[Bug 1783252] Comment bridged from LTC Bugzilla
bugproxy
bugproxy at us.ibm.com
Wed Aug 1 14:19:22 UTC 2018
------- Comment From wschmidt at us.ibm.com 2018-08-01 10:10 EDT-------
By the way, there *will* be a problem in greenlet if REGS_TO_SAVE includes r2. In that case you'll see the same error message, with r30 replaced by r2. That's because you aren't allowed to clobber the TOC register with an asm like this, so it's a feechur, not a bug. So please verify they aren't doing that.
--
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to gcc-5 in Ubuntu.
https://bugs.launchpad.net/bugs/1783252
Title:
greenlet 0.4.14 will not build on ppc64le
Status in The Ubuntu-power-systems project:
Won't Fix
Status in gcc-5 package in Ubuntu:
New
Bug description:
== Comment: #0 - William M. Edmonds <edmondsw at us.ibm.com> - 2018-07-23 16:21:41 ==
---Problem Description---
greenlet 0.4.14 will not build on ppc64le
Opened https://github.com/python-greenlet/greenlet/issues/136 because
attempting to build bdist_wheel for greenlet 0.4.14 on ppc64le Ubuntu
16.04 LTS yields the following error:
running build
running build_ext
building 'greenlet' extension
creating build
creating build/temp.linux-ppc64le-2.7
powerpc64le-linux-gnu-gcc -pthread -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -fno-strict-aliasing -Wdate-time -D_FORTIFY_SOURCE=2 -g -fstack-protector-strong -Wformat -Werror=format-security -fPIC -I/usr/include/python2.7 -c greenlet.c -o build/temp.linux-ppc64le-2.7/greenlet.o -fno-tree-dominator-opts
In file included from slp_platformselect.h:16:0,
from greenlet.c:343:
platform/switch_ppc64_linux.h: In function 'slp_switch':
platform/switch_ppc64_linux.h:80:5: error: PIC register clobbered by 'r30' in 'asm'
__asm__ volatile ("" : : : REGS_TO_SAVE);
^
platform/switch_ppc64_linux.h:95:5: error: PIC register clobbered by 'r30' in 'asm'
__asm__ volatile ("" : : : REGS_TO_SAVE);
^
error: command 'powerpc64le-linux-gnu-gcc' failed with exit status 1```
But the greenlet community is saying this is a gcc bug and it would not be safe to revert the greenlet change that exposed this issue. I have no idea whether that is true or not. Supposedly this should be fixed by https://gcc.gnu.org/git/?p=gcc.git;a=commitdiff;h=9213244550335bcb2b8590a0d7d58ac74c932361 but that fix is not present in the latest version of gcc (5.4.0) available for Ubuntu 16.04 LTS.
---uname output---
unavailable due to lab outage
Machine Type = unavailable due to lab outage
---Debugger---
A debugger is not configured
---Steps to Reproduce---
This should be reproducible with `git clone git://github.com/python-greenlet/greenlet; cd greenlet; python setup.py bdist_wheel`, though it was originally found using pypi rather than github and with a more complicated (scripted) command as detailed in the greenlet bug description.
Contact Information = Matthew Edmonds / edmondsw at us.ibm.com
Userspace tool common name: greenlet
The userspace tool has the following bit modes: no idea
Userspace rpm: N/A
Userspace tool obtained from project website: 0.4.14
*Additional Instructions for Matthew Edmonds / edmondsw at us.ibm.com:
-Attach ltrace and strace of userspace application.
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu-power-systems/+bug/1783252/+subscriptions
More information about the foundations-bugs
mailing list