Messages in this thread | | | Date | Tue, 19 Jan 2010 10:13:59 -0500 (EST) | From | Alan Stern <> | Subject | Re: [linux-pm] [RFC][PATCH] PM: Force GFP_NOIO during suspend/resume (was: Re: Memory allocations in .suspend became very unreliable) |
| |
On Mon, 18 Jan 2010, Rafael J. Wysocki wrote:
> On Monday 18 January 2010, Alan Stern wrote: > > On Mon, 18 Jan 2010, Oliver Neukum wrote: > > > > > Am Montag, 18. Januar 2010 00:00:23 schrieb Rafael J. Wysocki: > > > > On Sunday 17 January 2010, Benjamin Herrenschmidt wrote: > > > > > On Sun, 2010-01-17 at 14:27 +0100, Rafael J. Wysocki wrote: > > > > ... > > > > > However, it's hard to deal with the case of allocations that have > > > > > already started waiting for IOs. It might be possible to have some VM > > > > > hook to make them wakeup, re-evaluate the situation and get out of that > > > > > code path but in any case it would be tricky. > > > > > > > > In the second version of the patch I used an rwsem that made us wait for these > > > > allocations to complete before we changed gfp_allowed_mask. > > > > > > This will be a very, very hot semaphore. What's the impact on performance? > > > > Can it be replaced with something having lower overhead, such as SRCU? > > I'm not sure about that. In principle SRCU shouldn't be used if the reader can > sleep unpredictably long and the memory allocation sutiation is one of these.
I don't think this matters. Each SRCU usage has its own domain, and different domains don't affect one another. So the only thing that would be blocked by a long-sleeping reader would be the suspend process itself, and obviously you won't mind that.
Alan Stern
| |