mbox series

[v2,00/14] LUKS: encryption slot management using amend interface

Message ID 20200308151903.25941-1-mlevitsk@redhat.com (mailing list archive)
Headers show
Series LUKS: encryption slot management using amend interface | expand

Message

Maxim Levitsky March 8, 2020, 3:18 p.m. UTC
Hi!
Here is the updated series of my patches, incorporating all the feedback I received.

This implements the API interface that we agreed upon except that I merged the
LUKSKeyslotActive/LUKSKeyslotInactive union into a struct because otherwise
I need nested unions which are not supported currently by QAPI parser.
This didn't change the API and thus once support for nested unions is there,
it can always be implemented in backward compatible way.

I hope that this series will finally be considered for merging, since I am somewhat running
out of time to finish this task.

Patches are strictly divided by topic to 3 groups, and each group depends on former groups.

* Patches 1,2 implement qcrypto generic amend interface, including definition
  of structs used in crypto.json and implement this in luks crypto driver
  Nothing is exposed to the user at this stage

* Patches 3-9 use the code from patches 1,2 to implement qemu-img amend based encryption slot management
  for luks and for qcow2, and add a bunch of iotests to cover that.

* Patches 10-13 add x-blockdev-amend (I'll drop the -x prefix if you like), and wire it
  to luks and qcow2 driver to implement qmp based encryption slot management also using
  the code from patches 1,2, and also add a bunch of iotests to cover this.

Tested with -raw,-qcow2 and -luks iotests and 'make check'

Best regards,
        Maxim Levitsky

clone of "luks-keymgmnt-v2"

Maxim Levitsky (14):
  qcrypto/core: add generic infrastructure for crypto options amendment
  qcrypto/luks: implement encryption key management
  block/amend: add 'force' option
  block/amend: separate amend and create options for qemu-img
  block/amend: refactor qcow2 amend options
  block/crypto: rename two functions
  block/crypto: implement the encryption key management
  block/qcow2: extend qemu-img amend interface with crypto options
  iotests: filter few more luks specific create options
  iotests: qemu-img tests for luks key management
  block/core: add generic infrastructure for x-blockdev-amend qmp
    command
  block/crypto: implement blockdev-amend
  block/qcow2: implement blockdev-amend
  iotests: add tests for blockdev-amend

 block.c                          |   4 +-
 block/Makefile.objs              |   2 +-
 block/amend.c                    | 108 +++++++++
 block/crypto.c                   | 203 ++++++++++++++--
 block/crypto.h                   |  47 +++-
 block/qcow2.c                    | 314 ++++++++++++++----------
 crypto/block-luks.c              | 398 ++++++++++++++++++++++++++++++-
 crypto/block.c                   |  31 +++
 crypto/blockpriv.h               |   8 +
 docs/tools/qemu-img.rst          |   5 +-
 include/block/block.h            |   1 +
 include/block/block_int.h        |  24 +-
 include/crypto/block.h           |  22 ++
 qapi/block-core.json             |  68 ++++++
 qapi/crypto.json                 |  75 +++++-
 qapi/job.json                    |   4 +-
 qemu-img-cmds.hx                 |   4 +-
 qemu-img.c                       |  44 +++-
 tests/qemu-iotests/049.out       | 102 ++++----
 tests/qemu-iotests/061.out       |  12 +-
 tests/qemu-iotests/079.out       |  18 +-
 tests/qemu-iotests/082.out       | 176 ++++----------
 tests/qemu-iotests/085.out       |  38 +--
 tests/qemu-iotests/087.out       |   6 +-
 tests/qemu-iotests/115.out       |   2 +-
 tests/qemu-iotests/121.out       |   4 +-
 tests/qemu-iotests/125.out       | 192 +++++++--------
 tests/qemu-iotests/134.out       |   2 +-
 tests/qemu-iotests/144.out       |   4 +-
 tests/qemu-iotests/158.out       |   4 +-
 tests/qemu-iotests/182.out       |   2 +-
 tests/qemu-iotests/185.out       |   8 +-
 tests/qemu-iotests/188.out       |   2 +-
 tests/qemu-iotests/189.out       |   4 +-
 tests/qemu-iotests/198.out       |   4 +-
 tests/qemu-iotests/243.out       |  16 +-
 tests/qemu-iotests/250.out       |   2 +-
 tests/qemu-iotests/255.out       |   8 +-
 tests/qemu-iotests/263.out       |   4 +-
 tests/qemu-iotests/280.out       |   2 +-
 tests/qemu-iotests/284.out       |   6 +-
 tests/qemu-iotests/300           | 207 ++++++++++++++++
 tests/qemu-iotests/300.out       |  99 ++++++++
 tests/qemu-iotests/301           |  90 +++++++
 tests/qemu-iotests/301.out       |  30 +++
 tests/qemu-iotests/302           | 278 +++++++++++++++++++++
 tests/qemu-iotests/302.out       |  40 ++++
 tests/qemu-iotests/303           | 233 ++++++++++++++++++
 tests/qemu-iotests/303.out       |  33 +++
 tests/qemu-iotests/common.filter |   6 +-
 tests/qemu-iotests/group         |   6 +
 51 files changed, 2486 insertions(+), 516 deletions(-)
 create mode 100644 block/amend.c
 create mode 100755 tests/qemu-iotests/300
 create mode 100644 tests/qemu-iotests/300.out
 create mode 100755 tests/qemu-iotests/301
 create mode 100644 tests/qemu-iotests/301.out
 create mode 100755 tests/qemu-iotests/302
 create mode 100644 tests/qemu-iotests/302.out
 create mode 100755 tests/qemu-iotests/303
 create mode 100644 tests/qemu-iotests/303.out

Comments

Eric Blake March 12, 2020, 11:56 a.m. UTC | #1
On 3/8/20 10:18 AM, Maxim Levitsky wrote:
> Hi!
> Here is the updated series of my patches, incorporating all the feedback I received.
> 

> Patches are strictly divided by topic to 3 groups, and each group depends on former groups.
> 
> * Patches 1,2 implement qcrypto generic amend interface, including definition
>    of structs used in crypto.json and implement this in luks crypto driver
>    Nothing is exposed to the user at this stage
> 
> * Patches 3-9 use the code from patches 1,2 to implement qemu-img amend based encryption slot management
>    for luks and for qcow2, and add a bunch of iotests to cover that.
> 
> * Patches 10-13 add x-blockdev-amend (I'll drop the -x prefix if you like), and wire it
>    to luks and qcow2 driver to implement qmp based encryption slot management also using
>    the code from patches 1,2, and also add a bunch of iotests to cover this.

>   tests/qemu-iotests/284.out       |   6 +-
>   tests/qemu-iotests/300           | 207 ++++++++++++++++

Any reason why you skipped straight to test 300, rather than using an 
available slot like 290?  (Admittedly, our process for reserving slots 
is not very high-tech: manually scan the list for what other patches out 
there have claimed a slot, and be prepared to renumber when rebasing)
Maxim Levitsky March 12, 2020, 2:33 p.m. UTC | #2
On Thu, 2020-03-12 at 06:56 -0500, Eric Blake wrote:
> On 3/8/20 10:18 AM, Maxim Levitsky wrote:
> > Hi!
> > Here is the updated series of my patches, incorporating all the feedback I received.
> > 
> > Patches are strictly divided by topic to 3 groups, and each group depends on former groups.
> > 
> > * Patches 1,2 implement qcrypto generic amend interface, including definition
> >    of structs used in crypto.json and implement this in luks crypto driver
> >    Nothing is exposed to the user at this stage
> > 
> > * Patches 3-9 use the code from patches 1,2 to implement qemu-img amend based encryption slot management
> >    for luks and for qcow2, and add a bunch of iotests to cover that.
> > 
> > * Patches 10-13 add x-blockdev-amend (I'll drop the -x prefix if you like), and wire it
> >    to luks and qcow2 driver to implement qmp based encryption slot management also using
> >    the code from patches 1,2, and also add a bunch of iotests to cover this.
> >   tests/qemu-iotests/284.out       |   6 +-
> >   tests/qemu-iotests/300           | 207 ++++++++++++++++
> 
> Any reason why you skipped straight to test 300, rather than using an 
> available slot like 290?  (Admittedly, our process for reserving slots 
> is not very high-tech: manually scan the list for what other patches out 
> there have claimed a slot, and be prepared to renumber when rebasing)
The only reason I used these slots is that I know sadly that I'll have to resend and
rebase this patchset for a while, and every time a test with the number I use is added,
this causes relatively hard to fix conflict (or at least I don't know how to fix these conflicts effectively)

Thus I used safe numbers, but at the rate this task progresses I won't be surprised that when this is merged,
these will be test numbers to use...

TL;DR - these are placeholders, and once the patch set is blesssed for merging upstream I'll update this next
available numbers.

Best regards,
	Maxim Levitsky