public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Simon Baldwin <simonb@google.com>
To: Diego Novillo <dnovillo@google.com>
Cc: gcc-patches@gcc.gnu.org
Subject: Re: [LTO] PATCH: Fix for ICE vector VEC(cgraph_node_ptr,base) index  domain error
Date: Thu, 23 Oct 2008 15:56:00 -0000	[thread overview]
Message-ID: <490079A3.7000501@google.com> (raw)
In-Reply-To: <b798aad50810230553r628df29eiab6ea776af5b6230@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 848 bytes --]

Diego Novillo wrote:
> On Thu, Oct 23, 2008 at 05:38, Simon Baldwin <simonb@google.com> wrote:
>
>   
>> 2008-10-22  Simon Baldwin  <simonb@google.com>
>>
>>        * lto-lang.c: (input_node): Save the node reference, rather than the
>>        node pointer, in node->inlined_to.
>>        (fixup_cgraph_nodes): New function, convert node references into
>>        node pointers.
>>        (input_cgraph_1): Call fixup_cgraph_nodes.
>>     
>
> OK.
>   

Thanks.  However, changes between revs 141263 and 141308 mean that I've 
had to juggle it about a fair bit since emailing.

Would you mind taking a look at the revised version, appended below -- 
broadly the same, just a little more direct.  Thanks.

-- 
Google UK Limited | Registered Office: Belgrave House, 76 Buckingham
Palace Road, London SW1W 9TQ | Registered in England Number: 3977902


[-- Attachment #2: cgraph_fixup.patch --]
[-- Type: text/x-patch, Size: 2569 bytes --]

Callgraph nodes are not necessarily written out in such an order that an
inlined function can always locate the function that inlines it.  Sometimes,
the function that inlines it has not yet been read.

This patch addresses that by deferring node lookup by reference until all of
the callgraph nodes have been read, and fixes the following:

  extern inline void __attribute__((__always_inline__)) func(void) { }
  void f(void) { func(); }

  $ gcc -c -fwhopr -o f.o f.c
  $ gcc -shared -fwhopr -o f.so f.o
  lto1: internal compiler error: vector VEC(cgraph_node_ptr,base) index
  domain error, in input_node at lto/lto-lang.c:1216


gcc/lto/ChangeLog
2008-10-23  Simon Baldwin  <simonb@google.com>

	* lto-lang.c: (input_node): Save the node reference, rather than the
	node pointer, in node->inlined_to.
	(input_cgraph_1): Convert node references into node pointers.


Index: gcc/lto/lto-lang.c
===================================================================
--- gcc/lto/lto-lang.c	(revision 141320)
+++ gcc/lto/lto-lang.c	(working copy)
@@ -1185,10 +1185,10 @@ input_node (struct lto_file_decl_data* f
       node->global.estimated_stack_size = estimated_stack_size;
       node->global.stack_frame_offset = stack_frame_offset;
       node->global.insns = insns;
-      if (ref != LCC_NOT_FOUND)
-	node->global.inlined_to = VEC_index (cgraph_node_ptr, nodes, ref);
-      else
-	node->global.inlined_to = NULL;
+
+      /* Store a reference for now, and fix up later to be a pointer.  */
+      node->global.inlined_to = (cgraph_node_ptr) ref;
+
       node->global.estimated_growth = estimated_growth;
       node->global.inlined = inlined;
     }
@@ -1302,7 +1302,7 @@ input_cgraph_1 (struct lto_file_decl_dat
       LTO_DEBUG_INDENT (tag);
 
       if (tag == LTO_cgraph_edge)
-	  input_edge (ib, nodes);
+        input_edge (ib, nodes);
       else 
 	{
 	  node = input_node (file_data, ib, tag, nodes);
@@ -1315,6 +1315,20 @@ input_cgraph_1 (struct lto_file_decl_dat
       tag = lto_input_uleb128 (ib);
     }
 
+  if (flag_ltrans)
+    {
+      for (node = cgraph_nodes; node; node = node->next)
+        {
+          const int ref = (int) node->global.inlined_to;
+
+          /* Fixup inlined_to from reference to pointer.  */
+          if (ref != LCC_NOT_FOUND)
+            node->global.inlined_to = VEC_index (cgraph_node_ptr, nodes, ref);
+          else
+            node->global.inlined_to = NULL;
+        }
+    }
+
   for (node = cgraph_nodes; node; node = node->next)
     {
       tree prevailing = lto_symtab_prevailing_decl (node->decl);

  reply	other threads:[~2008-10-23 13:19 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-10-23 13:26 Simon Baldwin
2008-10-23 13:29 ` Diego Novillo
2008-10-23 15:56   ` Simon Baldwin [this message]
2008-10-23 16:35     ` Diego Novillo

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=490079A3.7000501@google.com \
    --to=simonb@google.com \
    --cc=dnovillo@google.com \
    --cc=gcc-patches@gcc.gnu.org \
    /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).