The phb3/phb4/phb5 root ports inherit from the default PCIE root port
implementation, which requests a LSI interrupt (#INTA). On real
hardware (POWER8/POWER9/POWER10), there is no such LSI. This patch
corrects it so that it matches the hardware.
As a consequence, the device tree previously generated was bogus, as
the root bridge LSI was not properly mapped. On some
implementation (powernv9), it was leading to inconsistent interrupt
controller (xive) data. With this patch, it is now clean.
Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220408131303.147840-3-fbarrat@linux.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
g_new(T, n) is neater than g_malloc(sizeof(T) * n). It's also safer,
for two reasons. One, it catches multiplication overflowing size_t.
Two, it returns T * rather than void *, which lets the compiler catch
more type errors.
This commit only touches allocations with size arguments of the form
sizeof(T).
Patch created mechanically with:
$ spatch --in-place --sp-file scripts/coccinelle/use-g_new-etc.cocci \
--macro-file scripts/cocci-macro-file.h FILES...
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Acked-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20220315144156.1595462-4-armbru@redhat.com>
Reviewed-by: Pavel Dovgalyuk <Pavel.Dovgalyuk@ispras.ru>
On a real system with POWER{8,9,10} processors, PHBs are sub-units of
the processor, they can be deactivated by firmware but not plugged in
or out like a PCI adapter on a slot. Nevertheless, having user-created
PHBs in QEMU seemed to be a good idea for testing purposes :
1. having a limited set of PHBs speedups boot time.
2. it is useful to be able to mimic a partially broken topology you
some time have to deal with during bring-up.
PowerNV is also used for distro install tests and having libvirt
support eases these tasks. libvirt prefers to run the machine with
-nodefaults to be sure not to drag unexpected devices which would need
to be defined in the domain file without being specified on the QEMU
command line. For this reason :
3. -nodefaults should not include default PHBs
User-created PHB{3,4,5} devices satisfied all these needs but reality
proves to be a bit more complex, internally when modeling such
devices, and externally when dealing with the user interface.
Req 1. and 2. can be simply addressed differently with a machine option:
"phb-mask=<uint>", which QEMU would use to enable/disable PHB device
nodes when creating the device tree.
For Req 3., we need to make sure we are taking the right approach. It
seems that we should expose a new type of user-created PHB device, a
generic virtualized one, that libvirt would use and not one depending
on the processor revision. This needs more thinking.
For now, remove user-created PHB{3,4,5} devices. All the cleanups we
did are not lost and they will be useful for the next steps.
Fixes: 5bc67b052b ("ppc/pnv: Introduce user creatable pnv-phb4 devices")
Fixes: 1f6a88fffc ("ppc/pnv: Introduce support for user created PHB3 devices")
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Frederic Barrat <fbarrat@linux.ibm.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220314130514.529931-1-clg@kaod.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
We already have the pnv-phb3 and pnv-phb4 devices for POWER8 and
POWER9 respectively. POWER10 uses version 5 of the PHB. It is very
close to the PHB4 from POWER9, at least in our model and we could
almost keep using the PHB4 model. However the matching root port
pnv-phb5-root-port is specific to POWER10 so to avoid confusion as
well as making it easy to introduce differences later, we create a
pnv-phb5 class, which is mostly an alias for pnv-phb4 for now.
With this patch, the command line for a user-created PHB on powernv10
becomes:
-machine powernv10 -nodefaults -device pnv-phb5 -device pnv-phb5-root-port
Fixes: 623575e16c ("ppc/pnv: Add model for POWER10 PHB5 PCIe Host bridge")
Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220310155101.294568-2-fbarrat@linux.ibm.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
When the Address-Based Interrupt Trigger mode is activated, the PHB
maps the interrupt source number into the interrupt command address.
The PHB directly triggers the IC ESB page of the interrupt number and
not the notify page of the IC anymore.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
The PQ_disable configuration bit disables the check done on the PQ
state bits when processing new MSI interrupts. When bit 9 is enabled,
the PHB forwards any MSI trigger to the XIVE interrupt controller
without checking the PQ state bits. The XIVE IC knows from the trigger
message that the PQ bits have not been checked and performs the check
locally.
This configuration bit only applies to MSIs and LSIs are still checked
on the PHB to handle the assertion level.
PQ_disable enablement is a requirement for StoreEOI.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
The trigger message coming from a HW source contains a special bit
informing the XIVE interrupt controller that the PQ bits have been
checked at the source or not. Depending on the value, the IC can
perform the check and the state transition locally using its own PQ
state bits.
The following changes add new accessors to the XiveRouter required to
query and update the PQ state bits. This only applies to the PowerNV
machine. sPAPR accessors are provided but the pSeries machine should
not be concerned by such complex configuration for the moment.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Commit 3f4c369ea6 ("ppc/pnv: make PECs create and realize PHB4s")
changed phb4_pec code to create the default PHB4 objects in
pnv_pec_default_phb_realize(). In this process the stacks[] PEC array was
removed and each PHB4 object is tied together with its PEC via the
phb->pec pointer.
This change also broke the previous QOM hierarchy - the PHB4 objects are
being created and not being parented to their respective chips. This can
be verified by 'info pic' in a powernv9 domain with default settings.
pnv_chip_power9_pic_print_info() will fail to find the PHBs because
object_child_foreach_recursive() won't find any.
The solution is to set the parent chip and the parent bus, in the same
way done for user created PHB4 devices, for all PHB4 devices.
Fixes: 3f4c369ea6 ("ppc/pnv: make PECs create and realize PHB4s")
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220218202804.413157-1-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
pnv_phb4_translate_tve() is quite similar to pnv_phb3_translate_tve(),
and that includes the fact that 'taddr' can be considered uninitialized
when throwing the "TCE access fault" error because, in theory, the loop
that sets 'taddr' can be skippable due to 'lev' being an signed int.
No one complained about this specific case yet, but since we took the
time to handle the same situtation in pnv_phb3_translate_tve(), let's
replicate it here as well.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Matheus Ferst <matheus.ferst@eldorado.org.br>
Message-Id: <20220127122234.842145-3-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
The 'taddr' variable is left unintialized, being set only inside the
"while ((lev--) >= 0)" loop where we get the TCE address. The 'lev' var
is an int32_t that is being initiliazed by the GETFIELD() macro, which
returns an uint64_t.
For a human reader this means that 'lev' will always be positive or zero.
But some compilers may beg to differ. 'lev' being an int32_t can in theory
be set as negative, and the "while ((lev--) >= 0)" loop might never be
reached, and 'taddr' will be left unitialized. This can cause phb3_error()
to use 'taddr' uninitialized down below:
if ((is_write & !(tce & 2)) || ((!is_write) && !(tce & 1))) {
phb3_error(phb, "TCE access fault at 0x%"PRIx64, taddr);
A quick way of fixing it is to use a do/while() loop. This will keep the
same semanting as the existing while() loop does and the compiler will
understand that 'taddr' will be initialized at least once.
Suggested-by: Matheus K. Ferst <matheus.ferst@eldorado.org.br>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/573
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220127122234.842145-2-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
If an iommu page has wrong permissions, an error message is displayed,
but the access is allowed, which is odd. This patch fixes it.
Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220121152350.381685-1-fbarrat@linux.ibm.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Grab the PHB version from the PEC class directly when needed.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220117122753.1655504-4-clg@kaod.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
PHB5 will introduce its own root port model. Prepare ground for it.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220117122753.1655504-3-clg@kaod.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
The root port device is currently created and attached to the PHB
early in pnv_phb4_realize(). Do it under pnv_pec_default_phb_realize()
after the PHB is fully realized. It's cleaner and avoids an extra
test on defaults_enabled().
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220117122753.1655504-2-clg@kaod.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
This function does not use 'stack' anymore. Rename it to
pnv_pec_phb_update_map().
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220114180719.52117-9-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
All the complexity that was scattered between PnvPhb4PecStack and
PnvPHB4 are now centered in the PnvPHB4 device. PnvPhb4PecStack does not
serve any purpose in the current code base.
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220114180719.52117-8-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
This patch changes the design of the PEC device to create and realize PHB4s
instead of PecStacks. After all the recent changes, PHB4s now contain all
the information needed for their proper functioning, not relying on PecStack
in any capacity.
All changes are being made in a single patch to avoid renaming parts of
the PecState and leaving the code in a strange way. E.g. rename
PecClass->num_stacks to num_phbs, which would then read a
pnv_pec_num_stacks[] array. To avoid mixing the old and new design more
than necessary it's clearer to do these changes in a single step.
The name changes made are:
- in PnvPhb4PecState:
* rename 'num_stacks' to 'num_phbs'
* remove the pec->stacks[] array. Current code relies on the
pec->stacks[] obj acting as a simple container, without ever accessing
pec->stacks[] for any other purpose. Instead of converting this into a
pec->phbs[] array, remove it
- in PnvPhb4PecClass, rename *num_stacks to *num_phbs;
- pnv_pec_num_stacks[] is renamed to pnv_pec_num_phbs[].
The logical changes:
- pnv_pec_default_phb_realize():
* init and set the properties of the PnvPHB4 qdev
* do not use stack->phb anymore;
- pnv_pec_realize():
* use the new default_phb_realize() to init/realize each PHB if
running with defaults;
- pnv_pec_instance_init(): removed since we're creating the PHBs during
pec_realize();
- pnv_phb4_get_stack():
* renamed to pnv_phb4_get_pec() and returns a PnvPhb4PecState*;
- pnv_phb4_realize(): use 'phb->pec' instead of 'stack'.
This design change shouldn't caused any behavioral change in the runtime
of the machine.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220114180719.52117-7-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
pnv_pec_default_phb_realize() stopped using it after the previous patch and
no one else is using it.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220114180719.52117-6-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Move the current pnv_pec_stk_default_phb_realize() call to
pec_realize(), renaming the function to pnv_pec_default_phb_realize(),
and set the PHB attributes using the PEC object directly.
This will be important to allow for PECs devices to handle PHB4s
directly later on.
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220114180719.52117-5-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
This pointer was being used for two reasons: pnv_phb4_update_regions()
was using it to access the PHB and phb4_realize() was using it as a way
to determine if the PHB was user created.
We can determine if the PHB is user created via phb->pec, introduced in
the previous patch, and pnv_phb4_update_regions() is no longer using
stack->phb.
Remove the pointer from the PnvPHB4 device.
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220114180719.52117-4-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
'stack->stack_no' represents the order that a stack appears in its PEC.
Its primary use is in XSCOM address space calculation in
pnv_phb4_xscom_realize() when calculating the memory region offset.
This attribute is redundant with phb->phb_id, which is calculated via
pnv_phb4_pec_get_phb_id() using stack->stack_no information. It'll also
be awkward to assign it when dealing with PECs and PHBs only in a future
patch.
A new pnv_phb4_get_phb_stack_no() helper is introduced to eliminate most
of the stack->stack_no uses we have. The only use left after this patch
is during pnv_pec_stk_default_phb_realize() when calculating phb_id,
which will also handled in the next patches.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220114180719.52117-3-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
This property will track the owner PEC of this PHB. For now it's
redundant since we can retrieve the PEC via phb->stack->pec but it
will not be redundant when we get rid of the stack device.
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220114180719.52117-2-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
After recent changes, this MemoryRegion can be migrated to PnvPHB4
without too much trouble.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220113192952.911188-11-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
We're now able to cleanly move nest_regs_mr to the PnvPHB4 device.
One thing of notice here is the need to use a phb->stack->pec pointer
because pnv_pec_stk_nest_xscom_write requires a PEC object. Another
thing that can be noticed in the use of 'stack->stack_no' that still
remains throughout the XSCOM code.
After moving all MemoryRegions to the PnvPHB4 object, this illustrates
what is the remaining role of the stack: provide a PEC pointer and the
'stack_no' information. If we can provide these in the PnvPHB4 object
instead (spoiler: we can, and we will), the PnvPhb4PecStack device will
be deprecated and can be removed.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220113192952.911188-10-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
stack->nest_regs_mr wasn't migrated to PnvPHB4 together with phb->nest_regs[] in
the previous patch. We were unable to cleanly convert its write MemoryRegionOps,
pnv_pec_stk_nest_xscom_write(), to use PnvPHB4 instead of PnvPhb4PecStack due to
pnv_pec_stk_update_map() using a stack. Thing is, we're now able to convert
pnv_pec_stk_update_map() because of what the did in previous patch.
The need for this intermediate step is a good example of the interconnected
relationship between stack and phb that we aim to cleanup.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220113192952.911188-9-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
stack->nest_regs[] is used in several XSCOM functions and it's one of
the main culprits of having to deal with stack->phb pointers around the
code.
Sure, we're having to add 2 extra stack->phb pointers to ease
nest_regs[] migration to PnvPHB4. They'll be dealt with shortly.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220113192952.911188-8-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
These 2 MemoryRegions, together with mmio(0|1)_base and mmio(0|1)_size
variables, are used together in the same functions. We're better of
moving them all in a single step.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220113192952.911188-7-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
The function does not rely on stack for anything it does anymore. This
is also one less instance of 'stack->phb' that we need to worry about.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220113192952.911188-6-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
This MemoryRegion can also be moved in a single step.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220113192952.911188-5-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
This MemoryRegion is simple enough to be moved in a single step.
A 'stack->phb' pointer had to be introduced in pnv_pec_stk_update_map()
because this function isn't ready to be fully converted to use a PnvPHB4
pointer instead. This will be dealt with in the following patches.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220113192952.911188-4-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Previous patch changed pnv_pec_stk_pci_xscom_read() and
pnv_pec_stk_pci_xscom_write() to use a PnvPHB4 opaque, making it easier
to move both pci_regs[] and the pci_regs_mr MemoryRegion to the PnvHB4
object.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220113192952.911188-3-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
The current relationship between PnvPhb4PecStack and PnvPHB4 objects is
overly complex. Recent work done in pnv_phb4.c and pnv_phb4_pec.c shows
that the stack obj role in the overall design is more of a placeholder for
its 'phb' object, having no atributes that stand on its own. This became
clearer after pnv-phb4 user creatable devices were implemented.
What remains now are a lot of stack->phb and phb->stack pointers
throughout .read and .write callbacks of MemoryRegionOps that are being
initialized in phb4_realize() time. stk_realize() is a no-op if the
machine is being run with -nodefaults.
The first step of trying to decouple the stack and phb relationship is
to move the MemoryRegionOps that belongs to PnvPhb4PecStack to PhbPHB4.
Unfortunately this can't be done without some preliminary steps to
change the usage of 'stack' and replace it with 'phb' in these
read/write callbacks.
This patch starts this process by using a PnvPHB4 opaque in
pnv_pec_stk_pci_xscom_ops instead of PnvPhb4PecStack.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220113192952.911188-2-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
pnv_pec_stk_pci_xscom_write() is pnv_pec_stk_pci_xscom_ops write
callback. It writes values into regs in the stack->nest_regs[] array.
The pnv_pec_stk_pci_xscom_read read callback, on the other hand, returns
values of the stack->pci_regs[]. In fact, at this moment, the only use
of stack->pci_regs[] is in pnv_pec_stk_pci_xscom_read(). There's no code
that is written anything in stack->pci_regs[], which is suspicious.
Considering that stack->nest_regs[] is widely used by the nested
MemoryOps pnv_pec_stk_nest_xscom_ops, in both read and write callbacks,
the conclusion is that we're writing the wrong array in
pnv_pec_stk_pci_xscom_write(). This function should write stack->pci_regs[]
instead.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Frederic Barrat <fbarrat@linux.ibm.com>
Message-Id: <20220111200132.633896-2-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Its only callers are inside pnv_phb4.c.
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220111131027.599784-6-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
This patch introduces pnv-phb4 user creatable devices that are created
in a similar manner as pnv-phb3 devices, allowing the user to interact
with the PHBs directly instead of creating PCI Express Controllers that
will create a certain amount of PHBs per controller index.
We accomplish this by doing the following:
- add a pnv_phb4_get_stack() helper to retrieve which stack an user
created phb4 would occupy;
- when dealing with an user created pnv-phb4 (detected by checking if
phb->stack is NULL at the start of phb4_realize()), retrieve its stack
and initialize its properties as done in stk_realize();
- use 'defaults_enabled()' in stk_realize() to avoid creating and
initializing a 'stack->phb' qdev that might be overwritten by an user
created pnv-phb4 device. This process is wrapped into a new helper
called pnv_pec_stk_default_phb_realize().
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220111131027.599784-5-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
At this moment, stack->phb is the plain PnvPHB4 device itself instead of
a pointer to the device. This will present a problem when adding user
creatable devices because we can't deal with this struct and the
realize() callback from the user creatable device.
We can't get rid of this attribute, similar to what we did when enabling
pnv-phb3 user creatable devices, because pnv_phb4_update_regions() needs
to access stack->phb to do its job. This function is called twice in
pnv_pec_stk_update_map(), which is one of the nested xscom write
callbacks (via pnv_pec_stk_nest_xscom_write()). In fact,
pnv_pec_stk_update_map() code comment is explicit about how the order of
the unmap/map operations relates with the PHB subregions.
All of this indicates that this code is tied together in a way that we
either go on a crusade, featuring lots of refactories and redesign and
considerable pain, to decouple stack and phb mapping, or we allow stack
update_map operations to access the associated PHB as it is today even
after introducing pnv-phb4 user devices.
This patch chooses the latter. Instead of getting rid of stack->phb,
turn it into a PHB pointer. This will allow us to assign an user created
PHB to an existing stack later. In this process,
pnv_pec_stk_instance_init() is removed because stack->phb is being
initialized in stk_realize() instead.
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220111131027.599784-4-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
The 'stack->phb_regs_mr' PHB4 passthrough XSCOM initialization relies on
'stack->phb' being not NULL. Moving 'stack->phb_regs_mr' region_init()
and add_subregion() to phb4_realize() time is a natural thing to do
since it's strictly PHB related.
The remaining XSCOM initialization is also related to 'stack->phb' but
in a different manner. For instance, 'stack->nest_regs_mr'
MemoryRegionOps, 'pnv_pec_stk_nest_xscom_ops', uses
pnv_pec_stk_nest_xscom_write() as a write callback. When trying to write
the PEC_NEST_STK_BAR_EN reg, pnv_pec_stk_update_map() is called. Inside
this function, pnv_phb4_update_regions() is called twice. This function
uses 'stack->phb' to manipulate memory regions of the phb.
This is not a problem now but, when enabling user creatable phb4s, a
stack that doesn't have an associated phb (i.e. stack->phb = NULL) it
will cause a SIGINT during boot in pnv_phb4_update_regions().
All this can be avoided if all XSCOM realize is moved to phb4_realize(),
when we have certainty about the existence of 'stack->phb'. A lot of
code was moved from pnv_phb4_pec.c to pnv_phb4.c due to static constant
and variables being used but the cleaner logic is worth the trouble.
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220111131027.599784-3-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Moving all phb4 properties setup to stk_realize() keeps this logic in
a single place instead of having it scattered between stk_realize() and
pec_realize().
'phb->index' can be retrieved using stack->stack_no and
pnv_phb4_pec_get_phb_id(), deprecating the use of 'phb-id' alias that
was being used for this purpose in pec_realize().
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220111131027.599784-2-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Relying on stack->phb to write the xscom DT of the PEC is something that
we won't be able to do with user creatable pnv-phb4 devices.
Hopefully, this can be done by using pnv_phb4_pec_get_phb_id(), which is
already used by pnv_pec_realize() to set the phb-id of the stack. Use
the same idea in pnv_pec_dt_xscom() to write ibm,phb-index without the
need to accessing stack->phb, since stack->phb is not granted to be !=
NULL when user creatable phbs are introduced.
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220110143346.455901-4-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
The logic inside pnv_pec_phb_offset() will be useful in the next patch
to determine the stack that should contain a PHB4 device.
Move the function to pnv_phb4.c and make it public since there's no
pnv_phb4_pec.h header. While we're at it, add 'stack_index' as a
parameter and make the function return 'phb-id' directly. And rename it
to pnv_phb4_pec_get_phb_id() to be even clearer about the function
intent.
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220110143346.455901-3-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
PHB3s ared SysBus devices and should be allowed to be dynamically
created.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220105212338.49899-9-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
The powernv machine uses the object hierarchy to populate the device
tree and each device should be parented to the chip it belongs to.
This is not the case for user created devices which are parented to
the container "/unattached".
Make sure a PHB3 device is parented to its chip by reparenting the
object if necessary.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Frederic Barrat <fbarrat@linux.ibm.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220105212338.49899-8-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
PHB3 devices and PCI devices can now be added to the powernv8 machine
using :
-device pnv-phb3,chip-id=0,index=1 \
-device nec-usb-xhci,bus=pci.1,addr=0x0
The 'index' property identifies the PHB3 in the chip. In case of user
created devices, a lookup on 'chip-id' is required to assign the
owning chip.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220105212338.49899-7-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
pnv_phb4_rc_config_read() and pnv_phb4_rc_config_write() are asserting
the existence of the root port. The root port is now optional, and there
will be cases where a pnv-phb4 device won't have a root port attached.
Instead of asserting, check if the root port exists before read/writing
into it.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220105212338.49899-6-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
We want to create only the absolutely minimal amount of devices when
running with -nodefaults. The root port is something that the machine
can boot up without. But, to do that, we need to provide a way for the
user to add them by hand.
This patch makes pnv-phb4-root-port user creatable and then uses the
pnv_phb_attach_root_port() helper to add a pnv_phb4_root_port only when
running with default settings.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220105212338.49899-5-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
This cleanups the PHB3 model a bit more since the root port is an
independent device and it will ease our task when adding user created
PHB3s.
pnv_phb_attach_root_port() is made public in pnv.c so it can be reused
with the pnv_phb4 root port later.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220105212338.49899-4-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
A similar situation as described previously with pnv_phb3_root_port
devices also happens with pnv_phb4_root_ports.
The solution is the same: assign an unique chassis/slot combo for them.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220105212338.49899-3-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
When creating a pnv_phb3_root_port using the command line, the first
root port is created successfully, but the second fails with the
following error:
qemu-system-ppc64: -device pnv-phb3-root-port,bus=phb3-root.0,id=pcie.3:
Can't add chassis slot, error -16
This error comes from the realize() function of its parent type,
rp_realize() from TYPE_PCIE_ROOT_PORT. pcie_chassis_add_slot() fails
with -EBUSY if there's an existing PCIESlot that has the same
chassis/slot value, regardless of being in a different bus.
One way to prevent this error is simply set chassis and slot values in
the command line. However, since phb3 root buses only supports a single
root port, we can just get an unique chassis/slot value by checking
which root bus the pnv_phb3_root_port is going to be attached, get the
equivalent phb3 device and use its chip-id and index values, which are
guaranteed to be unique.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20220105212338.49899-2-danielhb413@gmail.com>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
