On Tue, Aug 20 2013 at 5:57pm -0400, Frank Mayhar <fmay...@google.com> wrote:
> On Tue, 2013-08-20 at 17:47 -0400, Mikulas Patocka wrote: > > > > On Tue, 20 Aug 2013, Frank Mayhar wrote: > > > > > On Tue, 2013-08-20 at 17:22 -0400, Mikulas Patocka wrote: > > > > On Fri, 16 Aug 2013, Frank Mayhar wrote: > > > > > The device mapper and some of its modules allocate memory pools at > > > > > various points when setting up a device. In some cases, these pools > > > > > are > > > > > fairly large, for example the multipath module allocates a 256-entry > > > > > pool and the dm itself allocates three of that size. In a > > > > > memory-constrained environment where we're creating a lot of these > > > > > devices, the memory use can quickly become significant. > > > > > Unfortunately, > > > > > there's currently no way to change the size of the pools other than by > > > > > changing a constant and rebuilding the kernel. > > > > I think it would be better to set the values to some low value (1 > > > > should > > > > be enough, there is 16 in some places that is already low enough). > > > > There > > > > is no need to make it user-configurable, because the user will see no > > > > additional benefit from bigger mempools. > > > > > > > > Maybe multipath is the exception - but other dm targets don't really > > > > need > > > > big mempools so there is no need to make the size configurable. > > > > > > The point is not to make the mempools bigger, the point is to be able to > > > make them _smaller_ for memory-constrained environments. In some cases, > > > even 16 can be too big, especially when creating a large number of > > > devices. > > > > > > In any event, it seems unfortunate to me that these values are > > > hard-coded. One shouldn't have to rebuild the kernel to change them, > > > IMHO. > > > > So make a patch that changes 16 to 1 if it helps on your system (I'd like > > to ask - what is the configuration where this kind of change helps?). > > There is no need for that to be tunable, anyone can live with mempool size > > being 1. > > I reiterate: It seems unfortunate that these values are hard-coded. It > seems to me that a sysadmin should be able to tune them according to his > or her needs. Particularly when the same kernel is being run on many > machines that may have different constraints; building a special kernel > for each set of constraints doesn't scale. > > And as I said, it's a memory-constrained environment. Mikulas' point is that you cannot reduce the size to smaller than 1. And aside from rq-based DM, 1 is sufficient to allow for forward progress even when memory is completely consumed. A patch that simply changes them to 1 but makes the rq-based DM mempool's size configurable should actually be fine. Mike -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
