On Fri, 5 Oct 2012, Yinghai Lu wrote: > On Thu, Oct 4, 2012 at 2:54 PM, H. Peter Anvin <h...@zytor.com> 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 <ying...@kernel.org>
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 majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/