diff mbox series

[V1] migration: export fewer options

Message ID 1708708404-197951-1-git-send-email-steven.sistare@oracle.com (mailing list archive)
State New, archived
Headers show
Series [V1] migration: export fewer options | expand

Commit Message

Steven Sistare Feb. 23, 2024, 5:13 p.m. UTC
A small number of migration options are accessed by migration clients,
but to see them clients must include all of options.h, which is mostly
for migration core code.  migrate_mode() in particular will be needed by
multiple clients.

Refactor the option declarations so clients can see the necessary few via
misc.h, which already exports a portion of the client API.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
---
I suggest that eventually we should define a single file migration/client.h
which exports everything needed by the simpler clients: blockers, notifiers,
options, cpr, and state accessors.
---
---
 hw/vfio/migration.c             |  1 -
 hw/virtio/virtio-balloon.c      |  1 -
 include/migration/misc.h        |  1 +
 include/migration/options-pub.h | 24 ++++++++++++++++++++++++
 migration/options.h             |  6 +-----
 system/dirtylimit.c             |  1 -
 6 files changed, 26 insertions(+), 8 deletions(-)
 create mode 100644 include/migration/options-pub.h

Comments

Peter Xu Feb. 26, 2024, 2:40 a.m. UTC | #1
On Fri, Feb 23, 2024 at 09:13:24AM -0800, Steve Sistare wrote:
> A small number of migration options are accessed by migration clients,
> but to see them clients must include all of options.h, which is mostly
> for migration core code.  migrate_mode() in particular will be needed by
> multiple clients.
> 
> Refactor the option declarations so clients can see the necessary few via
> misc.h, which already exports a portion of the client API.
> 
> Signed-off-by: Steve Sistare <steven.sistare@oracle.com>

Sounds reasonable, queued, thanks.

> ---
> I suggest that eventually we should define a single file migration/client.h
> which exports everything needed by the simpler clients: blockers, notifiers,
> options, cpr, and state accessors.

What's the difference v.s. current migration/misc.h?
Markus Armbruster Feb. 26, 2024, 7:40 a.m. UTC | #2
Steve Sistare <steven.sistare@oracle.com> writes:

> A small number of migration options are accessed by migration clients,
> but to see them clients must include all of options.h, which is mostly
> for migration core code.  migrate_mode() in particular will be needed by
> multiple clients.
>
> Refactor the option declarations so clients can see the necessary few via
> misc.h, which already exports a portion of the client API.
>
> Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
> ---
> I suggest that eventually we should define a single file migration/client.h
> which exports everything needed by the simpler clients: blockers, notifiers,
> options, cpr, and state accessors.
> ---
> ---
>  hw/vfio/migration.c             |  1 -
>  hw/virtio/virtio-balloon.c      |  1 -
>  include/migration/misc.h        |  1 +
>  include/migration/options-pub.h | 24 ++++++++++++++++++++++++
>  migration/options.h             |  6 +-----

Unusual naming.  We have zero headers named -pub.h or -public.h, and
dozens named like -int.h or -internal.h.  Please stick to the existing
convention.

[...]
Steven Sistare Feb. 26, 2024, 2:41 p.m. UTC | #3
On 2/26/2024 2:40 AM, Markus Armbruster wrote:
> Steve Sistare <steven.sistare@oracle.com> writes:
> 
>> A small number of migration options are accessed by migration clients,
>> but to see them clients must include all of options.h, which is mostly
>> for migration core code.  migrate_mode() in particular will be needed by
>> multiple clients.
>>
>> Refactor the option declarations so clients can see the necessary few via
>> misc.h, which already exports a portion of the client API.
>>
>> Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
>> ---
>> I suggest that eventually we should define a single file migration/client.h
>> which exports everything needed by the simpler clients: blockers, notifiers,
>> options, cpr, and state accessors.
>> ---
>> ---
>>  hw/vfio/migration.c             |  1 -
>>  hw/virtio/virtio-balloon.c      |  1 -
>>  include/migration/misc.h        |  1 +
>>  include/migration/options-pub.h | 24 ++++++++++++++++++++++++
>>  migration/options.h             |  6 +-----
> 
> Unusual naming.  We have zero headers named -pub.h or -public.h, and
> dozens named like -int.h or -internal.h.  Please stick to the existing
> convention.

In the spirit of minimizing changes, I went that route to avoid renaming the 
existing migration/options.h and its references:

0 migration/block-dirty-bit 82 #include "options.h"
1 migration/block.c         32 #include "options.h"
2 migration/colo.c          37 #include "options.h"
3 migration/migration-hmp-c 35 #include "options.h"
4 migration/migration.c     68 #include "options.h"
5 migration/multifd-zlib.c  21 #include "options.h"
6 migration/multifd-zstd.c  21 #include "options.h"
7 migration/multifd.c       29 #include "options.h"
8 migration/options.c       30 #include "options.h"
9 migration/postcopy-ram.c  40 #include "options.h"
a migration/qemu-file.c     33 #include "options.h"
b migration/ram-compress.c  37 #include "options.h"
c migration/ram.c           63 #include "options.h"
d migration/rdma.c          40 #include "options.h"
e migration/savevm.c        71 #include "options.h"
f migration/socket.c        30 #include "options.h"
g migration/tls.c           25 #include "options.h"

But I take your point.

Peter, which do you prefer?

  A. rename: migration/options.h -> migration/options-internal.h 
     rename: include/migration/options-pub.h -> include/migration/options.h

  B. rename: include/migration/options.h -> include/migration/client-options.h

I prefer B. If you prefer B, but want a different file name, please choose the
final name.

- Steve
Steven Sistare Feb. 26, 2024, 4:45 p.m. UTC | #4
On 2/25/2024 9:40 PM, Peter Xu wrote:
> On Fri, Feb 23, 2024 at 09:13:24AM -0800, Steve Sistare wrote:
>> A small number of migration options are accessed by migration clients,
>> but to see them clients must include all of options.h, which is mostly
>> for migration core code.  migrate_mode() in particular will be needed by
>> multiple clients.
>>
>> Refactor the option declarations so clients can see the necessary few via
>> misc.h, which already exports a portion of the client API.
>>
>> Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
> 
> Sounds reasonable, queued, thanks.
> 
>> ---
>> I suggest that eventually we should define a single file migration/client.h
>> which exports everything needed by the simpler clients: blockers, notifiers,
>> options, cpr, and state accessors.
> 
> What's the difference v.s. current migration/misc.h?

This file would be sufficient for most clients:

diff --git a/include/migration/client.h b/include/migration/client.h
new file mode 100644
index 0000000..a55e504
--- /dev/null
+++ b/include/migration/client.h
@@ -0,0 +1,6 @@
+#ifndef MIGRATION_CLIENT_H
+#define MIGRATION_CLIENT_H
+#include "migration/misc.h"
+#include "migration/blocker.h"
+#include "migration/client-options.h"
+#endif

Or, we could rename misc.h -> client.h and include blocker.h and client-options.h in it.

I just like the idea that most clients could include a single, obviously named file to
use the most-common exported API.  "misc.h" is not obvious, and not complete.

- Steve
Peter Xu Feb. 27, 2024, 8:08 a.m. UTC | #5
On Mon, Feb 26, 2024 at 09:41:15AM -0500, Steven Sistare wrote:
> On 2/26/2024 2:40 AM, Markus Armbruster wrote:
> > Steve Sistare <steven.sistare@oracle.com> writes:
> > 
> >> A small number of migration options are accessed by migration clients,
> >> but to see them clients must include all of options.h, which is mostly
> >> for migration core code.  migrate_mode() in particular will be needed by
> >> multiple clients.
> >>
> >> Refactor the option declarations so clients can see the necessary few via
> >> misc.h, which already exports a portion of the client API.
> >>
> >> Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
> >> ---
> >> I suggest that eventually we should define a single file migration/client.h
> >> which exports everything needed by the simpler clients: blockers, notifiers,
> >> options, cpr, and state accessors.
> >> ---
> >> ---
> >>  hw/vfio/migration.c             |  1 -
> >>  hw/virtio/virtio-balloon.c      |  1 -
> >>  include/migration/misc.h        |  1 +
> >>  include/migration/options-pub.h | 24 ++++++++++++++++++++++++
> >>  migration/options.h             |  6 +-----
> > 
> > Unusual naming.  We have zero headers named -pub.h or -public.h, and
> > dozens named like -int.h or -internal.h.  Please stick to the existing
> > convention.
> 
> In the spirit of minimizing changes, I went that route to avoid renaming the 
> existing migration/options.h and its references:
> 
> 0 migration/block-dirty-bit 82 #include "options.h"
> 1 migration/block.c         32 #include "options.h"
> 2 migration/colo.c          37 #include "options.h"
> 3 migration/migration-hmp-c 35 #include "options.h"
> 4 migration/migration.c     68 #include "options.h"
> 5 migration/multifd-zlib.c  21 #include "options.h"
> 6 migration/multifd-zstd.c  21 #include "options.h"
> 7 migration/multifd.c       29 #include "options.h"
> 8 migration/options.c       30 #include "options.h"
> 9 migration/postcopy-ram.c  40 #include "options.h"
> a migration/qemu-file.c     33 #include "options.h"
> b migration/ram-compress.c  37 #include "options.h"
> c migration/ram.c           63 #include "options.h"
> d migration/rdma.c          40 #include "options.h"
> e migration/savevm.c        71 #include "options.h"
> f migration/socket.c        30 #include "options.h"
> g migration/tls.c           25 #include "options.h"
> 
> But I take your point.
> 
> Peter, which do you prefer?

From statistics, "-internal.h" wins "-int.h":

$ git grep "\-internal.h" | wc -l
135
$ git grep "\-int.h" | wc -l
3

> 
>   A. rename: migration/options.h -> migration/options-internal.h 
>      rename: include/migration/options-pub.h -> include/migration/options.h
> 
>   B. rename: include/migration/options.h -> include/migration/client-options.h
> 
> I prefer B. If you prefer B, but want a different file name, please choose the
> final name.

Personally I don't have a strong opinion on the name.  I'll see whether
Markus has any comment.

[and of course, I removed this patch from -staging queue to keep the
 discussion going..]
Steven Sistare Feb. 27, 2024, 1:15 p.m. UTC | #6
On 2/27/2024 3:08 AM, Peter Xu wrote:
> On Mon, Feb 26, 2024 at 09:41:15AM -0500, Steven Sistare wrote:
>> On 2/26/2024 2:40 AM, Markus Armbruster wrote:
>>> Steve Sistare <steven.sistare@oracle.com> writes:
>>>
>>>> A small number of migration options are accessed by migration clients,
>>>> but to see them clients must include all of options.h, which is mostly
>>>> for migration core code.  migrate_mode() in particular will be needed by
>>>> multiple clients.
>>>>
>>>> Refactor the option declarations so clients can see the necessary few via
>>>> misc.h, which already exports a portion of the client API.
>>>>
>>>> Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
>>>> ---
>>>> I suggest that eventually we should define a single file migration/client.h
>>>> which exports everything needed by the simpler clients: blockers, notifiers,
>>>> options, cpr, and state accessors.
>>>> ---
>>>> ---
>>>>  hw/vfio/migration.c             |  1 -
>>>>  hw/virtio/virtio-balloon.c      |  1 -
>>>>  include/migration/misc.h        |  1 +
>>>>  include/migration/options-pub.h | 24 ++++++++++++++++++++++++
>>>>  migration/options.h             |  6 +-----
>>>
>>> Unusual naming.  We have zero headers named -pub.h or -public.h, and
>>> dozens named like -int.h or -internal.h.  Please stick to the existing
>>> convention.
>>
>> In the spirit of minimizing changes, I went that route to avoid renaming the 
>> existing migration/options.h and its references:
>>
>> 0 migration/block-dirty-bit 82 #include "options.h"
>> 1 migration/block.c         32 #include "options.h"
>> 2 migration/colo.c          37 #include "options.h"
>> 3 migration/migration-hmp-c 35 #include "options.h"
>> 4 migration/migration.c     68 #include "options.h"
>> 5 migration/multifd-zlib.c  21 #include "options.h"
>> 6 migration/multifd-zstd.c  21 #include "options.h"
>> 7 migration/multifd.c       29 #include "options.h"
>> 8 migration/options.c       30 #include "options.h"
>> 9 migration/postcopy-ram.c  40 #include "options.h"
>> a migration/qemu-file.c     33 #include "options.h"
>> b migration/ram-compress.c  37 #include "options.h"
>> c migration/ram.c           63 #include "options.h"
>> d migration/rdma.c          40 #include "options.h"
>> e migration/savevm.c        71 #include "options.h"
>> f migration/socket.c        30 #include "options.h"
>> g migration/tls.c           25 #include "options.h"
>>
>> But I take your point.
>>
>> Peter, which do you prefer?
> 
> From statistics, "-internal.h" wins "-int.h":
> 
> $ git grep "\-internal.h" | wc -l
> 135
> $ git grep "\-int.h" | wc -l
> 3

Yes, I did the same search to choose the name for option A below.
But in option B, I keep the existing name for the private file,
and choose a new name for the public file.  There is no suffix
in use in qemu to denote a public file; we just use names indicating
the functionality, so I chose client-options.h.

Let's use client-options.h and move on.  I am preparing another cleanup
series that I think you will like.

- Steve

>>   A. rename: migration/options.h -> migration/options-internal.h 
>>      rename: include/migration/options-pub.h -> include/migration/options.h
>>
>>   B. rename: include/migration/options.h -> include/migration/client-options.h
>>
>> I prefer B. If you prefer B, but want a different file name, please choose the
>> final name.
> 
> Personally I don't have a strong opinion on the name.  I'll see whether
> Markus has any comment.
> 
> [and of course, I removed this patch from -staging queue to keep the
>  discussion going..]
>
diff mbox series

Patch

diff --git a/hw/vfio/migration.c b/hw/vfio/migration.c
index 86d8c9e..f1f6878 100644
--- a/hw/vfio/migration.c
+++ b/hw/vfio/migration.c
@@ -18,7 +18,6 @@ 
 #include "sysemu/runstate.h"
 #include "hw/vfio/vfio-common.h"
 #include "migration/migration.h"
-#include "migration/options.h"
 #include "migration/savevm.h"
 #include "migration/vmstate.h"
 #include "migration/qemu-file.h"
diff --git a/hw/virtio/virtio-balloon.c b/hw/virtio/virtio-balloon.c
index d004cf2..74d419f 100644
--- a/hw/virtio/virtio-balloon.c
+++ b/hw/virtio/virtio-balloon.c
@@ -32,7 +32,6 @@ 
 #include "qemu/error-report.h"
 #include "migration/misc.h"
 #include "migration/migration.h"
-#include "migration/options.h"
 
 #include "hw/virtio/virtio-bus.h"
 #include "hw/virtio/virtio-access.h"
diff --git a/include/migration/misc.h b/include/migration/misc.h
index 916b65f..627726c 100644
--- a/include/migration/misc.h
+++ b/include/migration/misc.h
@@ -17,6 +17,7 @@ 
 #include "qemu/notify.h"
 #include "qapi/qapi-types-migration.h"
 #include "qapi/qapi-types-net.h"
+#include "migration/options-pub.h"
 
 /* migration/ram.c */
 
diff --git a/include/migration/options-pub.h b/include/migration/options-pub.h
new file mode 100644
index 0000000..54bd678
--- /dev/null
+++ b/include/migration/options-pub.h
@@ -0,0 +1,24 @@ 
+/*
+ * QEMU public migration capabilities
+ *
+ * Copyright (c) 2012-2023 Red Hat Inc
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef QEMU_MIGRATION_OPTIONS_PUB_H
+#define QEMU_MIGRATION_OPTIONS_PUB_H
+
+/* capabilities */
+
+bool migrate_background_snapshot(void);
+bool migrate_dirty_limit(void);
+bool migrate_postcopy_ram(void);
+bool migrate_switchover_ack(void);
+
+/* parameters */
+
+MigMode migrate_mode(void);
+
+#endif
diff --git a/migration/options.h b/migration/options.h
index 246c160..84c08e1 100644
--- a/migration/options.h
+++ b/migration/options.h
@@ -16,6 +16,7 @@ 
 
 #include "hw/qdev-properties.h"
 #include "hw/qdev-properties-system.h"
+#include "migration/options-pub.h"
 
 /* migration properties */
 
@@ -24,12 +25,10 @@  extern Property migration_properties[];
 /* capabilities */
 
 bool migrate_auto_converge(void);
-bool migrate_background_snapshot(void);
 bool migrate_block(void);
 bool migrate_colo(void);
 bool migrate_compress(void);
 bool migrate_dirty_bitmaps(void);
-bool migrate_dirty_limit(void);
 bool migrate_events(void);
 bool migrate_ignore_shared(void);
 bool migrate_late_block_activate(void);
@@ -37,11 +36,9 @@  bool migrate_multifd(void);
 bool migrate_pause_before_switchover(void);
 bool migrate_postcopy_blocktime(void);
 bool migrate_postcopy_preempt(void);
-bool migrate_postcopy_ram(void);
 bool migrate_rdma_pin_all(void);
 bool migrate_release_ram(void);
 bool migrate_return_path(void);
-bool migrate_switchover_ack(void);
 bool migrate_validate_uuid(void);
 bool migrate_xbzrle(void);
 bool migrate_zero_blocks(void);
@@ -83,7 +80,6 @@  uint8_t migrate_max_cpu_throttle(void);
 uint64_t migrate_max_bandwidth(void);
 uint64_t migrate_avail_switchover_bandwidth(void);
 uint64_t migrate_max_postcopy_bandwidth(void);
-MigMode migrate_mode(void);
 int migrate_multifd_channels(void);
 MultiFDCompression migrate_multifd_compression(void);
 int migrate_multifd_zlib_level(void);
diff --git a/system/dirtylimit.c b/system/dirtylimit.c
index 495c7a7..696eaab 100644
--- a/system/dirtylimit.c
+++ b/system/dirtylimit.c
@@ -26,7 +26,6 @@ 
 #include "trace.h"
 #include "migration/misc.h"
 #include "migration/migration.h"
-#include "migration/options.h"
 
 /*
  * Dirtylimit stop working if dirty page rate error