qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Gerd Hoffmann <kraxel@redhat.com>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Dr. David Alan Gilbert" <dgilbert@redhat.com>,
	Thomas Huth <thuth@redhat.com>,
	Juan Quintela <quintela@redhat.com>, John Snow <jsnow@redhat.com>,
	qemu-devel@nongnu.org, "Michael S. Tsirkin" <mst@redhat.com>
Subject: Re: [Qemu-devel] Deprecating old machine types
Date: Thu, 30 Mar 2017 10:04:29 +0200	[thread overview]
Message-ID: <1490861069.11091.55.camel@redhat.com> (raw)
In-Reply-To: <0eebf55d-aeb9-5426-001d-175b0cbfc60c@redhat.com>

On Mi, 2017-03-29 at 18:58 +0200, Paolo Bonzini wrote:
> 
> On 29/03/2017 18:46, Dr. David Alan Gilbert wrote:
> >> By the way, what about old machine types like "pc-0.10" ? Do we want to
> >> carry them along forever (e.g. since it is not too complicated to
> >> maintain?), or shall we get rid of those one day (e.g. with QEMU 3.0), too?
> > It seems reasonable to slowly deprecate them.
> > I'm just not sure how slowly.
> 
> Some data:
> 
> - dropping 0.12, 0.13 _and_ isapc would let us kill the code for
> rombar=0 (i.e. where QEMU copies ROM BARs directly to low memory).
> 
> - the oldest versions in use are probably 0.12 (CentOS 6) and 1.0
> (Ubuntu 12.04)
> 
> - migration from old versions is broken in various ways from at least
> QEMU 1.2 and older.

Maybe it is useful to discuss this more in terms of code we want drop
instead of version numbers ...

So, here is my (x86-centric) wishlist:

  * Drop support for isapc (already added that to the wiki a few days
    ago).  Also isa-* devices where we have pci variants (isa-vga,
    isa-cirrus-vga, ne2k_isa).

    Any old guests without pci support (such as ms-dos) should be
    happy with "pc" having vga and ide ioports on the standard isa
    locations for backward compatibility and seabios handing all pci
    stuff.  Plugging isa devices (soundblaster for example) into "pc"
    works too.  So I can't see a compelling use case for isapc.  IIRC
    we even had a release with broken isapc with nobody noticing during
    the -rc phase ...

  * Drop support for rombar, and also the code for the linear vbe
    framebuffer magically showing up at 0xe0000000.

  * Drop backward-compatibility with pre-memory-api qemu versions.
    I guess the memory api switch is one of the big reasons why live
    migration to older versions is broken.

The memory api was merged in 1.0, so the above would imply dropping
support for all pc-0.x machine types and isapc.

The 1.0 release was tagged on Dec 1st 2011, more than five years ago.

cheers,
  Gerd

  parent reply	other threads:[~2017-03-30  8:04 UTC|newest]

Thread overview: 52+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-03-08  8:26 [Qemu-devel] What's the next QEMU version after 2.9 ? (or: when is a good point in time to get rid of old interfaces) Thomas Huth
2017-03-08 10:03 ` Peter Maydell
2017-03-08 11:22   ` Thomas Huth
2017-03-08 11:24     ` Daniel P. Berrange
2017-03-09 12:33       ` Markus Armbruster
2017-03-09  2:21     ` Jason Wang
2017-03-09  8:50       ` Thomas Huth
2017-03-09  9:53         ` Jason Wang
2017-03-09 10:20           ` Yongbok Kim
2017-03-10 11:07             ` Jason Wang
2017-03-10 11:22               ` Peter Maydell
2017-03-10 11:53                 ` Thomas Huth
2017-03-10 11:58                   ` Yongbok Kim
2018-04-24 19:45                     ` Philippe Mathieu-Daudé
2017-03-09 10:11         ` [Qemu-devel] external snapshots freezes block device since qemu 2.8 Piotr Rybicki
2017-03-09 12:26           ` Dr. David Alan Gilbert
2017-04-05 22:18             ` John Snow
2017-04-06  9:25               ` Dr. David Alan Gilbert
2017-03-10 14:49           ` Kashyap Chamarthy
2017-03-10 15:44             ` Piotr Rybicki
2017-03-08 10:20 ` [Qemu-devel] What's the next QEMU version after 2.9 ? (or: when is a good point in time to get rid of old interfaces) Daniel P. Berrange
2017-03-08 11:19   ` Gerd Hoffmann
2017-04-12 13:47     ` Marc-André Lureau
2017-04-12 14:10       ` Gerd Hoffmann
2017-03-09 16:00 ` Kevin Wolf
2017-03-24 22:10 ` John Snow
2017-03-27  8:06   ` Thomas Huth
2017-03-27 12:01     ` Stefan Hajnoczi
2017-03-27 12:49       ` Peter Maydell
2017-04-03 14:19         ` Stefan Hajnoczi
2017-04-11 12:53           ` Markus Armbruster
2017-04-18  9:51             ` Stefan Hajnoczi
2017-04-18 11:57               ` Gerd Hoffmann
2017-04-18 17:18                 ` John Snow
2017-04-19  5:53                   ` Markus Armbruster
2017-04-19 10:35                     ` Gerd Hoffmann
2017-04-19 10:15                   ` Gerd Hoffmann
2017-04-19 23:08                     ` John Snow
2017-04-20  5:40                       ` Gerd Hoffmann
2017-04-20 11:10                         ` Philippe Mathieu-Daudé
2017-03-27 12:56       ` [Qemu-devel] Deprecating the -net option (was: What's the next QEMU version after 2.9 ? (or: when is a good point in time to get rid of old interfaces)) Thomas Huth
2017-03-27 13:09         ` [Qemu-devel] Deprecating the -net option Thomas Huth
2017-03-27 15:04           ` Paolo Bonzini
2017-03-27 19:04     ` [Qemu-devel] What's the next QEMU version after 2.9 ? (or: when is a good point in time to get rid of old interfaces) John Snow
2017-03-27 19:46       ` Thomas Huth
2017-03-29 16:21       ` [Qemu-devel] Deprecating old machine types Thomas Huth
2017-03-29 16:46         ` Dr. David Alan Gilbert
2017-03-29 16:54           ` Thomas Huth
2017-03-29 16:58           ` Paolo Bonzini
2017-03-29 21:42             ` Michael S. Tsirkin
2017-03-30  8:04             ` Gerd Hoffmann [this message]
2017-03-28 17:18     ` [Qemu-devel] Deprecating the -drive option is a good point in time to get rid of old interfaces) Kevin Wolf

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=1490861069.11091.55.camel@redhat.com \
    --to=kraxel@redhat.com \
    --cc=dgilbert@redhat.com \
    --cc=jsnow@redhat.com \
    --cc=mst@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=qemu-devel@nongnu.org \
    --cc=quintela@redhat.com \
    --cc=thuth@redhat.com \
    /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).