| Message ID | 1528688464-17289-1-git-send-email-thuth@redhat.com |
|---|---|
| State | New |
| Headers | show |
| Series | hw/i386: Deprecate the machine types pc-0.10 and pc-0.11 | expand |
On Mon, Jun 11, 2018 at 05:41:04AM +0200, Thomas Huth wrote: > The oldest machine type which is still used in a maintained distribution > is a pc-0.12 based machine type in RHEL6, so everything that is older > than pc-0.12 should not be used anymore. Thus let's deprecate pc-0.10 > and pc-0.11 so that we can finally remove them in a future release. > > Signed-off-by: Thomas Huth <thuth@redhat.com> > --- > This is based on a patch that I already sent in 2017. But back then, we > were still in progress of discussing our deprecation policies (e.g. auto- > matic deprecation for old machine types), and there was no clear consensus > whether we should deprecate 0.10 - 0.11, all 0.x or even up to version 1.2. > After some iterations and too much discussion, I've forgotten about this > patch. Anyway, I think we agreed that at least 0.10 and 0.11 can certainly > be removed nowadays, so let's finally get at least those two machine types > marked as deprecated! If that works fine and we will finally have removed > these two types in v3.2, we can resume the discussion about newer machine > types afterwards. Thanks! > > hw/i386/pc_piix.c | 2 ++ > include/hw/boards.h | 1 + > qemu-doc.texi | 5 +++++ > vl.c | 9 +++++++-- > 4 files changed, 15 insertions(+), 2 deletions(-) > > diff --git a/hw/i386/pc_piix.c b/hw/i386/pc_piix.c > index 3d81136..fa61dc3 100644 > --- a/hw/i386/pc_piix.c > +++ b/hw/i386/pc_piix.c > @@ -955,6 +955,8 @@ static void pc_i440fx_0_11_machine_options(MachineClass *m) > { > pc_i440fx_0_12_machine_options(m); > m->hw_version = "0.11"; > + m->deprecation_msg = "Old and unsupported machine version, " > + "use a newer machine type instead."; Sounds simple enough to me, but see comment about QMP below. > SET_MACHINE_COMPAT(m, PC_COMPAT_0_11); > } > > diff --git a/include/hw/boards.h b/include/hw/boards.h > index ef7457f..e1bd25f 100644 > --- a/include/hw/boards.h > +++ b/include/hw/boards.h > @@ -166,6 +166,7 @@ struct MachineClass { > char *name; > const char *alias; > const char *desc; > + const char *deprecation_msg; Let's document guidelines for deprecation_msg to avoid the inconsistencies we saw in previous proposals? > > void (*init)(MachineState *state); > void (*reset)(void); > diff --git a/qemu-doc.texi b/qemu-doc.texi > index 2effe66..2a597d6 100644 > --- a/qemu-doc.texi > +++ b/qemu-doc.texi > @@ -2965,6 +2965,11 @@ support page sizes < 4096 any longer. > > @section System emulator machines > > +@subsection pc-0.10 and pc-0.11 (since 3.0) > + > +These machine types are very old and likely can not be used for life-migration ^^^^ "live" I never know if the right way to spell it is "live-migration" or "live migration". > +from old QEMU versions anymore. A newer machine type should be used instead. > + > @section Block device options > > @subsection "backing": "" (since 2.12.0) > diff --git a/vl.c b/vl.c > index 0603171..096814c 100644 > --- a/vl.c > +++ b/vl.c > @@ -2560,8 +2560,9 @@ static gint machine_class_cmp(gconstpointer a, gconstpointer b) > if (mc->alias) { > printf("%-20s %s (alias of %s)\n", mc->alias, mc->desc, mc->name); > } > - printf("%-20s %s%s\n", mc->name, mc->desc, > - mc->is_default ? " (default)" : ""); > + printf("%-20s %s%s%s\n", mc->name, mc->desc, > + mc->is_default ? " (default)" : "", > + mc->deprecation_msg ? " (deprecated)" : ""); > } > } > > @@ -3952,6 +3953,10 @@ int main(int argc, char **argv, char **envp) > } > > machine_class = select_machine(); > + if (machine_class->deprecation_msg) { > + error_report("Machine type '%s' is deprecated: %s", > + machine_class->name, machine_class->deprecation_msg); > + } Do you plan to add this info to 'query-machines' QMP command? If we do that, maybe we should represent the common "this machine type is too old, but there's a new version" case in a more machine-friendly way? Maybe a 'deprecation_reason' enum would be better than a 'deprecation_msg' field? (Note that I don't think any discussions about the QMP interface should block this patch from being merged. We can deprecate the machines first, and decide about QMP later.) > > set_memory_options(&ram_slots, &maxram_size, machine_class); > > -- > 1.8.3.1 >
On 12.06.2018 00:18, Eduardo Habkost wrote: > On Mon, Jun 11, 2018 at 05:41:04AM +0200, Thomas Huth wrote: >> The oldest machine type which is still used in a maintained distribution >> is a pc-0.12 based machine type in RHEL6, so everything that is older >> than pc-0.12 should not be used anymore. Thus let's deprecate pc-0.10 >> and pc-0.11 so that we can finally remove them in a future release. [...] >> @@ -3952,6 +3953,10 @@ int main(int argc, char **argv, char **envp) >> } >> >> machine_class = select_machine(); >> + if (machine_class->deprecation_msg) { >> + error_report("Machine type '%s' is deprecated: %s", >> + machine_class->name, machine_class->deprecation_msg); >> + } > > Do you plan to add this info to 'query-machines' QMP command? No, I'm not planning to add this. We'd need a request from upper layers (i.e. libvirt) for this first, otherwise it's just a dead interface that nobody is using. Thomas
(CCing Markus and libvir-list) On Wed, Jun 20, 2018 at 08:40:38PM +0200, Thomas Huth wrote: > On 12.06.2018 00:18, Eduardo Habkost wrote: > > On Mon, Jun 11, 2018 at 05:41:04AM +0200, Thomas Huth wrote: > >> The oldest machine type which is still used in a maintained distribution > >> is a pc-0.12 based machine type in RHEL6, so everything that is older > >> than pc-0.12 should not be used anymore. Thus let's deprecate pc-0.10 > >> and pc-0.11 so that we can finally remove them in a future release. > [...] > >> @@ -3952,6 +3953,10 @@ int main(int argc, char **argv, char **envp) > >> } > >> > >> machine_class = select_machine(); > >> + if (machine_class->deprecation_msg) { > >> + error_report("Machine type '%s' is deprecated: %s", > >> + machine_class->name, machine_class->deprecation_msg); > >> + } > > > > Do you plan to add this info to 'query-machines' QMP command? > > No, I'm not planning to add this. We'd need a request from upper layers > (i.e. libvirt) for this first, otherwise it's just a dead interface that > nobody is using. I believe that useful information being available only through stderr is at least as bad as being available only through HMP. Should we extend QMP more proactively in cases like this, too? (In either case, I don't think this should block your series)
Eduardo Habkost <ehabkost@redhat.com> writes: > On Mon, Jun 11, 2018 at 05:41:04AM +0200, Thomas Huth wrote: >> The oldest machine type which is still used in a maintained distribution >> is a pc-0.12 based machine type in RHEL6, so everything that is older >> than pc-0.12 should not be used anymore. Thus let's deprecate pc-0.10 >> and pc-0.11 so that we can finally remove them in a future release. >> >> Signed-off-by: Thomas Huth <thuth@redhat.com> >> --- >> This is based on a patch that I already sent in 2017. But back then, we >> were still in progress of discussing our deprecation policies (e.g. auto- >> matic deprecation for old machine types), and there was no clear consensus >> whether we should deprecate 0.10 - 0.11, all 0.x or even up to version 1.2. >> After some iterations and too much discussion, I've forgotten about this >> patch. Anyway, I think we agreed that at least 0.10 and 0.11 can certainly >> be removed nowadays, so let's finally get at least those two machine types >> marked as deprecated! If that works fine and we will finally have removed >> these two types in v3.2, we can resume the discussion about newer machine >> types afterwards. > > Thanks! > > >> >> hw/i386/pc_piix.c | 2 ++ >> include/hw/boards.h | 1 + >> qemu-doc.texi | 5 +++++ >> vl.c | 9 +++++++-- >> 4 files changed, 15 insertions(+), 2 deletions(-) >> >> diff --git a/hw/i386/pc_piix.c b/hw/i386/pc_piix.c >> index 3d81136..fa61dc3 100644 >> --- a/hw/i386/pc_piix.c >> +++ b/hw/i386/pc_piix.c >> @@ -955,6 +955,8 @@ static void pc_i440fx_0_11_machine_options(MachineClass *m) >> { >> pc_i440fx_0_12_machine_options(m); >> m->hw_version = "0.11"; >> + m->deprecation_msg = "Old and unsupported machine version, " >> + "use a newer machine type instead."; > > Sounds simple enough to me, but see comment about QMP below. > > >> SET_MACHINE_COMPAT(m, PC_COMPAT_0_11); >> } >> [...] >> diff --git a/vl.c b/vl.c >> index 0603171..096814c 100644 >> --- a/vl.c >> +++ b/vl.c >> @@ -2560,8 +2560,9 @@ static gint machine_class_cmp(gconstpointer a, gconstpointer b) >> if (mc->alias) { >> printf("%-20s %s (alias of %s)\n", mc->alias, mc->desc, mc->name); >> } >> - printf("%-20s %s%s\n", mc->name, mc->desc, >> - mc->is_default ? " (default)" : ""); >> + printf("%-20s %s%s%s\n", mc->name, mc->desc, >> + mc->is_default ? " (default)" : "", >> + mc->deprecation_msg ? " (deprecated)" : ""); >> } >> } >> >> @@ -3952,6 +3953,10 @@ int main(int argc, char **argv, char **envp) >> } >> >> machine_class = select_machine(); >> + if (machine_class->deprecation_msg) { >> + error_report("Machine type '%s' is deprecated: %s", >> + machine_class->name, machine_class->deprecation_msg); >> + } > > Do you plan to add this info to 'query-machines' QMP command? > > If we do that, maybe we should represent the common "this machine > type is too old, but there's a new version" case in a more > machine-friendly way? Maybe a 'deprecation_reason' enum would be > better than a 'deprecation_msg' field? QMP needs a generic mechanism to communicate "FOO is deprecated, use BAR instead". > (Note that I don't think any discussions about the QMP interface > should block this patch from being merged. We can deprecate the > machines first, and decide about QMP later.) Yes. >> >> set_memory_options(&ram_slots, &maxram_size, machine_class); >> >> -- >> 1.8.3.1 >>
diff --git a/hw/i386/pc_piix.c b/hw/i386/pc_piix.c index 3d81136..fa61dc3 100644 --- a/hw/i386/pc_piix.c +++ b/hw/i386/pc_piix.c @@ -955,6 +955,8 @@ static void pc_i440fx_0_11_machine_options(MachineClass *m) { pc_i440fx_0_12_machine_options(m); m->hw_version = "0.11"; + m->deprecation_msg = "Old and unsupported machine version, " + "use a newer machine type instead."; SET_MACHINE_COMPAT(m, PC_COMPAT_0_11); } diff --git a/include/hw/boards.h b/include/hw/boards.h index ef7457f..e1bd25f 100644 --- a/include/hw/boards.h +++ b/include/hw/boards.h @@ -166,6 +166,7 @@ struct MachineClass { char *name; const char *alias; const char *desc; + const char *deprecation_msg; void (*init)(MachineState *state); void (*reset)(void); diff --git a/qemu-doc.texi b/qemu-doc.texi index 2effe66..2a597d6 100644 --- a/qemu-doc.texi +++ b/qemu-doc.texi @@ -2965,6 +2965,11 @@ support page sizes < 4096 any longer. @section System emulator machines +@subsection pc-0.10 and pc-0.11 (since 3.0) + +These machine types are very old and likely can not be used for life-migration +from old QEMU versions anymore. A newer machine type should be used instead. + @section Block device options @subsection "backing": "" (since 2.12.0) diff --git a/vl.c b/vl.c index 0603171..096814c 100644 --- a/vl.c +++ b/vl.c @@ -2560,8 +2560,9 @@ static gint machine_class_cmp(gconstpointer a, gconstpointer b) if (mc->alias) { printf("%-20s %s (alias of %s)\n", mc->alias, mc->desc, mc->name); } - printf("%-20s %s%s\n", mc->name, mc->desc, - mc->is_default ? " (default)" : ""); + printf("%-20s %s%s%s\n", mc->name, mc->desc, + mc->is_default ? " (default)" : "", + mc->deprecation_msg ? " (deprecated)" : ""); } } @@ -3952,6 +3953,10 @@ int main(int argc, char **argv, char **envp) } machine_class = select_machine(); + if (machine_class->deprecation_msg) { + error_report("Machine type '%s' is deprecated: %s", + machine_class->name, machine_class->deprecation_msg); + } set_memory_options(&ram_slots, &maxram_size, machine_class);
The oldest machine type which is still used in a maintained distribution is a pc-0.12 based machine type in RHEL6, so everything that is older than pc-0.12 should not be used anymore. Thus let's deprecate pc-0.10 and pc-0.11 so that we can finally remove them in a future release. Signed-off-by: Thomas Huth <thuth@redhat.com> --- This is based on a patch that I already sent in 2017. But back then, we were still in progress of discussing our deprecation policies (e.g. auto- matic deprecation for old machine types), and there was no clear consensus whether we should deprecate 0.10 - 0.11, all 0.x or even up to version 1.2. After some iterations and too much discussion, I've forgotten about this patch. Anyway, I think we agreed that at least 0.10 and 0.11 can certainly be removed nowadays, so let's finally get at least those two machine types marked as deprecated! If that works fine and we will finally have removed these two types in v3.2, we can resume the discussion about newer machine types afterwards. hw/i386/pc_piix.c | 2 ++ include/hw/boards.h | 1 + qemu-doc.texi | 5 +++++ vl.c | 9 +++++++-- 4 files changed, 15 insertions(+), 2 deletions(-)