Pure LFS Hint - good to go

Bill's LFS Login lfsbill at wlmcs.com
Wed Feb 19 07:17:30 PST 2003


On Sun, 16 Feb 2003, Greg Schafer wrote:

> On Sat, Feb 15, 2003 at 08:01:05PM -0500, Bill's LFS Login wrote:
> ><snip>

> > Chapter 5 goes pretty good except for the following three items.
> > == Item 1 ======================================================
> > ### Glibc  - initial - Sat Feb 15 12:18:46 EST 2003 ###
> >  o Configure OK
> >  o Build OK
> > XXXXXX glibc make check NOT OK - CHECK LOGS (glibc-init.log) XXXXXX
> > -Installing Glibc
> >  o install OK
> >  o ALL OK
> >
> > Which is a math (double) failure. I modified the script to keep running,
> > based on some memory of posts that said those types were not critical,
> > IIRC. Probably because I'm running versions of something out-of-sync
> > with your testing. At least I hope that's all it is. I know this will need
> > to be resolved later.
>
> The Ch 5 glibc is not as critical as the one in Ch 6. If that still fails
> then definitely need to investigate. Could be processor or arch specific.
> Should be more clues in the *.out file e.g. test-double.out. It may be
> explained if you are using glibc-2.2.5 instead of 2.3.1.

These results do not include your very latest changes. The system is
essentially from your Saturday hint.

Well, the math errors are still there, same ones as before, IIRC. If we
need to confirm if they differ, I have the chap 5 stuff tarred up. Since
you said you would have concern if they still existed in chapter 6, I
thought I would post.

Googled for a couple minimal combinations of glibc and "... math ...
ldouble". A short thread involving Roland McGrath (looks like it might
apply, but I lack enough understanding, v 2.1.94?) and H. J. Liu(sp?)
about ia64. Since I'm on an IBM 6x86-II PR-300, the latter seemed to be
not applicable. I've attached math/test-ldouble.out.

In one of those threads they mentioned a new ulp(?) file, which I
surmise are the expected results to act as comparands. I wonder if that
is what I need.

As I said before, these are very small differences with which I have
little concern (no scientific or math apps) and I can live with them
AFAIK. I am dropping the '&&' from the make check invocation and
proceeding.

If you've some major concerns, or words of wisdom/patch, just holler. If
you now feel it is not an issue, silence is understood.

OH! BTW, I'm using the gcc-core, gcc-g++ archives rather than the unified
gcc archive, if it makes any difference.

A couple of displays from w/i chroot environ:

   root:/# gcc -v
   Reading specs from /stage1/lib/gcc-lib/i686-pc-linux-gnu/3.2.1/specs
   Configured with: ../gcc-3.2.1/configure --prefix=/stage1
   --with-local-prefix=/stage1 --enable-languages=c,c++ --disable-nls
   --enable-shared --enable-threads=posix --enable-__cxa_atexit
   --enable-version-specific-runtime-libs
   Thread model: posix
   gcc version 3.2.1

   On the following, I'm still doing make localedata/install-locales
   and so the info is from the end of Chap 5 (Ryan's scripts).

   root:/# /lib/libc-2.3.1.so
   GNU C Library stable release version 2.3.1, by Roland McGrath et al.
   <snip>
   Compiled by GNU CC version 3.2.1.
   Compiled on a Linux 2.4.20 system on 2003-02-18.
   Available extensions:
           GNU libio by Per Bothner
           crypt add-on version 2.1 by Michael Glad and others
           linuxthreads-0.10 by Xavier Leroy
           BIND-8.2.3-T5B
           <snip>

   readelf -a /stage1/bin/gcc|grep nterpret
         [Requesting program interpreter: /stage1/lib/ld-linux.so.2]

   root:/# cat /proc/cpuinfo
   processor       : 0
   vendor_id       : CyrixInstead
   cpu family      : 6
   model           : 2
   model name      : M II 3.5x Core/Bus Clock
   stepping        : 8
   cpu MHz         : 233.862
   fdiv_bug        : no
   hlt_bug         : no
   f00f_bug        : no
   coma_bug        : no
   fpu             : yes
   fpu_exception   : yes
   cpuid level     : 1
   wp              : yes
   flags           : fpu de tsc msr cx8 pge cmov mmx cyrix_arr
   bogomips        : 466.94

> ><snip>

> Greg

-- 
Bill Maltby
lfsbill at wlmcs.com
-------------- next part --------------
testing long double (without inline functions)
Failure: Test: sinh (0x8p-32) == 1.86264514923095703232705808926175479e-9
Result:
 is:          1.86264514923095703246e-09   0x8.00000000000000600000p-32
 should be:   1.86264514923095703226e-09   0x8.00000000000000500000p-32
 difference:  2.01948391736579022185e-28   0x8.00000000000000000000p-95
 ulp       :  1.0000
 max.ulp   :  0.0000
Maximal error of `sinh'
 is      : 1 ulp
 accepted: 0 ulp
Failure: Test: Real part of: cpow (2 + 3 i, 4 + 0 i) == -119.0 - 120.0 i
Result:
 is:         -1.19000000000000000007e+02  -0xe.e0000000000000100000p+3
 should be:  -1.19000000000000000000e+02  -0xe.e0000000000000000000p+3
 difference:  6.93889390390722837765e-18   0x8.00000000000000000000p-60
 ulp       :  1.0000
 max.ulp   :  0.0000
Failure: Test: Imaginary part of: cpow (2 + 3 i, 4 + 0 i) == -119.0 - 120.0 i
Result:
 is:         -1.19999999999999999972e+02  -0xe.ffffffffffffffc00000p+3
 should be:  -1.20000000000000000000e+02  -0xf.00000000000000000000p+3
 difference:  2.77555756156289135106e-17   0x8.00000000000000000000p-58
 ulp       :  4.0000
 max.ulp   :  2.0000
Maximal error of real part of: cpow
 is      : 6 ulp
 accepted: 763 ulp
Maximal error of imaginary part of: cpow
 is      : 4 ulp
 accepted: 2 ulp
Failure: Test: j0 (4.0) == -3.9714980986384737228659076845169804197562E-1
Result:
 is:         -3.97149809863847372228e-01  -0xc.b573849f4a7cb9800000p-5
 should be:  -3.97149809863847372282e-01  -0xc.b573849f4a7cb9a00000p-5
 difference:  5.42101086242752217004e-20   0x8.00000000000000000000p-67
 ulp       :  2.0000
 max.ulp   :  1.0000
Failure: Test: j0 (-4.0) == -3.9714980986384737228659076845169804197562E-1
Result:
 is:         -3.97149809863847372228e-01  -0xc.b573849f4a7cb9800000p-5
 should be:  -3.97149809863847372282e-01  -0xc.b573849f4a7cb9a00000p-5
 difference:  5.42101086242752217004e-20   0x8.00000000000000000000p-67
 ulp       :  2.0000
 max.ulp   :  1.0000
Maximal error of `j0'
 is      : 2 ulp
 accepted: 1 ulp
Failure: Test: jn (0, 4.0) == -3.9714980986384737228659076845169804197562E-1
Result:
 is:         -3.97149809863847372228e-01  -0xc.b573849f4a7cb9800000p-5
 should be:  -3.97149809863847372282e-01  -0xc.b573849f4a7cb9a00000p-5
 difference:  5.42101086242752217004e-20   0x8.00000000000000000000p-67
 ulp       :  2.0000
 max.ulp   :  1.0000
Failure: Test: jn (0, -4.0) == -3.9714980986384737228659076845169804197562E-1
Result:
 is:         -3.97149809863847372228e-01  -0xc.b573849f4a7cb9800000p-5
 should be:  -3.97149809863847372282e-01  -0xc.b573849f4a7cb9a00000p-5
 difference:  5.42101086242752217004e-20   0x8.00000000000000000000p-67
 ulp       :  2.0000
 max.ulp   :  1.0000

Test suite completed:
  2520 test cases plus 2308 tests for exception flags executed.
  10 errors occurred.


More information about the lfs-dev mailing list