Messages in this thread | | | From | "Rafael J. Wysocki" <> | Subject | Re: Linux 3.19-rc5 | Date | Fri, 30 Jan 2015 02:49:40 +0100 |
| |
On Thursday, January 29, 2015 05:12:11 PM Linus Torvalds wrote: > On Thu, Jan 29, 2015 at 5:22 PM, Rafael J. Wysocki <rjw@rjwysocki.net> wrote: > > On Wednesday, January 21, 2015 05:54:00 PM Peter Hurley wrote: > >> > >> Yeah, but the debug check is triggering worse behavior, requiring > >> bisecting back to the debug commit. > > > > Yes, it is. > > > > So I'm wondering is anyone is working on fixing this in any way? > > > > It kind of sucks when this is happening on an otherwise perfectly usable > > old(ish) machine ... > > The WARN() was already changed to a WARN_ONCE(). > > So that debug check doesn't cause problems any more. If somebody is > bisecting something else, and the WARN() is a problem for those > intermediate kernels, then just disabling CONFIG_DEBUG_ATOMIC_SLEEP > should get you past that point. > > IOW, this really shouldn't be an issue. > > Does the pccard thing still not work?
Interestingly enough, if the kernel is built with CONFIG_DEBUG_ATOMIC_SLEEP unset, the problem with 99+% CPU load from pccardd goes away, so thanks for the hint.
Rafael
| |