Re: [Patch] Allocate sparse vmemmap block above 4G

2007-11-09 Thread Mel Gorman
On (09/11/07 09:28), Zou, Nanhai didst pronounce: > > Wrong markup there I believe. The __meminit markup is for functions > > that are needed at runtime when memory is hot-added or hot-removed. > > Bootmem functions do not qualify. __init is sufficient. > > > __meminit here is to avoid section li

RE: [Patch] Allocate sparse vmemmap block above 4G

2007-11-08 Thread Christoph Lameter
On Fri, 9 Nov 2007, Zou, Nanhai wrote: > > More magic values, both the 4GiB address here and the magic "1" at the > > end are problems. > > > Yes, the 4UL*1024*1024*1024 could be a define here. The 4GB boundary here is MAX_DMA32_ADDRESS I guess? We are only having this problem because of the tw

RE: [Patch] Allocate sparse vmemmap block above 4G

2007-11-08 Thread Zou, Nanhai
Re: [Patch] Allocate sparse vmemmap block above 4G > > On (08/11/07 08:52), Zou Nan hai didst pronounce: > > Resend the patch for more people to review > > > > On some single node x64 system with huge amount of physical memory e.g > > > 64G. the memmap size maybe

Re: [Patch] Allocate sparse vmemmap block above 4G

2007-11-08 Thread Mel Gorman
On (08/11/07 08:52), Zou Nan hai didst pronounce: > Resend the patch for more people to review > > On some single node x64 system with huge amount of physical memory e.g > > 64G. the memmap size maybe very big. > > If the memmap is allocated from low pages, it may occupies too much > memory belo

Re: [Patch] Allocate sparse vmemmap block above 4G

2007-11-07 Thread Dave Hansen
On Wed, 2007-11-07 at 15:07 +0800, Zou Nan hai wrote: > Try to allocate sparse vmemmap block above 4G on x64 system. > > On some single node x64 system with huge amount of physical memory e.g > 64G. the memmap size maybe very big. Could we just change the default bootmem behavior to allocate top