was using rock64 for python3 development, but after running
apt-get update
apt-get upgrade
notice that running python3 just gives me segmentation fault
anyone had this experience on debian stretcg python3.5?
root@rock64:/mnt/home/rock64# uname -a
Linux rock64 4.4.77-rockchip-ayufan-136 #1 SMP Thu Oct 12 09:14:48 UTC 2017 aarch64 GNU/Linux
root@rock64:/mnt/home/rock64# cat /etc/os-release
PRETTY_NAME="Debian GNU/Linux 9 (stretch)"
03-30-2018, 06:57 PM
(This post was last modified: 03-30-2018, 06:59 PM by edwinbmiller.)
reinstalled debian stretch 4.4.77 and did a
apt-get update -y
apt-get install linux-rock64 -y (instead of pt-get upgrade)
then ran the following python
root@rock64:/mnt/home/rock64/0636920028154-master-c581427d3e9fc7ef210dd46d112d2eae4043d557/lp5e-code-1.0-jun1813/code# python3 pybench_cases.py
Segmentation fault
root@rock64:/mnt/home/rock64/0636920028154-master-c581427d3e9fc7ef210dd46d112d2eae4043d557/lp5e-code-1.0-jun1813/code# python3 pybench_cases.py
Segmentation fault
root@rock64:/mnt/home/rock64/0636920028154-master-c581427d3e9fc7ef210dd46d112d2eae4043d557/lp5e-code-1.0-jun1813/code# python3 pybench_cases.py
3.5.3 (default, Jan 19 2017, 14:11:04)
[GCC 6.3.0 20170118]
2.1335 ['[x ** 2 for x in range(1000)]']
2.5179 ['res=[]\nfor x in range(1000): res.append(x ** 2)']
3.0498 ['list(map(lambda x: x ** 2, range(1000)))']
2.3948 ['list(x ** 2 for x in range(1000))']
4.7236 ["s = 'spam' * 2500\nx = [s[i] for i in range(10000)]"]
6.6688 ["s = '?'\nfor i in range(10000): s += '?'"]
root@rock64:/mnt/home/rock64/0636920028154-master-c581427d3e9fc7ef210dd46d112d2eae4043d557/lp5e-code-1.0-jun1813/code# cat pybench_cases.py
"""
pybench_cases.py: Run pybench on a set of pythons and statements.
Select modes by editing this script or using command-line arguments (in
sys.argv): e.g., run a "C:\python27\python pybench_cases.py" to test just
one specific version on stmts, "pybench_cases.py -a" to test all pythons
listed, or a "py -3 pybench_cases.py -a -t" to trace command lines too.
"""
import pybench, sys
pythons = [ # (ispy3?, path)
(1, 'C:\python33\python'),
(0, 'C:\python27\python'),
(0, 'C:\pypy\pypy-1.9\pypy')
]
stmts = [ # (num,rpt,stmt)
(0, 0, "[x ** 2 for x in range(1000)]"), # Iterations
(0, 0, "res=[]\nfor x in range(1000): res.append(x ** 2)"), # \n=multistmt
(0, 0, "$listif3(map(lambda x: x ** 2, range(1000)))"), # \n\t=indent
(0, 0, "list(x ** 2 for x in range(1000))"), # $=list or ''
(0, 0, "s = 'spam' * 2500\nx = [s[i] for i in range(10000)]"), # String ops
(0, 0, "s = '?'\nfor i in range(10000): s += '?'"),
]
tracecmd = '-t' in sys.argv # -t: trace commmand lines?
pythons = pythons if '-a' in sys.argv else None # -a: all in list, else one?
pybench.runner(stmts, pythons, tracecmd)
it gives Segmentation Fault first 2 times i try to run but on 3rd try runs fine
03-30-2018, 10:17 PM
(This post was last modified: 03-30-2018, 10:26 PM by edwinbmiller.)
switched over to xenial and no longer have problems with python3
root@rock64:~# uname -a
Linux rock64 4.4.77-rockchip-ayufan-136 #1 SMP Thu Oct 12 09:14:48 UTC 2017 aarch64 aarch64 aarch64 GNU/Linux
root@rock64:~# cat /etc/os-release
NAME="Ubuntu"
VERSION="16.04.3 LTS (Xenial Xerus)"
ID=ubuntu
ID_LIKE=debian
PRETTY_NAME="Ubuntu 16.04.3 LTS"
more good news was able to install pip & upgrade it successfully on xenial minimal with kernel 4.4.77
apt-get install python3-pip
pip3 install --upgrade pip
What did you use before? Bionic? Bionic is still not released and in beta, so if you find issues there, report them to the Ubuntu upstream bugtracker...
(03-31-2018, 07:23 AM)xalius Wrote: What did you use before? Bionic? Bionic is still not released and in beta, so if you find issues there, report them to the Ubuntu upstream bugtracker...
debian stretch  So I'm guessing it was an 'old' bug in a python C library that has since been resolved...
also did try this with python3
https://github.com/ayufan-rock64/linux-b...m64.img.xz
but is this release of xenial is broken in regard to python3, so went back to 0.5.15: jenkins-linux-build-rock-64-136 which seems to work fine.
also did try artful but had lots of issues.
just tried bionic prebuild with perl and python3 and works fine
https://github.com/ayufan-rock64/linux-b...mhf.img.xz
so the 2 builds i've had the best luck with for python3 are:
release xenial-minimal-rock64-0.5.15-136-armhf.img.xz
prerelease bionic-minimal-rock64-0.6.30-208-armhf.img.xz
debian stetch in general seems broken with python3?
04-02-2018, 06:20 AM
(This post was last modified: 04-02-2018, 06:30 AM by prof7bit.)
same problem here.
Is there any way to fix it?
The upgrade that broke it instralled these things (taken from /var/log/apt/history.log):
Install:
linux-headers-4.4.120-rockchip-ayufan-208:arm64 (0.6.30, automatic),
linux-image-4.4.120-rockchip-ayufan-208:arm64 (0.6.30, automatic),
device-tree-compiler:arm64 (1.4.2-1, automatic),
debsums:arm64 (2.2.2, automatic),
u-boot-rock64:arm64 (0.6.30, automatic),
libfile-fnmatch-perl:arm64 (0.02-2+b3, automatic)
Upgrade:
openssl:arm64 (1.1.0f-3+deb9u1, 1.1.0f-3+deb9u2),
libsystemd0:arm64 (232-25+deb9u2, 232-25+deb9u3),
libicu57:arm64 (57.1-6+deb9u1, 57.1-6+deb9u2),
udev:arm64 (232-25+deb9u2, 232-25+deb9u3),
libudev1:arm64 (232-25+deb9u2, 232-25+deb9u3),
systemd-sysv:arm64 (232-25+deb9u2, 232-25+deb9u3),
linux-rock64:arm64 (0.6.25, 0.6.30),
libpam-systemd:arm64 (232-25+deb9u2, 232-25+deb9u3),
systemd:arm64 (232-25+deb9u2, 232-25+deb9u3),
linux-rock64-package:arm64 (0.6.25, 0.6.30),
libssl1.1:arm64 (1.1.0f-3+deb9u1, 1.1.0f-3+deb9u2),
libssl1.0.2:arm64 (1.0.2l-2+deb9u2, 1.0.2l-2+deb9u3),
tzdata:arm64 (2018c-0+deb9u1, 2018d-0+deb9u1)
downgrading the kernel to
linux-image-4.4.77-rockchip-ayufan-136
seems to make the problem disappear. Also with 4.4.114-rockchip-ayufan-193 it seems to have worked, that was the version I have been using for a few months now, only today when I upgraded from -193 to -208 it broke. So it must somehow be related to something that changed in the kernel.
It also segfaulted not always immediately and not deterministically, in 2 out of 5 attempts it was possible to start a python shell without segfault before the prompt but trying to run a script was still impossible.
I use latest pre-release img,
stretch-minimal-rock64-0.6.32-210-arm64.img.xz
Segmentation fault also come up with high probability.
I'm not sure if this issue is only show in arm64 while works fine in armhf.
|
