[Bug 1705538] Comment bridged from LTC Bugzilla

bugproxy bugproxy at us.ibm.com
Thu Apr 12 13:30:39 UTC 2018


------- Comment From hegdevasant at in.ibm.com 2018-04-12 09:24 EDT-------
Steve,

(In reply to comment #29)
> > Also we have unit test cases in libservicelog package .. which is enough to cover basic
> > validation.
>
> If I'm not mistaken, we still have this testsuite disabled in the package
> because while I sent some fixes, servicelog_open() still fails if
> /var/lib/servicelog does not exist.  Does this mean you will plan to run
> this testsuite manually against the packages in the SRU?

Yes. Unfortunately we still need to disable test suites.

Its in my TODO list .. but never got their due to various reasons. So we
have 2 issues that blocks us enabling test suite by default.

1 - if servicelog_open() returns a failure, test_libservicelog segfaults
because it tries to dereference a null pointer.

I've applied your patch. I think we still left with one more place. I
have patch locally to fix this issue.

2 - if /var/lib/libservicelog/ does not already exist on the system,
servicelog_open() will *always* fail.

I'm thinking of creating database under current dir itself and make it
work. I will get this done in a day or two.

-Vasant

------- Comment From hegdevasant at in.ibm.com 2018-04-12 09:28 EDT-------
> > Update Package request for libservicelog. Some upstream patches have been requested to be included in both bionic and xenial.
>
> This is not a user impact explanation as described at
> https://wiki.ubuntu.com/StableReleaseUpdates#Procedure and I can't find any
> other justification or need for this to go to Xenial.
>
> Please could someone update the reason this need to go into Xenial from a
> user perspective and update the bug description accordingly?

Patches mentioned in this bugzilla can potentially impact customer
environment. Hence we wanted those fixes in 16.04 LTS release.

But when we added test cases we did broke it in build environment.
Anyway this was new stuff added later.

Let me know whether you can do some manually test and push changes?
-OR- give test build so that IBM can test it before you merge? and I promise to test it ASAP .

-Vasant

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to libservicelog in Ubuntu.
https://bugs.launchpad.net/bugs/1705538

Title:
  Update Package request for libservicelog

Status in The Ubuntu-power-systems project:
  Incomplete
Status in libservicelog package in Ubuntu:
  Fix Released
Status in libservicelog source package in Xenial:
  Incomplete

Bug description:
  [Impact]

  Update Package request for libservicelog. Some upstream patches have
  been requested to be included in both bionic and xenial.

  [Test Case]

  The packages will be validated by the requesting party (see original
  description).

  [Regression Potential]

  Hard to assess but the requested patches have been in upstream trunk
  since at least a year and no issues have been reported - changes also
  present in Debian and Ubuntu bionic since last year.

  [Original Description]

  ---Problem Description---
  Update Package request for libservicelog

  Machine Type = lpar

  ---Steps to Reproduce---
   servicelog --dump

  ---uname output---
  Linux tuleta4u-lp9 4.10.0-27-generic #30~16.04.2-Ubuntu SMP Thu Jun 29 16:06:52 UTC 2017 ppc64le ppc64le ppc64le GNU/Linux

  Please pull below mentioned patches for libservicelog package

  commit 4fe9d9239f172604e16268ca1fb6fff1c06632b2
  Author: Ankit Kumar <ankit at linux.vnet.ibm.com>
  Date:   Sat May 20 01:38:41 2017 +0530

      Validate text string before and after bind call

      While binding string to query, if string is NULL then bind call gets ignored.
      While fetching data from string, if data is NULL then current code makes query
      as NULL and returns.

      This patch validates text string before and after bind call. It assigns string
      to "" incase data is NULL points to NULL pointer. After this we will be able
      to display information even if some of string data is NULL.

      It also does NULL checks for compulsory string.

      Signed-off-by: Ankit Kumar <ankit at linux.vnet.ibm.com>
      [Killed redundant goto statements - Vasant]
      Signed-off-by: Vasant Hegde <hegdevasant at linux.vnet.ibm.com>

  commit 787594814eb88e3149cd660bcb6aaa6d3dd1347c
  Author: Ankit Kumar <ankit at linux.vnet.ibm.com>
  Date:   Sat May 20 01:10:49 2017 +0530

      Correct string length calculation and validates destination buffer
  size before strncpy

      This patch corrects string length calculation logic and validates destination
      buffer size before calling strncpy to avoid memory corruption.

      Signed-off-by: Ankit Kumar <ankit at linux.vnet.ibm.com>
      [Moved memset to right place and removed redundant condition check -
       Vasant]
      Signed-off-by: Vasant Hegde <hegdevasant at linux.vnet.ibm.com>

  commit 48875ee8614eeefaa3d5d8ff92fb424915738169
  Author: Ankit Kumar <ankit at linux.vnet.ibm.com>
  Date:   Thu Sep 15 16:16:49 2016 +0530

      NULL check before strdup call

      This patch does "NULL checks" before passing argument to strdup
  call.

      Signed-off-by: Ankit Kumar <ankit at linux.vnet.ibm.com>
      Signed-off-by: Vasant Hegde <hegdevasant at linux.vnet.ibm.com>

  commit 40b4f7a52e61fb9da30b4cb9b5de9a85673da262
  Author: Ankit Kumar <ankit at linux.vnet.ibm.com>
  Date:   Thu Sep 15 16:16:48 2016 +0530

      NULL check before strlen call

      This patch checks NULL pointer before strlen call.

      Signed-off-by: Ankit Kumar <ankit at linux.vnet.ibm.com>
      [Fixed build warning - Vasant]
      Signed-off-by: Vasant Hegde <hegdevasant at linux.vnet.ibm.com>

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu-power-systems/+bug/1705538/+subscriptions



More information about the foundations-bugs mailing list