Bug 239003

Created attachment 154115 [details]
bit-ops.c (pre-processed)

Also happens with this command line:

gcc -DHAVE_CONFIG_H -I. -I. -I../.. -I../../include -I../../include
-Disfinite=finite -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions
-fstack-protector --param=ssp-buffer-size=4 -m64 -mminimal-toc -c bit-ops.c 
-fPIC -DPIC -o .libs/bit-ops.o

i.e. without -O6.

Reduced testcase (gcc -O2 -m64 -g on ppc*):
void
foo (int l, const unsigned char *i, unsigned char *o0, unsigned char *o1,
     unsigned char *o2, unsigned char *o3, unsigned char *o4,
     unsigned char *o5, unsigned char *o6, unsigned char *o7,
     unsigned char *o8, unsigned char *o9, unsigned char *o10,
     unsigned char *o11, unsigned char *o12, unsigned char *o13,
     unsigned char *o14, unsigned char *o15)
{
  unsigned char u;
  unsigned char t[16];

  __builtin_memset (t, 0, 16);
  for (; l > 0; l--)
    {
      u = *i++;
      if (u & 192)
        t[0] |= (u & 192);
      *o0++ = t[0];
      *o1++ = t[1];
      *o2++ = t[2];
      *o3++ = t[3];
      *o4++ = t[4];
      *o5++ = t[5];
      *o6++ = t[6];
      *o7++ = t[7];
      *o8++ = t[8];
      *o9++ = t[9];
      *o10++ = t[10];
      *o11++ = t[11];
      *o12++ = t[12];
      *o13++ = t[13];
      *o14++ = t[14];
      *o15++ = t[15];
    }
}

Reload changes:
239003.i.36.lreg-(insn:HI 56 53 47 2 239003.i:17 (set (reg:SI 151)
239003.i.36.lreg:        (ior:SI (reg:SI 148)
239003.i.36.lreg-            (subreg:SI (reg:QI 185 [ t ]) 0))) 104
{*boolsi3_internal1} (insn_list:REG_DEP_TRUE 53 (nil))
239003.i.36.lreg-    (expr_list:REG_DEAD (reg:SI 148)
239003.i.36.lreg-        (nil)))
into:
239003.i.37.greg-(insn:HI 56 53 47 2 239003.i:17 (set (reg:SI 0 0 [151])
239003.i.37.greg:        (ior:SI (reg:SI 0 0 [148])
239003.i.37.greg-            (reg:SI 6 6 [orig:185 t+-3 ] [185]))) 104
{*boolsi3_internal1} (insn_list:REG_DEP_TRUE 53 (nil))
239003.i.37.greg-    (nil))
as this is big endian (note, t is a char array and so t+-3 makes some sense,
as only the low 8 bits of the 32-bit register contain t (== t+0).
This is generated all the way through current trunk, but only 4.1-rh ICEs on
this in var_tracking, though so far I haven't seen anything that would look like
a fix for this in 4.2+.

I have also just encountered this bug when building libtomcrypt for ppc64. 
Here's the build log:

http://koji.fedoraproject.org/koji/getfile?taskID=61676&name=build.log

I was able to use Tim's build trick and get the output dumped to the log.  This
build log should be more useful.

http://koji.fedoraproject.org/koji/getfile?taskID=61751&name=build.log

Based on the date this bug was created, it appears to have been reported
against rawhide during the development of a Fedora release that is no
longer maintained. In order to refocus our efforts as a project we are
flagging all of the open bugs for releases which are no longer
maintained. If this bug remains in NEEDINFO thirty (30) days from now,
we will automatically close it.

If you can reproduce this bug in a maintained Fedora version (7, 8, or
rawhide), please change this bug to the respective version and change
the status to ASSIGNED. (If you're unable to change the bug's version
or status, add a comment to the bug and someone will change it for you.)

Thanks for your help, and we apologize again that we haven't handled
these issues to this point.

The process we're following is outlined here:
http://fedoraproject.org/wiki/BugZappers/F9CleanUp

We will be following the process here:
http://fedoraproject.org/wiki/BugZappers/HouseKeeping to ensure this
doesn't happen again.

This bug has been in NEEDINFO for more than 30 days since feedback was
first requested. As a result we are closing it.

If you can reproduce this bug in the future against a maintained Fedora
version please feel free to reopen it against that version.

The process we're following is outlined here:
http://fedoraproject.org/wiki/BugZappers/F9CleanUp