Re: 2.6.39.1 immediately reboots/resets on EFI system

[Jim Bos <jim876@xs4all.nl>]

[-- Attachment #1: Type: text/plain, Size: 3070 bytes --]

On 06/08/2011 12:25 AM, Yinghai Lu wrote:
> On 06/07/2011 05:22 AM, Maarten Lankhorst wrote:
>> It seems the error still occurs when calling free_bootmem_late
>> even if I only reserve blocks that haven't been reserved yet.
> 
> please check this one.
> 
> ---
>  arch/x86/platform/efi/efi.c |    2 +-
>  include/linux/mm.h          |    2 ++
>  mm/page_alloc.c             |   32 ++++++++++++++++++++++++++++++++
>  3 files changed, 35 insertions(+), 1 deletion(-)
> 
> Index: linux-2.6/arch/x86/platform/efi/efi.c
> ===================================================================
> --- linux-2.6.orig/arch/x86/platform/efi/efi.c
> +++ linux-2.6/arch/x86/platform/efi/efi.c
> @@ -368,7 +368,7 @@ static void __init efi_free_boot_service
>  		    md->type != EFI_BOOT_SERVICES_DATA)
>  			continue;
>  
> -		free_bootmem_late(start, size);
> +		free_bootmem_late_with_active_regions(start, size);
>  	}
>  }
>  
> Index: linux-2.6/include/linux/mm.h
> ===================================================================
> --- linux-2.6.orig/include/linux/mm.h
> +++ linux-2.6/include/linux/mm.h
> @@ -1322,6 +1322,8 @@ extern void get_pfn_range_for_nid(unsign
>  extern unsigned long find_min_pfn_with_active_regions(void);
>  extern void free_bootmem_with_active_regions(int nid,
>  						unsigned long max_low_pfn);
> +void free_bootmem_late_with_active_regions(unsigned long addr,
> +						unsigned long size);
>  int add_from_early_node_map(struct range *range, int az,
>  				   int nr_range, int nid);
>  u64 __init find_memory_core_early(int nid, u64 size, u64 align,
> Index: linux-2.6/mm/page_alloc.c
> ===================================================================
> --- linux-2.6.orig/mm/page_alloc.c
> +++ linux-2.6/mm/page_alloc.c
> @@ -3828,6 +3828,38 @@ void __init free_bootmem_with_active_reg
>  	}
>  }
>  
> +/**
> + * free_bootmem_late_with_active_regions - Call free_bootmem_late for each active range
> + * @addr: starting address of the range
> + * @size: size of the range in bytes
> + *
> + * this function make sure on active regions only
> + */
> +void __init free_bootmem_late_with_active_regions(unsigned long addr,
> +						unsigned long size)
> +{
> +	int i;
> +	int nid = MAX_NUMNODES;
> +	unsigned long start_pfn = PFN_UP(addr);
> +	unsigned long end_pfn = PFN_DOWN(addr + size);
> +
> +	if (start_pfn >= end_pfn)
> +		return;
> +
> +	for_each_active_range_index_in_nid(i, nid) {
> +		unsigned long common_start, common_end;
> +
> +		common_start = max(start_pfn, early_node_map[i].start_pfn);
> +		common_end = min(end_pfn, early_node_map[i].end_pfn);
> +
> +		if (common_start >= common_end)
> +			continue;
> +
> +		free_bootmem_late(common_start << PAGE_SHIFT,
> +				  (common_end - common_start) << PAGE_SHIFT);
> +	}
> +}
> +
>  #ifdef CONFIG_HAVE_MEMBLOCK
>  /*
>   * Basic iterator support. Return the last range of PFNs for a node
> 

I applied this on top of 3.0-rc1 (with offsets), kernels boots but see
now in dmesg output a new 'BUG: Bad page state in process swapper
pfn:01800' see attached dmesg output.

Jim




[-- Attachment #2: bug.txt --]
[-- Type: text/plain, Size: 1075 bytes --]

[    2.630101] BUG: Bad page state in process swapper  pfn:01800
[    2.631771] page:ffffea0000054000 count:0 mapcount:-127 mapping:          (null) index:0x0
[    2.633472] page flags: 0x4000000000000000()
[    2.635172] Pid: 1, comm: swapper Not tainted 3.0.0-rc1 #3
[    2.636856] Call Trace:
[    2.638515]  [<ffffffff8107e50d>] ? bad_page+0xd0/0xe2
[    2.640176]  [<ffffffff8107ea4c>] ? free_pages_prepare+0x67/0xa3
[    2.641846]  [<ffffffff8107db1e>] ? get_pageblock_flags_group+0x3f/0x76
[    2.643511]  [<ffffffff8107ebbf>] ? free_hot_cold_page+0x25/0x193
[    2.645177]  [<ffffffff81021ac9>] ? change_page_attr_set+0x23/0x28
[    2.646858]  [<ffffffff8101ee85>] ? free_init_pages+0xfb/0x110
[    2.648550]  [<ffffffff810001d8>] ? init_post+0xe/0xb6
[    2.650237]  [<ffffffff817b9bfc>] ? kernel_init+0x122/0x126
[    2.651926]  [<ffffffff81444564>] ? kernel_thread_helper+0x4/0x10
[    2.653602]  [<ffffffff817b9ada>] ? start_kernel+0x34f/0x34f
[    2.655268]  [<ffffffff81444560>] ? gs_change+0x13/0x13
[    2.656920] Disabling lock debugging due to kernel taint

[-- Attachment #3: dmesg-3.0.0-rc1.gz --]
[-- Type: application/gzip, Size: 14862 bytes --]