[PATCHv2 2/3] arm64: Add support for ARCH_SUPPORTS_DEBUG_PAGEALLOC

Laura Abbott labbott at redhat.com
Fri Feb 5 16:08:05 PST 2016


On 02/05/2016 06:20 AM, Mark Rutland wrote:
> On Thu, Feb 04, 2016 at 11:43:36AM -0800, Laura Abbott wrote:
>>
>> ARCH_SUPPORTS_DEBUG_PAGEALLOC provides a hook to map and unmap
>> pages for debugging purposes. This requires memory be mapped
>> with PAGE_SIZE mappings since breaking down larger mappings
>> at runtime will lead to TLB conflicts. Check if debug_pagealloc
>> is enabled at runtime and if so, map everyting with PAGE_SIZE
>> pages. Implement the functions to actually map/unmap the
>> pages at runtime.
>>
>> Signed-off-by: Laura Abbott <labbott at fedoraproject.org>
>
> I've given this a spin on Juno, with and without the config option
> selected, and with and without the command line option. I've also given
> it a spin on Seattle with inline KASAN also enabled.
>
> I wasn't sure how to deliberately trigger a failure, but those all
> booted fine, and the dumepd page tables looks right, so FWIW:
>

I wrote a test that does a write after free on an allocated page
for my testing. Might be worth it to look into adding a test to
the lkdtm module.

I also did testing with 64K pages but couldn't test on 16K due
to lack of hardware and QEMU running off into the weeds.
  
> Tested-by: Mark Rutland <mark.rutland at arm.com>
>
> I have a few minor comments below, and with those fixed up:
>
> Reviewed-by: Mark Rutland <mark.rutland at arm.com>
>
>> ---
>>   arch/arm64/Kconfig       |  3 +++
>>   arch/arm64/mm/mmu.c      | 25 +++++++++++++++++++++----
>>   arch/arm64/mm/pageattr.c | 46 ++++++++++++++++++++++++++++++++++++----------
>>   3 files changed, 60 insertions(+), 14 deletions(-)
>>
>> diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
>> index 8cc6228..0f33218 100644
>> --- a/arch/arm64/Kconfig
>> +++ b/arch/arm64/Kconfig
>> @@ -537,6 +537,9 @@ config HOTPLUG_CPU
>>   source kernel/Kconfig.preempt
>>   source kernel/Kconfig.hz
>>
>> +config ARCH_SUPPORTS_DEBUG_PAGEALLOC
>> +	def_bool y
>> +
>>   config ARCH_HAS_HOLES_MEMORYMODEL
>>   	def_bool y if SPARSEMEM
>>
>> diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c
>> index ef0d66c..be81a59 100644
>> --- a/arch/arm64/mm/mmu.c
>> +++ b/arch/arm64/mm/mmu.c
>> @@ -180,8 +180,14 @@ static void alloc_init_pmd(pud_t *pud, unsigned long addr, unsigned long end,
>>   	pmd = pmd_set_fixmap_offset(pud, addr);
>>   	do {
>>   		next = pmd_addr_end(addr, end);
>> -		/* try section mapping first */
>> -		if (((addr | next | phys) & ~SECTION_MASK) == 0) {
>> +		/*
>> +		 * try a section mapping first
>> +		 *
>> +		 * See comment in use_1G_block for why we need the check
>> +		 * for !pgtable_alloc with !debug_pagealloc
>> +		 */
>> +		if (((addr | next | phys) & ~SECTION_MASK) == 0 &&
>> +		      (!debug_pagealloc_enabled() || !pgtable_alloc)) {
>>   			pmd_t old_pmd =*pmd;
>>   			set_pmd(pmd, __pmd(phys |
>>   					   pgprot_val(mk_sect_prot(prot))));
>> @@ -208,8 +214,19 @@ static void alloc_init_pmd(pud_t *pud, unsigned long addr, unsigned long end,
>>   }
>>
>>   static inline bool use_1G_block(unsigned long addr, unsigned long next,
>> -			unsigned long phys)
>> +			unsigned long phys, phys_addr_t (*pgtable_alloc)(void))
>>   {
>> +	/*
>> +	 * If debug_page_alloc is enabled we don't want to be using sections
>> +	 * since everything needs to be mapped with pages. The catch is
>> +	 * that we only want to force pages if we can allocate the next
>> +	 * layer of page tables. If there is no pgtable_alloc function,
>> +	 * it's too early to allocate another layer and we should use
>> +	 * section mappings.
>> +	 */
>
> I'm not sure this quite captures the rationale, as we only care about
> the linear map using pages (AFAIK), and the earliness only matters
> w.r.t. the DTB mapping. How about:
>
> 	/*
> 	 * If debug_page_alloc is enabled we must map the linear map
> 	 * using pages. However, other mappings created by
> 	 * create_mapping_noalloc must use sections in some cases. Allow
> 	 * sections to be used in those cases, where no pgtable_alloc
> 	 * function is provided.
> 	 */
>
> Does that sound ok to you?
>
> As a future optimisation, I think we can allow sections when mapping
> permanent kernel chunks (.e.g .rodata and .text), as these shouldn't
> contain pages available for dynamic allocation. That would require using
> something other than the presence of pgtable_alloc to determine when we
> should force page usage.
>
>> +	if (pgtable_alloc && debug_pagealloc_enabled())
>> +		return false;
>> +
>>   	if (PAGE_SHIFT != 12)
>>   		return false;
>>
>> @@ -241,7 +258,7 @@ static void alloc_init_pud(pgd_t *pgd, unsigned long addr, unsigned long end,
>>   		/*
>>   		 * For 4K granule only, attempt to put down a 1GB block
>>   		 */
>> -		if (use_1G_block(addr, next, phys)) {
>> +		if (use_1G_block(addr, next, phys, pgtable_alloc)) {
>>   			pud_t old_pud = *pud;
>>   			set_pud(pud, __pud(phys |
>>   					   pgprot_val(mk_sect_prot(prot))));
>> diff --git a/arch/arm64/mm/pageattr.c b/arch/arm64/mm/pageattr.c
>> index 1360a02..57877af 100644
>> --- a/arch/arm64/mm/pageattr.c
>> +++ b/arch/arm64/mm/pageattr.c
>> @@ -37,14 +37,31 @@ static int change_page_range(pte_t *ptep, pgtable_t token, unsigned long addr,
>>   	return 0;
>>   }
>>
>> +/*
>> + * This function assumes that the range is mapped with PAGE_SIZE pages.
>> + */
>> +static int __change_memory_common(unsigned long start, unsigned long size,
>> +				pgprot_t set_mask, pgprot_t clear_mask)
>> +{
>> +	struct page_change_data data;
>> +	int ret;
>> +
>> +	data.set_mask = set_mask;
>> +	data.clear_mask = clear_mask;
>> +
>> +	ret = apply_to_page_range(&init_mm, start, size, change_page_range,
>> +					&data);
>> +
>> +	flush_tlb_kernel_range(start, start+size);
>
> Nit: spaces around '+' please.
>
>> +	return ret;
>> +}
>> +
>>   static int change_memory_common(unsigned long addr, int numpages,
>>   				pgprot_t set_mask, pgprot_t clear_mask)
>>   {
>>   	unsigned long start = addr;
>>   	unsigned long size = PAGE_SIZE*numpages;
>>   	unsigned long end = start + size;
>> -	int ret;
>> -	struct page_change_data data;
>>   	struct vm_struct *area;
>>
>>   	if (!PAGE_ALIGNED(addr)) {
>> @@ -72,14 +89,7 @@ static int change_memory_common(unsigned long addr, int numpages,
>>   	    !(area->flags & VM_ALLOC))
>>   		return -EINVAL;
>>
>> -	data.set_mask = set_mask;
>> -	data.clear_mask = clear_mask;
>> -
>> -	ret = apply_to_page_range(&init_mm, start, size, change_page_range,
>> -					&data);
>> -
>> -	flush_tlb_kernel_range(start, end);
>> -	return ret;
>> +	return __change_memory_common(start, size, set_mask, clear_mask);
>>   }
>>
>>   int set_memory_ro(unsigned long addr, int numpages)
>> @@ -111,3 +121,19 @@ int set_memory_x(unsigned long addr, int numpages)
>>   					__pgprot(PTE_PXN));
>>   }
>>   EXPORT_SYMBOL_GPL(set_memory_x);
>> +
>> +#ifdef CONFIG_DEBUG_PAGEALLOC
>> +void __kernel_map_pages(struct page *page, int numpages, int enable)
>> +{
>> +	unsigned long addr = (unsigned long) page_address(page);
>> +
>> +	if (enable)
>> +		__change_memory_common(addr, PAGE_SIZE*numpages,
>> +					__pgprot(PTE_VALID),
>> +					__pgprot(0));
>> +	else
>> +		__change_memory_common(addr, PAGE_SIZE*numpages,
>> +					__pgprot(0),
>> +					__pgprot(PTE_VALID));
>
> Nit: spaces around each '*' here, please.
>
> Thanks,
> Mark.
>
>> +}
>> +#endif
>> --
>> 2.5.0
>>




More information about the linux-arm-kernel mailing list