Re: GCC fails check on x86_64
Matthias Klose wrote:
Alex Perry writes:
Attached message is my most recent posting to the AMD64 porting
list. GCC appears to build successfully, but checks fail as shown
in the attachment.
"fails"? the test results look pretty good.
I was assuming that this aspect of the test result was a bad thing.
Is it bad, or is it normal that the checks give these unexpected items ?
$ grep "unexpected" gcc-3.3.test
# of unexpected successes 14
# of unexpected failures 4
# of unexpected failures 8
# of unexpected successes 6
# of unexpected failures 1
# of unexpected successes 1
# of unexpected failures 1
# of unexpected successes 1
# of unexpected failures 8
# of unexpected successes 6
# of unexpected failures 4
# of unexpected successes 14
anyway, please build the
gcc-3.4 package from experimental for a comparision.
Good idea. In progress.
I note in passing that some of the patches have rejects.
Any other software that makes heavy use of floating point, such as
blas or atlas, also fails to complete its selftest routines
successfully. I don't know how to tell whether the problem is the
GCC itself, a system library or a kernel corruption.
is this really needed for a first base system?
Nonono; I'm mentioning BLAS (and others) as
(a) how I first noticed that there was a problem, and
(b) an indication of what areas the symptom is affecting.
I have a pure64 environment and a pure32 environment only ... at this time.
I'm in the process of setting up a cross toolchain for pure64 that runs
on pure32.
Maybe ask Arnd Bergmann who added biarch support for gcc? Look at
debian/rules.defs how to enable the biarch i386.
Thank you, I will do.
Reply to: