mbox

[PULL,00/33] migration queue

Message ID 20220623092810.96234-1-dgilbert@redhat.com
State New
Headers show

Pull-request

https://gitlab.com/dagrh/qemu.git tags/pull-migration-20220623a

Message

Dr. David Alan Gilbert June 23, 2022, 9:27 a.m. UTC 
From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>

The following changes since commit 2b049d2c8dc01de750410f8f1a4eac498c04c723:

  Merge tag 'pull-aspeed-20220622' of https://github.com/legoater/qemu into staging (2022-06-22 07:27:06 -0700)

are available in the Git repository at:

  https://gitlab.com/dagrh/qemu.git tags/pull-migration-20220623a

for you to fetch changes up to 5df0eaff8e24223974bf2516e6dc773695603017:

  tests: Add dirty page rate limit test (2022-06-23 10:18:14 +0100)

----------------------------------------------------------------
Migration pull 2022-06-23

This replaces my pull from yesterday, and Juan's from the day before.

Compared to my pull eysterday:
  A one character fix in get_buffer patch spotted by Peter Xu

Compared to Juan's pull:
  a) Hopefully fixed non-Linux builds
    (Local build test on mingw64 works
    Note: the zero-copy capability is now
    defined on non-Linux systems)
  b) Added Hyman's series - it had been
    on queue for a while (sorry for the delay)

In this today migration PULL request:
- Dainiel Berrangé - qemufileops cleanup
- Leonardo Bras  - cleanups for zero copy
- Juan Quintela  - RDMA cleanups
- Hyman Huang - per-vcpu dirty ring work

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>

----------------------------------------------------------------
Daniel P. Berrangé (21):
      io: add a QIOChannelNull equivalent to /dev/null
      migration: switch to use QIOChannelNull for dummy channel
      migration: remove unreachble RDMA code in save_hook impl
      migration: rename rate limiting fields in QEMUFile
      migration: rename 'pos' field in QEMUFile to 'bytes_processed'
      migration: rename qemu_ftell to qemu_file_total_transferred
      migration: rename qemu_update_position to qemu_file_credit_transfer
      migration: rename qemu_file_update_transfer to qemu_file_acct_rate_limit
      migration: introduce a QIOChannel impl for BlockDriverState VMState
      migration: convert savevm to use QIOChannelBlock for VMState
      migration: stop passing 'opaque' parameter to QEMUFile hooks
      migration: hardcode assumption that QEMUFile is backed with QIOChannel
      migration: introduce new constructors for QEMUFile
      migration: remove unused QEMUFileGetFD typedef / qemu_get_fd method
      migration: remove the QEMUFileOps 'shut_down' callback
      migration: remove the QEMUFileOps 'set_blocking' callback
      migration: remove the QEMUFileOps 'close' callback
      migration: remove the QEMUFileOps 'get_buffer' callback
      migration: remove the QEMUFileOps 'writev_buffer' callback
      migration: remove the QEMUFileOps 'get_return_path' callback
      migration: remove the QEMUFileOps abstraction

Hyman Huang (8):
      accel/kvm/kvm-all: Refactor per-vcpu dirty ring reaping
      cpus: Introduce cpu_list_generation_id
      migration/dirtyrate: Refactor dirty page rate calculation
      softmmu/dirtylimit: Implement vCPU dirtyrate calculation periodically
      accel/kvm/kvm-all: Introduce kvm_dirty_ring_size function
      softmmu/dirtylimit: Implement virtual CPU throttle
      softmmu/dirtylimit: Implement dirty page rate limit
      tests: Add dirty page rate limit test

Juan Quintela (1):
      migration: Remove RDMA_UNREGISTRATION_EXAMPLE

Leonardo Bras (3):
      QIOChannelSocket: Introduce assert and reduce ifdefs to improve readability
      QIOChannelSocket: Fix zero-copy send so socket flush works
      migration: Change zero_copy_send from migration parameter to migration capability

 accel/kvm/kvm-all.c               |  46 ++-
 accel/stubs/kvm-stub.c            |   5 +
 cpus-common.c                     |   8 +
 hmp-commands-info.hx              |  13 +
 hmp-commands.hx                   |  32 ++
 include/exec/cpu-common.h         |   1 +
 include/exec/memory.h             |   5 +-
 include/hw/core/cpu.h             |   6 +
 include/io/channel-null.h         |  55 ++++
 include/monitor/hmp.h             |   3 +
 include/sysemu/dirtylimit.h       |  37 +++
 include/sysemu/dirtyrate.h        |  28 ++
 include/sysemu/kvm.h              |   2 +
 io/channel-null.c                 | 237 +++++++++++++++
 io/channel-socket.c               |  19 +-
 io/meson.build                    |   1 +
 io/trace-events                   |   3 +
 migration/block.c                 |  10 +-
 migration/channel-block.c         | 195 +++++++++++++
 migration/channel-block.h         |  59 ++++
 migration/channel.c               |   4 +-
 migration/colo.c                  |   5 +-
 migration/dirtyrate.c             | 227 ++++++++------
 migration/dirtyrate.h             |   7 +-
 migration/meson.build             |   2 +-
 migration/migration.c             |  68 ++---
 migration/multifd.c               |   4 +-
 migration/qemu-file-channel.c     | 194 ------------
 migration/qemu-file-channel.h     |  32 --
 migration/qemu-file.c             | 193 ++++++------
 migration/qemu-file.h             | 125 ++++----
 migration/ram.c                   |   8 +-
 migration/rdma.c                  | 185 +++---------
 migration/savevm.c                |  55 +---
 migration/vmstate.c               |   5 +-
 monitor/hmp-cmds.c                |   6 -
 qapi/migration.json               | 113 +++++--
 softmmu/dirtylimit.c              | 601 ++++++++++++++++++++++++++++++++++++++
 softmmu/meson.build               |   1 +
 softmmu/trace-events              |   7 +
 tests/qtest/migration-helpers.c   |  22 ++
 tests/qtest/migration-helpers.h   |   2 +
 tests/qtest/migration-test.c      | 255 ++++++++++++++++
 tests/qtest/qmp-cmd-test.c        |   2 +
 tests/unit/meson.build            |   1 +
 tests/unit/test-io-channel-null.c |  95 ++++++
 tests/unit/test-vmstate.c         |   5 +-
 47 files changed, 2219 insertions(+), 770 deletions(-)
 create mode 100644 include/io/channel-null.h
 create mode 100644 include/sysemu/dirtylimit.h
 create mode 100644 include/sysemu/dirtyrate.h
 create mode 100644 io/channel-null.c
 create mode 100644 migration/channel-block.c
 create mode 100644 migration/channel-block.h
 delete mode 100644 migration/qemu-file-channel.c
 delete mode 100644 migration/qemu-file-channel.h
 create mode 100644 softmmu/dirtylimit.c
 create mode 100644 tests/unit/test-io-channel-null.c

Comments

Richard Henderson June 23, 2022, 2:26 p.m. UTC | #1 
On 6/23/22 02:27, Dr. David Alan Gilbert (git) wrote:
> From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
> 
> The following changes since commit 2b049d2c8dc01de750410f8f1a4eac498c04c723:
> 
>    Merge tag 'pull-aspeed-20220622' of https://github.com/legoater/qemu into staging (2022-06-22 07:27:06 -0700)
> 
> are available in the Git repository at:
> 
>    https://gitlab.com/dagrh/qemu.git tags/pull-migration-20220623a
> 
> for you to fetch changes up to 5df0eaff8e24223974bf2516e6dc773695603017:
> 
>    tests: Add dirty page rate limit test (2022-06-23 10:18:14 +0100)
> 
> ----------------------------------------------------------------
> Migration pull 2022-06-23
> 
> This replaces my pull from yesterday, and Juan's from the day before.
> 
> Compared to my pull eysterday:
>    A one character fix in get_buffer patch spotted by Peter Xu
> 
> Compared to Juan's pull:
>    a) Hopefully fixed non-Linux builds
>      (Local build test on mingw64 works
>      Note: the zero-copy capability is now
>      defined on non-Linux systems)
>    b) Added Hyman's series - it had been
>      on queue for a while (sorry for the delay)
> 
> In this today migration PULL request:
> - Dainiel Berrangé - qemufileops cleanup
> - Leonardo Bras  - cleanups for zero copy
> - Juan Quintela  - RDMA cleanups
> - Hyman Huang - per-vcpu dirty ring work

New build failure on all 32-bit hosts:

https://gitlab.com/qemu-project/qemu/-/jobs/2631167210

../softmmu/dirtylimit.c: In function ‘vcpu_dirty_rate_get’:
/builds/qemu-project/qemu/include/qemu/compiler.h:74:36: error: static assertion failed: 
"not expecting: sizeof(*&rates[cpu_index].dirty_rate) > ATOMIC_REG_SIZE"
    74 | #define QEMU_BUILD_BUG_MSG(x, msg) _Static_assert(!(x), msg)
       |                                    ^~~~~~~~~~~~~~
/builds/qemu-project/qemu/include/qemu/compiler.h:76:30: note: in expansion of macro 
‘QEMU_BUILD_BUG_MSG’
    76 | #define QEMU_BUILD_BUG_ON(x) QEMU_BUILD_BUG_MSG(x, "not expecting: " #x)
       |                              ^~~~~~~~~~~~~~~~~~
/builds/qemu-project/qemu/include/qemu/atomic.h:136:5: note: in expansion of macro 
‘QEMU_BUILD_BUG_ON’
   136 |     QEMU_BUILD_BUG_ON(sizeof(*ptr) > ATOMIC_REG_SIZE); \
       |     ^~~~~~~~~~~~~~~~~
../softmmu/dirtylimit.c:120:12: note: in expansion of macro ‘qatomic_read’
   120 |     return qatomic_read(&rates[cpu_index].dirty_rate);
       |            ^~~~~~~~~~~~

Still not fixed on non-linux:

https://gitlab.com/qemu-project/qemu/-/jobs/2631167373

../tests/qtest/migration-test.c:2294:34: error: use of undeclared identifier 
'DIRTYLIMIT_TOLERANCE_RANGE'
         if (rate < (quota_rate + DIRTYLIMIT_TOLERANCE_RANGE)) {
                                  ^
../tests/qtest/migration-test.c:2316:34: error: use of undeclared identifier 
'DIRTYLIMIT_TOLERANCE_RANGE'
         if (rate > (quota_rate + DIRTYLIMIT_TOLERANCE_RANGE)) {
                                  ^

r~
Dr. David Alan Gilbert June 23, 2022, 4:37 p.m. UTC | #2 
* Richard Henderson (richard.henderson@linaro.org) wrote:
> On 6/23/22 02:27, Dr. David Alan Gilbert (git) wrote:
> > From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
> > 
> > The following changes since commit 2b049d2c8dc01de750410f8f1a4eac498c04c723:
> > 
> >    Merge tag 'pull-aspeed-20220622' of https://github.com/legoater/qemu into staging (2022-06-22 07:27:06 -0700)
> > 
> > are available in the Git repository at:
> > 
> >    https://gitlab.com/dagrh/qemu.git tags/pull-migration-20220623a
> > 
> > for you to fetch changes up to 5df0eaff8e24223974bf2516e6dc773695603017:
> > 
> >    tests: Add dirty page rate limit test (2022-06-23 10:18:14 +0100)
> > 
> > ----------------------------------------------------------------
> > Migration pull 2022-06-23
> > 
> > This replaces my pull from yesterday, and Juan's from the day before.
> > 
> > Compared to my pull eysterday:
> >    A one character fix in get_buffer patch spotted by Peter Xu
> > 
> > Compared to Juan's pull:
> >    a) Hopefully fixed non-Linux builds
> >      (Local build test on mingw64 works
> >      Note: the zero-copy capability is now
> >      defined on non-Linux systems)
> >    b) Added Hyman's series - it had been
> >      on queue for a while (sorry for the delay)
> > 
> > In this today migration PULL request:
> > - Dainiel Berrangé - qemufileops cleanup
> > - Leonardo Bras  - cleanups for zero copy
> > - Juan Quintela  - RDMA cleanups
> > - Hyman Huang - per-vcpu dirty ring work
> 
> New build failure on all 32-bit hosts:
> 
> https://gitlab.com/qemu-project/qemu/-/jobs/2631167210
> 
> ../softmmu/dirtylimit.c: In function ‘vcpu_dirty_rate_get’:
> /builds/qemu-project/qemu/include/qemu/compiler.h:74:36: error: static
> assertion failed: "not expecting: sizeof(*&rates[cpu_index].dirty_rate) >
> ATOMIC_REG_SIZE"
>    74 | #define QEMU_BUILD_BUG_MSG(x, msg) _Static_assert(!(x), msg)
>       |                                    ^~~~~~~~~~~~~~
> /builds/qemu-project/qemu/include/qemu/compiler.h:76:30: note: in expansion
> of macro ‘QEMU_BUILD_BUG_MSG’
>    76 | #define QEMU_BUILD_BUG_ON(x) QEMU_BUILD_BUG_MSG(x, "not expecting: " #x)
>       |                              ^~~~~~~~~~~~~~~~~~
> /builds/qemu-project/qemu/include/qemu/atomic.h:136:5: note: in expansion of
> macro ‘QEMU_BUILD_BUG_ON’
>   136 |     QEMU_BUILD_BUG_ON(sizeof(*ptr) > ATOMIC_REG_SIZE); \
>       |     ^~~~~~~~~~~~~~~~~
> ../softmmu/dirtylimit.c:120:12: note: in expansion of macro ‘qatomic_read’
>   120 |     return qatomic_read(&rates[cpu_index].dirty_rate);
>       |            ^~~~~~~~~~~~
> 
> Still not fixed on non-linux:
> 
> https://gitlab.com/qemu-project/qemu/-/jobs/2631167373
> 
> ../tests/qtest/migration-test.c:2294:34: error: use of undeclared identifier
> 'DIRTYLIMIT_TOLERANCE_RANGE'
>         if (rate < (quota_rate + DIRTYLIMIT_TOLERANCE_RANGE)) {
>                                  ^
> ../tests/qtest/migration-test.c:2316:34: error: use of undeclared identifier
> 'DIRTYLIMIT_TOLERANCE_RANGE'
>         if (rate > (quota_rate + DIRTYLIMIT_TOLERANCE_RANGE)) {
>                                  ^

Sorry about that; hmm these are both from the extra series I added that
weren't in Juan's first version.  I'll drop it.

Dave

> r~
>