[Bug c/17126] New: ICE (segfault)

[Bug c/17126] New: ICE (segfault)

[bommar33 at msu dot edu]

GCC Version: 3.5.0 20040815 (experimental)
uname -srm: Linux 2.6.8.1 i686
GCC Configure Flags: --with-languages=c,c++

Command: make bzImage V=1
...
make -f scripts/Makefile.build obj=drivers/usb/core
  gcc -Wp,-MD,drivers/usb/core/.inode.o.d -nostdinc -iwithprefix include
-D__KERNEL__ -Iinclude  -Wall -Wstrict-prototypes -Wno-trigraphs
-fno-strict-aliasing -fno-common -O3 -march=athlon-xp -pipe -msoft-float
-mpreferred-stack-boundary=2 -fno-unit-at-a-time -march=athlon
-Iinclude/asm-i386/mach-default -O2 -fomit-frame-pointer
-Wdeclaration-after-statement   -O3 -march=athlon-xp -DKBUILD_BASENAME=inode
-DKBUILD_MODNAME=usbcore -c -o drivers/usb/core/inode.o drivers/usb/core/inode.c

Compiler Output:
drivers/usb/core/inode.c: In function `get_dentry':
drivers/usb/core/inode.c:463: warning: pointer targets in assignment differ in
signedness
drivers/usb/core/inode.c:465: warning: pointer targets in passing arg 1 of
`full_name_hash' differ in signedness
drivers/usb/core/inode.c: In function `usbfs_add_device':
drivers/usb/core/inode.c:745: internal compiler error: Segmentation fault

Using the latest FTP snapshot from the 3.5 branch, I tried to build the latest
mm3 patch to the kernel.  Upon further investigation, the unpatched source from
the initial 2.6.8.1 release also fails to build, so feel free to skip the step
of applying the patch in reproducing the error.

As the compiler segfaults without any dump, and I stripped the binaries, I don't
have the skill to trace the error.  On that note, due to the lack of information
I have at this point, apologies if this has already been reported in another
bug, but I found no other segfault reported for this snapshot, and the 2.6.8.1
release compiled cleanly on last week's snapshot.

I'll attach the bz'ed preprocessed source ASAP.

-- 
           Summary: ICE (segfault)
           Product: gcc
           Version: 3.5.0
            Status: UNCONFIRMED
          Severity: critical
          Priority: P1
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: bommar33 at msu dot edu
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: i686-pc-linux
  GCC host triplet: i686-pc-linux
GCC target triplet: i686-pc-linux


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug c/17126] ICE (segfault)

[bommar33 at msu dot edu]

------- Additional Comments From bommar33 at msu dot edu  2004-08-20 20:45 -------
Created an attachment (id=6966)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=6966&action=view)
Bzip2'ed preprocessed inode.c 


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug c/17126] ICE (segfault)

[bangerth at dealii dot org]

------- Additional Comments From bangerth at dealii dot org  2004-08-20 21:00 -------
Confirmed. Actually the following are the only flags necessary: 
 
g/x> /home/bangerth/bin/gcc-3.5-pre/bin/gcc -fno-unit-at-a-time 
-fomit-frame-pointer -O3 -march=athlon-xp -c x.i  
x.i: In function `get_dentry': 
x.i:12583: warning: pointer targets in assignment differ in signedness 
x.i:12585: warning: pointer targets in passing arg 1 of `full_name_hash' 
differ in signedness 
x.i: In function `usbfs_add_device': 
x.i:12882: internal compiler error: Segmentation fault 
Please submit a full bug report, 
with preprocessed source if appropriate. 
See <URL:http://gcc.gnu.org/bugs.html> for instructions. 
 
I will be working on a reduced testcase. 
 
W. 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|ICE (segfault)              |ICE (segfault)


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |middle-end
           Keywords|                            |ice-on-valid-code
            Summary|ICE (segfault)              |[3.5 Regression] ICE in GC
                   |                            |with non-unit-at-a-time
   Target Milestone|---                         |3.5.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[bangerth at dealii dot org]

------- Additional Comments From bangerth at dealii dot org  2004-08-20 21:53 -------
I can't seem to get anything off this darn testcase. I'll attach it 
at least without the line indicators. The ICE is a regression from 3.4, 
btw. 
 
W. 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
      Known to fail|                            |3.5.0
      Known to work|                            |3.4.1
   Last reconfirmed|0000-00-00 00:00:00         |2004-08-20 21:53:41
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[bommar33 at msu dot edu]

------- Additional Comments From bommar33 at msu dot edu  2004-08-20 23:51 -------
(In reply to comment #3)
> I can't seem to get anything off this darn testcase. I'll attach it 
> at least without the line indicators. The ICE is a regression from 3.4, 
> btw. 
>  
> W. 

Took a break for a meal and sat back down to look at this, and I noticed one
difference between your testcase and the kernel source.  It struck me as
peculiar when I was testing the 2.6.8.1 release that both the inode.c from
2.6.8.1 and 2.6.8.1-mm3, although they fail in different functions,
(usbfs_add_device for vanilla and usbfs_remove_device in -mm3, adjacent in the
source) they generate the exact same assembly source.  

Both assembly sources fail after this block:
.L330:
	movl	$.LC9, 4(%esp)
	movl	$.LC25, (%esp)
	call	printk
	addl	$44, %esp
	popl	%ebx
	popl	%esi
	popl	%edi
	popl	%ebp
	ret
	.size	usbfs_add_device, .-usbfs_add_device

Upon further investigation, the following becomes apparent:

$] gcc -E -fno-unit-at-a-time -fomit-frame-pointer -O3 -march=athlon-xp inode.i
SUCCEEDS


$] gcc -S -fno-unit-at-a-time -fomit-frame-pointer -O3 -march=athlon-xp inode.i
FAILS

Therefore, we know the segfault occurs past preprocessing.

Now, to determine where which flags truly trigger the segfault.

$] gcc -S -fno-unit-at-a-time -fomit-frame-pointer -O3 -march=athlon-xp inode.i
FAILS

$] gcc -S -fno-unit-at-a-time -fomit-frame-pointer -O2 -march=athlon-xp inode.i
SUCCEEDS

$] gcc -S -fno-unit-at-a-time -fomit-frame-pointer -O3 inode.i
FAILS

$] gcc -S -fomit-frame-pointer -O3 inode.i
SUCCEEDS

$] gcc -S -fno-unit-at-a-time -O3 inode.i
FAILS

$] gcc -S -fno-unit-at-a-time -O2 inode.i
SUCCEEDS

By a rather ugly process of elimination, this "narrows" the problem down to the
code handling the third level of optimization and its interaction with
-fno-unit-at-a-time.  Hope this helps someone with more knowledge of the real
internals save a few minutes and get to the problem at hand.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-08-20 23:54 -------
I would try with "--param ggc-min-expand=0 --param ggc-min-heapsize=0" as it is segfaulting in 
the GC.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[bommar33 at msu dot edu]

------- Additional Comments From bommar33 at msu dot edu  2004-08-21 00:11 -------
(In reply to comment #6)
> I would try with "--param ggc-min-expand=0 --param ggc-min-heapsize=0" as it
is segfaulting in 
> the GC.

$] gcc -S -fno-unit-at-a-time -O3 --param ggc-min-expand=0 --param
ggc-min-heapsize=0 inode.i
FAILS

$] gcc -S -fno-unit-at-a-time -O3 --param ggc-min-expand=0 --param
ggc-min-heapsize=1 inode.i
FAILS

...

$] gcc -S -fno-unit-at-a-time -O3 --param ggc-min-expand=0 --param
ggc-min-heapsize=8192 inode.i
FAILS

$] gcc -S -fno-unit-at-a-time -O3 --param ggc-min-expand=0 --param
ggc-min-heapsize=16384 inode.i
SUCCEEDS

Hope that helps.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[reichelt at gcc dot gnu dot org]

------- Additional Comments From reichelt at gcc dot gnu dot org  2004-08-21 01:40 -------
Here's a reduced testcase (it was really a tough one).
Compile with "gcc -fno-unit-at-a-time -O3 --param ggc-min-expand=0
--param ggc-min-heapsize=0" to see the crash.

=============================================================
typedef struct { volatile int i; } atomic_t;

struct S { atomic_t a; };

static void down(struct S* s)
{
 __asm__ __volatile__("" :"=m" (s->a) :"c" (s) :"memory");
}

static void foo1(struct S *s)
{
  down(s);
}

static void foo2(struct S *s)
{
  foo3();
}

static void foo4() {}

typedef void (*fp) (struct S*);

static fp f1 = { foo1 };
static fp f2 = { foo2 };

static void foo5()
{
  foo4();
}

void foo6()
{
  foo3();
  foo7(foo5);
}
=============================================================

bug.c: In function `foo5':
bug.c:28: internal compiler error: Segmentation fault
Please submit a full bug report, [etc.]


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |reichelt at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-08-21 01:56 -------
: Search converges between 2004-04-01-trunk (#447) and 2004-04-10-trunk (#448).

Maybe related to (or just that the other bug you filed also has the same dates which they fail on) PR 
17128

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-08-21 02:02 -------
I think it was caused by this change:
2004-04-02  Jan Hubicka  <jh@suse.cz>

        * cgraph.c: Add overall comment.
        (cgraph_inline_hash): New global variable.
        (cgraph_create_node): Break out from ...
        (cgraph_node): ... here.
        (cgraph_edge): New function.
        .....

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jh at suse dot cz


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[reichelt at gcc dot gnu dot org]

------- Additional Comments From reichelt at gcc dot gnu dot org  2004-08-21 11:44 -------
Here's a shorter example with asm's and volatiles:

======================================0
static void foo0() {}

static void foo1()
{
  foo0();
}

static void foo2()
{
  foo3(foo0);
}

typedef void (*fp)();
static fp f1 = foo1, f2 = foo2;

static void foo4()
{
  foo0();
}

void foo5()
{
  foo3(foo4);
}
======================================0


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[bommar33 at msu dot edu]

------- Additional Comments From bommar33 at msu dot edu  2004-08-21 13:54 -------
Here's what I get with diagnostics turned on.

$] ./cc1 -O3 -fno-unit-at-a-time --param ggc-min-expand=0 --param
ggc-min-heapsize=0 a.i
 foo0
 {GC 583k -> 496k} {GC 496k -> 496k} foo1
 {GC 498k -> 497k} {GC 497k -> 497k} foo2
 {GC 499k -> 499k} {GC 499k -> 499k} {GC 500k -> 499k} {GC 500k -> 500k} foo4
 {GC 505k -> 505k} {GC 506k -> 506k} {GC 506k -> 506k} {GC 506k -> 506k} {GC
506k -> 506k} {GC 506k -> 506k} {GC 506k -> 506k} {GC 506k -> 506k} {GC 506k ->
506k} {GC 506k -> 506k} {GC 506k -> 506k} {GC 506k -> 506k} {GC 506k -> 506k}
{GC 506k -> 506k} {GC 506k -> 506k} {GC 506k -> 506k} {GC 506k -> 506k} {GC 506k
-> 506k} {GC 506k -> 506k} {GC 506k -> 506k} {GC 506k -> 506k} {GC 506k -> 506k}
{GC 506k -> 506k} {GC 506k -> 506k} {GC 506k -> 506k} {GC 506k -> 506k} {GC 507k
-> 506k} {GC 506k -> 506k} {GC 508k -> 507k} {GC 509k -> 508k} {GC 508k -> 508k}
{GC 508k -> 508k} {GC 508k -> 508k} {GC 508k -> 508k} {GC 508k -> 508k} {GC 508k
-> 508k} {GC 508k -> 508k} {GC 508k -> 508k} {GC 508k -> 508k} {GC 509k -> 508k}
{GC 509k -> 508k} {GC 509k -> 508k} {GC 509k -> 509k} {GC 510k -> 509k} {GC 509k
-> 509k} {GC 509k -> 509k} {GC 509k -> 509k} {GC 509k -> 509k} {GC 509k -> 509k}
{GC 512k -> 511k} {GC 512k -> 511k} {GC 512k -> 511k} {GC 511k -> 511k} {GC 512k
-> 511k} {GC 511k -> 511k} {GC 511k -> 511k} {GC 511k -> 511k} {GC 511k -> 511k}
{GC 511k -> 511k} {GC 511k -> 511k} {GC 512k -> 511k} {GC 511k -> 511k} {GC 511k
-> 511k} {GC 511k -> 511k} {GC 511k -> 511k} {GC 511k -> 511k} {GC 511k -> 511k}
{GC 512k -> 511k} {GC 511k -> 511k} {GC 511k -> 511k} {GC 511k -> 511k} {GC 512k
-> 511k} {GC 511k -> 511k} {GC 511k -> 511k} {GC 511k -> 511k} {GC 512k -> 511k}
{GC 511k -> 511k} {GC 512k -> 512k} {GC 513k -> 511k} {GC 511k -> 511k} {GC 511k
-> 511k} {GC 511k -> 511k} {GC 511k -> 511k} {GC 511k -> 511k} {GC 511k -> 511k}
{GC 511k -> 511k} {GC 511k -> 511k} {GC 511k -> 511k} {GC 511k -> 511k} {GC 511k
-> 511k} {GC 511k -> 511k} {GC 512k -> 511k} {GC 512k -> 511k} {GC 511k -> 511k}
{GC 511k -> 511k} {GC 512k -> 511k} {GC 511k -> 511k} {GC 511k -> 511k} {GC 515k
-> 514k} {GC 515k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k}
{GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k
-> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k}
{GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k
-> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k}
{GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 515k -> 514k} {GC 514k
-> 514k} {GC 515k -> 515k} {GC 516k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k}
{GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k
-> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k}
{GC 514k -> 514k} {GC 515k -> 514k} {GC 515k -> 514k} {GC 514k -> 514k} {GC 514k
-> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k} {GC 514k -> 514k}
foo5
 {GC 519k -> 518k} {GC 520k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC
519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 519k ->
519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k}
{GC 519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 519k
-> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k}
{GC 519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 520k
-> 519k} {GC 519k -> 519k} {GC 520k -> 520k} {GC 521k -> 519k} {GC 519k -> 519k}
{GC 519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 519k
-> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k}
{GC 519k -> 519k} {GC 519k -> 519k} {GC 520k -> 519k} {GC 521k -> 520k} {GC 520k
-> 520k} {GC 520k -> 520k} {GC 520k -> 519k} {GC 519k -> 519k} {GC 519k -> 519k}
{GC 524k ->
a.i: In function `foo4':
a.i:17: internal compiler error: Segmentation fault

Based on a little ugly printf debugging in the cgraph_node, cgraph_create_node,
and ggc_alloc_cleared, it looks like the majority of the allocation going on
there is the 84-byte cgraph_node struct.  

Another note: The calls to dump_cgraph_node fail generate no output when called
just prior to segfault.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[bommar33 at msu dot edu]

------- Additional Comments From bommar33 at msu dot edu  2004-08-21 16:27 -------
gtype-desc.c:L200

while (x != xlimit)
{
  gt_ggc_m_9tree_node ((*x).decl);
  ...
}

It looks like the error occurs when both x is equal to NULL and xlimit is equal
to 0.  The problem appears to originate from the previous block:

for (;;)
{
  struct cgraph_node * const xprev = ((*x).previous);
  if (xprev == NULL) break;
  x = xprev;
  (void) ggc_test_and_set_mark (xprev);
}

This is just speculation, and I'm still obviously very much in the dark and
rather bogged down in foreign code, but perhaps this rambling will trigger a
thought.

//0x082ac394 <gt_ggc_mx_cgraph_node+84>:  mov    0xfffffff4(%ebp),%eax
//0x082ac397 <gt_ggc_mx_cgraph_node+87>:  mov    %eax,0xfffffffc(%ebp)
x = xprev;

This assigns the pointer x the address of the pointer xprev.  xprev, however, is
declared only within the scope of the for loop.  gcc_test_and_set_mark declares
its argument as const, so I assume that no alteration would occur therein.  Is
xprev's address getting overwritten once the compiler passes its scope and
thinks it's safe to mange its memory?

Either way, the segfault occurs on this op where eax = 0x0:

0x082ac3c7 <gt_ggc_mx_cgraph_node+135>: cmpl   $0x0,(%eax)

...which, I believe, matches up to the NULL check in the macro's declaration.

Anyway, I'm off to lunch, so I help this was more helpful than a waste of space.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[bommar33 at msu dot edu]

------- Additional Comments From bommar33 at msu dot edu  2004-08-21 17:19 -------
Someone with more knowledge of graph theory and callgraphs will know whether or
not this is actually valid or just a hack to obscure some larger and more
consequential underlying error...

gtype-desc.c - line 200
- while (x != xlimit)
+ while (x != xlimit && x != NULL)

Of course, proper regression should performed, but that seems to fix the problem
for me.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[reichelt at gcc dot gnu dot org]

------- Additional Comments From reichelt at gcc dot gnu dot org  2004-08-23 10:08 -------
Anrew's guess was right. The bug appears with Jan's patch
http://gcc.gnu.org/ml/gcc-cvs/2004-04/msg00137.html


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

Re: [Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[Christian Ehrhardt]

On Mon, Aug 23, 2004 at 10:08:38AM -0000, reichelt at gcc dot gnu dot org wrote:
> 
> ------- Additional Comments From reichelt at gcc dot gnu dot org  2004-08-23 10:08 -------
> Anrew's guess was right. The bug appears with Jan's patch
> http://gcc.gnu.org/ml/gcc-cvs/2004-04/msg00137.html

Maybe I'm missing something but this patch seems to make the assumption
that DECL_ASSEMBLER_NAME is a tree_node. However, it is simply a char *:

>From the patch: 

cgraphunit.c:

+      void **slot;
+      if (!cgraph_inline_hash)
+        cgraph_inline_hash = htab_create_ggc (42, htab_hash_pointer,
+                                             htab_eq_pointer, NULL);
+      slot = htab_find_slot (cgraph_inline_hash,
+                            DECL_ASSEMBLER_NAME (e->callee->decl), INSERT);
+      *slot = DECL_ASSEMBLER_NAME (e->callee->decl);
       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

but in cgraph.h we have:

+extern GTY((param_is (union tree_node))) htab_t cgraph_inline_hash;
             ^^^^^^^^^^^^^^^^^^^^^^^^^^

    regards  Christian

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[ehrhardt at mathematik dot uni-ulm dot de]

------- Additional Comments From ehrhardt at mathematik dot uni-ulm dot de  2004-08-23 11:49 -------
Subject: Re:  [3.5 Regression] ICE in GC with non-unit-at-a-time

On Mon, Aug 23, 2004 at 10:08:38AM -0000, reichelt at gcc dot gnu dot org wrote:
> 
> ------- Additional Comments From reichelt at gcc dot gnu dot org  2004-08-23 10:08 -------
> Anrew's guess was right. The bug appears with Jan's patch
> http://gcc.gnu.org/ml/gcc-cvs/2004-04/msg00137.html

Maybe I'm missing something but this patch seems to make the assumption
that DECL_ASSEMBLER_NAME is a tree_node. However, it is simply a char *:

>From the patch: 

cgraphunit.c:

+      void **slot;
+      if (!cgraph_inline_hash)
+        cgraph_inline_hash = htab_create_ggc (42, htab_hash_pointer,
+                                             htab_eq_pointer, NULL);
+      slot = htab_find_slot (cgraph_inline_hash,
+                            DECL_ASSEMBLER_NAME (e->callee->decl), INSERT);
+      *slot = DECL_ASSEMBLER_NAME (e->callee->decl);
       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

but in cgraph.h we have:

+extern GTY((param_is (union tree_node))) htab_t cgraph_inline_hash;
             ^^^^^^^^^^^^^^^^^^^^^^^^^^

    regards  Christian



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[ehrhardt at mathematik dot uni-ulm dot de]

------- Additional Comments From ehrhardt at mathematik dot uni-ulm dot de  2004-08-23 11:52 -------
Subject: Re:  [3.5 Regression] ICE in GC with non-unit-at-a-time

On Mon, Aug 23, 2004 at 11:49:26AM -0000, ehrhardt at mathematik dot uni-ulm dot de wrote:
> Maybe I'm missing something but this patch seems to make the assumption
> that DECL_ASSEMBLER_NAME is a tree_node. However, it is simply a char *:

I did miss something :-(. DECL_ASSEMBLER_NAME IS in fact a tree_node.
Sorry for the confusion.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

Re: [Bug middle-end/17126] [3.5 Regression] ICE in GC with non-unit-at-a-time

[Christian Ehrhardt]

On Mon, Aug 23, 2004 at 11:49:26AM -0000, ehrhardt at mathematik dot uni-ulm dot de wrote:
> Maybe I'm missing something but this patch seems to make the assumption
> that DECL_ASSEMBLER_NAME is a tree_node. However, it is simply a char *:

I did miss something :-(. DECL_ASSEMBLER_NAME IS in fact a tree_node.
Sorry for the confusion.

[Bug middle-end/17126] [3.5 Regression] ICE with -fno-unit-at-a-time and -O3

[reichelt at gcc dot gnu dot org]

------- Additional Comments From reichelt at gcc dot gnu dot org  2004-09-10 13:23 -------
The testcases in comment #11 and #8 do not fail any more.
However, the original testcase still fails.
Here's a reduced testcase for this failure:

==================================
void foo()
{
    struct { int i; } *p;
    p->i = 0;
}

void bar()
{
    foo();
}

void baz()
{
    bar();
}
==================================

Compiling this with "gcc -fno-unit-at-a-time -O3" causes an ICE:

PR17126B.c:15: error: Inlined_to pointer is wrong
foo/11: (inline copy in baz/9) 1 insns tree inlinable asm_written
  called by: bar/10 (inlined) 
  calls: 
PR17126B.c:15: internal compiler error: verify_cgraph_node failed.
Please submit a full bug report, [etc.]


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|4.0                         |4.0.0
            Summary|[3.5 Regression] ICE in GC  |[3.5 Regression] ICE with -
                   |with non-unit-at-a-time     |fno-unit-at-a-time and -O3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [4.0 Regression] ICE with -fno-unit-at-a-time and -O3

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-09-17 21:50 -------
Fixed by the patch for PR 17509.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [4.0 Regression] ICE with -fno-unit-at-a-time and -O3

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-09-17 21:51 -------
I had meant to close this bug.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [4.0 Regression] ICE with -fno-unit-at-a-time and -O3

[pinskia at gcc dot gnu dot org]

-- 
Bug 17126 depends on bug 17419, which changed state.

Bug 17419 Summary: [4.0 Regression] Bootstrap ICE in stage2: verify_cgraph_node failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17419

           What    |Old Value                   |New Value
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [4.0 Regression] ICE with -fno-unit-at-a-time and -O3

[reichelt at gcc dot gnu dot org]

-- 
Bug 17126 depends on bug 17419, which changed state.

Bug 17419 Summary: [4.0 Regression] Bootstrap ICE in stage2: verify_cgraph_node failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17419

           What    |Old Value                   |New Value
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [4.0 Regression] ICE with -fno-unit-at-a-time and -O3

[reichelt at gcc dot gnu dot org]

------- Additional Comments From reichelt at gcc dot gnu dot org  2004-09-18 21:21 -------
Still not fixed. The reduced testcases do compile now,
but the original one still fails when compiled with
-fno-unit-at-a-time -fomit-frame-pointer -O3 -march=athlon-xp


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|2004-08-20 21:53:41         |2004-09-18 21:21:32
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [4.0 Regression] ICE with -fno-unit-at-a-time and -O3

[reichelt at gcc dot gnu dot org]

------- Additional Comments From reichelt at gcc dot gnu dot org  2004-09-18 21:22 -------
Thus, reopen PR.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [4.0 Regression] ICE with -fno-unit-at-a-time and -O3

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-09-21 00:09 -------
Subject: Bug 17126

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	hubicka@gcc.gnu.org	2004-09-21 00:08:54

Modified files:
	gcc            : ChangeLog tree-optimize.c 

Log message:
	PR middle-end/17126
	* tree-optimize.c (update_inlined_to_pointers): Fix the recursion.
	(tree_rest_of_compilation): Cleanup.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.5539&r2=2.5540
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/tree-optimize.c.diff?cvsroot=gcc&r1=2.49&r2=2.50



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126

[Bug middle-end/17126] [4.0 Regression] ICE with -fno-unit-at-a-time and -O3

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-09-21 04:14 -------
Fixed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
Bug 17126 depends on bug 17419, which changed state.

Bug 17419 Summary: [4.0 Regression] Bootstrap ICE in stage2: verify_cgraph_node failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17419

           What    |Old Value                   |New Value
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXEDBug 17126 depends on bug 17419, which changed state.

Bug 17419 Summary: [4.0 Regression] Bootstrap ICE in stage2: verify_cgraph_node failed
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17419

           What    |Old Value                   |New Value
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17126