Re: [Lightning] Trouble on compiling lightning-1.2 for F-7 for ppc (koji)

[Ludovic Courtès]

Hi,

Jochen Schmitt <address@hidden> writes:

> It will be nice, if you have a look on the build log you may find at
>
> http://koji.fedoraproject.org/koji/getfile?taskID=48772&name=build.log

You're getting a lot of warnings, probably because you asked for them
through CFLAGS.  Warnings about aliasing might be worrying: you'd better
compile with `-fno-string-aliasing' to make sure.

FWIW, on my PPC32 machine with lightning 1.2b (which I assume is the one
you're using), "only" the `divi' test fails out of the 19 tests (which
is much better than what you get).

I vaguely remember fixing something related to this in the past.  The
bug does not always show up but Valgrind catches it (segfault).  Here's
what happens:

  (gdb) bt
  #0  0x100111f4 in buffer.3056 ()
  #1  0x10000830 in test_divider (divisor=8, operands=0x10001058, 
operand_count=3) at divi.c:54
  #2  0x10000a30 in main (argc=<value optimized out>, argv=<value optimized 
out>) at divi.c:75
  (gdb) disassemble
  Dump of assembler code for function buffer.3056:
  0x100111f4 <buffer.3056+0>:     mflr    r0
  0x100111f8 <buffer.3056+4>:     stw     r0,4(r1)
  0x100111fc <buffer.3056+8>:     stwu    r1,-96(r1)
  0x10011200 <buffer.3056+12>:    stmw    r23,56(r1)
  0x10011204 <buffer.3056+16>:    mr      r23,r3
  0x10011208 <buffer.3056+20>:    mr      r10,r23
  0x1001120c <buffer.3056+24>:    li      r24,8
  0x10011210 <buffer.3056+28>:    divw    r11,r10,r24
  0x10011214 <buffer.3056+32>:    mr      r3,r11
  0x10011218 <buffer.3056+36>:    lwz     r0,100(r1)
  0x1001121c <buffer.3056+40>:    mtlr    r0
  0x10011220 <buffer.3056+44>:    lmw     r23,56(r1)
  0x10011224 <buffer.3056+48>:    addi    r1,r1,96
  0x10011228 <buffer.3056+52>:    blr
  (gdb) info registers 
  r0             0x10011240       268505664
  r1             0x7ecd13d0       2127369168
  r2             0x4027270        67269232
  r3             0x28     40
  r4             0xff     255
  r5             0x2000   8192
  r6             0x100111fc       268505596
  r7             0x10011220       268505632
  r8             0x10011260       268505696
  r9             0x10011240       268505664
  r10            0x20     32
  r11            0x10011260       268505696
  r12            0x22000424       570426404
  r13            0x100191e8       268538344
  r14            0x0      0
  r15            0x0      0
  r16            0x0      0
  r17            0x0      0
  r18            0x0      0
  r19            0x0      0
  r20            0x0      0
  r21            0x10010000       268500992
  r22            0x100111f4       268505588
  r23            0x3      3
  r24            0x10001058       268439640
  r25            0x0      0
  r26            0x10000000       268435456
  r27            0x0      0
  r28            0x60     96
  r29            0x1001122c       268505644
  r30            0x8      8
  r31            0x1      1
  pc             0x100111f4       268505588
  cr             0x42000424       1107297316
  lr             0x10000830       268437552
  ctr            0x100111f4       268505588
  xer            0x0      0

It's not clear to me what's wrong.  I'll try to investigate it further
later if no one beats me at it.  ;-)

Thanks,
Ludovic.