Date:	Wed, 23 Mar 2016 13:40:19 +0900 (KST)
From:	SeongJae Park <sj38.park@...il.com>
To:	Vaishali Thakkar <vaishali.thakkar@...cle.com>
cc:	SeongJae Park <sj38.park@...il.com>, akpm@...ux-foundation.org,
	linux-kernel@...r.kernel.org, linux-mm@...ck.org,
	Mike Kravetz <mike.kravetz@...cle.com>,
	Naoya Horiguchi <n-horiguchi@...jp.nec.com>,
	Hillf Danton <hillf.zj@...baba-inc.com>,
	Michal Hocko <mhocko@...e.com>,
	Yaowei Bai <baiyaowei@...s.chinamobile.com>,
	Dominik Dingel <dingel@...ux.vnet.ibm.com>,
	"Kirill A. Shutemov" <kirill.shutemov@...ux.intel.com>,
	Paul Gortmaker <paul.gortmaker@...driver.com>,
	Dave Hansen <dave.hansen@...ux.intel.com>
Subject: Re: [PATCH 1/2] mm/hugetlb: Introduce hugetlb_bad_size



On Wed, 23 Mar 2016, Vaishali Thakkar wrote:

>
>
> On Wednesday 23 March 2016 04:57 AM, SeongJae Park wrote:
>> Hello Vaishali,
>>
>>
>> The patch looks good to me.  However, I have few trivial questions.
>>
>> On Tue, 22 Mar 2016, Vaishali Thakkar wrote:
>>
>>> When any unsupported hugepage size is specified, 'hugepagesz=' and
>>> 'hugepages=' should be ignored during command line parsing until any
>>> supported hugepage size is found. But currently incorrect number of
>>> hugepages are allocated when unsupported size is specified as it fails
>>> to ignore the 'hugepages=' command.
>>>
>>> Test case:
>>>
>>> Note that this is specific to x86 architecture.
>>>
>>> Boot the kernel with command line option 'hugepagesz=256M hugepages=X'.
>>> After boot, dmesg output shows that X number of hugepages of the size 2M
>>> is pre-allocated instead of 0.
>>>
>>> So, to handle such command line options, introduce new routine
>>> hugetlb_bad_size. The routine hugetlb_bad_size sets the global variable
>>> parsed_valid_hugepagesz. We are using parsed_valid_hugepagesz to save the
>>> state when unsupported hugepagesize is found so that we can ignore the
>>> 'hugepages=' parameters after that and then reset the variable when
>>> supported hugepage size is found.
>>>
>>> The routine hugetlb_bad_size can be called while setting 'hugepagesz='
>>> parameter in an architecture specific code.
>>>
>>> Signed-off-by: Vaishali Thakkar <vaishali.thakkar@...cle.com>
>>> Cc: Mike Kravetz <mike.kravetz@...cle.com>
>>> Cc: Naoya Horiguchi <n-horiguchi@...jp.nec.com>
>>> Cc: Hillf Danton <hillf.zj@...baba-inc.com>
>>> Cc: Michal Hocko <mhocko@...e.com>
>>> Cc: Yaowei Bai <baiyaowei@...s.chinamobile.com>
>>> Cc: Dominik Dingel <dingel@...ux.vnet.ibm.com>
>>> Cc: Kirill A. Shutemov <kirill.shutemov@...ux.intel.com>
>>> Cc: Paul Gortmaker <paul.gortmaker@...driver.com>
>>> Cc: Dave Hansen <dave.hansen@...ux.intel.com>
>>> ---
>>> The patch is having 2 checkpatch.pl warnings. I have just followed
>>> the current code to maintain consistency. If we decide to silent
>>> these warnings then may be we should silent those warnings as well.
>>> I am fine with any option whichever works best for everyone else.
>>> ---
>>> include/linux/hugetlb.h |  1 +
>>> mm/hugetlb.c            | 14 +++++++++++++-
>>> 2 files changed, 14 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
>>> index 7d953c2..e44c578 100644
>>> --- a/include/linux/hugetlb.h
>>> +++ b/include/linux/hugetlb.h
>>> @@ -338,6 +338,7 @@ int huge_add_to_page_cache(struct page *page, struct address_space *mapping,
>>> /* arch callback */
>>> int __init alloc_bootmem_huge_page(struct hstate *h);
>>>
>>> +void __init hugetlb_bad_size(void);
>>> void __init hugetlb_add_hstate(unsigned order);
>>> struct hstate *size_to_hstate(unsigned long size);
>>>
>>> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
>>> index 06058ea..44fae6a 100644
>>> --- a/mm/hugetlb.c
>>> +++ b/mm/hugetlb.c
>>> @@ -51,6 +51,7 @@ __initdata LIST_HEAD(huge_boot_pages);
>>> static struct hstate * __initdata parsed_hstate;
>>> static unsigned long __initdata default_hstate_max_huge_pages;
>>> static unsigned long __initdata default_hstate_size;
>>> +static bool __initdata parsed_valid_hugepagesz = true;
>>>
>>> /*
>>>  * Protects updates to hugepage_freelists, hugepage_activelist, nr_huge_pages,
>>> @@ -2659,6 +2660,11 @@ static int __init hugetlb_init(void)
>>> subsys_initcall(hugetlb_init);
>>>
>>> /* Should be called on processing a hugepagesz=... option */
>>> +void __init hugetlb_bad_size(void)
>>> +{
>>> +    parsed_valid_hugepagesz = false;
>>> +}
>>> +
>>> void __init hugetlb_add_hstate(unsigned int order)
>>> {
>>>     struct hstate *h;
>>> @@ -2691,11 +2697,17 @@ static int __init hugetlb_nrpages_setup(char *s)
>>>     unsigned long *mhp;
>>>     static unsigned long *last_mhp;
>>>
>>> +    if (!parsed_valid_hugepagesz) {
>>> +        pr_warn("hugepages = %s preceded by "
>>> +            "an unsupported hugepagesz, ignoring\n", s);
>>
>> How about concatenating the format string?  `CodingStyle` now suggests to
>> _never_ break every user-visible strings.
>>
>
> As I said above, I just followed the pattern of the current code to maintain the
> consistency. Probably a separate change would be good for solving all those
> warnings. :)

Understood and agreed. :)


>
>>> +        parsed_valid_hugepagesz = true;
>>> +        return 1;
>>> +    }
>>>     /*
>>>      * !hugetlb_max_hstate means we haven't parsed a hugepagesz= parameter yet,
>>>      * so this hugepages= parameter goes to the "default hstate".
>>>      */
>>> -    if (!hugetlb_max_hstate)
>>> +    else if (!hugetlb_max_hstate)
>>
>> Because the upper `if` statement will do `return`, above change looks not
>> significantly necessary.  Is this intended?
>>
>
> I think above change is necessary for the cases like "hugepages=X" because in that
> case the X hugepages of the default size [like 2M for x86] should be allocated.

Looks like my poor English made some confusion, sorry.  I was just saying
about the addition of `else` in the line, not whole change.

Because the upper `if` statement that catching wrong `hugepagesz=` case
does `return`, below statements will not be executed at all in the case.
So, the result will not be changed even if the `else` is not added though
the addition of `else` may help readability for someone.
That's why I said the change (addition of `else`) looks not significantly
necessary.

If I am missing something wrong, please let me know.  Or, if this question
bothers you, just ignore it because I also know that this is just a
trivial question. ;)


Thanks,
SeongJae Park

>
>>>         mhp = &default_hstate_max_huge_pages;
>>>     else
>>>         mhp = &parsed_hstate->max_huge_pages;
>>> --
>>> 2.1.4
>>>
>>> --
>>> To unsubscribe, send a message with 'unsubscribe linux-mm' in
>>> the body to majordomo@...ck.org.  For more info on Linux MM,
>>> see: http://www.linux-mm.org/ .
>>> Don't email: <a href=mailto:"dont@...ck.org"> email@...ck.org </a>
>>>
>
> -- 
> Vaishali
>
>

Powered by blists - more mailing lists