From: Paolo Bonzini <pbonzini@redhat.com>
To: qemu-devel@nongnu.org
Cc: jsnow@redhat.com, berrange@redhat.com, ehabkost@redhat.com,
imammedo@redhat.com
Subject: [RFC PATCH 00/22] cleanup qemu_init and make sense of command line processing
Date: Wed, 21 Oct 2020 16:56:54 -0400 [thread overview]
Message-ID: <20201021205716.2359430-1-pbonzini@redhat.com> (raw)
This series cleans up qemu_init, grouping together code that does
a similar function. In particular:
* generic initialization of various subsystemd is placed in separate
functions
* code that was randomly placed in vl.c is moved to machine.c
* early options that affect the whole program are grouped
together (-sandbox, -name, -trace, -L)
* pure syntactic sugar options (-mem-prealloc, -watchdog)
are grouped together
* the creation of backends is split into an "early" (before -machine)
and a "late" (after -machine) part
The remaining code that I have not split deals mostly with setting
up the machine object; this includes -smp, -boot, -m, -numa. With
this refactoring, one can (almost) see some clearly delimited phases:
* before the machine is created, only backends can be created
(e.g. early -object processing). In principle more initialization
could be moved before the machine is created: the machine class is
available and it can be used to decide which default devices to create.
* after the machine is created, the accelerator needs to be established
* after the accelerator is established, board setup can
proceed and devices can be created
Apart from the obvious absence of commands like machine-set and accel-set
(and making chardev-add, blockdev-add etc. available in preconfig mode),
the following complications are the blockers for pure-QMP configuration
of machines:
* as usual, there is no hook into board creation code to communicate
the backends for on-board devices. This is because serial_hd,
nd_table etc. don't have a QAPI/QMP representation.
* the current "preconfig" loop does not let you configure enough, since it
ends before devices can be created and the regular monitor loop only
starts after qdev_machine_creation_done. Perhaps x-exit-preconfig would
execute everything up to (and excluding) qdev_machine_creation_done,
while the rest would be deferred to the first "cont" command using a vm
state change notifier.
Paolo
Paolo Bonzini (22):
semihosting: fix order of initialization functions
machine: remove deprecated -machine enforce-config-section option
machine: move UP defaults to class_base_init
machine: move SMP initialization from vl.c
vl: extract validation of -smp to machine.c
vl: remove bogus check
trace: remove argument from trace_init_file
vl: split various early command line options to a separate function
vl: move various initialization routines out of qemu_init
vl: extract qemu_init_subsystems
vl: move prelaunch part of qemu_init to a new function
vl: move bios_name out of softmmu/vl.c
vl: extract various command line validation snippets to a new function
vl: preconfig and loadvm are mutually exclusive
vl: extract various command line desugaring snippets to a new function
vl: create "-net nic -net user" default earlier
vl: load plugins as late as possible
vl: move semihosting command line fallback to qemu_finish_machine_init
vl: extract default devices to separate functions
vl: move CHECKPOINT_INIT after preconfig
vl: separate qemu_create_early_backends
vl: separate qemu_create_late_backends
bsd-user/main.c | 6 +-
docs/system/deprecated.rst | 12 +-
hw/core/machine.c | 68 +-
include/hw/boards.h | 2 +-
include/hw/qdev-core.h | 8 -
include/sysemu/sysemu.h | 1 +
linux-user/main.c | 6 +-
migration/migration.c | 12 +-
qemu-img.c | 6 +-
qemu-io.c | 6 +-
qemu-nbd.c | 6 +-
qemu-options.hx | 8 -
scsi/qemu-pr-helper.c | 6 +-
softmmu/qdev-monitor.c | 6 -
softmmu/vl.c | 1249 +++++++++++++-------------
storage-daemon/qemu-storage-daemon.c | 9 +-
trace/control.c | 10 +-
trace/control.h | 12 +-
18 files changed, 704 insertions(+), 729 deletions(-)
--
2.26.2
next reply other threads:[~2020-10-21 20:59 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-21 20:56 Paolo Bonzini [this message]
2020-10-21 20:56 ` [PATCH 01/22] semihosting: fix order of initialization functions Paolo Bonzini
2020-10-27 11:18 ` Alex Bennée
2020-10-27 13:32 ` Paolo Bonzini
2020-10-21 20:56 ` [PATCH 02/22] machine: remove deprecated -machine enforce-config-section option Paolo Bonzini
2020-10-22 5:09 ` Thomas Huth
2020-10-22 6:54 ` Paolo Bonzini
2020-10-21 20:56 ` [PATCH 03/22] machine: move UP defaults to class_base_init Paolo Bonzini
2020-10-22 5:11 ` Thomas Huth
2020-10-21 20:56 ` [PATCH 04/22] machine: move SMP initialization from vl.c Paolo Bonzini
2020-10-22 7:16 ` Philippe Mathieu-Daudé
2020-10-21 20:56 ` [PATCH 05/22] vl: extract validation of -smp to machine.c Paolo Bonzini
2020-10-21 20:57 ` [PATCH 06/22] vl: remove bogus check Paolo Bonzini
2020-10-21 20:57 ` [PATCH 07/22] trace: remove argument from trace_init_file Paolo Bonzini
2020-10-21 20:57 ` [PATCH 08/22] vl: split various early command line options to a separate function Paolo Bonzini
2020-10-21 20:57 ` [PATCH 09/22] vl: move various initialization routines out of qemu_init Paolo Bonzini
2020-10-21 20:57 ` [PATCH 10/22] vl: extract qemu_init_subsystems Paolo Bonzini
2020-10-21 20:57 ` [PATCH 11/22] vl: move prelaunch part of qemu_init to a new function Paolo Bonzini
2020-10-21 20:57 ` [PATCH 12/22] vl: move bios_name out of softmmu/vl.c Paolo Bonzini
2020-10-22 7:22 ` Philippe Mathieu-Daudé
2020-10-21 20:57 ` [PATCH 13/22] vl: extract various command line validation snippets to a new function Paolo Bonzini
2020-10-21 20:57 ` [PATCH 14/22] vl: preconfig and loadvm are mutually exclusive Paolo Bonzini
2020-10-21 20:57 ` [PATCH 15/22] vl: extract various command line desugaring snippets to a new function Paolo Bonzini
2020-10-21 20:57 ` [PATCH 16/22] vl: create "-net nic -net user" default earlier Paolo Bonzini
2020-10-21 20:57 ` [PATCH 17/22] vl: load plugins as late as possible Paolo Bonzini
2020-10-22 7:28 ` Philippe Mathieu-Daudé
2020-10-21 20:57 ` [PATCH 18/22] vl: move semihosting command line fallback to qemu_finish_machine_init Paolo Bonzini
2020-10-21 20:57 ` [PATCH 19/22] vl: extract default devices to separate functions Paolo Bonzini
2020-10-21 20:57 ` [PATCH 20/22] vl: move CHECKPOINT_INIT after preconfig Paolo Bonzini
2020-10-22 7:29 ` Philippe Mathieu-Daudé
2020-10-21 20:57 ` [PATCH 21/22] vl: separate qemu_create_early_backends Paolo Bonzini
2020-10-21 20:57 ` [PATCH 22/22] vl: separate qemu_create_late_backends Paolo Bonzini
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20201021205716.2359430-1-pbonzini@redhat.com \
--to=pbonzini@redhat.com \
--cc=berrange@redhat.com \
--cc=ehabkost@redhat.com \
--cc=imammedo@redhat.com \
--cc=jsnow@redhat.com \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).