rpms/linuxptp
7
0
Fork 0
Code Issues Pull Requests Releases Activity

fix high CPU usage by ptp4l when link is down (RHEL-178071)

Browse Source 
Resolves: RHEL-178071
This commit is contained in:
Miroslav Lichvar 2026-05-20 10:09:58 +02:00
parent 1ea2129850
commit 8ea86ea011
1 changed files with 33 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
linuxptp-rtnlinit.patch
View File

@ -73,3 +73,36 @@ index 7f945ac..1bb407c 100644
}
}
commit 84c2d7f4ef7c0ff9353909194ae03e5a09aa0a12
Author: Miroslav Lichvar (via linuxptp-devel Mailing List) <linuxptp-devel@lists.nwtime.org>
Date: Mon May 4 15:32:46 2026 +0200
port: Don't request sde on repeated link down event.
When the link goes down, request a state decision event only when the
link status actually changed (the same status can be received multiple
times due to unrelated events or refresh).
This fixes an infinite loop when the link is down and BMCA is set to
"noop" due to bmc_state_decision() copying the FAULTY state and
causing port_dispatch() to be called with EV_FAULT_DETECTED, which
triggers an rntl_link_status() call. The netlink response (with no
change in the link status) no longer forces a new state decision event.
Fixes: a1eae149f519 ("port: Refresh link status on faults.")
Reported-by: Martin Pecka <peckama2@fel.cvut.cz>
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
diff --git a/port.c b/port.c
index 90b0860..05aabea 100644
--- a/port.c
+++ b/port.c
@@ -3042,7 +3042,7 @@ void port_link_status(void *ctx, int linkup, int ts_index)
* A port going down can affect the BMCA result.
* Force a state decision event.
*/
- if (p->link_status & LINK_DOWN)
+ if (p->link_status & LINK_STATE_CHANGED && p->link_status & LINK_DOWN)
clock_set_sde(p->clock, 1);
}