[lfs-dev] perl tests loop at 100% CPU in current SVN

Ken Moffat zarniwhoop at ntlworld.com
Mon Aug 8 13:26:55 PDT 2016


On Mon, Aug 08, 2016 at 02:04:10AM +0100, Ken Moffat wrote:
> That box used to be my development machine, so it has several
> filesystems for building.  And I urgently need a fresh system to
> look at texlive re the tlmgr ticket.  So, since my 20160527 build
> worked on the haswell, I'll use a fresh partition and try that
> version for LFS.
> 
A (minimal) progress update -

1. perl tests are *supposed* to take 100% CPU if they don't do I/O.
But they are not supposed to run for ever.

2. The 20160527 build seemed to be stuck at the same place (running
a 4.7.0 kernel).

3. I rebooted to a 4.6.0 kernel and retried, but after about 30 min
I built gdb, saw it seemed to be in libpthread (a lack of debugging
info), killed it.

4. On the host (svn from June last year), still using 4.6.0, I
tried building and testing perl-5.24.0 : the tests completed in
45m03, but with the following failures -

Failed 3 tests out of 2245, 99.87% okay.
        ../dist/Time-HiRes/t/itimer.t
        ../lib/Benchmark.t
        op/time.t

 and for the first two -

t/itimer.t: overall time allowed for tests (360s) exceeded!
FAILED--expected 2 tests, saw 1

lib/Benchmark .................................................
Illegal division by zero at ../lib/Benchmark.t line 77.
# Looks like you planned 213 tests but ran 10.
# Looks like your test exited with 255 just after 10.
FAILED--expected 213 tests, saw 10

5. At that point I went back to the haswell, which is runnning the
20160527 build with a 4.7.0 kernel, to try building and testing
perl: all fine, less than 8 minutes for the tests.

6. Came back to the AMD, running 4.6.0, retried perl in chroot -
the tests have been running for 80 minutes so far, still in
lib/Benchmark - I'll let it run for a bit longer.

At the moment I suspect some of it might be down to an inopportune
kernel config, but I suppose I got lucky in the manual run (test
terminated after division by zero).  I'll give that another go, then
think about changing my kernel config to be closer to the haswell.

ĸen
-- 
`I shall take my mountains', said Lu-Tze. `The climate will be good
for them.'     -- Small Gods


More information about the lfs-dev mailing list