Bug 1509936
| Summary: | Disable python-dmidecode build for s390x | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux 7 | Reporter: | Lukáš Doktor <ldoktor> |
| Component: | python-dmidecode | Assignee: | lijiang |
| Status: | CLOSED ERRATA | QA Contact: | Mike Gahagan <mgahagan> |
| Severity: | unspecified | Docs Contact: | |
| Priority: | unspecified | ||
| Version: | 7.4-Alt | CC: | jstodola, ldoktor, ruyang, tkasparek, tlestach |
| Target Milestone: | rc | ||
| Target Release: | --- | ||
| Hardware: | s390x | ||
| OS: | Unspecified | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | If docs needed, set a value | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2018-10-30 11:48:42 UTC | Type: | Bug |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
| Bug Depends On: | 1617942 | ||
| Bug Blocks: | |||
[root@ibm-z-10 ~]# dmesg|grep "/dev/mem" [ 739.354298] Program python tried to access /dev/mem between f0000->100000. [ 744.269732] Program python tried to access /dev/mem between f0000->100000. The python-dmidecode will access "/dev/mem", looks like it has no permission to access "/dev/mem", furthermore, we can also see some warnings about this. ** COLLECTED WARNINGS ** /dev/mem (mmap): Operation not permitted No SMBIOS nor DMI entry point found, sorry. /dev/mem (mmap): Operation not permitted ** END OF WARNINGS ** This warning should probably related to the kernel enables the option "CONFIG_STRICT_DEVMEM". I'm not sure if another packages will depend on the python-dmidecode. That could need to talk about disabling build for s390x whether it is reasonable. By the way, upstream hasn't maintained this package. Thanks. x86 is about the only arch left where /dev/mem is readable in this manner. (In reply to Mike Gahagan from comment #3) > x86 is about the only arch left where /dev/mem is readable in this manner. Sure, we previously disabled the kernel option(CONFIG_STRICT_DEVMEM), it doesn't have WARNINGS, but it returns the null string, in fact, that might make no sense. (In reply to lijiang from comment #2) > [root@ibm-z-10 ~]# dmesg|grep "/dev/mem" > [ 739.354298] Program python tried to access /dev/mem between f0000->100000. > [ 744.269732] Program python tried to access /dev/mem between f0000->100000. > > The python-dmidecode will access "/dev/mem", looks like it has no permission > to access "/dev/mem", furthermore, we can also see some warnings about this. > > ** COLLECTED WARNINGS ** > /dev/mem (mmap): Operation not permitted > No SMBIOS nor DMI entry point found, sorry. > /dev/mem (mmap): Operation not permitted > ** END OF WARNINGS ** > > This warning should probably related to the kernel enables the option > "CONFIG_STRICT_DEVMEM". > > I'm not sure if another packages will depend on the python-dmidecode. That > could need to talk about disabling build for s390x whether it is reasonable. > By the way, upstream hasn't maintained this package. > > Thanks. I'm not sure what info should I provide. I'm not aware of any program depending on python-dmesg but if there was it should be fixed not to use it as it won't work anyway. built rpm package: https://brewweb.engineering.redhat.com/brew/taskinfo?taskID=17337993 I've confirmed there are no s/390x builds of python-dmidecode-3.12.2-3.el7 there are s/390 builds of it however and it probably makes sense to disable those. Note also that alpha 1.2 has python-dmidecode-3.12.2-2.el7.s390x.rpm so as long as we bump up the version of python-dmidecode in future composes we should be fine. Marking Verified based on the 3.12.2-3.el7 version. Since there is no dmidecode for s390, I suppose python-dmidecode should not be built for s390 either. Actually, python-dmidecode.s390 (and it's dependency libxml2-python.s390) got pulled into compose RHEL-7.6-20180802.2 instead of python-dmidecode.s390x as a dependency of rhn-client-tools - is this what we want? Since rhn-client-tools requires python-dmidecode, there would be a missing dependency on s390x if there is no python-dmidecode for s390/s390x. Adding tkasparek (rhn-client-tools maintainer) to CC to be aware of this possible dependency issue. Moving this bug back to ASSIGNED to clarify what will happen with s390 build of python-dmidecode. Satellite 5 (and thus rhn-client-tools) does NOT support RHEL ALT, as long as these changes end up in regular RHEL 7 we're okay in this.
If these changes end up in both versions of RHEL 7 (regular and ALT), let me know and we should work together on some solution for this.
However I was not able to even run reproducer from the OP.
>>> import dmidecode
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/lib/python2.7/site-packages/dmidecode.py", line 29, in <module>
from dmidecodemod import *
ImportError: /usr/lib/python2.7/site-packages/dmidecodemod.so: wrong ELF class: ELFCLASS32
[root@ibm-z-35 ~]# uname -a
Linux ibm-z-35.rhts.eng.bos.redhat.com 3.10.0-931.el7.s390x #1 SMP Tue Jul 31 17:55:54 EDT 2018 s390x s390x s390x GNU/Linux
[root@ibm-z-35 ~]# cat /etc/redhat-release
Red Hat Enterprise Linux Server release 7.6 Beta (Maipo)
Looking into the issue more.
RHEL and RHEL-ALT contain mostly the same packages, so this change affects both RHEL and RHEL-ALT s390x composes. (In reply to Tomáš Kašpárek from comment #10) > Satellite 5 (and thus rhn-client-tools) does NOT support RHEL ALT, as long > as these changes end up in regular RHEL 7 we're okay in this. I meant 'as long as these changes DOES NOT end up', sorry for confusion :( Hi, Tomáš Kašpárek Do you have a plan to remove the dependency for the rhn-client-tools? Actually, the python-dmidecode can't get the any data on s390x. If there's a way how to stop requiring python-dmidecode (and to keep the rhn-client-tools a noarch package), I'm ok to proceed with this BZ. Background info: dmidecode is actually missing not only in RHEL8, but also in RHEL7, RHEL6 and RHEL5 for the s390x architecture. Note that this is not only s390x problem, for example here is bz related to aarch64 https://bugzilla.redhat.com/show_bug.cgi?id=1509938 and (don't remember where) but I heard dmidecode should only work on x86_64. It'd be nice to get rid of this dependency. (In reply to Lukas Doktor from comment #15) > Note that this is not only s390x problem, for example here is bz related to > aarch64 ... Sure, it's just I'm talking for Satellite 5 product and aarch64 is not relevant for us ... while s390x is ... a lot. :-) There is another bug, Lianbo mentioned python-dmidecode upstream is no response for long time, maybe not maitained: https://bugzilla.redhat.com/show_bug.cgi?id=1606950 Does anyone see problems if we drop it? Hi, Mike and Jan This bug only disables s390x on rhel-7.6 branch, if anyone also wants to disable s390 on rhel-7.6 branch, he can create a new bug. So i move the status to ON_QA again, Would you like to check it? Thanks. confirmed RHEL-7.6-20180907.0 does not contain python-dmidecode for s/390x or s/390 Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2018:3314 |
Description of problem: The `dmidecode` for s390x does not exists, but this module is available for python, which makes no sense as it actually can't get any data. Version-Release number of selected component (if applicable): python-dmidecode-3.12.2-1.el7.s390x How reproducible: Always Steps to Reproduce: >>> import dmidecode >>> dmidecode.bios() Actual results: Traceback (most recent call last): File "<stdin>", line 1, in <module> RuntimeError: [src/dmidecodemodule.c:328] Error decoding DMI data Expected results: dmidecode should not be available to install (or it should behave correctly) Additional info: