[Bug driver/63837] [5 Regression] r217391 causes kernel build errors with GCC_COMPARE_DEBUG=1

["trippels at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63837

--- Comment #7 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
(In reply to Manuel López-Ibáñez from comment #6)
> Thanks for the testcase. It seems that the GCC_COMPARE_DEBUG=0 uses a
> temporary file 
> 
>  ./cc1 -quiet -iprefix
> /home/manuel/test1/217394M/build/gcc/../lib/gcc/x86_64-unknown-linux-gnu/5.0.
> 0/ -isystem ./include -isystem ./include-fixed /dev/null -quiet -dumpbase
> null "-mtune=generic" "-march=x86-64" -auxbase null -Wpointer-sign
> "-fcompare-debug=-gtoggle" "-frandom-seed=0x6daada6b94b95048"
> "-fdump-final-insns=/tmp/cc3xSGvK.gkd" -o /tmp/cc3L00jg.s
> 
> but the GCC_COMPARE_DEBUG=1 uses /dev/null:
> 
>  ./cc1 -quiet -iprefix
> /home/manuel/test1/217394M/build/gcc/../lib/gcc/x86_64-unknown-linux-gnu/5.0.
> 0/ -isystem ./include -isystem ./include-fixed /dev/null -quiet -dumpbase
> null.gk "-mtune=generic" "-march=x86-64" -auxbase null -gtoggle
> -Wpointer-sign -w "-fcompare-debug=-gtoggle" -fcompare-debug-second -o
> /dev/null "-frandom-seed=0x6daada6b94b95048"
> "-fdump-final-insns=/tmp/ccMDEp8L.gk.gkd"
> 
> which triggers the heuristic. I can simply ignore /dev/null when checking
> for input==output. I'm testing this patch:
> 
> Index: gcc.c
> ===================================================================
> --- gcc.c       (revision 217457)
> +++ gcc.c       (working copy)
> @@ -4047,11 +4047,12 @@ process_command (unsigned int decoded_op
>        read_cmdline_option (&global_options, &global_options_set,
>                            decoded_options + j, UNKNOWN_LOCATION,
>                            CL_DRIVER, &handlers, global_dc);
>      }
>  
> -  if (output_file && strcmp (output_file, "-"))
> +  if (output_file && strcmp (output_file, "-")
> +      && strcmp (output_file, HOST_BIT_BUCKET))
>      {
>        int i;
>        for (i = 0; i < n_infiles; i++)
>         if ((!infiles[i].language || infiles[i].language[0] != '*')
>             && canonical_filename_eq (infiles[i].name, output_file))
> Index: toplev.c
> ===================================================================
> --- toplev.c    (revision 217457)
> +++ toplev.c    (working copy)
> @@ -940,11 +940,12 @@ init_asm_output (const char *name)
>           strcat (dumpname, ".s");
>           asm_file_name = dumpname;
>         }
>        if (!strcmp (asm_file_name, "-"))
>         asm_out_file = stdout;
> -      else if (!canonical_filename_eq (asm_file_name, name))
> +      else if (!canonical_filename_eq (asm_file_name, name)
> +              || !strcmp (asm_file_name, HOST_BIT_BUCKET))
>         asm_out_file = fopen (asm_file_name, "w");
>        else
>         /* Use fatal_error (UNKOWN_LOCATION) instead of just fatal_error to
>            prevent gcc from printing the first line in the current file. */
>         fatal_error (UNKNOWN_LOCATION,
> 
> If you can test it on your side, it would be helpful.

Your patch fixes the issue. Thank you.
>From gcc-bugs-return-466645-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Thu Nov 13 13:56:43 2014
Return-Path: <gcc-bugs-return-466645-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 30741 invoked by alias); 13 Nov 2014 13:56:43 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 30708 invoked by uid 48); 13 Nov 2014 13:56:39 -0000
From: "robert.suchanek at imgtec dot com" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug rtl-optimization/63823] [5.0 Regression] MIPS will not build due to LRA ICE with 64 bit ABI
Date: Thu, 13 Nov 2014 13:56:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: rtl-optimization
X-Bugzilla-Version: 5.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: robert.suchanek at imgtec dot com
X-Bugzilla-Status: UNCONFIRMED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 5.0
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-63823-4-2ACP9HnGhN@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-63823-4@http.gcc.gnu.org/bugzilla/>
References: <bug-63823-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2014-11/txt/msg01117.txt.bz2
Content-length: 1753

https://gcc.gnu.org/bugzilla/show_bug.cgi?idc823

--- Comment #5 from Robert Suchanek <robert.suchanek at imgtec dot com> ---
It appears that enabling the debug info can trigger the ICE. In the testcase,
after the patch, an instruction 1136 gets deleted and all references to pseudo
704 meant to be updated.

The following change in process_bb_lives () triggers the assertion later in the
pass.

+             EXECUTE_IF_SET_IN_BITMAP
+               (&lra_reg_info[dst_regno].insn_bitmap, 0, uid, bi)
+               {
+                 insn = lra_insn_recog_data[uid]->insn;
+                 lra_substitute_pseudo_within_insn (insn, dst_regno,
+                                                    SET_SRC (set));
+                 lra_update_insn_regno_info (insn);
+               }

On the first iteration, the list of insns using pseudo 704 is correct:

(gdb) call debug_bitmap(&lra_reg_info[704].insn_bitmap)

first = 0x188b720 current = 0x188b770 indx = 3
        0x188b720 next = 0x188b770 prev = (nil) indx = 2
                bits = { 343 }
        0x188b770 next = (nil) prev = 0x188b720 indx = 3
                bits = { 384 }

Insn 343 gets updated, pseudo 704 is replaced with original pseudo 234.
However, the call to lra_update_insn_regno_info destroys the remaining
references to pseudo 704 via invalidate_insn_data_regno_info so we end up with
the following:

(gdb) call debug_bitmap(&lra_reg_info[704].insn_bitmap)

first = 0x1855770 current = 0x1855770 indx = 3
        0x1855770 next = (nil) prev = (nil) indx = 3
                bits = { 384 }

The next iteration(s) in EXECUTE_IF_SET_IN_BITMAP will not happen as
bmp_iter_set () returns false. Insn 384 is left unchanged and the compiler
fails later on assertion.