Thread (3 messages) 3 messages, 2 authors, 2018-07-11

Re: Why the mce reverts in 4.16-rt?

From: Steven Rostedt <rostedt@goodmis.org>
Date: 2018-07-11 12:51:07
Also in: lkml

On Wed, 11 Jul 2018 09:23:11 +0200
Sebastian Andrzej Siewior [off-list ref] wrote:
On 2018-07-10 20:56:24 [-0400], Steven Rostedt wrote:
quoted
Hi Sebastian,  
Hi Steven,
quoted
I'm looking at backporting patches from 4.16-rt and noticed that you
have:

Revert "x86: Convert mce timer to hrtimer"
Revert "x86/mce: use swait queue for mce wakeups"

With no explanation to why they were reverted. When did things change?
Should this be 4.14-rt material?  
from the announce email:
|    Changes since v4.16.8-rt2:
|    
|      - Drop two MCE related patches. They are no longer required since the
|        timer wheel rework.
Thanks, still would have preferred a comment in the revert commit
message ;-)
The timer wheel rework made it possible to have a raw_spin_lock_t in the
timer_list timer. The problem with MCE was that it enqueued a timer from
hardirq context which is now possible. So the patches are no longer
required. This revert does not fix a bug in v4.14 nor does it break
anything.
OK, even though this doesn't fix anything, I think I'll still
backport it.

-- Steve
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help