Re: [PATCH] staging/lustre/llite: Use memdup_user() rather than duplicating its implementation

From: Vaishali Thakkar
Date: Sun Aug 21 2016 - 06:55:49 EST




On Sunday 21 August 2016 04:01 PM, Julia Lawall wrote:
>
>
> On Sun, 21 Aug 2016, Christophe JAILLET wrote:
>
>> Le 21/08/2016 à 11:45, SF Markus Elfring a écrit :
>>> From: Markus Elfring <elfring@xxxxxxxxxxxxxxxxxxxxx>
>>> Date: Sun, 21 Aug 2016 11:30:57 +0200
>>>
>>> Reuse existing functionality from memdup_user() instead of keeping
>>> duplicate source code.
>>>
>>> This issue was detected by using the Coccinelle software.
>>>
>>> Signed-off-by: Markus Elfring <elfring@xxxxxxxxxxxxxxxxxxxxx>
>>> ---
>>> drivers/staging/lustre/lustre/llite/dir.c | 12 +++---------
>>> 1 file changed, 3 insertions(+), 9 deletions(-)
>>>
>>> diff --git a/drivers/staging/lustre/lustre/llite/dir.c
>>> b/drivers/staging/lustre/lustre/llite/dir.c
>>> index 031c9e4..8b70e42 100644
>>> --- a/drivers/staging/lustre/lustre/llite/dir.c
>>> +++ b/drivers/staging/lustre/lustre/llite/dir.c
>>> @@ -1676,14 +1676,9 @@ out_poll:
>>> case LL_IOC_QUOTACTL: {
>>> struct if_quotactl *qctl;
>>> - qctl = kzalloc(sizeof(*qctl), GFP_NOFS);
>> Same as previously reported in another patch, GFP_NOFS has not the same
>> meaning than GPF_KERNEL.
>> So your proposed clean-up is not 100% equivalent.
>>
>> Are your sure that GPF_KERNEL instead of GFP_NOFS is right in this code?
>>
>> Maybe, the coccinelle check should be tweak to only spot "kzalloc(...,
>> GFP_KERNEL)" allocation?
>
> To my dim recollection, GFP_NOFS is not actually allowed in a place where
> copy_from_user is being used. copy_from_user can block due to page
> faults, and GFP_NOFS is used when a certain kind of blocking is not
> allowed. So if the code really needs GFP_NOFS, then something else is
> wrong.
>
> The semantic patch intentionally does not specify GFP_KERNEL for this
> reason, ie so that these issues will come up and be discussed. On the
> ther hand I agree about the GFP_DMA case, since that doesn't relate to
> blocking, as far as I know. The semantic patch should be updated to not
> make/propose the change in that case.

I think semantic patch should be updated for all possible flags except
GFP_NOFS and GFP_ATOMIC. Because only using these 2 flags with
copy_from_user can cause blocking.

> julia
>
>>
>>> - if (!qctl)
>>> - return -ENOMEM;
>>> -
>>> - if (copy_from_user(qctl, (void __user *)arg, sizeof(*qctl))) {
>>> - rc = -EFAULT;
>>> - goto out_quotactl;
>>> - }
>>> + qctl = memdup_user((void __user *)arg, sizeof(*qctl));
>>> + if (IS_ERR(qctl))
>>> + return PTR_ERR(qctl);
>>> rc = quotactl_ioctl(sbi, qctl);
>>> @@ -1691,7 +1686,6 @@ out_poll:
>>> sizeof(*qctl)))
>>> rc = -EFAULT;
>>> -out_quotactl:
>>> kfree(qctl);
>>> return rc;
>>> }
>>
>>
>>
>> ---
>> L'absence de virus dans ce courrier électronique a été vérifiée par le
>> logiciel antivirus Avast.
>> https://www.avast.com/antivirus
>>
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe kernel-janitors" in
>> the body of a message to majordomo@xxxxxxxxxxxxxxx
>> More majordomo info at http://vger.kernel.org/majordomo-info.html
>>
>

--
Vaishali