Re: [PATCH v3 1/6] block: add disk sequence number

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, 2021-06-23 at 17:02 +0200, Hannes Reinecke wrote:
> On 6/23/21 4:12 PM, Lennart Poettering wrote:
> > On Mi, 23.06.21 16:01, Hannes Reinecke (hare@xxxxxxx) wrote:
> > 
> > > > Thus: a global instead of local sequence number counter is absolutely
> > > > *key* for the problem this is supposed to solve
> > > > 
> > > Well ... except that you'll need to keep track of the numbers (otherwise you
> > > wouldn't know if the numbers changed, right?).
> > > And if you keep track of the numbers you probably will have to implement an
> > > uevent listener to get the events in time.
> > 
> > Hmm? This is backwards. The goal here is to be able to safely match up
> > uevents to specific uses of a block device, given that block device
> > names are agressively recycled.
> > 
> > you imply it was easy to know which device use a uevent belongs
> > to. But that's the problem: it is not possible to do so safely. if i
> > see a uevent for a block device "loop0" I cannot tell if it was from
> > my own use of the device or for some previous user of it.
> > 
> > And that's what we'd like to see fixed: i.e. we query the block device
> > for the seqeno now used and then we can use that to filter the uevents
> > and ignore the ones that do not carry the same sequence number as we
> > got assigned for our user.
> > 
> 
> It is notoriously tricky to monitor the intended use-case for kernel 
> devices, precisely because we do _not_ attach any additional information 
> to it.
> I have send a proposal for LSF to implement block-namespaces, the prime 
> use-case of which is indeed attaching cgroup/namespace information to 
> block devices such that we _can_ match (block) devices to specific contexts.

Having namespaces for block devices would be an awesome feature, very
much looking forward to have that, as it solves a number of other
issues we have.
And while it could maybe be used in some instances of this particular
problem, unfortunately I don't think it can solve all of them - in some
real cases, we have to work in the root namespace as we are setting
things up for it.

> Which I rather prefer than adding sequence numbers to block devices; 
> incidentally you could solve the same problem by _not_ reusing numbers 
> aggressively but rather allocate the next free one after the most 
> recently allocated one.
> Will give you much the same thing without having to burden others with it.

If I understood this right, you are proposing to move the
monothonically increasing sequence id to the device name, rather than
as internal metadata? So that, eg, loop0 gets used exactly once and
never again, and so on? Wouldn't that be a much more visible and
disruptive change, potentially backward incompatible and breaking
userspace left and right?

> The better alternative here would be to extend the loop ioctl to pass in 
> an UUID when allocating the device.
> That way you can easily figure out whether the loop device has been 
> modified.

A UUID solves the problem we are currently talking about. But a
monothonically increasing sequence number has additional great
properties compared to a UUID, that we very much want to make use of
(again these are all _real_ use cases), and were described  in detail
here:

https://lore.kernel.org/linux-fsdevel/20210315201331.GA2577561@xxxxxxxxxxxxxxxxxxxx/t/#m3b1ffdffcc70a7c3ef4d7f13d0c2d5b9e4dde35a

> But in the end, it's the loop driver and I'm not particular bothered 
> with it. I am, though, if you need to touch all drivers just to support 
> one particular use-case in one particular device driver.
> 
> Incidentally, we don't have this problem in SCSI as we _can_ identify 
> devices here. So in the end we couldn't care less on which /dev/sdX 
> device it ends up.
> And I guess that's what we should attempt for loop devices, too.

Sorry, I'm not sure what you mean here by "touch all drivers"? This
series changes only drivers/block/loop.c, no other device drivers code
is touched?

-- 
Kind regards,
Luca Boccassi

Attachment: signature.asc
Description: This is a digitally signed message part


[Index of Archives]     [Linux Ext4 Filesystem]     [Union Filesystem]     [Filesystem Testing]     [Ceph Users]     [Ecryptfs]     [AutoFS]     [Kernel Newbies]     [Share Photos]     [Security]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux Cachefs]     [Reiser Filesystem]     [Linux RAID]     [Samba]     [Device Mapper]     [CEPH Development]

  Powered by Linux