From: Jason Lunz Date: Thu, 29 Mar 2007 08:20:31 +0000 (-0700) Subject: [PATCH] uml: fix LVM crash X-Git-Tag: v2.6.21-rc6~62 X-Git-Url: https://err.no/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=af84eab208916acad91f6342bbd57bc865caf150;p=linux-2.6 [PATCH] uml: fix LVM crash Permit lvm to create logical volumes without crashing UML. When device-mapper's DM_DEV_CREATE_CMD ioctl is called to create a new device, dev_create()->dm_create()->alloc_dev()-> blk_queue_bounce_limit(md->queue, BLK_BOUNCE_ANY) is called. blk_queue_bounce_limit(BLK_BOUNCE_ANY) calls init_emergency_isa_pool() if blk_max_pfn < blk_max_low_pfn. This is the case on UML, but init_emergency_isa_pool() hits BUG_ON(!isa_page_pool) because there doesn't seem to be a dma zone on UML for mempool_create() to allocate from. Most architectures seem to have max_low_pfn == max_pfn, but UML doesn't because of the uml_reserved chunk it keeps for itself. From what I can see, max_pfn and max_low_pfn don't get much use after the bootmem-allocator stops being used anyway, except that they initialize the block layer's blk_max_low_pfn/blk_max_pfn. This ensures init_emergency_isa_pool() doesn't crash uml in this situation by setting max_low_pfn == max_pfn in mem_init(). Signed-off-by: Jason Lunz Signed-off-by: Jeff Dike Cc: Paolo 'Blaisorblade' Giarrusso Cc: Alasdair G Kergon Cc: Jens Axboe Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- diff --git a/arch/um/kernel/mem.c b/arch/um/kernel/mem.c index e85d65deea..df7d662b98 100644 --- a/arch/um/kernel/mem.c +++ b/arch/um/kernel/mem.c @@ -64,8 +64,6 @@ static void setup_highmem(unsigned long highmem_start, void mem_init(void) { - max_low_pfn = (high_physmem - uml_physmem) >> PAGE_SHIFT; - /* clear the zero-page */ memset((void *) empty_zero_page, 0, PAGE_SIZE); @@ -80,6 +78,7 @@ void mem_init(void) /* this will put all low memory onto the freelists */ totalram_pages = free_all_bootmem(); + max_low_pfn = totalram_pages; #ifdef CONFIG_HIGHMEM totalhigh_pages = highmem >> PAGE_SHIFT; totalram_pages += totalhigh_pages;