In page_alloc.c in the function capture_page_range there is a call to try_to_migrate_pages(), which is not defined unless CONFIG_MEMORY_MIGRATE is turned on. My first hackish fix was to #define try_to_migrate_pages(p1) to (NULL) in hopes that will break us out of the loop. This hackish fix is to just #define out the whole loop. I am leaning toward the real fix being to make CONFIG_MEMORY_HOTPLUG depend on CONFIG_MEMORY_MIGRATE, but if we don't want to tie them together this might still be useful. CONFIG_MEMORY_HOTPLUG depends on CONFIG_MEMORY_MIGRATE is the right fix in the long run but, for now, just use the ifdef. This patch will not be sumbitted upstream - daveh Signed-off-by: Joel Schopp Signed-off-by: Dave Hansen --- memhotplug-dave/mm/page_alloc.c | 3 +++ 1 files changed, 3 insertions(+) diff -puN mm/page_alloc.c~Y1-ifdef-try_to_migrate_pages-FIXME mm/page_alloc.c --- memhotplug/mm/page_alloc.c~Y1-ifdef-try_to_migrate_pages-FIXME 2004-11-12 16:52:34.000000000 -0800 +++ memhotplug-dave/mm/page_alloc.c 2004-11-12 16:52:34.000000000 -0800 @@ -762,6 +762,8 @@ retry: remove_page_freearea(page, order); rest = grab_capturing_pages(&page_list, start_pfn, nr_pages); + +#ifdef CONFIG_MEMORY_MIGRATE remains = try_to_migrate_pages(&page_list); if (rest || !list_empty(&page_list)) { @@ -772,6 +774,7 @@ retry: msleep(1000); goto retry; } +#endif /* * storing the last result (fp) keeps up from having _