Re: [PATCH 04/13] x86, mm: Revert back good_end setting for 64bit

[Stefano Stabellini]

On Fri, 5 Oct 2012, Yinghai Lu wrote:
> On Thu, Oct 4, 2012 at 2:54 PM, H. Peter Anvin <hpa@xxxxxxxxx> wrote:
> >
> > See my other post.  This is bringing up the Kernel Summit algorithm again.
> >
> 
> sure. please check if you are ok with attached one on top of x86/mm2
> 
> Subject: [PATCH] x86: get early page table from BRK
> 
> set pgt_buf early from BRK, and use it to map page table at first.
> 
> also use the left at first, then use new extend one.
> 
> Signed-off-by: Yinghai Lu <yinghai@xxxxxxxxxx>

If I read the patch correctly, it wouldn't actually change the pagetable
allocation flow or implement Peter's suggestion.
However it would pre-map (pgt_buf_start-pgt_buf_top) using brk memory.

So, if that's correct, we could remove the early_memremap call from
alloc_low_page and map_low_page, right?

Also the patch introduces an additional range of pagetable pages
(early_pgt_buf_start-early_pgt_buf_top) and that would need to be
communicated somehow to Xen (that at the moment assumes that the range
is pgt_buf_start-pgt_buf_top. Pretty bad).

Overall I still prefer Peter's suggestion :)


> ---
>  arch/x86/include/asm/init.h    |    4 ++++
>  arch/x86/include/asm/pgtable.h |    1 +
>  arch/x86/kernel/setup.c        |    2 ++
>  arch/x86/mm/init.c             |   23 +++++++++++++++++++++++
>  arch/x86/mm/init_32.c          |    8 ++++++--
>  arch/x86/mm/init_64.c          |    8 ++++++--
>  6 files changed, 42 insertions(+), 4 deletions(-)
> 
> Index: linux-2.6/arch/x86/include/asm/pgtable.h
> ===================================================================
> --- linux-2.6.orig/arch/x86/include/asm/pgtable.h
> +++ linux-2.6/arch/x86/include/asm/pgtable.h
> @@ -599,6 +599,7 @@ static inline int pgd_none(pgd_t pgd)
>  
>  extern int direct_gbpages;
>  void init_mem_mapping(void);
> +void early_alloc_pgt_buf(void);
>  
>  /* local pte updates need not use xchg for locking */
>  static inline pte_t native_local_ptep_get_and_clear(pte_t *ptep)
> Index: linux-2.6/arch/x86/kernel/setup.c
> ===================================================================
> --- linux-2.6.orig/arch/x86/kernel/setup.c
> +++ linux-2.6/arch/x86/kernel/setup.c
> @@ -950,6 +950,8 @@ void __init setup_arch(char **cmdline_p)
>  
>  	reserve_ibft_region();
>  
> +	early_alloc_pgt_buf();
> +
>  	/*
>  	 * Need to conclude brk, before memblock_x86_fill()
>  	 *  it could use memblock_find_in_range, could overlap with
> Index: linux-2.6/arch/x86/mm/init.c
> ===================================================================
> --- linux-2.6.orig/arch/x86/mm/init.c
> +++ linux-2.6/arch/x86/mm/init.c
> @@ -21,6 +21,10 @@ unsigned long __initdata pgt_buf_start;
>  unsigned long __meminitdata pgt_buf_end;
>  unsigned long __meminitdata pgt_buf_top;
>  
> +unsigned long __initdata early_pgt_buf_start;
> +unsigned long __meminitdata early_pgt_buf_end;
> +unsigned long __meminitdata early_pgt_buf_top;
> +
>  int after_bootmem;
>  
>  int direct_gbpages
> @@ -291,6 +295,11 @@ static void __init find_early_table_spac
>  	if (!base)
>  		panic("Cannot find space for the kernel page tables");
>  
> +	init_memory_mapping(base, base + tables);
> +	printk(KERN_DEBUG "kernel direct mapping tables from %#llx to %#llx @ [mem %#010lx-%#010lx]\n",
> +		base, base + tables - 1, early_pgt_buf_start << PAGE_SHIFT,
> +		(early_pgt_buf_end << PAGE_SHIFT) - 1);
> +
>  	pgt_buf_start = base >> PAGE_SHIFT;
>  	pgt_buf_end = pgt_buf_start;
>  	pgt_buf_top = pgt_buf_start + (tables >> PAGE_SHIFT);
> @@ -437,6 +446,20 @@ void __init init_mem_mapping(void)
>  	early_memtest(0, max_pfn_mapped << PAGE_SHIFT);
>  }
>  
> +RESERVE_BRK(early_pgt_alloc, 16384);
> +
> +void  __init early_alloc_pgt_buf(void)
> +{
> +	unsigned long tables = 13864;
> +	phys_addr_t base;
> +
> +	base = __pa(extend_brk(tables, PAGE_SIZE));
> +
> +	early_pgt_buf_start = base >> PAGE_SHIFT;
> +	early_pgt_buf_end = early_pgt_buf_start;
> +	early_pgt_buf_top = early_pgt_buf_start + (tables >> PAGE_SHIFT);
> +}
> +
>  /*
>   * devmem_is_allowed() checks to see if /dev/mem access to a certain address
>   * is valid. The argument is a physical page number.
> Index: linux-2.6/arch/x86/mm/init_32.c
> ===================================================================
> --- linux-2.6.orig/arch/x86/mm/init_32.c
> +++ linux-2.6/arch/x86/mm/init_32.c
> @@ -61,10 +61,14 @@ bool __read_mostly __vmalloc_start_set =
>  
>  static __init void *alloc_low_page(void)
>  {
> -	unsigned long pfn = pgt_buf_end++;
> +	unsigned long pfn;
>  	void *adr;
>  
> -	if (pfn >= pgt_buf_top)
> +	if (early_pgt_buf_end < early_pgt_buf_top)
> +		pfn = early_pgt_buf_end++;
> +	else if (pgt_buf_end < pgt_buf_top)
> +		pfn = pgt_buf_end++;
> +	else
>  		panic("alloc_low_page: ran out of memory");
>  
>  	adr = __va(pfn * PAGE_SIZE);
> Index: linux-2.6/arch/x86/mm/init_64.c
> ===================================================================
> --- linux-2.6.orig/arch/x86/mm/init_64.c
> +++ linux-2.6/arch/x86/mm/init_64.c
> @@ -318,7 +318,7 @@ void __init cleanup_highmap(void)
>  
>  static __ref void *alloc_low_page(unsigned long *phys)
>  {
> -	unsigned long pfn = pgt_buf_end++;
> +	unsigned long pfn;
>  	void *adr;
>  
>  	if (after_bootmem) {
> @@ -328,7 +328,11 @@ static __ref void *alloc_low_page(unsign
>  		return adr;
>  	}
>  
> -	if (pfn >= pgt_buf_top)
> +	if (early_pgt_buf_end < early_pgt_buf_top)
> +		pfn = early_pgt_buf_end++;
> +	else if (pgt_buf_end < pgt_buf_top)
> +		pfn = pgt_buf_end++;
> +	else
>  		panic("alloc_low_page: ran out of memory");
>  
>  	adr = early_memremap(pfn * PAGE_SIZE, PAGE_SIZE);
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/