[Bug c/21840] New: Linux 2.6.11.1.11: drivers/char/keyboard.c:619: internal compiler error: Segmentation fault

[Bug c/21840] New: Linux 2.6.11.1.11: drivers/char/keyboard.c:619: internal compiler error: Segmentation fault

[mclinden at informed dot net]

The relevant code is here:
CONFIG_GRKERNSEC_PROC and 
CONFIG_GRKERNSEC_PROC_MEMMAP)

are defined 

static void k_spec(struct vc_data *vc, unsigned char value, char up_flag, 
struct pt_regs *regs)
{
        if (up_flag)
                return;
        if (value >= ARRAY_SIZE(fn_handler))
                return;
        if ((kbd->kbdmode == VC_RAW ||
             kbd->kbdmode == VC_MEDIUMRAW) &&
             value != KVAL(K_SAK))
                return;         /* SAK is allowed even in raw mode */

#if defined(CONFIG_GRKERNSEC_PROC) || defined(CONFIG_GRKERNSEC_PROC_MEMMAP)
        {
                void *func = fn_handler[value];
                if (func == fn_show_state || func == fn_show_ptregs ||
                    func == fn_show_mem)
                        return;
        }
#endif

        fn_handler[value](vc, regs); /* line 619 */
}

-- 
           Summary: Linux 2.6.11.1.11: drivers/char/keyboard.c:619: internal
                    compiler error: Segmentation fault
           Product: gcc
           Version: 4.1.0
            Status: UNCONFIRMED
          Severity: minor
          Priority: P2
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: mclinden at informed dot net
                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=21840

[Bug middle-end/21840] Linux 2.6.11.1.11: drivers/char/keyboard.c:619: internal compiler error: Segmentation fault

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-05-31 14:04 -------
Again, we need the preprocessed source.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |WAITING
          Component|c                           |middle-end
           Keywords|                            |ice-on-valid-code


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21840

[Bug middle-end/21840] Linux 2.6.11.1.11: drivers/char/keyboard.c:619: internal compiler error: Segmentation fault

[mclinden at informed dot net]

------- Additional Comments From mclinden at informed dot net  2005-05-31 14:27 -------
gcc: warning: -pipe ignored because -save-temps specified
Using built-in specs.
Target: i686-pc-linux-gnu
Configured with: ../gcc-4.1-20050528/configure --prefix=/usr --with-mpfr=/usr --
with-mpfr-dir=/usr/src/otai/mpfr-2.1.1 --with-gmp=/usr --with-gmp-
dir=/usr/src/otai/gmp-4.1.4 --enable-shared --enable-__cxa_atexit --enable-
clocale --enable-multilib=no --enable-threads=posix --libexecdir=/usr/lib --
enable-languages=c,c++
Thread model: posix
gcc version 4.1.0 20050528 (experimental)
 /usr/lib/gcc/i686-pc-linux-gnu/4.1.0/cc1 -E -quiet -nostdinc -v -Iinclude -
Iinclude/asm-i386/mach-default -D__KERNEL__ -DKBUILD_BASENAME=keyboard -
DKBUILD_MODNAME=keyboard -isystem /usr/lib/gcc/i686-pc-linux-gnu/4.1.0/include -
MD drivers/char/.keyboard.o.d drivers/char/keyboard.c -msoft-float -mpreferred-
stack-boundary=2 -march=i686 -mtune=pentium4 -Wall -Wstrict-prototypes -Wno-
trigraphs -Wdeclaration-after-statement -Wno-pointer-sign -fno-strict-aliasing -
fno-common -ffreestanding -fomit-frame-pointer -fno-unit-at-a-time -O2 -fpch-
preprocess -o keyboard.i
#include "..." search starts here:
#include <...> search starts here:
 include
 include/asm-i386/mach-default
 /usr/lib/gcc/i686-pc-linux-gnu/4.1.0/include
End of search list.
 /usr/lib/gcc/i686-pc-linux-gnu/4.1.0/cc1 -fpreprocessed keyboard.i -quiet -
dumpbase keyboard.c -msoft-float -mpreferred-stack-boundary=2 -march=i686 -
mtune=pentium4 -auxbase-strip drivers/char/keyboard.o -O2 -Wall -Wstrict-
prototypes -Wno-trigraphs -Wdeclaration-after-statement -Wno-pointer-sign -
version -fno-strict-aliasing -fno-common -ffreestanding -fomit-frame-pointer -
fno-unit-at-a-time -o keyboard.s
GNU C version 4.1.0 20050528 (experimental) (i686-pc-linux-gnu)
        compiled by GNU C version 4.1.0 20050528 (experimental).
GGC heuristics: --param ggc-min-expand=30 --param ggc-min-heapsize=4096
Compiler executable checksum: 1b490aabfa7a74d21cf5375dd255b543
drivers/char/keyboard.c: In function 'k_spec':
drivers/char/keyboard.c:619: internal compiler error: Segmentation fault


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21840

[Bug middle-end/21840] Linux 2.6.11.1.11: drivers/char/keyboard.c:619: internal compiler error: Segmentation fault

[pluto at agmk dot net]

------- Additional Comments From pluto at agmk dot net  2005-05-31 14:40 -------
[ PR21839, PR21840 ]

Sean, Do the sigsegvs occur in random places during kernel build?
If yes then You must check other RAM chips. Memtest86 is not enough
to be 100% sure. You should also try the parallel kernel build (eg. `make -j8`).

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21840

[Bug middle-end/21840] Linux 2.6.11.1.11: drivers/char/keyboard.c:619: internal compiler error: Segmentation fault

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-05-31 14:46 -------
The backtrace:
#0  0x083f1b29 in ix86_struct_value_rtx (type=0xb7c0657c, incoming=0) at /home/peshtigo/pinskia/
src/gnu/gcc/src/gcc/config/i386/i386.c:3184
#1  0x081974c9 in expand_call (exp=0xb778e9d8, target=0x0, ignore=1) at /home/peshtigo/pinskia/
src/gnu/gcc/src/gcc/calls.c:1909
#2  0x08203cf8 in expand_expr_real_1 (exp=0xb778e9d8, target=0x0, tmode=Variable "tmode" is not 
available.
) at /home/peshtigo/pinskia/src/gnu/gcc/src/gcc/expr.c:7344
#3  0x0820e497 in expand_expr_real (exp=0xb778e9d8, target=0xb7bf4210, tmode=VOIDmode, 
modifier=EXPAND_NORMAL, alt_rtl=0x0)
    at /home/peshtigo/pinskia/src/gnu/gcc/src/gcc/expr.c:6417
#4  0x083a3ac6 in expand_expr_stmt (exp=0xb778e9d8) at expr.h:489
#5  0x083e0bee in expand_gimple_basic_block (bb=0xb7807ca8, dump_file=Variable "dump_file" is 
not available.
) at /home/peshtigo/pinskia/src/gnu/gcc/src/gcc/cfgexpand.c:955
#6  0x083e1853 in tree_expand_cfg () at /home/peshtigo/pinskia/src/gnu/gcc/src/gcc/cfgexpand.c:
1307
#7  0x080d2cbd in execute_pass_list (pass=0x8574b40) at /home/peshtigo/pinskia/src/gnu/gcc/src/
gcc/tree-optimize.c:623
#8  0x080d30d3 in tree_rest_of_compilation (fndecl=0xb7767798) at /home/peshtigo/pinskia/src/
gnu/gcc/src/gcc/tree-optimize.c:793
#9  0x0805dc12 in c_expand_body (fndecl=0xb7767798) at /home/peshtigo/pinskia/src/gnu/gcc/
src/gcc/c-decl.c:6593
#10 0x08415b82 in cgraph_expand_function (node=0xb779b0d8) at /home/peshtigo/pinskia/src/
gnu/gcc/src/gcc/cgraphunit.c:949
#11 0x084179d5 in cgraph_optimize () at /home/peshtigo/pinskia/src/gnu/gcc/src/gcc/cgraphunit.c:
1013
#12 0x083aff76 in toplev_main (argc=33, argv=0xbff263b4) at /home/peshtigo/pinskia/src/gnu/gcc/
src/gcc/toplev.c:1008
#13 0x0035dad4 in __libc_start_main () from /lib/tls/libc.so.6
#14 0x08049b21 in _start ()


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21840

[Bug middle-end/21840] [4.1 Regression] Linux 2.6.11.1.11: drivers/char/keyboard.c:619: internal compiler error: Segmentation fault

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-05-31 14:57 -------
Confirmed, reduced testcase:
void fn_show_state(void);
typedef void (*fn_handler_fn)(void);
static fn_handler_fn fn_handler[1];

void k_spec(unsigned char value)
{
  void *func = fn_handler[value];
  if (func == fn_show_state )
    return;
  fn_handler[value]();
}

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|minor                       |normal
             Status|WAITING                     |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2005-05-31 14:57:52
               date|                            |
            Summary|Linux 2.6.11.1.11:          |[4.1 Regression] Linux
                   |drivers/char/keyboard.c:619:|2.6.11.1.11:
                   |internal compiler error:    |drivers/char/keyboard.c:619:
                   |Segmentation fault          |internal compiler error:
                   |                            |Segmentation fault
   Target Milestone|---                         |4.1.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21840

[Bug middle-end/21840] [4.1 Regression] Linux 2.6.11.1.11: drivers/char/keyboard.c:619: internal compiler error: Segmentation fault

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-05-31 15:05 -------
Compiling with -O1, we seg fault but compiling with -O1 -fno-tree-fre we don't because we don't 
"FRE", the load and get the type correct.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21840

[Bug middle-end/21840] [4.1 Regression] Linux 2.6.11.1.11: drivers/char/keyboard.c:619: internal compiler error: Segmentation fault

[dberlin at gcc dot gnu dot org]

------- Additional Comments From dberlin at gcc dot gnu dot org  2005-05-31 15:28 -------
Not really an FRE bug.
FRE eliminates the load, and everything we have tells it it can (it asks
may_propagate_copy, etc).

However, eliminating the load replaces a void (*foo) type with a void * type (IE
poitner to function replaced with pointer to void).

Once FRE has eliminated the load, something later copy-props the void * func_4
into the call and then it gets transforms it into a void.
So we end up with a call of a void expression, which makes things angry.
I can't fix this in FRE, it's not doing anything wrong, it's just removing a
load, validly :)

Whatever is copy-prop'ing into the call and transforming it into a call of a
void is wrong, and that's not FRE.





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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21840

[Bug middle-end/21840] [4.1 Regression] Linux 2.6.11.1.11: drivers/char/keyboard.c:619: internal compiler error: Segmentation fault

[pluto at agmk dot net]

------- Additional Comments From pluto at agmk dot net  2005-06-16 23:53 -------
(In reply to comment #7) 
> Compiling with -O1, we seg fault but compiling with -O1 -fno-tree-fre 
> we don't because we don't "FRE", the load and get the type correct. 
 
with -O2 -fno-tree-fre gcc (from mainline) segfaults too. 
 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21840

[Bug middle-end/21840] [4.1 Regression] ICE while building Linux kernel (seg fault), missing cast

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-07-12 06:44 -------
With the patch in PR 22368, we get the following ICE instead:
t.c:6: error: statement types mismatch
D.1273_6 = funcD.1271_4;

voidD.28 (*<T2ca>) (void)
voidD.28 *

I will look into fixing this after I resubmit my other patch which fixes one of those issues.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
OtherBugsDependingO|                            |22368
              nThis|                            |
   Last reconfirmed|2005-05-31 14:57:52         |2005-07-12 06:44:00
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21840

[Bug middle-end/21840] [4.1 Regression] ICE while building Linux kernel (seg fault), missing cast

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-07-12 16:58 -------
Created an attachment (id=9250)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=9250&action=view)
patch which should fix this without any really bad side effects

This patch should fix it.  I have not tested it yet though.  We always have the
correct type if we have *rhs_p as an SSA_NAME because of may_propagate_copy
check.	And since *rhs_p is always the right hand side of a MODIFY_EXPR and
fold_convert produces a NOP_EXPR/CONVERT_EXPR for SSA_NAMEs operands and
CONSTANTS (ADDR_EXPR will produce a cast).

ChangeLog:
* tree-ssa-pre.c (eliminate): Convert the sprime to the correct type if *rhs_p
is not a SSA_NAME.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |pinskia at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21840

[Bug middle-end/21840] [4.1 Regression] ICE while building Linux kernel (seg fault), missing cast

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-07-13 01:37 -------
Patch posted here: <http://gcc.gnu.org/ml/gcc-patches/2005-07/msg00883.html>.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-
                   |                            |patches/2005-
                   |                            |07/msg00883.html
           Keywords|                            |patch


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21840

[Bug middle-end/21840] [4.1 Regression] ICE while building Linux kernel (seg fault), missing cast

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-07-13 02:33 -------
Fixed.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21840

[Bug middle-end/21840] [4.1 Regression] ICE while building Linux kernel (seg fault), missing cast

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-07-13 02:33 -------
Subject: Bug 21840

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	pinskia@gcc.gnu.org	2005-07-13 02:33:11

Modified files:
	gcc            : ChangeLog tree-ssa-pre.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/gcc.c-torture/compile: pr21840.c 

Log message:
	2005-07-12  Andrew Pinski  <pinskia@physics.uc.edu>
	
	PR tree-opt/21840
	* tree-ssa-pre.c (eliminate): Convert the sprime to the correct type
	if *rhs_p is not a SSA_NAME.
	
	2005-07-12  Andrew Pinski  <pinskia@physics.uc.edu>
	
	PR tree-opt/21840
	* gcc.c-torture/compile/pr21840.c: New test.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.9425&r2=2.9426
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/tree-ssa-pre.c.diff?cvsroot=gcc&r1=2.95&r2=2.96
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&r1=1.5758&r2=1.5759
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/compile/pr21840.c.diff?cvsroot=gcc&r1=NONE&r2=1.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21840