On Tue, Apr 2, 2013 at 7:17 AM, Vivek Goyal <vgo...@redhat.com> wrote: > On Tue, Apr 02, 2013 at 09:50:01AM -0400, Vivek Goyal wrote: > > [..] >> To achieve the behavior where we want to enforce that memory either >> comes from low or high area only otherwise allocation fails, we could >> probably use. >> >> crashkernel=X,high_only >> crashkernel=X,low_only > > Thinking more about it. We have following existing syntax. > > crashkernel=range1:size1[,range2:size2,...][@offset] > > Which uses ',' as delimiter for range:size pairs. > > May be we can use a different delimiter, say ';'. > > crashkernel=<amount_of_memory_to_reserve>;<option1>;<option2> > > All the existing crashkernel= options should fall into the category of > <amount_of_memory_to_reserve>. > > option1 could specify whether to search for memory from higher addresses > or from lower addresses. (valid values are high or low) > > option2 could specify the range of memory search should be performed in. > syntax to specify range could be XM:[YM]. So one could possibly specify. > > 4G-8G > <4G >>4G:<8G >>8G > > Now crashkernel_low=0 could be emulated by > > crashkernel=0M;;<4G > > If we want to reserve 128MB of memory between 4G and 8G and starting scan > from high, we could say. > > crashkernel=128M;high;>4G:<8G > > If this is deemed too generic and not worth it. Then we could simplify > option 2 to take values "high_only" and "low_only" and that leaves the > scope of implementing proper ranges down the line if need be. > > So crashkernel_low=0 will be emulated by. > > crashkernel=0M;;low_only
Oh, no. the grammar for crashkernel= looks crazy now. You should not burden user like this. Thanks Yinghai -- 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/
