[linux-yocto] [PATCH 12/57] arch/arm/mach-axxia: Updates to the VMFS File System

[Zumeng Chen]

On 2014年03月19日 05:43, Jacques, John M (John) wrote:
> Build failure.
>
> You guys are going to teach me to write comments yet!

Hi Charlie,

I think you might mis-understand Bruce for his comments,
in most situation, you had better go through your patch again
because we can get any useful information from your original
commit log.

Actually, build failure is not good for comments, and I think
there are many stuffs were missed for vmfs-self as well, like
Kconfig, and especially docs in Documentation/fs.

As for this patch, I think we can use __kuid_val/__kgid_val to
finish what your patch wants to do if I'm right here, cause it's
Linux style to avoid too much #ifdef in codes

Cheers,
Zumeng

>
>> -----Original Message-----
>> From: Bruce Ashfield [mailto:bruce.ashfield at windriver.com]
>> Sent: Tuesday, March 18, 2014 4:32 PM
>> To: Charlie Paul; linux-yocto at yoctoproject.org
>> Cc: charlie.paul at windriver.com; Jacques, John M (John)
>> Subject: Re: [PATCH 12/57] arch/arm/mach-axxia: Updates to the VMFS File
>> System
>>
>> Minor 'nit as well. What was the symptom when it was broken ? A build
>> failure ? runtime issue ? We should capture that in the commit message.
>>
>> Bruce
>>
>> On 14-03-18 12:56 AM, Charlie Paul wrote:
>>> From: John Jacques <john.jacques at lsi.com>
>>>
>>> Updated to work when CONFIG_UIDGID_STRICT_TYPE_CHECKS is not
>> defined.
>>> Signed-off-by: John Jacques <john.jacques at lsi.com>
>>> ---
>>>    fs/vmfs/inode.c |   18 ++++++++++++++++--
>>>    fs/vmfs/ioctl.c |    8 ++++++++
>>>    2 files changed, 24 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/fs/vmfs/inode.c b/fs/vmfs/inode.c index 713da0d..099bd76
>>> 100644
>>> --- a/fs/vmfs/inode.c
>>> +++ b/fs/vmfs/inode.c
>>> @@ -486,11 +486,25 @@ int vmfs_notify_change(struct dentry *dentry,
>> struct iattr *attr)
>>>    		goto out;
>>>
>>>    	error = -EPERM;
>>> -	if ((attr->ia_valid & ATTR_UID) && ((attr->ia_uid).val != (server-
>>> mnt->uid).val))
>>> +#ifdef CONFIG_UIDGID_STRICT_TYPE_CHECKS
>>> +	if ((attr->ia_valid & ATTR_UID) &&
>>> +	    (attr->ia_uid.val != (server->mnt->uid).val))
>>>    		goto out;
>>> +#else
>>> +	if ((attr->ia_valid & ATTR_UID) &&
>>> +	    (attr->ia_uid != server->mnt->uid))
>>> +		goto out;
>>> +#endif
>>>
>>> -	if ((attr->ia_valid & ATTR_GID) && ((attr->ia_gid).val != (server-
>>> mnt->gid).val))
>>> +#ifdef CONFIG_UIDGID_STRICT_TYPE_CHECKS
>>> +	if ((attr->ia_valid & ATTR_GID) &&
>>> +	    ((attr->ia_gid).val != (server->mnt->gid).val))
>>>    		goto out;
>>> +#else
>>> +	if ((attr->ia_valid & ATTR_GID) &&
>>> +	    (attr->ia_gid != server->mnt->gid))
>>> +		goto out;
>>> +#endif
>>>
>>>    	if ((attr->ia_valid & ATTR_MODE) && (attr->ia_mode & ~mask))
>>>    		goto out;
>>> diff --git a/fs/vmfs/ioctl.c b/fs/vmfs/ioctl.c index 988e69c..798df63
>>> 100644
>>> --- a/fs/vmfs/ioctl.c
>>> +++ b/fs/vmfs/ioctl.c
>>> @@ -34,11 +34,19 @@ long vmfs_unlocked_ioctl(struct file *filp, unsigned
>> int cmd,
>>>    		uid16_t uid16;
>>>    		uid_t uid32;
>>>    	case VMFS_IOC_GETMOUNTUID:
>>> +#ifdef CONFIG_UIDGID_STRICT_TYPE_CHECKS
>>>    		SET_UID(uid16, (server->mnt->mounted_uid).val);
>>> +#else
>>> +		SET_UID(uid16, server->mnt->mounted_uid); #endif
>>>    		result = put_user(uid16, (uid16_t __user *) arg);
>>>    		break;
>>>    	case VMFS_IOC_GETMOUNTUID32:
>>> +#ifdef CONFIG_UIDGID_STRICT_TYPE_CHECKS
>>>    		SET_UID(uid32, (server->mnt->mounted_uid).val);
>>> +#else
>>> +		SET_UID(uid32, server->mnt->mounted_uid); #endif
>>>    		result = put_user(uid32, (uid_t __user *) arg);
>>>    		break;
>>>    	default:
>>>