https://jira.whamcloud.com This file is an XML representation of an issue en-us 9.4.14 940014 05-12-2023 https://jira.whamcloud.com/browse/LU-9397 Lustre <p>"<a href="https://jira.whamcloud.com/browse/LU-6245" title="Untangle userland and kernel space support for libcfs" class="issue-link" data-issue-key="LU-6245"><del>LU-6245</del></a> libcfs: add ktime_get_real_seconds support" introduced a problem where sometimes we use cfs_time_current() and others ktime_get_real_seconds(). This leads to wrong comparison, leading to strange timing issues.</p> <p>One behavior observed were routers considering peers dead, due to these inconsistent comparison.</p> <p>We'll need to make sure that the LNet and LND code follows a consistent and correct approach.</p> LU-9397

Inconsistence use of cfs_time_current() and ktime_get_real_seconds()

Bug Critical Resolved Fixed James A Simmons Amir Shehata Tue, 25 Apr 2017 18:56:41 +0000 Fri, 2 Mar 2018 15:55:57 +0000 Sat, 3 Jun 2017 04:21:11 +0000 Lustre 2.10.0 0 6 <p>I see what happened. The ksocklnd stuff landed first with the intent of last_alive being moved over to time64_t. Later we change the behavior of LNetEQPoll and lnet ping so it went back to the jiffies based system. The ksocklnd code was left behind.</p> <p>James Simmons (uja.ornl@yahoo.com) uploaded a new patch: <a href="https://review.whamcloud.com/26813" class="external-link" target="_blank" rel="nofollow noopener">https://review.whamcloud.com/26813</a><br/> Subject: <a href="https://jira.whamcloud.com/browse/LU-9397" title="Inconsistence use of cfs_time_current() and ktime_get_real_seconds()" class="issue-link" data-issue-key="LU-9397"><del>LU-9397</del></a> ksocklnd: make last_alive in units of jiffies<br/> Project: fs/lustre-release<br/> Branch: master<br/> Current Patch Set: 1<br/> Commit: 7006890fcc484a8074a033517a6607e6ace15b8b</p> <p>It looks like a similar issue exists for <tt>ksock_conn::ksnc_tx_last_post</tt>.</p> <p>And in <tt>ksocknal_destroy_conn()</tt> the calculation in the <tt>CERROR()</tt> involving <tt>last_rcv</tt> is wrong. <tt>last_rcv</tt> is in jiffies, is subtracted from <tt>ktime_get_real_seconds()</tt>, with the result then being converted from jiffies to seconds.</p> <p>aside from Olaf's comments, I tested the fix and I don't see the issue any longer</p> <p>Good we are on the right track. I have to look over the ksocklnd driver and do a proper 64 bit time port.</p> <p>I have been looking at the socklnd code and it looks like everything is pretty much needs only second precision. It looks like everything could be changed into time64_t. Olaf do you see cases where we will need higher resolution? When we want to go to jiffies I can use cfs_time_seconds() as the case for the call to lnet_notify() and ksocknal_query(). Other than that I don't see the need for ktime_t.</p> <p>As far as I can tell, the main argument for jiffies is that they're cheaper to get and use. The extra resolution doesn't matter for these cases.</p> <p>Using <tt>ktime_get_real_seconds()</tt> in particular should get you (assuming proper time sync in place) time stamp values that can (easily) be compared between nodes. That by itself might justify this.</p> <p>Okay just wanted to verify that was the case for jiffies. The issue with jiffies they can also vary between nodes (<a href="https://jira.whamcloud.com/browse/LU-5443" title="Remove references to defined constant HZ" class="issue-link" data-issue-key="LU-5443"><del>LU-5443</del></a>) depending on what the user selected for HZ so I like to migrate away as much as possible from them. Currently I'm working on the patch but I have to go over it very carefully. To easy to break things.</p> <p>Oleg Drokin (oleg.drokin@intel.com) merged in patch <a href="https://review.whamcloud.com/26813/" class="external-link" target="_blank" rel="nofollow noopener">https://review.whamcloud.com/26813/</a><br/> Subject: <a href="https://jira.whamcloud.com/browse/LU-9397" title="Inconsistence use of cfs_time_current() and ktime_get_real_seconds()" class="issue-link" data-issue-key="LU-9397"><del>LU-9397</del></a> ksocklnd: move remaining time handling to 64 bits<br/> Project: fs/lustre-release<br/> Branch: master<br/> Current Patch Set: <br/> Commit: 59c25356208706b4dd0920bc7754f21e2db14a0f</p> <p>Landed for 2.10</p> Related LU-10707 Development Rank 1|hzzb73: Rank (Obsolete) 9223372036854775807 Severity