachment #2: pr104498.patch --]
[-- Type: text/plain, Size: 1152 bytes --]

diff --git a/gcc/alias.cc b/gcc/alias.cc
index 3fd71cff2e2b488bc39fcf7d937e118b96f491ab..8c08452e0acfcbf1bfd8fd2e8cd420b5b929d6b4 100644
--- a/gcc/alias.cc
+++ b/gcc/alias.cc
@@ -2195,6 +2195,7 @@ compare_base_symbol_refs (const_rtx x_base, const_rtx y_base,
   tree x_decl = SYMBOL_REF_DECL (x_base);
   tree y_decl = SYMBOL_REF_DECL (y_base);
   bool binds_def = true;
+  bool swap = false;
 
   if (XSTR (x_base, 0) == XSTR (y_base, 0))
     return 1;
@@ -2204,6 +2205,7 @@ compare_base_symbol_refs (const_rtx x_base, const_rtx y_base,
     {
       if (!x_decl)
 	{
+	  swap = true;
 	  std::swap (x_decl, y_decl);
 	  std::swap (x_base, y_base);
 	}
@@ -2238,8 +2240,8 @@ compare_base_symbol_refs (const_rtx x_base, const_rtx y_base,
       if (SYMBOL_REF_BLOCK (x_base) != SYMBOL_REF_BLOCK (y_base))
 	return 0;
       if (distance)
-	*distance += (SYMBOL_REF_BLOCK_OFFSET (y_base)
-		      - SYMBOL_REF_BLOCK_OFFSET (x_base));
+	*distance += (swap ? -1 : 1) * (SYMBOL_REF_BLOCK_OFFSET (y_base)
+					- SYMBOL_REF_BLOCK_OFFSET (x_base));
       return binds_def ? 1 : -1;
     }
   /* Either the symbols are equal (via aliasing) or they refer to

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [PATCH][gcc][middle-end] PR104498: Fix comparing symbol reference
  2022-02-16 11:12 [PATCH][gcc][middle-end] PR104498: Fix comparing symbol reference Andre Vieira (lists)
@ 2022-02-17  9:54 ` Richard Biener
  0 siblings, 0 replies; 2+ messages in thread
From: Richard Biener @ 2022-02-17  9:54 UTC (permalink / raw)
  To: Andre Vieira (lists); +Cc: gcc-patches, Richard Sandiford

On Wed, 16 Feb 2022, Andre Vieira (lists) wrote:

> Hi,
> 
> As reported on PR104498, the issue here is that when compare_base_symbol_refs
> swaps x and y but doesn't take that into account when computing the distance.
> This patch makes sure that if x and y are swapped, we correct the distance
> computation by multiplying it by -1 to end up with the correct expected result
> of the original Y_BASE - X_BASE.
> 
> Bootstrapped and regression tested on aarch64-none-linux.
> 
> OK for trunk?

OK.

Thanks,
Richard.

> gcc/ChangeLog:
> 
>         PR middle-end/104498
>         * alias.cc (compare_base_symbol_refs): Correct distance 
> computation when
>         swapping x and y.
> 

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2022-02-17  9:54 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-02-16 11:12 [PATCH][gcc][middle-end] PR104498: Fix comparing symbol reference Andre Vieira (lists)
2022-02-17  9:54 ` Richard Biener

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).