[Bug target/42774] New: [4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[Bug target/42774] New: [4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[debian-gcc at lists dot debian dot org]

works with 4.3 branch, fails with 4.4 branch and trunk 20100109, not seen with
-O1

  Matthias

$ gcc-4.4 -c -g -O2 fs_ntfs.i
fs_ntfs.c: In function 'ntfs_getinfo':
fs_ntfs.c:102: internal compiler error: in get_aligned_mem, at
config/alpha/alpha.c:1466
Please submit a full bug report,
with preprocessed source if appropriate.


-- 
           Summary: [4.5 Regression] ICE in get_aligned_mem, at
                    config/alpha/alpha.c:1484
           Product: gcc
           Version: 4.5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: debian-gcc at lists dot debian dot org
GCC target triplet: alpha-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[debian-gcc at lists dot debian dot org]

------- Comment #1 from debian-gcc at lists dot debian dot org  2010-01-17 11:22 -------
Created an attachment (id=19633)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=19633&action=view)
preprocessed source


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[rguenth at gcc dot gnu dot org]

-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.4.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[rguenth at gcc dot gnu dot org]

-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[mikpe at it dot uu dot se]

------- Comment #2 from mikpe at it dot uu dot se  2010-01-17 19:08 -------
Reproduced with an alpha-unknown-linux cross hosted on i686, built from
gcc-4.4-20100112.


-- 

mikpe at it dot uu dot se changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mikpe at it dot uu dot se


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[ubizjak at gmail dot com]

------- Comment #3 from ubizjak at gmail dot com  2010-01-17 19:47 -------
Well, gcc would just generate invalid code without this assert.

Reduced test:

--cut here--
typedef unsigned short int uint16_t;
typedef unsigned int uint32_t;
typedef unsigned long int uint64_t;
typedef uint16_t u16;
typedef uint32_t u32;
typedef uint64_t u64;
enum
{ FSYSHEADKEY_NULL =
    0, FSYSHEADKEY_FILESYSTEM, FSYSHEADKEY_MNTPATH, FSYSHEADKEY_BYTESTOTAL,
    FSYSHEADKEY_MINFSAVERSION, FSYSHEADKEY_MOUNTINFO, FSYSHEADKEY_ORIGDEV,
    FSYSHEADKEY_FSEXTEOPTRAIDSTRIDE };
typedef struct s_dico
{
}
cdico;
struct s_ntfsinfo
{
  u32 bytes_per_sector;
  u64 uuid;
};
enum
{ MSG_FORCE = 0, MSG_VERB1 = 1, MSG_VERB2 = 2, MSG_STACK = 3, MSG_DEBUG1 =
    4, MSG_DEBUG2 = 5, MSG_DEBUG3 = 6, MSG_DEBUG4 = 7, MSG_DEBUG5 = 8 };
ntfs_getinfo (cdico * d, char *devname)
{
  struct s_ntfsinfo info;
  char bootsect[512];
  int fd = -1;

  info.bytes_per_sector = ((u16) (*((u16 *) (bootsect + 0xB))));

  foo (info.bytes_per_sector);
  info.uuid = ((u64) (*((u64 *) (bootsect + 0x48))));
  fsaprintf (MSG_VERB2, 0, MSG_VERB2 >= 3, "fs_ntfs.c", __FUNCTION__, 107,
             "bytes_per_sector=[%lld]\n", (long long) info.bytes_per_sector);
  fsaprintf (MSG_VERB2, 0, MSG_VERB2 >= 3, "fs_ntfs.c", __FUNCTION__, 109,
             "uuid=[%016llX]\n", (long long unsigned int) info.uuid);
  dico_add_u64 (d, 0, FSYSHEADKEY_MINFSAVERSION,
                ((u64)
                 ((((u64) 0 & 0xFFFF) << 48) + (((u64) 6 & 0xFFFF) << 32) +
                  (((u64) 4 & 0xFFFF) << 16) + (((u64) 0 & 0xFFFF) << 0))));
}
--cut here--

Problematic line is bootsect + 0xB that crosses SImode aligned load.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[mikpe at it dot uu dot se]

------- Comment #4 from mikpe at it dot uu dot se  2010-01-17 19:52 -------
Created an attachment (id=19638)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=19638&action=view)
reduced test case

I'm attaching a really tiny test case. The trigger appears to be a 2-byte load
from the stack that crosses the boundary between two adjacent 4-byte ints.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[hjl dot tools at gmail dot com]

------- Comment #5 from hjl dot tools at gmail dot com  2010-01-17 19:57 -------
Is this related to PR 39954?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[ubizjak at gmail dot com]

------- Comment #6 from ubizjak at gmail dot com  2010-01-17 20:25 -------
Created an attachment (id=19640)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=19640&action=view)
Patch to teach {,un}aligned_memory_operand about unaligned offsets

Mikael, can you please test attached patch on your target? Unaligned offsets
should be classified as unaligned memory access.


-- 

ubizjak at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |ubizjak at gmail dot com
                   |dot org                     |
             Status|UNCONFIRMED                 |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[mikpe at it dot uu dot se]

------- Comment #7 from mikpe at it dot uu dot se  2010-01-17 21:36 -------
Uros' proposed patch fixes the ICE in gcc-4.5. It doesn't apply to 4.4 though.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[ubizjak at gmail dot com]

------- Comment #8 from ubizjak at gmail dot com  2010-01-17 21:50 -------
Created an attachment (id=19642)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=19642&action=view)
Patch for 4.3 and 4.4 branch.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[mikpe at it dot uu dot se]

------- Comment #9 from mikpe at it dot uu dot se  2010-01-17 22:32 -------
Uros' proposed fix for 4.4 and 4.3 fixes the ICE on both branches.

The original report stated that 4.3 doesn't ICE. A vanilla 4.3-20100103
definitely ICEs for me on the reduced test case.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[ubizjak at gmail dot com]

------- Comment #10 from ubizjak at gmail dot com  2010-01-18 08:42 -------
Patch at http://gcc.gnu.org/ml/gcc-patches/2010-01/msg00958.html


-- 

ubizjak at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-
                   |                            |patches/2010-
                   |                            |01/msg00958.html


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[uros at gcc dot gnu dot org]

------- Comment #11 from uros at gcc dot gnu dot org  2010-01-18 17:04 -------
Subject: Bug 42774

Author: uros
Date: Mon Jan 18 17:04:29 2010
New Revision: 156014

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=156014
Log:
        PR target/42774
        * config/alpha/predicates.md (aligned_memory_operand): Return 0 for
        memory references with unaligned offsets.  Remove CQImode handling.
        (unaligned_memory_operand): Return 1 for memory references with
        unaligned offsets.  Remove CQImode handling.

testsuite/ChangeLog:

        PR target/42774
        * gcc.target/alpha/pr42774.c: New test.


Added:
    branches/gcc-4_4-branch/gcc/testsuite/gcc.target/alpha/pr42774.c
Modified:
    branches/gcc-4_4-branch/gcc/ChangeLog
    branches/gcc-4_4-branch/gcc/config/alpha/predicates.md
    branches/gcc-4_4-branch/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[uros at gcc dot gnu dot org]

------- Comment #12 from uros at gcc dot gnu dot org  2010-01-18 17:46 -------
Subject: Bug 42774

Author: uros
Date: Mon Jan 18 17:46:17 2010
New Revision: 156017

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=156017
Log:
        PR target/42774
        * config/alpha/predicates.md (aligned_memory_operand): Return 0 for
        memory references with unaligned offsets.  Remove CQImode handling.
        (unaligned_memory_operand): Return 1 for memory references with
        unaligned offsets.  Remove CQImode handling.

testsuite/ChangeLog:

        PR target/42774
        * gcc.target/alpha/pr42774.c: New test.


Added:
    trunk/gcc/testsuite/gcc.target/alpha/pr42774.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/alpha/predicates.md
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[uros at gcc dot gnu dot org]

------- Comment #13 from uros at gcc dot gnu dot org  2010-01-18 21:44 -------
Subject: Bug 42774

Author: uros
Date: Mon Jan 18 21:44:32 2010
New Revision: 156024

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=156024
Log:
        PR target/42774
        * config/alpha/predicates.md (aligned_memory_operand): Return 0 for
        memory references with unaligned offsets.  Remove CQImode handling.
        (unaligned_memory_operand): Return 1 for memory references with
        unaligned offsets.  Remove CQImode handling.

testsuite/ChangeLog:

        PR target/42774
        * gcc.target/alpha/pr42774.c: New test.


Added:
    branches/gcc-4_3-branch/gcc/testsuite/gcc.target/alpha/pr42774.c
Modified:
    branches/gcc-4_3-branch/gcc/ChangeLog
    branches/gcc-4_3-branch/gcc/config/alpha/predicates.md
    branches/gcc-4_3-branch/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774

[Bug target/42774] [4.4/4.5 Regression] ICE in get_aligned_mem, at config/alpha/alpha.c:1484

[ubizjak at gmail dot com]

------- Comment #14 from ubizjak at gmail dot com  2010-01-18 21:48 -------
Fixed.


-- 

ubizjak at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
      Known to fail|4.4.3 4.5.0                 |4.3.4 4.4.3 4.5.0
      Known to work|4.3.4                       |
         Resolution|                            |FIXED
   Target Milestone|4.4.3                       |4.3.5


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42774