[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20190729054706.GB94090@jaegeuk-macbookpro.roam.corp.google.com>
Date: Sun, 28 Jul 2019 22:47:06 -0700
From: Jaegeuk Kim <jaegeuk@...nel.org>
To: Chao Yu <yuchao0@...wei.com>
Cc: Chao Yu <chao@...nel.org>, linux-kernel@...r.kernel.org,
linux-f2fs-devel@...ts.sourceforge.net
Subject: Re: [f2fs-dev] [PATCH] f2fs: fix to do sanity with enabled features
in image
On 07/23, Chao Yu wrote:
> On 2019/7/23 9:35, Jaegeuk Kim wrote:
> > On 07/16, Chao Yu wrote:
> >> Hi Jaegeuk,
> >>
> >> On 2019/5/9 9:15, Chao Yu wrote:
> >>> On 2019/5/5 10:51, Chao Yu wrote:
> >>>> On 2019/5/1 11:22, Jaegeuk Kim wrote:
> >>>>> On 04/29, Chao Yu wrote:
> >>>>>> On 2019-4-28 21:38, Jaegeuk Kim wrote:
> >>>>>>> On 04/24, Chao Yu wrote:
> >>>>>>>> This patch fixes to do sanity with enabled features in image, if
> >>>>>>>> there are features kernel can not recognize, just fail the mount.
> >>>>>>>
> >>>>>>> We need to figure out per-feature-based rejection, since some of them can
> >>>>>>> be set without layout change.
> >>
> >> What about adding one field in superblock for compatible features in future?
> >>
> >> sb.feature(F2FS_FEATURE_LAST, max] stores uncompatible features
> >> sb.compatible_feature stores compatible features
> >>
> >> If we follow above rule when adding one feature, then, we can fail the mount if
> >> sb.feature(F2FS_FEATURE_LAST, max] is valid.
> >
> > How about adding required_features flag in sb to check part of features only?
>
> You mean all incompatible features can be add into sb.required_features later
> like this?
>
> __le32 required_features; /* incompatible feature to old kernel */
>
> And we can check required_features with supported features in current kernel?
Yeah, I think so.
>
> if (le32_to_cpu(raw_super->required_features) &
> (~NOW_SUPPORTED_FEATURES_IN_CURRENT_KERNEL)) {
> print msg & ret error;
> }
>
> Thanks,
>
> >
> >>
> >> Thanks,
> >>
> >>>>>>
> >>>>>> So any suggestion on how to implement this?
> >>>>>
> >>>>> Which features do we need to disallow? When we introduce new features, they
> >>>>
> >>>> I guess it should be the new features.
> >>>>
> >>>>> didn't hurt the previous flow by checking f2fs_sb_has_###().
> >>>>
> >>>> Yes, but new features may use new disk layout, if old kernel handled it with old
> >>>> disk layout, there must be problematic.
> >>>>
> >>>> e.g. format image with -O extra_attr, and mount it with kernel who don't
> >>>> recognize new inode layout.
> >>>
> >>> Jaegeuk,
> >>>
> >>> Any thoughts?
> >>>
> >>> Thanks,
> >>>
> >>>>
> >>>> Thanks,
> >>>>
> >>>>>
> >>>>>>
> >>>>>> Maybe:
> >>>>>>
> >>>>>> if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 14, 0))
> >>>>>> check 4.14+ features
> >>>>>> else if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 9, 0))
> >>>>>> check 4.9+ features
> >>>>>> else if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 4, 0))
> >>>>>> check 4.4+ features
> >>>>>>
> >>>>>> Thanks,
> >>>>>>
> >>>>>>>
> >>>>>>>>
> >>>>>>>> Signed-off-by: Chao Yu <yuchao0@...wei.com>
> >>>>>>>> ---
> >>>>>>>> fs/f2fs/f2fs.h | 13 +++++++++++++
> >>>>>>>> fs/f2fs/super.c | 9 +++++++++
> >>>>>>>> 2 files changed, 22 insertions(+)
> >>>>>>>>
> >>>>>>>> diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
> >>>>>>>> index f5ffc09705eb..15b640967e12 100644
> >>>>>>>> --- a/fs/f2fs/f2fs.h
> >>>>>>>> +++ b/fs/f2fs/f2fs.h
> >>>>>>>> @@ -151,6 +151,19 @@ struct f2fs_mount_info {
> >>>>>>>> #define F2FS_FEATURE_VERITY 0x0400 /* reserved */
> >>>>>>>> #define F2FS_FEATURE_SB_CHKSUM 0x0800
> >>>>>>>>
> >>>>>>>> +#define F2FS_ALL_FEATURES (F2FS_FEATURE_ENCRYPT | \
> >>>>>>>> + F2FS_FEATURE_BLKZONED | \
> >>>>>>>> + F2FS_FEATURE_ATOMIC_WRITE | \
> >>>>>>>> + F2FS_FEATURE_EXTRA_ATTR | \
> >>>>>>>> + F2FS_FEATURE_PRJQUOTA | \
> >>>>>>>> + F2FS_FEATURE_INODE_CHKSUM | \
> >>>>>>>> + F2FS_FEATURE_FLEXIBLE_INLINE_XATTR | \
> >>>>>>>> + F2FS_FEATURE_QUOTA_INO | \
> >>>>>>>> + F2FS_FEATURE_INODE_CRTIME | \
> >>>>>>>> + F2FS_FEATURE_LOST_FOUND | \
> >>>>>>>> + F2FS_FEATURE_VERITY | \
> >>>>>>>> + F2FS_FEATURE_SB_CHKSUM)
> >>>>>>>> +
> >>>>>>>> #define __F2FS_HAS_FEATURE(raw_super, mask) \
> >>>>>>>> ((raw_super->feature & cpu_to_le32(mask)) != 0)
> >>>>>>>> #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask)
> >>>>>>>> diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c
> >>>>>>>> index 4f8e9ab48b26..57f2fc6d14ba 100644
> >>>>>>>> --- a/fs/f2fs/super.c
> >>>>>>>> +++ b/fs/f2fs/super.c
> >>>>>>>> @@ -2573,6 +2573,15 @@ static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
> >>>>>>>> return 1;
> >>>>>>>> }
> >>>>>>>>
> >>>>>>>> + /* check whether kernel supports all features */
> >>>>>>>> + if (le32_to_cpu(raw_super->feature) & (~F2FS_ALL_FEATURES)) {
> >>>>>>>> + f2fs_msg(sb, KERN_INFO,
> >>>>>>>> + "Unsupported feature:%u: supported:%u",
> >>>>>>>> + le32_to_cpu(raw_super->feature),
> >>>>>>>> + F2FS_ALL_FEATURES);
> >>>>>>>> + return 1;
> >>>>>>>> + }
> >>>>>>>> +
> >>>>>>>> /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
> >>>>>>>> if (sanity_check_area_boundary(sbi, bh))
> >>>>>>>> return 1;
> >>>>>>>> --
> >>>>>>>> 2.18.0.rc1
> >>>>> .
> >>>>>
> >>>>
> >>>>
> >>>> _______________________________________________
> >>>> Linux-f2fs-devel mailing list
> >>>> Linux-f2fs-devel@...ts.sourceforge.net
> >>>> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
> >>>> .
> >>>>
> >>>
> >>>
> >>> _______________________________________________
> >>> Linux-f2fs-devel mailing list
> >>> Linux-f2fs-devel@...ts.sourceforge.net
> >>> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
> >>> .
> >>>
> > .
> >
Powered by blists - more mailing lists