Replace DeviceState dependency with VMStateIf on vmstate API.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Acked-by: Halil Pasic <pasic@linux.ibm.com>
Leaking the drc_name while preparing the DT properties.
Fixing that.
Also, remove the const qualifier from spapr_drc_name().
Signed-off-by: Shivaprasad G Bhat <sbhat@linux.ibm.com>
Message-Id: <156335159028.82682.5404622104535818162.stgit@lep8c.aus.stglabs.ibm.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
In my "build everything" tree, changing hw/qdev-properties.h triggers
a recompile of some 2700 out of 6600 objects (not counting tests and
objects that don't depend on qemu/osdep.h).
Many places including hw/qdev-properties.h (directly or via hw/qdev.h)
actually need only hw/qdev-core.h. Include hw/qdev-core.h there
instead.
hw/qdev.h is actually pointless: all it does is include hw/qdev-core.h
and hw/qdev-properties.h, which in turn includes hw/qdev-core.h.
Replace the remaining uses of hw/qdev.h by hw/qdev-properties.h.
While there, delete a few superfluous inclusions of hw/qdev-core.h.
Touching hw/qdev-properties.h now recompiles some 1200 objects.
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Daniel P. Berrangé" <berrange@redhat.com>
Cc: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20190812052359.30071-22-armbru@redhat.com>
In my "build everything" tree, changing migration/vmstate.h triggers a
recompile of some 2700 out of 6600 objects (not counting tests and
objects that don't depend on qemu/osdep.h).
hw/hw.h supposedly includes it for convenience. Several other headers
include it just to get VMStateDescription. The previous commit made
that unnecessary.
Include migration/vmstate.h only where it's still needed. Touching it
now recompiles only some 1600 objects.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20190812052359.30071-16-armbru@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
In my "build everything" tree, changing sysemu/reset.h triggers a
recompile of some 2600 out of 6600 objects (not counting tests and
objects that don't depend on qemu/osdep.h).
The main culprit is hw/hw.h, which supposedly includes it for
convenience.
Include sysemu/reset.h only where it's needed. Touching it now
recompiles less than 200 objects.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Tested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20190812052359.30071-9-armbru@redhat.com>
This makes some minor cleanups to spapr_drc_populate_dt(), renaming it to
the shorter and more idiomatic spapr_dt_drc() along the way.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
The qemu coding standard is to use CamelCase for type and structure names,
and the pseries code follows that... sort of. There are quite a lot of
places where we bend the rules in order to preserve the capitalization of
internal acronyms like "PHB", "TCE", "DIMM" and most commonly "sPAPR".
That was a bad idea - it frequently leads to names ending up with hard to
read clusters of capital letters, and means they don't catch the eye as
type identifiers, which is kind of the point of the CamelCase convention in
the first place.
In short, keeping type identifiers look like CamelCase is more important
than preserving standard capitalization of internal "words". So, this
patch renames a heap of spapr internal type names to a more standard
CamelCase.
In addition to case changes, we also make some other identifier renames:
VIOsPAPR* -> SpaprVio*
The reverse word ordering was only ever used to mitigate the capital
cluster, so revert to the natural ordering.
VIOsPAPRVTYDevice -> SpaprVioVty
VIOsPAPRVLANDevice -> SpaprVioVlan
Brevity, since the "Device" didn't add useful information
sPAPRDRConnector -> SpaprDrc
sPAPRDRConnectorClass -> SpaprDrcClass
Brevity, and makes it clearer this is the same thing as a "DRC"
mentioned in many other places in the code
This is 100% a mechanical search-and-replace patch. It will, however,
conflict with essentially any and all outstanding patches touching the
spapr code.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Hotplugging PHBs is a machine-level operation, but PHBs reside on the
main system bus, so we register spapr machine as the handler for the
main system bus.
Provide the usual pre-plug, plug and unplug-request handlers.
Move the checking of the PHB index to the pre-plug handler. It is okay
to do that and assert in the realize function because the pre-plug
handler is always called, even for the oldest machine types we support.
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
(Fixed interrupt controller phandle in "interrupt-map" and
TCE table size in "ibm,dma-window" FDT fragment, Greg Kurz)
Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <155059672926.1466090.13612804072190051439.stgit@bahia.lab.toulouse-stg.fr.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <155059670389.1466090.10015601248906623076.stgit@bahia.lab.toulouse-stg.fr.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
All DRC subtypes have been converted to generate the FDT fragment at
configure connector time instead of attach time. The fdt and fdt_offset
arguments of spapr_drc_attach() aren't needed anymore. Drop them and
make the implementation of the dt_populate() method mandatory.
Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <155059667853.1466090.16527852453054217565.stgit@bahia.lab.toulouse-stg.fr.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <155059667346.1466090.326696113231137772.stgit@bahia.lab.toulouse-stg.fr.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <155059666839.1466090.3833376527523126752.stgit@bahia.lab.toulouse-stg.fr.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <155059666331.1466090.6766540766297333313.stgit@bahia.lab.toulouse-stg.fr.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The current logic is to provide the FDT fragment when attaching a device
to a DRC. This works perfectly fine for our current hotplug support, but
soon we will add support for PHB hotplug which has some constraints, that
CPU, PCI and LMB devices don't seem to have.
The first constraint is that the "ibm,dma-window" property of the PHB
node requires the IOMMU to be configured, ie, spapr_tce_table_enable()
has been called, which happens during PHB reset. It is okay in the case
of hotplug since the device is reset before the hotplug handler is
called. On the contrary with coldplug, the hotplug handler is called
first and device is only reset during the initial system reset. Trying
to create the FDT fragment on the hotplug path in this case, would
result in somthing like this:
ibm,dma-window = < 0x80000000 0x00 0x00 0x00 0x00 >;
This will cause linux in the guest to panic, by simply removing and
re-adding the PHB using the drmgr command:
page = alloc_pages_node(nid, GFP_KERNEL, get_order(sz));
if (!page)
panic("iommu_init_table: Can't allocate %ld bytes\n", sz);
The second and maybe more problematic constraint is that the
"interrupt-map" property needs to reference the interrupt controller
node using the very same phandle that SLOF has already exposed to the
guest. QEMU requires SLOF to call the private KVMPPC_H_UPDATE_DT hcall
at some point to know about this phandle. With the latest QEMU and SLOF,
this happens when SLOF gets quiesced. This means that if the PHB gets
hotplugged after CAS but before SLOF quiesce, then we're sure that the
phandle is not known when the hotplug handler is called.
The FDT is only needed when the guest first invokes RTAS to configure
the connector actually, long after SLOF quiesce. Let's postpone the
creation of FDT fragments for PHBs to rtas_ibm_configure_connector().
Since we only need this for PHBs, introduce a new method in the base
DRC class for that. DRC subtypes will be converted to use it in
subsequent patches.
Allow spapr_drc_attach() to be passed a NULL fdt argument if the method
is available. When all DRC subtypes have been converted, the fdt argument
will eventually disappear.
Signed-off-by: Greg Kurz <groug@kaod.org>
Message-Id: <155059665823.1466090.18358845122627355537.stgit@bahia.lab.toulouse-stg.fr.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Use error_report() + abort() instead of error_setg(&error_abort),
as suggested by the "qapi/error.h" documentation:
Please don't error_setg(&error_fatal, ...), use error_report() and
exit(), because that's more obvious.
Likewise, don't error_setg(&error_abort, ...), use assert().
Use abort() instead of the suggested assert() because the error message
already got displayed.
Suggested-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Now that we can safely call QOBJECT() on QObject * as well as its
subtypes, we can have macros qobject_ref() / qobject_unref() that work
everywhere instead of having to use QINCREF() / QDECREF() for QObject
and qobject_incref() / qobject_decref() for its subtypes.
The replacement is mechanical, except I broke a long line, and added a
cast in monitor_qmp_cleanup_req_queue_locked(). Unlike
qobject_decref(), qobject_unref() doesn't accept void *.
Note that the new macros evaluate their argument exactly once, thus no
need to shout them.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20180419150145.24795-4-marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
[Rebased, semantic conflict resolved, commit message improved]
Signed-off-by: Markus Armbruster <armbru@redhat.com>
The macro expansions of qdict_put_TYPE() and qlist_append_TYPE() need
qbool.h, qnull.h, qnum.h and qstring.h to compile. We include qnull.h
and qnum.h in the headers, but not qbool.h and qstring.h. Works,
because we include those wherever the macros get used.
Open-coding these helpers is of dubious value. Turn them into
functions and drop the includes from the headers.
This cleanup makes the number of objects depending on qapi/qmp/qnum.h
from 4551 (out of 4743) to 46 in my "build everything" tree. For
qapi/qmp/qnull.h, the number drops from 4552 to 21.
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20180201111846.21846-10-armbru@redhat.com>
A DRC with a pending unplug request releases its associated device at
machine reset time.
In the case of LMB, when all DRCs for a DIMM device have been reset,
the DIMM gets unplugged, causing guest memory to disappear. This may
be very confusing for anything still using this memory.
This is exactly what happens with vhost backends, and QEMU aborts
with:
qemu-system-ppc64: used ring relocated for ring 2
qemu-system-ppc64: qemu/hw/virtio/vhost.c:649: vhost_commit: Assertion
`r >= 0' failed.
The issue is that each DRC registers a QEMU reset handler, and we
don't control the order in which these handlers are called (ie,
a LMB DRC will unplug a DIMM before the virtio device using the
memory on this DIMM could stop its vhost backend).
To avoid such situations, let's reset DRCs after all devices
have been reset.
Reported-by: Mallesh N. Koti <mallesh@linux.vnet.ibm.com>
Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Daniel Henrique Barboza <danielhb@linux.vnet.ibm.com>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
DRC objects attach themselves to an owner as a child
property. unref afterward to allow them to be finalized
when their owner is finalized.
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Greg Kurz <groug@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
When hot-unplugging a PHB, all its PCI DRC connectors get unrealized. This
patch adds an unrealize method to the physical DRC class, in order to undo
registrations performed in realize_physical().
Signed-off-by: Greg Kurz <groug@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Passing a stack allocated buffer of arbitrary length to snprintf()
without checking the return value can cause the resultant strings
to be silently truncated.
Signed-off-by: Greg Kurz <groug@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This patch is a follow up on the discussions made in patch
"hw/ppc: disable hotplug before CAS is completed" that can be
found at [1].
At this moment, we do not support CPU/memory hotplug in early
boot stages, before CAS. When a hotplug occurs, the event is logged
in an internal RTAS event log queue and an IRQ pulse is fired. In
regular conditions, the guest handles the interrupt by executing
check_exception, fetching the generated hotplug event and enabling
the device for use.
In early boot, this IRQ isn't caught (SLOF does not handle hotplug
events), leaving the event in the rtas event log queue. If the guest
executes check_exception due to another hotplug event, the re-assertion
of the IRQ ends up de-queuing the first hotplug event as well. In short,
a device hotplugged before CAS is considered coldplugged by SLOF.
This leads to device misbehavior and, in some cases, guest kernel
Ooops when trying to unplug the device.
A proper fix would be to turn every device hotplugged before CAS
as a colplugged device. This is not trivial to do with the current
code base though - the FDT is written in the guest memory at
ppc_spapr_reset and can't be retrieved without adding extra state
(fdt_size for example) that will need to managed and migrated. Adding
the hotplugged DT in the middle of CAS negotiation via the updated DT
tree works with CPU devs, but panics the guest kernel at boot. Additional
analysis would be necessary for LMBs and PCI devices. There are
questions to be made in QEMU/SLOF/kernel level about how we can make
this change in a sustainable way.
With Linux guests, a fix would be the kernel executing check_exception
at boot time, de-queueing the events that happened in early boot and
processing them. However, even if/when the newer kernels start
fetching these events at boot time, we need to take care of older
kernels that won't be doing that.
This patch works around the situation by issuing a CAS reset if a hotplugged
device is detected during CAS:
- the DRC conditions that warrant a CAS reset is the same as those that
triggers a DRC migration - the DRC must have a device attached and
the DRC state is not equal to its ready_state. With that in mind, this
patch makes use of 'spapr_drc_needed' to determine if a CAS reset
is needed.
- In the middle of CAS negotiations, the function
'spapr_hotplugged_dev_before_cas' goes through all the DRCs to see
if there are any DRC that requires a reset, using spapr_drc_needed. If
that happens, returns '1' in 'spapr_h_cas_compose_response' which will set
spapr->cas_reboot to true, causing the machine to reboot.
No changes are made for coldplug devices.
[1] http://lists.nongnu.org/archive/html/qemu-devel/2017-08/msg02855.html
Signed-off-by: Daniel Henrique Barboza <danielhb@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This patch makes a small fix in 'spapr_drc_needed' to change how we detect
if a DRC has a device attached. Previously it used dr_entity_sense for this,
which works for physical DRCs.
However, for logical DRCs, it didn't cover the case where a logical DRC has
a drc->dev but the state is LOGICAL_UNUSABLE (e.g. a hotplugged CPU before
CAS). In this case, the dr_entity_sense of this DRC returns UNUSABLE and the
code was considering that there were no dev attached, making spapr_drc_needed
return 'false' when in fact we would like to migrate the DRC.
Changing it to check for drc->dev instead works for all DRC types.
Signed-off-by: Daniel Henrique Barboza <danielhb@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
In case of in-kernel memory hot unplug, when the guest is not able
to remove all the LMBs that are requested for removal, it will add back
any LMBs that have been successfully removed. The DR Connectors of
these LMBs wouldn't have been unconfigured and hence the addition of
these LMBs will result in configure-connector call being issued on
LMB DR connectors that are already in configured state. Such
configure-connector calls will fail resulting in a DIMM which is
partially unplugged.
This however worked till recently before we overhauled the DRC
implementation in QEMU. Commit 9d4c0f4f0a: "spapr: Consolidate
DRC state variables" is the first commit where this problem shows up
as per git bisect.
Ideally guest shouldn't be issuing configure-connector call on an
already configured DR connector. However for now, work around this in
QEMU by allowing configure-connector to be called multiple times for
all types of DR connectors.
Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
[dwg: Corrected buglet that would have initialized fdt pointers ready
for reading on a device not present at reset]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
object_property_add_child() can only fail in two cases:
- the child already has a parent, which shouldn't happen since the DRC was
allocated a few lines above
- the parent already has a child with the same name, which would mean the
caller tries to create a DRC that already exists
In both case, this is a QEMU bug and we should abort.
Signed-off-by: Greg Kurz <groug@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
If object_property_add_alias() returns an error in realize(), we should
propagate it to the caller and certainly not unref the DRC.
Same thing goes for unrealize(). Since object_property_del() is the last
call, we can even get rid of the intermediate Error *.
And finally, unrealize() should undo all registrations performed by
realize().
Signed-off-by: Greg Kurz <groug@kaod.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Make visit_type_null() take an @obj argument like its buddies. This
helps keep the next commit simple.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
According to PAPR, the DR-indicator should only be valid for physical DRCs,
not logical DRCs. At the moment we implement it for all DRCs, so restrict
it to physical ones only.
We move the state to the physical DRC subclass, which means adding some
QOM boilerplate to handle the newly distinct type.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Daniel Barboza <danielhb@linux.vnet.ibm.com>
Tested-by: Daniel Barboza <danielhb@linux.vnet.ibm.com>
Most of the time, the state of a DRC object is contained in the single
'state' variable. However, during the transition from UNISOLATE to
CONFIGURED state requires multiple calls to the ibm,configure-connector
RTAS call to retrieve the device tree for the attached device. We need
some extra state to keep track of where we're up to in delivering the
device tree information to the guest.
Currently that extra state is in a sPAPRConfigureConnectorState
substructure which is only allocated when we're in the middle of the
configure connector process. That sounds like a good idea, but the extra
state is only two integers - on many platforms that will take up the same
room as the (maybe NULL) ccs pointer even before malloc() overhead. Plus
it's another object whose lifetime we need to manage. In short, it's not
worth it.
So, fold the sPAPRConfigureConnectorState substructure directly into the
DRC object.
Previously the structure was allocated lazily when the configure-connector
call discovers it's not there. Now, we need to initialize the subfields
pre-emptively, as soon as we enter UNISOLATE state.
Although it's not strictly necessary (the field values should only ever
be consulted when in UNISOLATE state), we try to keep them at -1 when in
other states, as a debugging aid.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Daniel Barboza <danielhb@linux.vnet.ibm.com>
Tested-by: Daniel Barboza <danielhb@linux.vnet.ibm.com>
Each DRC has three fields describing its state: isolation_state,
allocation_state and configured. At first this seems like a reasonable
representation, since its based directly on the PAPR defined
isolation-state and allocation-state indicators. However:
* Only a few combinations of the two fields' values are permitted
* allocation_state isn't used at all for physical DRCs
* The indicators are write only so they don't really have a well
defined current value independent of each other
This replaces these variables with a single state variable, whose names
and numbers are based on the diagram in LoPAPR section 13.4. Along with
this we add code to check the current state on various operations and make
sure the requested transition is permitted.
Strictly speaking, this makes guest visible changes to behaviour (since we
probably allowed some transitions we shouldn't have before). However, a
hypothetical guest broken by that wasn't PAPR compliant, and probably
wouldn't have worked under PowerVM.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Daniel Barboza <danielhb@linux.vnet.ibm.com>
Tested-by: Daniel Barboza <danielhb@linux.vnet.ibm.com>
'awaiting_release' indicates that the host has requested an unplug of the
device attached to the DRC, but the guest has not (yet) put the device
into a state where it is safe to complete removal.
1. Rename it to 'unplug_requested' which to me at least is clearer
2. Remove the ->release_pending() method used to check this from outside
spapr_drc.c. The method only plausibly has one implementation, so use
a plain function (spapr_drc_unplug_requested()) instead.
3. Remove it from the migration stream. Attempting to migrate mid-unplug
is broken not just for spapr - in general management has no good way to
determine if the device should be present on the destination or not. So,
until that's fixed, there's no point adding extra things to the stream.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Tested-by: Daniel Barboza <danielhb@linux.vnet.ibm.com>
This function has two unused parameters - remove them.
It also sets awaiting_release on all paths, except one. On that path
setting it is harmless, since it will be immediately cleared by
spapr_drc_release(). So factor it out of the if statements.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Tested-by: Daniel Barboza <danielhb@linux.vnet.ibm.com>
We currently ignore errors from the object_property_del() in
spapr_drc_release(). But the only way that could fail is if the property
doesn't exist, in which case it's a bug that we're in spapr_drc_release()
at all. So change from ignoring to abort()ing on errors.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The awaiting_allocation flag in the DRC was introduced by aab9913
"spapr_drc: Prevent detach racing against attach for CPU DR", allegedly to
prevent a guest crash on racing attach and detach. Except.. information
from the BZ actually suggests a qemu crash, not a guest crash. And there
shouldn't be a problem here anyway: if the guest has already moved the DRC
away from UNUSABLE state, the detach would already be deferred, and if it
hadn't it should be safe to detach it (the guest should fail gracefully
when it attempts to change the allocation state).
I think this was probably just a bandaid for some other problem in the
state management. So, remove awaiting_allocation and associated code.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Tested-by: Greg Kurz <groug@kaod.org>
Tested-by: Daniel Barboza <danielhb@linux.vnet.ibm.com>
When migrating a guest which has already had devices hotplugged,
libvirt typically starts the destination qemu with -incoming defer,
adds those hotplugged devices with qmp, then initiates the incoming
migration.
This causes problems for the management of spapr DRC state. Because
the device is treated as hotplugged, it goes into a DRC state for a
device immediately after it's plugged, but before the guest has
acknowledged its presence. However, chances are the guest on the
source machine *has* acknowledged the device's presence and configured
it.
If the source has fully configured the device, then DRC state won't be
sent in the migration stream: for maximum migration compatibility with
earlier versions we don't migrate DRCs in coldplug-equivalent state.
That means that the DRC effectively changes state over the migrate,
causing problems later on.
In addition, logging hotplug events for these devices isn't what we
want because a) those events should already have been issued on the
source host and b) the event queue should get wiped out by the
incoming state anyway.
In short, what we really want is to treat devices added before an
incoming migration as if they were coldplugged.
To do this, we first add a spapr_drc_hotplugged() helper which
determines if the device is hotplugged in the sense relevant for DRC
state management. We only send hotplug events when this is true.
Second, when we add a device which isn't hotplugged in this sense, we
force a reset of the DRC state - this ensures the DRC is in a
coldplug-equivalent state (there isn't usually a system reset between
these device adds and the incoming migration).
This is based on an earlier patch by Laurent Vivier, cleaned up and
extended.
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Tested-by: Daniel Barboza <danielhb@linux.vnet.ibm.com>
All the DRC subtypes explicitly list instance_size in TypeInfo (all as
sizeof(sPAPRDRConnector). This isn't necessary, since if it's not listed
it will be derived from the parent type.
Worse, this is dangerous, because if a subtype is changed in future to
have a larger structure, then subtypes of that subtype also need to have
instance_size changed, or it will lead to hard to track memory corruption
bugs.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
spapr_drc_attach() has a 'coldplug' parameter which sets the DRC into
configured state initially, instead of the usual ISOLATED/UNUSABLE state.
It turns out this is unnecessary: although coldplugged devices do need to
be in CONFIGURED state once the guest starts, that will already be
accomplished by the reset code which will move DRCs for already plugged
devices into a coldplug equivalent state.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
At the moment, spapr_drc_release() has an ugly switch on the DRC type to
call the right, device-specific release function. This cleans it up by
doing that via a proper QOM method.
It's still arguably an abstraction violation for the DRC code to call into
the specific device code, but one mess at a time.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
DRC objects have a regular device reset method. However, it only gets
called in the usual way for PCI DRCs. Because of where CPU and LMB DRCs
are in the QOM tree, their device reset method isn't automatically called.
So, the machine manually registers reset handlers to call device_reset().
This patch removes the device reset method, and instead always explicitly
registers the reset handler from realize(). This means the callers don't
have to worry about the two cases, and we always get proper resets.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Laurent Vivier <lvivier@redhat.com>
The DR-indicator is essentially a "virtual LED" attached to a hotpluggable
device, which the guest can set to various states for the attention of
the operator or management layers.
It's mostly guest managed, except that we once-off set it to
ACTIVE/INACTIVE in the attach/detach path. While that makes certain sense,
there's no indication in PAPR that the hypervisor should do this, and the
drmgr code on the guest side doesn't appear to need it (it will already set
the indicator to ACTIVE on hotplug, and INACTIVE on remove).
So, leave the DR-indicator entirely to the guest; the only thing we need
to do is ensure it's in a sane state on reset.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
There are substantial differences in the various paths through
set_isolation_state(), both for setting to ISOLATED versus UNISOLATED
state and for logical versus physical DRCs.
So, split the set_isolation_state() method into isolate() and unisolate()
methods, and give it different implementations for the two DRC types.
Factor some minimal common checks, including for valid indicator values
(which we weren't previously checking) into rtas_set_isolation_state().
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
The allocation-state indicator should only actually be implemented for
"logical" DRCs, not physical ones. Factor a check for this, and also for
valid indicator state values into rtas_set_allocation_state(). Because
they don't exist for physical DRCs, there's no reason that we'd ever want
more than one method implementation, so it can just be a plain function.
In addition, the setting to USABLE and setting to UNUSABLE paths in
set_allocation_state() don't actually have much in common. So, split the
method separate functions for each parameter value (drc_set_usable()
and drc_set_unusable()).
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
The reset handler for DRCs attempts several state transitions which are
subject to various checks and restrictions. But at reset time we know
there is no guest, so we can ignore most of the usual sequencing rules and
just set the DRC back to a known state. In fact, it's safer to do so.
The existing code also has several redundant checks for
drc->awaiting_release inside a block which has already tested that. This
patch removes those and sets the DRC to a fixed initial state based only
on whether a device is currently plugged or not.
With DRCs correctly reset to a state based on device presence, we don't
need to force state transitions as cold plugged devices are processed.
This allows us to remove all the callers of the set_*_state() methods from
outside spapr_drc.c.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
spapr_drc_detach() is called when qemu generic code requests a device be
unplugged. It makes a number of tests, which could well delay further
action until later, before actually detach the device from the DRC.
This splits out the part which actually removes the device from the DRC
into spapr_drc_release(). This will be useful for further cleanups.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
The 'signalled' field in the DRC appears to be entirely a torturous
workaround for the fact that PCI devices were started in UNISOLATED state
for unclear reasons.
1) 'signalled' is already meaningless for logical (so far, all non PCI)
DRCs. It's always set to true (at least at any point it might be tested),
and can't be assigned any real meaning due to the way signalling works for
logical DRCs.
2) For PCI DRCs, the only time signalled would be false is when non-zero
functions of a multifunction device are hotplugged, followed by function
zero (the other way around is explicitly not permitted). In that case the
secondary function DRCs are attached, but the notification isn't sent to
the guest until function 0 is plugged.
3) signalled being false is used to allow a DRC detach to switch mode
back to ISOLATED state, which allows a secondary function to be hotplugged
then unplugged with function 0 never inserted. Without this a secondary
function starting in UNISOLATED state couldn't be detached again without
function 0 being inserted, all the functions configured by the guest, then
sent back to ISOLATED state.
4) But now that PCI DRCs start in ISOLATED state, there's nothing to be
done. If the guest doesn't get the notification, it won't switch the
device to UNISOLATED state, so nothing prevents it from being unplugged.
If the guest does move it to UNISOLATED state without the signal (due to
a manual drmgr call, for instance) then it really isn't safe to unplug it.
So, this patch removes the signalled variable and all code related to it.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
PCI DRCs, and only PCI DRCs, are immediately moved to UNISOLATED isolation
state once the device is attached. This has been there from the initial
implementation, and it's not clear why.
The state diagram in PAPR 13.4 suggests PCI devices should start in
ISOLATED state until the guest moves them into UNISOLATED, and the code in
the guest-side drmgr tool seems to work that way too.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
This reverts commit fe6824d126.
Conflicts hw/ppc/spapr_drc.c, because get_index() has been renamed
spapr_get_index().
This didn't fix the problem. Once the hotplug has been started
some memory is allocated and some structures are allocated.
We don't free it when we ignore the unplug, and we can't because
they can be in use by the kernel.
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Tested-by: Daniel Barboza <danielhb@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
DRC objects have a get_name method which returns the DRC name generated
when the DRC is created. Replace that with a fixed spapr_drc_name()
function which generates the name on the fly from other information. This
means:
* We get rid of a method with only one implementation, and only local
callers
* We don't have to carry the name string around for the lifetime of the
DRC
* We use information added to the class structure to generate the name
in standard format, so we don't need an explicit switch on drc type
any more
We also eliminate the 'name' property; it's basically useless since the
only information in it can easily be deduced from other things.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Acked-by: Michael Roth <mdroth@linux.vnet.ibm.com>
DRC objects have attach & detach methods, but there's only one
implementation. Although there are some differences in its behaviour for
different DRC types, the overall structure is the same, so while we might
want different method implementations for some parts, we're unlikely to
want them for the top-level functions.
So, replace them with direct function calls.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Acked-by: Michael Roth <mdroth@linux.vnet.ibm.com>
There are 3 types of "indicator" associated with hotplug in the PAPR spec
the "allocation state", "isolation state" and "DR-indicator". The first
two are intimately tied to the various state transitions associated with
hotplug. The DR-indicator, however, is different and simpler.
It's basically just a guest controlled variable which can be used by the
guest to flag state or problems associated with a device. The idea is that
the hypervisor can use it to present information back on management
consoles (on some machines with PowerVM it may even control physical LEDs
on the machine case associated with the relevant device).
For that reason, there's only ever likely to be a single update
implementation so the set_indicator_state method isn't useful. Replace it
with a direct function call.
While we're there, make some small associated cleanups:
* PAPR doesn't use the term "indicator state", just "DR-indicator" and
the allocation state and isolation state are also considered "indicators".
Rename things to be less confusing
* Fold set_indicator_state() and rtas_set_indicator_state() into a single
rtas_set_dr_indicator() function.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Michael Roth <mdroth@linux.vnet.ibm.com>
Acked-by: Michael Roth <mdroth@linux.vnet.ibm.com>
