projid32bit=1 default in xfsprogs-3.2.0

projid32bit=1 default in xfsprogs-3.2.0

[Eric Sandeen]

The commit:

f3edb66a mkfs.xfs: validate options for CRCs up front.

unintentionally (?) defaulted to projid32bit=1 for all filesystems
at mkfs time.

Support for 32 bit project IDs has been present since kernel v2.6.37.

Is everybody ok with making this change for 3.2.0?  Speak now or
forever hold your peace, as they say.

-Eric

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: projid32bit=1 default in xfsprogs-3.2.0

[Dave Chinner]

On Mon, Oct 14, 2013 at 11:30:13AM -0500, Eric Sandeen wrote:
> The commit:
> 
> f3edb66a mkfs.xfs: validate options for CRCs up front.
> 
> unintentionally (?) defaulted to projid32bit=1 for all filesystems
> at mkfs time.

It wasn't unintentional, I just forgot to document it.

> Support for 32 bit project IDs has been present since kernel v2.6.37.

Right, so taking the usual approach to these things, all the
distro's now ship kernels that support 32 bit project IDs, so there
is no reason for mkfs.xfs in a new xfsprogs release not to set it
by default.

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: projid32bit=1 default in xfsprogs-3.2.0

[Ben Myers]

On Tue, Oct 15, 2013 at 08:16:59AM +1100, Dave Chinner wrote:
> On Mon, Oct 14, 2013 at 11:30:13AM -0500, Eric Sandeen wrote:
> > The commit:
> > 
> > f3edb66a mkfs.xfs: validate options for CRCs up front.
> > 
> > unintentionally (?) defaulted to projid32bit=1 for all filesystems
> > at mkfs time.
> 
> It wasn't unintentional, I just forgot to document it.
> 
> > Support for 32 bit project IDs has been present since kernel v2.6.37.
> 
> Right, so taking the usual approach to these things, all the
> distro's now ship kernels that support 32 bit project IDs, so there
> is no reason for mkfs.xfs in a new xfsprogs release not to set it
> by default.

I'm not seeing much downside to turning on 32 bit project IDs by default in
3.2.0.  Sounds good to me.

-Ben

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: projid32bit=1 default in xfsprogs-3.2.0

[Eric Sandeen]

On 10/14/13 4:23 PM, Ben Myers wrote:
> On Tue, Oct 15, 2013 at 08:16:59AM +1100, Dave Chinner wrote:
>> On Mon, Oct 14, 2013 at 11:30:13AM -0500, Eric Sandeen wrote:
>>> The commit:
>>>
>>> f3edb66a mkfs.xfs: validate options for CRCs up front.
>>>
>>> unintentionally (?) defaulted to projid32bit=1 for all filesystems
>>> at mkfs time.
>>
>> It wasn't unintentional, I just forgot to document it.
>>
>>> Support for 32 bit project IDs has been present since kernel v2.6.37.
>>
>> Right, so taking the usual approach to these things, all the
>> distro's now ship kernels that support 32 bit project IDs, so there
>> is no reason for mkfs.xfs in a new xfsprogs release not to set it
>> by default.
> 
> I'm not seeing much downside to turning on 32 bit project IDs by default in
> 3.2.0.  Sounds good to me.
> 
> -Ben
> 

Ok.  Default it is!   Thanks guys, just wanted to make sure we were
all in agreement there.

-Eric

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: projid32bit=1 default in xfsprogs-3.2.0

[Christoph Hellwig]

On Mon, Oct 14, 2013 at 04:24:24PM -0500, Eric Sandeen wrote:
> Ok.  Default it is!   Thanks guys, just wanted to make sure we were
> all in agreement there.

Can we make sure this gets documented in doc/CHANGES?

I'm also pretty sure the entry for 3.2.0-alpha1 is severly lacking,
can we make sure the one for the real release is as detailed as the ones
for the previous releases?

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: projid32bit=1 default in xfsprogs-3.2.0

[Christoph Hellwig]

On Tue, Oct 15, 2013 at 12:46:09PM -0700, Christoph Hellwig wrote:
> On Mon, Oct 14, 2013 at 04:24:24PM -0500, Eric Sandeen wrote:
> > Ok.  Default it is!   Thanks guys, just wanted to make sure we were
> > all in agreement there.
> 
> Can we make sure this gets documented in doc/CHANGES?
> 
> I'm also pretty sure the entry for 3.2.0-alpha1 is severly lacking,
> can we make sure the one for the real release is as detailed as the ones
> for the previous releases?

Btw, I also noticed releases after 3.1.9 aren't tagged in git, any
chance we could fix that?  The release.sh script was supposed to take
care of this automatically.

Also we have tons of fixes in the xfsprogs repo that really should go
out into a new 3.1.12 release:

    xfs_io: v8 add the lseek() SEEK_DATA/SEEK_HOLE support
    xfs_db: add header to freesp -d output
    xfs_repair: zero out unused parts of superblocks
    xfs_repair: add prototype for alloc_ex_data()
    xfsprogs: fix Out-of-bounds access in repair/dinode.c
    xfsprogs: fix inode crash in xfs_repair
    xfsprogs: fix unint var in repair phase6
    xfsprogs: fix agcnts leak in xfs_repair's scan_ags
    xfsprogs:free bp in xlog_find_tail() error path
    xfsprogs: free bp in xlog_find_zeroed() error path
    xfsprogs: fix buffer leak in xlog_print_find_oldest
    xfsprogs: avoid double-free in xfs_attr_node_addname
    xfsprogs/io: add readdir command
    mkfs.xfs: fix protofile name create block reservation
    logprint: fix wrapped log dump issue
    xfs_metadump: manpage fix regarding frozen fs
    mkfs: add missing noalign suboption information
    xfsprogs: fix make deb
    xfsprogs: define umode_t for build if not defined already
    xfs_logprint: fix continuation transactions
    xfs_fsr: file reads should be O_DIRECT
    xfs_logprint: print all AGI unlinked buckets

+ a few more pending on the list.  I'd volunteer to prepare and test
thos backports once we get the remaining patches from the list in.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: projid32bit=1 default in xfsprogs-3.2.0

[Ben Myers]

Hey Christoph,

On Tue, Oct 15, 2013 at 01:43:42PM -0700, Christoph Hellwig wrote:
> On Tue, Oct 15, 2013 at 12:46:09PM -0700, Christoph Hellwig wrote:
> > On Mon, Oct 14, 2013 at 04:24:24PM -0500, Eric Sandeen wrote:
> > > Ok.  Default it is!   Thanks guys, just wanted to make sure we were
> > > all in agreement there.
> > 
> > Can we make sure this gets documented in doc/CHANGES?
> > 
> > I'm also pretty sure the entry for 3.2.0-alpha1 is severly lacking,
> > can we make sure the one for the real release is as detailed as the ones
> > for the previous releases?
> 
> Btw, I also noticed releases after 3.1.9 aren't tagged in git, any
> chance we could fix that?  The release.sh script was supposed to take
> care of this automatically.

The tags are out there:
http://oss.sgi.com/cgi-bin/gitweb.cgi?p=xfs/cmds/xfsprogs.git;a=tags

The release script doesn't push tags, just reminds you to do so.  ;)

IMO it's probably a good idea not automate pushing tags, so you have
opportunity to hit the breaks and double check your work before posting a tag
publicly.

> Also we have tons of fixes in the xfsprogs repo that really should go
> out into a new 3.1.12 release:
> 
>     xfs_io: v8 add the lseek() SEEK_DATA/SEEK_HOLE support
>     xfs_db: add header to freesp -d output
>     xfs_repair: zero out unused parts of superblocks
>     xfs_repair: add prototype for alloc_ex_data()
>     xfsprogs: fix Out-of-bounds access in repair/dinode.c
>     xfsprogs: fix inode crash in xfs_repair
>     xfsprogs: fix unint var in repair phase6
>     xfsprogs: fix agcnts leak in xfs_repair's scan_ags
>     xfsprogs:free bp in xlog_find_tail() error path
>     xfsprogs: free bp in xlog_find_zeroed() error path
>     xfsprogs: fix buffer leak in xlog_print_find_oldest
>     xfsprogs: avoid double-free in xfs_attr_node_addname
>     xfsprogs/io: add readdir command
>     mkfs.xfs: fix protofile name create block reservation
>     logprint: fix wrapped log dump issue
>     xfs_metadump: manpage fix regarding frozen fs
>     mkfs: add missing noalign suboption information
>     xfsprogs: fix make deb
>     xfsprogs: define umode_t for build if not defined already
>     xfs_logprint: fix continuation transactions
>     xfs_fsr: file reads should be O_DIRECT
>     xfs_logprint: print all AGI unlinked buckets
> 
> + a few more pending on the list.  I'd volunteer to prepare and test
> thos backports once we get the remaining patches from the list in.

That's an interesting idea.  We hadn't discussed a 3.1.12 release.  No
particular objection to doing that, it's just that we didn't branch for the
v3.0.5->v3.1.0 set of releases.  Maybe it is better if we all focus on 3.2.0.

Regards,
	Ben

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: projid32bit=1 default in xfsprogs-3.2.0

[Christoph Hellwig]

On Tue, Oct 15, 2013 at 04:22:14PM -0500, Ben Myers wrote:
> The tags are out there:
> http://oss.sgi.com/cgi-bin/gitweb.cgi?p=xfs/cmds/xfsprogs.git;a=tags

I managed to somehow miss them.  Sorry!

> That's an interesting idea.  We hadn't discussed a 3.1.12 release.  No
> particular objection to doing that, it's just that we didn't branch for the
> v3.0.5->v3.1.0 set of releases.  Maybe it is better if we all focus on 3.2.0.

With git you can easily branch from a past commit, e.g.:

hch@brick:~/work/xfsprogs$ git checkout -b release-3.1 v3.1.11
Switched to a new branch 'release-3.1'

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: projid32bit=1 default in xfsprogs-3.2.0

[Ben Myers]

Hey,

On Tue, Oct 15, 2013 at 02:31:10PM -0700, Christoph Hellwig wrote:
> On Tue, Oct 15, 2013 at 04:22:14PM -0500, Ben Myers wrote:
> > The tags are out there:
> > http://oss.sgi.com/cgi-bin/gitweb.cgi?p=xfs/cmds/xfsprogs.git;a=tags
> 
> I managed to somehow miss them.  Sorry!

np
 
> > That's an interesting idea.  We hadn't discussed a 3.1.12 release.  No
> > particular objection to doing that, it's just that we didn't branch for the
> > v3.0.5->v3.1.0 set of releases.  Maybe it is better if we all focus on 3.2.0.
> 
> With git you can easily branch from a past commit, e.g.:
> 
> hch@brick:~/work/xfsprogs$ git checkout -b release-3.1 v3.1.11
> Switched to a new branch 'release-3.1'

Yep.  If folks want a 3.1.12 we should certainly kick the idea around a bit.
I'm just guessing that most would prefer to focus on a 3.2.0 in the near term.
Maybe I'm mistaken about that though.  ;)

Thanks,
	Ben

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: projid32bit=1 default in xfsprogs-3.2.0

[Eric Sandeen]

On 10/15/13 4:49 PM, Ben Myers wrote:
> Hey,
> 
> On Tue, Oct 15, 2013 at 02:31:10PM -0700, Christoph Hellwig wrote:
>> On Tue, Oct 15, 2013 at 04:22:14PM -0500, Ben Myers wrote:
>>> The tags are out there:
>>> http://oss.sgi.com/cgi-bin/gitweb.cgi?p=xfs/cmds/xfsprogs.git;a=tags
>>
>> I managed to somehow miss them.  Sorry!
> 
> np
>  
>>> That's an interesting idea.  We hadn't discussed a 3.1.12 release.  No
>>> particular objection to doing that, it's just that we didn't branch for the
>>> v3.0.5->v3.1.0 set of releases.  Maybe it is better if we all focus on 3.2.0.
>>
>> With git you can easily branch from a past commit, e.g.:
>>
>> hch@brick:~/work/xfsprogs$ git checkout -b release-3.1 v3.1.11
>> Switched to a new branch 'release-3.1'
> 
> Yep.  If folks want a 3.1.12 we should certainly kick the idea around a bit.
> I'm just guessing that most would prefer to focus on a 3.2.0 in the near term.
> Maybe I'm mistaken about that though.  ;)

I'm not super keen to divide the focus; I've already updated the distros I
care about to 3.2.0-alpha1, so I wouldn't go backwards to a 3.1.12.

hch, do you have a specific need for it?

-Eric

> Thanks,
> 	Ben
> 

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: projid32bit=1 default in xfsprogs-3.2.0

[Dave Chinner]

On Tue, Oct 15, 2013 at 04:57:36PM -0500, Eric Sandeen wrote:
> On 10/15/13 4:49 PM, Ben Myers wrote:
> > Hey,
> > 
> > On Tue, Oct 15, 2013 at 02:31:10PM -0700, Christoph Hellwig wrote:
> >> On Tue, Oct 15, 2013 at 04:22:14PM -0500, Ben Myers wrote:
> >>> The tags are out there:
> >>> http://oss.sgi.com/cgi-bin/gitweb.cgi?p=xfs/cmds/xfsprogs.git;a=tags
> >>
> >> I managed to somehow miss them.  Sorry!
> > 
> > np
> >  
> >>> That's an interesting idea.  We hadn't discussed a 3.1.12 release.  No
> >>> particular objection to doing that, it's just that we didn't branch for the
> >>> v3.0.5->v3.1.0 set of releases.  Maybe it is better if we all focus on 3.2.0.
> >>
> >> With git you can easily branch from a past commit, e.g.:
> >>
> >> hch@brick:~/work/xfsprogs$ git checkout -b release-3.1 v3.1.11
> >> Switched to a new branch 'release-3.1'
> > 
> > Yep.  If folks want a 3.1.12 we should certainly kick the idea around a bit.
> > I'm just guessing that most would prefer to focus on a 3.2.0 in the near term.
> > Maybe I'm mistaken about that though.  ;)
> 
> I'm not super keen to divide the focus; I've already updated the distros I
> care about to 3.2.0-alpha1, so I wouldn't go backwards to a 3.1.12.

I'd prefer we focus on getting stuff reviewed and integrated into
3.2.0 more quickly than we are now.  At this point in the cycle, we
really need the 3.2 branch and xfstests to be updated daily with the
changes that were reviewed in the past 24 hours so that we can
iterate test cycles with the latest fixes more easily.

This will give us a clearer idea of the problems we still need to
fix before the release can progress without everyone having to keep
their own private trees up to date with what everyone else is
changing/fixing...

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: projid32bit=1 default in xfsprogs-3.2.0

[Eric Sandeen]

On 10/15/13 5:48 PM, Dave Chinner wrote:
> On Tue, Oct 15, 2013 at 04:57:36PM -0500, Eric Sandeen wrote:
>> On 10/15/13 4:49 PM, Ben Myers wrote:
>>> Hey,
>>>
>>> On Tue, Oct 15, 2013 at 02:31:10PM -0700, Christoph Hellwig wrote:
>>>> On Tue, Oct 15, 2013 at 04:22:14PM -0500, Ben Myers wrote:
>>>>> The tags are out there:
>>>>> http://oss.sgi.com/cgi-bin/gitweb.cgi?p=xfs/cmds/xfsprogs.git;a=tags
>>>>
>>>> I managed to somehow miss them.  Sorry!
>>>
>>> np
>>>  
>>>>> That's an interesting idea.  We hadn't discussed a 3.1.12 release.  No
>>>>> particular objection to doing that, it's just that we didn't branch for the
>>>>> v3.0.5->v3.1.0 set of releases.  Maybe it is better if we all focus on 3.2.0.
>>>>
>>>> With git you can easily branch from a past commit, e.g.:
>>>>
>>>> hch@brick:~/work/xfsprogs$ git checkout -b release-3.1 v3.1.11
>>>> Switched to a new branch 'release-3.1'
>>>
>>> Yep.  If folks want a 3.1.12 we should certainly kick the idea around a bit.
>>> I'm just guessing that most would prefer to focus on a 3.2.0 in the near term.
>>> Maybe I'm mistaken about that though.  ;)
>>
>> I'm not super keen to divide the focus; I've already updated the distros I
>> care about to 3.2.0-alpha1, so I wouldn't go backwards to a 3.1.12.
> 
> I'd prefer we focus on getting stuff reviewed and integrated into
> 3.2.0 more quickly than we are now.  At this point in the cycle, we
> really need the 3.2 branch and xfstests to be updated daily with the
> changes that were reviewed in the past 24 hours so that we can
> iterate test cycles with the latest fixes more easily.

Agreed, I'm sitting on things that I don't know if I need to resend or not.

I've offered it before, and I'll offer again: if it's a burden for SGI to
keep up with xfstests merging, we can help - but the delays are getting to
be problematic again in the current process.

Thanks,
-Eric

> This will give us a clearer idea of the problems we still need to
> fix before the release can progress without everyone having to keep
> their own private trees up to date with what everyone else is
> changing/fixing...
> 
> Cheers,
> 
> Dave.
> 

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: projid32bit=1 default in xfsprogs-3.2.0

[Rich Johnston]

On 10/15/2013 03:43 PM, Christoph Hellwig wrote:
> On Tue, Oct 15, 2013 at 12:46:09PM -0700, Christoph Hellwig wrote:
>> On Mon, Oct 14, 2013 at 04:24:24PM -0500, Eric Sandeen wrote:
>>> Ok.  Default it is!   Thanks guys, just wanted to make sure we were
>>> all in agreement there.
>>
>> Can we make sure this gets documented in doc/CHANGES?
>>
>> I'm also pretty sure the entry for 3.2.0-alpha1 is severly lacking,
>> can we make sure the one for the real release is as detailed as the ones
>> for the previous releases?
>
> Btw, I also noticed releases after 3.1.9 aren't tagged in git, any
> chance we could fix that?  The release.sh script was supposed to take
> care of this automatically.
>
> Also we have tons of fixes in the xfsprogs repo that really should go
> out into a new 3.1.12 release:
>
>      xfs_io: v8 add the lseek() SEEK_DATA/SEEK_HOLE support
>      xfs_db: add header to freesp -d output
>      xfs_repair: zero out unused parts of superblocks
>      xfs_repair: add prototype for alloc_ex_data()
>      xfsprogs: fix Out-of-bounds access in repair/dinode.c
>      xfsprogs: fix inode crash in xfs_repair
>      xfsprogs: fix unint var in repair phase6
>      xfsprogs: fix agcnts leak in xfs_repair's scan_ags
>      xfsprogs:free bp in xlog_find_tail() error path
>      xfsprogs: free bp in xlog_find_zeroed() error path
>      xfsprogs: fix buffer leak in xlog_print_find_oldest
>      xfsprogs: avoid double-free in xfs_attr_node_addname
>      xfsprogs/io: add readdir command
>      mkfs.xfs: fix protofile name create block reservation
>      logprint: fix wrapped log dump issue
>      xfs_metadump: manpage fix regarding frozen fs
>      mkfs: add missing noalign suboption information
>      xfsprogs: fix make deb
>      xfsprogs: define umode_t for build if not defined already
>      xfs_logprint: fix continuation transactions
>      xfs_fsr: file reads should be O_DIRECT
>      xfs_logprint: print all AGI unlinked buckets
>
> + a few more pending on the list.  I'd volunteer to prepare and test
> thos backports once we get the remaining patches from the list in.
>
> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs
>
Hey Christoph,

If your still interested in a 3.1.12 release, a 'for-3.1.12' branch has 
been created.
http://oss.sgi.com/archives/xfs/2014-01/msg00056.html

Thanks
--Rich

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: projid32bit=1 default in xfsprogs-3.2.0

[Christoph Hellwig]

On Fri, Jan 10, 2014 at 08:45:58AM -0600, Rich Johnston wrote:
> If your still interested in a 3.1.12 release, a 'for-3.1.12' branch
> has been created.
> http://oss.sgi.com/archives/xfs/2014-01/msg00056.html

I still think we need it, but I'm not having a lot of time at the
moment.  I should be able to put something together in the next couple
days, but I will need some help on the QA side.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs