[Starlingx-discuss] Opinion wanted on STX 2.0 kernel memory leak bug 1836638
Hi Folks, I have identified 3 kernel patches to fix the observed memory leak in the RT kernel. Namely listed here in comment 53: https://bugs.launchpad.net/starlingx/+bug/1836638/comments/53 While the leak is only really seen in RT, the fixes are not RT specific. The RT kernel has ferreted out many linux bugs in the past mainly due to its different scheduling points, and running irq handlers as kernel threads, causing things to run in a different order from std. So should these patches be applied to both of our kernels, even though we only see the leaking on RT? I would say yes, but don't want to waste my, and possibly Yi's, time if approvers disagree. Thanks, -Jim
Nice work ! +1 on doing both kernels. Brent Sent from my iPhone
On Aug 23, 2019, at 5:58 AM, Jim Somerville <jim.somerville@windriver.com> wrote:
Hi Folks,
I have identified 3 kernel patches to fix the observed memory leak in the RT kernel. Namely listed here in comment 53: https://bugs.launchpad.net/starlingx/+bug/1836638/comments/53
While the leak is only really seen in RT, the fixes are not RT specific. The RT kernel has ferreted out many linux bugs in the past mainly due to its different scheduling points, and running irq handlers as kernel threads, causing things to run in a different order from std.
So should these patches be applied to both of our kernels, even though we only see the leaking on RT? I would say yes, but don't want to waste my, and possibly Yi's, time if approvers disagree.
Thanks,
-Jim
Yes, I think we should do on both kernels, the chances that we didn't see std kernel leak might due to our test cases coverage, but as we now have root-causes, we shall get it fixed. -----Original Message----- From: Rowsell, Brent [mailto:Brent.Rowsell@windriver.com] Sent: Friday, August 23, 2019 5:54 AM To: Somerville, Jim <Jim.Somerville@windriver.com> Cc: starlingx <starlingx-discuss@lists.starlingx.io>; Troyer, Dean <dean.troyer@intel.com>; Saul Wold <sgw@linux.intel.com>; Wang, Yi C <yi.c.wang@intel.com> Subject: Re: [Starlingx-discuss] Opinion wanted on STX 2.0 kernel memory leak bug 1836638 Nice work ! +1 on doing both kernels. Brent Sent from my iPhone
On Aug 23, 2019, at 5:58 AM, Jim Somerville <jim.somerville@windriver.com> wrote:
Hi Folks,
I have identified 3 kernel patches to fix the observed memory leak in the RT kernel. Namely listed here in comment 53: https://bugs.launchpad.net/starlingx/+bug/1836638/comments/53
While the leak is only really seen in RT, the fixes are not RT specific. The RT kernel has ferreted out many linux bugs in the past mainly due to its different scheduling points, and running irq handlers as kernel threads, causing things to run in a different order from std.
So should these patches be applied to both of our kernels, even though we only see the leaking on RT? I would say yes, but don't want to waste my, and possibly Yi's, time if approvers disagree.
Thanks,
-Jim
_______________________________________________ Starlingx-discuss mailing list Starlingx-discuss@lists.starlingx.io http://lists.starlingx.io/cgi-bin/mailman/listinfo/starlingx-discuss
Jim: Great work on tracking down these issues, I am curious are they addressed in the newer 3.10 series from CentOS or only from other upstream sources? If they are from other uptream sources, has it been reported to CentOS/Redhat? More curious than anything else, not a requirement. I do agree that we should make the changes to both kernel. Sau! On 8/22/19 1:58 PM, Jim Somerville wrote:
Hi Folks,
I have identified 3 kernel patches to fix the observed memory leak in the RT kernel. Namely listed here in comment 53: https://bugs.launchpad.net/starlingx/+bug/1836638/comments/53
While the leak is only really seen in RT, the fixes are not RT specific. The RT kernel has ferreted out many linux bugs in the past mainly due to its different scheduling points, and running irq handlers as kernel threads, causing things to run in a different order from std.
So should these patches be applied to both of our kernels, even though we only see the leaking on RT? I would say yes, but don't want to waste my, and possibly Yi's, time if approvers disagree.
Thanks,
-Jim
_______________________________________________ Starlingx-discuss mailing list Starlingx-discuss@lists.starlingx.io http://lists.starlingx.io/cgi-bin/mailman/listinfo/starlingx-discuss
participants (4)
-
Jim Somerville
-
Rowsell, Brent
-
Saul Wold
-
Xie, Cindy