mbox

[PULL,v9,00/16] xen: xen-domid-restrict improvements

Message ID 1524758187-9351-1-git-send-email-ian.jackson@eu.citrix.com
State New
Headers show

Pull-request

https://xenbits.xen.org/git-http/people/iwj/qemu.git for-upstream.depriv-2

Message

Ian Jackson April 26, 2018, 3:56 p.m. UTC 
From: Ian Jackson <Ian.Jackson@eu.citrix.com>

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

The following changes since commit b8846a4d6352b2a1d2012f8b3b9115640524aeda:

  vl.c: new function serial_max_hds() (2018-04-26 13:58:29 +0100)

are available in the git repository at:

  https://xenbits.xen.org/git-http/people/iwj/qemu.git for-upstream.depriv-2

for you to fetch changes up to 8bbe05d77360b73c1834808023016a778ccf55ca:

  configure: do_compiler: Dump some extra info under bash (2018-04-26 16:29:52 +0100)

This series provides necessary support for running qemu as a Xen
device model without power equivalent to root.  In particular, it
makes -xen-domid-restrict effective.

The patches are unchanged since v8, apart from additional reviews, and
rebasing onto master.  All the patches are now reviewed and/or acked
apart from the patch to configure, which has a Tested-by from Thomas
Huth.  I have stripped off the per-patch-series-verson update
comments.

- ----------------------------------------------------------------

Anthony PERARD (1):
      xen: link against xentoolcore

Ian Jackson (13):
      checkpatch: Add xendevicemodel_handle to the list of types
      AccelClass: Introduce accel_setup_post
      xen: restrict: use xentoolcore_restrict_all
      xen: defer call to xen_restrict until just before os_setup_post
      xen: destroy_hvm_domain: Move reason into a variable
      xen: move xc_interface compatibility fallback further up the file
      xen: destroy_hvm_domain: Try xendevicemodel_shutdown
      os-posix: cleanup: Replace fprintfs with error_report in change_process_uid
      os-posix: Provide new -runas <uid>:<gid> facility
      xen: Remove now-obsolete xen_xc_domain_add_to_physmap
      os-posix: cleanup: Replace fprintf with error_report in remaining call sites
      os-posix: cleanup: Replace perror with error_report
      configure: do_compiler: Dump some extra info under bash

Ross Lagerwall (2):
      xen: Use newly added dmops for mapping VGA memory
      xen: Expect xenstore write to fail when restricted

 accel/accel.c               |   9 ++++
 configure                   |  33 +++++++++++-
 hw/i386/xen/xen-hvm.c       |  75 ++++++++++++++------------
 hw/xen/xen-common.c         |  21 +++++++-
 include/hw/xen/xen_common.h | 125 +++++++++++++++++++++-----------------------
 include/sysemu/accel.h      |   3 ++
 os-posix.c                  |  91 +++++++++++++++++++++++++-------
 qemu-options.hx             |   3 +-
 scripts/checkpatch.pl       |   1 +
 vl.c                        |   1 +
 10 files changed, 242 insertions(+), 120 deletions(-)


Anthony PERARD (1):
  xen: link against xentoolcore

Ian Jackson (13):
  checkpatch: Add xendevicemodel_handle to the list of types
  AccelClass: Introduce accel_setup_post
  xen: restrict: use xentoolcore_restrict_all
  xen: defer call to xen_restrict until just before os_setup_post
  xen: destroy_hvm_domain: Move reason into a variable
  xen: move xc_interface compatibility fallback further up the file
  xen: destroy_hvm_domain: Try xendevicemodel_shutdown
  os-posix: cleanup: Replace fprintfs with error_report in
    change_process_uid
  os-posix: Provide new -runas <uid>:<gid> facility
  xen: Remove now-obsolete xen_xc_domain_add_to_physmap
  os-posix: cleanup: Replace fprintf with error_report in remaining call
    sites
  os-posix: cleanup: Replace perror with error_report
  configure: do_compiler: Dump some extra info under bash

Ross Lagerwall (2):
  xen: Use newly added dmops for mapping VGA memory
  xen: Expect xenstore write to fail when restricted

 accel/accel.c               |   9 ++++
 configure                   |  33 +++++++++++-
 hw/i386/xen/xen-hvm.c       |  75 ++++++++++++++------------
 hw/xen/xen-common.c         |  21 +++++++-
 include/hw/xen/xen_common.h | 125 +++++++++++++++++++++-----------------------
 include/sysemu/accel.h      |   3 ++
 os-posix.c                  |  91 +++++++++++++++++++++++++-------
 qemu-options.hx             |   3 +-
 scripts/checkpatch.pl       |   1 +
 vl.c                        |   1 +
 10 files changed, 242 insertions(+), 120 deletions(-)

- -- 
2.1.4

-----BEGIN PGP SIGNATURE-----

iQFUBAEBCAA+FiEEVZrkbC1rbTJl58uh4+M5I0i1DTkFAlrh9pkgHGlqYWNrc29u
QGNoaWFyay5ncmVlbmVuZC5vcmcudWsACgkQ4+M5I0i1DTm6dwgAx4Y2BlEvGZZF
1ZM+ocUu97egBCRb8WOJl1zvfeaxIkLfM+zP2TNFazjWUSKHJJUmzUE6IdaJ3ddX
2XVV0u9LtERI7Si0UVHF/2n3UhpJk4fCBKNyiv5HTPDySLIBdOuONaKLg5XlFUxm
ZJ3s5S/JncgPj4/Q475OOglJyMtfRRDfFHXM5ajAExncMAKjDrX1r+VNfCU1Eg/q
lFzaB6tt8GlVXu7y0ur2B1JsF5lm/Oluo4k4paXEHi2+OJVAvlAAZvkky6laL6DI
8jONOJBhL5rVuAtQCJ7ShdJaTR7z/9Z2MWzjjBzTMY1JhslyUfPS7oPM13m+oHyY
OYzrEfZJOA==
=gdCR
-----END PGP SIGNATURE-----

Comments

Peter Maydell April 26, 2018, 5:27 p.m. UTC | #1 
On 26 April 2018 at 16:56, Ian Jackson <ian.jackson@eu.citrix.com> wrote:
> From: Ian Jackson <Ian.Jackson@eu.citrix.com>
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA256
>
> The following changes since commit b8846a4d6352b2a1d2012f8b3b9115640524aeda:
>
>   vl.c: new function serial_max_hds() (2018-04-26 13:58:29 +0100)
>
> are available in the git repository at:
>
>   https://xenbits.xen.org/git-http/people/iwj/qemu.git for-upstream.depriv-2

Hi -- my apply-pullreq script thinks this isn't a signed tag:

error: remotes/iwj/for-upstream.depriv-2: cannot verify a non-tag object of
type commit. Does not appear to be a signed tag

We only apply signed-tag pull requests these days; could you
create and sign a tag and resend the pullreq, please? (You
only need to resend the cover letter, not the whole set of
patches.)

(https://wiki.qemu.org/Contribute/SubmitAPullRequest
is our process documentation on pull requests, but I think
everything else with this one is fine.)

thanks
-- PMM
Ian Jackson April 26, 2018, 5:36 p.m. UTC | #2 
Peter Maydell writes ("Re: [PULL v9 00/16] xen: xen-domid-restrict improvements"):
> Hi -- my apply-pullreq script thinks this isn't a signed tag:
> 
> error: remotes/iwj/for-upstream.depriv-2: cannot verify a non-tag object of
> type commit. Does not appear to be a signed tag

It isn't.  It's not a tag at all.  It's a branch.

> We only apply signed-tag pull requests these days; could you
> create and sign a tag and resend the pullreq, please? (You
> only need to resend the cover letter, not the whole set of
> patches.)

I can do that.  Alternatively you could manually verify the signature
on my email and observe that it mentions the correct commit hash.

> (https://wiki.qemu.org/Contribute/SubmitAPullRequest
> is our process documentation on pull requests, but I think
> everything else with this one is fine.)

I read that.  It does not say that the pull request must refer to a
signed tag.  It says that the pull request must be signed, so I signed
my pull request by feeding the email body to gnupg.

A detailed explanation of why you want the pull request to be signed,
rather than for it to be an unsigned email referring to a signed tag,
is beyond the scope of this memo :-).

Anyway, would you like me to make a signed tag and send you an
unsigned email referring to it, then ?  Or will you use the signed
email ?

Regards,
Ian.
Peter Maydell April 26, 2018, 5:41 p.m. UTC | #3 
On 26 April 2018 at 18:36, Ian Jackson <ian.jackson@citrix.com> wrote:
> Peter Maydell writes ("Re: [PULL v9 00/16] xen: xen-domid-restrict improvements"):
>> Hi -- my apply-pullreq script thinks this isn't a signed tag:
>>
>> error: remotes/iwj/for-upstream.depriv-2: cannot verify a non-tag object of
>> type commit. Does not appear to be a signed tag
>
> It isn't.  It's not a tag at all.  It's a branch.
>
>> We only apply signed-tag pull requests these days; could you
>> create and sign a tag and resend the pullreq, please? (You
>> only need to resend the cover letter, not the whole set of
>> patches.)
>
> I can do that.  Alternatively you could manually verify the signature
> on my email and observe that it mentions the correct commit hash.

I could, but I hope you will appreciate that I'm not terribly
interested in trying to fathom the mysteries of gpg in order
to deal with a single pull request that doesn't appear in
the form that my tooling expects it to be and that every other
pull request I deal with comes in...

>> (https://wiki.qemu.org/Contribute/SubmitAPullRequest
>> is our process documentation on pull requests, but I think
>> everything else with this one is fine.)
>
> I read that.  It does not say that the pull request must refer to a
> signed tag.  It says that the pull request must be signed, so I signed
> my pull request by feeding the email body to gnupg.

I have fixed the wiki page to be clearer about what we mean by
"signed pull requests".

> Anyway, would you like me to make a signed tag and send you an
> unsigned email referring to it, then ?

Yes, please. (You can sign the email as well if you like, but I'm
not going to be checking that, so it will be wasted effort :-)).

thanks
-- PMM
Ian Jackson April 26, 2018, 6:13 p.m. UTC | #4 
Peter Maydell writes ("Re: [PULL v9 00/16] xen: xen-domid-restrict improvements"):
> On 26 April 2018 at 18:36, Ian Jackson <ian.jackson@citrix.com> wrote:
> > I can do that.  Alternatively you could manually verify the signature
> > on my email and observe that it mentions the correct commit hash.
> 
> I could, but I hope you will appreciate that I'm not terribly
> interested in trying to fathom the mysteries of gpg in order
> to deal with a single pull request that doesn't appear in
> the form that my tooling expects it to be and that every other
> pull request I deal with comes in...

Fair enough.

> I have fixed the wiki page to be clearer about what we mean by
> "signed pull requests".

Thanks :-).

> > Anyway, would you like me to make a signed tag and send you an
> > unsigned email referring to it, then ?
> 
> Yes, please. (You can sign the email as well if you like, but I'm
> not going to be checking that, so it will be wasted effort :-)).

I will do this in a moment.

Ian.