IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
We are passing pointers to the firmware for reads, we need to properly
convert the result as OPAL is always BE.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
opal_xscom_read uses a pointer to return the data so we need
to byteswap it on LE builds.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
A couple more device tree properties that need byte swapping.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The MSI code is miscalculating quotas in little endian mode.
Add required byteswaps to fix this.
Before we claimed a quota of 65536, after the patch we
see the correct value of 256.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We need to byteswap ibm,pcie-link-speed-stats.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The NVRAM code has a number of endian issues. I noticed a very
confused error log count:
RTAS: 100663330 -------- RTAS event begin --------
100663330 == 0x06000022. 0x6 LE error logs and 0x22 BE error logs.
The pstore code has similar issues - if we write an oops in one
endian and attempt to read it in another we get junk.
Make both of these formats big endian, and byteswap as required.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
cpu_to_core_id() is missing a byteswap:
cat /sys/devices/system/cpu/cpu63/topology/core_id
201326592
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
During on LE boot we see:
Partition configured for 1073741824 cpus, operating system maximum is 2048.
Clearly missing a byteswap here.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
There is a bug in using ptrace to access FPRs via PTRACE_PEEKUSR /
PTRACE_POKEUSR. In effect, trying to access any of the FPRs always
really accesses FPR0, which does seriously break debugging :-)
The problem seems to have been introduced by commit 3ad26e5c4459d
(Merge branch 'for-kvm' into next).
[ It is indeed a merge conflict between Paul's FPU/VSX state rework
and my LE patches - Anton ]
Signed-off-by: Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Commit ce11e48b7fdd256ec68b932a89b397a790566031 ("KVM: PPC: E500: Add
userspace debug stub support") added "struct thread_struct" to the
stack of kvmppc_vcpu_run(). thread_struct is 1152 bytes on my build,
compared to 48 bytes for the recently-introduced "struct debug_reg".
Use the latter instead.
This fixes the following error:
cc1: warnings being treated as errors
arch/powerpc/kvm/booke.c: In function 'kvmppc_vcpu_run':
arch/powerpc/kvm/booke.c:760:1: error: the frame size of 1424 bytes is larger than 1024 bytes
make[2]: *** [arch/powerpc/kvm/booke.o] Error 1
make[1]: *** [arch/powerpc/kvm] Error 2
make[1]: *** Waiting for unfinished jobs....
Signed-off-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
The current logic sets the kdump base to min of 2G or ppc64_rma_size/2.
On PowerNV kernel the first memory block 'memory@0' can be very large,
equal to the DIMM size with ppc64_rma_size value capped to 1G. Hence on
PowerNV, kdump base is set to 512M resulting kdump to fail while allocating
paca array. This is because, paca need its memory from RMA region capped
at 256M (see allocate_pacas()).
This patch lowers the kdump base cap to 128M so that kdump kernel can
successfully get memory below 256M for paca allocation.
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
I have recently found out that no iommu_groups could be found under
/sys/ on a P8. That prevents PCI passthrough from working.
During my investigation, I found out there seems to be a missing
iommu_register_group for PHB3. The following patch seems to fix the
problem. After applying it, I see iommu_groups under
/sys/kernel/iommu_groups/, and can also bind vfio-pci to an adapter,
which gives me a device at /dev/vfio/.
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The bestcomm driver has been moved to drivers/dma, so to select
this driver by default additionally CONFIG_DMADEVICES has to be
enabled. Currently it is not enabled in the config despite existing
CONFIG_PPC_BESTCOMM=y in the config files. Fix it.
Signed-off-by: Anatolij Gustschin <agust@denx.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
At least some distros expect it these days; turn it on. Also, random
churn from doing a savedefconfig for the first time in a year or so.
Signed-off-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
In pte_alloc_one(), pgtable_page_ctor() is passed an address that has
not been converted by page_address() to the newly allocated PTE page.
When the PTE is freed, __pte_free_tlb() calls pgtable_page_dtor()
with an address to the PTE page that has been converted by page_address().
The mismatch in the PTE's page address causes pgtable_page_dtor() to access
invalid memory, so resources for that PTE (such as the page lock) is not
properly cleaned up.
On PPC32, only SMP kernels are affected.
On PPC64, only SMP kernels with 4K page size are affected.
This bug was introduced by commit d614bb041209fd7cb5e4b35e11a7b2f6ee8f62b8
"powerpc: Move the pte free routines from common header".
On a preempt-rt kernel, a spinlock is dynamically allocated for each
PTE in pgtable_page_ctor(). When the PTE is freed, calling
pgtable_page_dtor() with a mismatched page address causes a memory leak,
as the pointer to the PTE's spinlock is bogus.
On mainline, there isn't any immediately obvious symptoms, but the
problem still exists here.
Fixes: d614bb041209fd7c "powerpc: Move the pte free routes from common header"
Cc: Paul Mackerras <paulus@samba.org>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: linux-stable <stable@vger.kernel.org> # v3.10+
Signed-off-by: Hong H. Pham <hong.pham@windriver.com>
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Allocate enough memory for the ocm_block structure, not just a pointer
to it.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
A kernel configured with PPC_EARLY_DEBUG_BOOTX=y but PPC_PMAC=n and
PPC_MAPLE=n will fail to link:
btext.c:(.text+0x2d0fc): undefined reference to `.rmci_off'
btext.c:(.text+0x2d214): undefined reference to `.rmci_on'
Fix it by making the build of rmci_on/off() depend on
PPC_EARLY_DEBUG_BOOTX, which also enable the only code that uses them.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Now that the svcpu sync is interrupt aware we can enable interrupts
earlier in the exit code path again, moving 32bit and 64bit closer
together.
While at it, document the fact that we're always executing the exit
path with interrupts enabled so that the next person doesn't trap
over this.
Signed-off-by: Alexander Graf <agraf@suse.de>
As soon as we get back to our "highmem" handler in virtual address
space we may get preempted. Today the reason we can get preempted is
that we replay interrupts and all the lazy logic thinks we have
interrupts enabled.
However, it's not hard to make the code interruptible and that way
we can enable and handle interrupts even earlier.
This fixes random guest crashes that happened with CONFIG_PREEMPT=y
for me.
Signed-off-by: Alexander Graf <agraf@suse.de>
The kvmppc_copy_{to,from}_svcpu functions are publically visible,
so we should also export them in a header for others C files to
consume.
So far we didn't need this because we only called it from asm code.
The next patch will introduce a C caller.
Signed-off-by: Alexander Graf <agraf@suse.de>
We call a C helper to save all svcpu fields into our vcpu. The C
ABI states that r12 is considered volatile. However, we keep our
exit handler id in r12 currently.
So we need to save it away into a non-volatile register instead
that definitely does get preserved across the C call.
This bug usually didn't hit anyone yet since gcc is smart enough
to generate code that doesn't even need r12 which means it stayed
identical throughout the call by sheer luck. But we can't rely on
that.
Signed-off-by: Alexander Graf <agraf@suse.de>
Get the memory errors reported by opal and plumb it into memory poison
infrastructure. This patch uses new messaging channel infrastructure to
pull the fsp memory errors to linux.
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We steal the _PAGE_COHERENCE bit and use that for indicating NUMA ptes.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We want to make sure we don't use these function when updating a pte
or pmd entry that have a valid hpte entry, because these functions
don't invalidate them. So limit the check to _PAGE_PRESENT bit.
Numafault core changes use these functions for updating _PAGE_NUMA bits.
That should be ok because when _PAGE_NUMA is set we can be sure that
hpte entries are not present.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Set memory coherence always on hash64 config. If
a platform cannot have memory coherence always set they
can infer that from _PAGE_NO_CACHE and _PAGE_WRITETHRU
like in lpar. So we dont' really need a separate bit
for tracking _PAGE_COHERENCE.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Even though we have same value for linux PTE bits and hash PTE pits
use the hash pte bits wen updating hash pte
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Currently, the slb_shadow buffer is our largest symbol:
[jk@pablo linux]$ nm --size-sort -r -S obj/vmlinux | head -1
c000000000da0000 0000000000040000 d slb_shadow
- we allocate 128 bytes per cpu; so 256k with NR_CPUS=2048. As we have
constant initialisers, it's allocated in .text, causing a larger vmlinux
image. We may also allocate unecessary slb_shadow buffers (> no. pacas),
since we use the build-time NR_CPUS rather than the run-time nr_cpu_ids.
We could move this to the bss, but then we still have the NR_CPUS vs
nr_cpu_ids potential for overallocation.
This change dynamically allocates the slb_shadow array, during
initialise_pacas(). At a cost of 104 bytes of text, we save 256k of
data:
[jk@pablo linux]$ size obj/vmlinux{.orig,}
text data bss dec hex filename
9202795 5244676 1169576 15617047 ee4c17 obj/vmlinux.orig
9202899 4982532 1169576 15355007 ea4c7f obj/vmlinux
Tested on pseries.
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The only external user of slb_shadow is the pseries lpar code, and it
can access through the paca array instead.
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Use PCI standard marco dev_is_pci() instead of directly compare
pci_bus_type to check whether it is pci device.
Signed-off-by: Yijing Wang <wangyijing@huawei.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Currently we continue to poll get/set_rtc_time even when we know they
are not working.
This changes it so that if it fails at boot time we remove the ppc_md
get/set_rtc_time hooks so that we don't end up polling known broken
calls.
Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
These accessors allow us to do cache inhibited accesses when in real
mode. They should only be used in real mode.
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
In order to support concurrent adapter firmware download
to SR-IOV adapters on pSeries, each VF will see an EEH event
where the slot will remain in the unavailable state for
the duration of the adapter firmware update, which can take
as long as 5 minutes. Extend the EEH recovery timeout to
account for this.
Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Acked-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
When hitting frozen PE or fenced PHB, it's always indicative to
have dumped PHB diag-data for further analysis and diagnosis.
However, we never dump that for the cases. The patch intends to
dump PHB diag-data at the backend of eeh_ops::get_log() for PowerNV
platform.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Prior to the completion of PCI enumeration, we actively detects
EEH errors on PCI config cycles and dump PHB diag-data if necessary.
The EEH backend also dumps PHB diag-data in case of frozen PE or
fenced PHB. However, we are using different functions to dump the
PHB diag-data for those 2 cases.
The patch merges the functions for dumping PHB diag-data to one so
that we can avoid duplicate code. Also, we never dump PHB3 diag-data
during PCI config cycles with frozen PE. The patch fixes it as well.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The current implementation of IOMMU on sPAPR does not use iommu_ops
and therefore does not call IOMMU API's bus_set_iommu() which
1) sets iommu_ops for a bus
2) registers a bus notifier
Instead, PCI devices are added to IOMMU groups from
subsys_initcall_sync(tce_iommu_init) which does basically the same
thing without using iommu_ops callbacks.
However Freescale PAMU driver (https://lkml.org/lkml/2013/7/1/158)
implements iommu_ops and when tce_iommu_init is called, every PCI device
is already added to some group so there is a conflict.
This patch does 2 things:
1. removes the loop in which PCI devices were added to groups and
adds explicit iommu_add_device() calls to add devices as soon as they get
the iommu_table pointer assigned to them.
2. moves a bus notifier to powernv code in order to avoid conflict with
the notifier from Freescale driver.
iommu_add_device() and iommu_del_device() are public now.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Move SG list and entry structure to header file so that
it can be used in other places as well.
Signed-off-by: Vasant Hegde <hegdevasant@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Opal now has a new messaging infrastructure to push the messages to
linux in a generic format for different type of messages using only one
event bit. The format of the opal message is as below:
struct opal_msg {
uint32_t msg_type;
uint32_t reserved;
uint64_t params[8];
};
This patch allows clients to subscribe for notification for specific
message type. It is upto the subscriber to decipher the messages who showed
interested in receiving specific message type.
The interface to subscribe for notification is:
int opal_message_notifier_register(enum OpalMessageType msg_type,
struct notifier_block *nb)
The notifier will fetch the opal message when available and notify the
subscriber with message type and the opal message. It is subscribers
responsibility to copy the message data before returning from notifier
callback.
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Add basic error handling in machine check exception handler.
- If MSR_RI isn't set, we can not recover.
- Check if disposition set to OpalMCE_DISPOSITION_RECOVERED.
- Check if address at fault is inside kernel address space, if not then send
SIGBUS to process if we hit exception when in userspace.
- If address at fault is not provided then and if we get a synchronous machine
check while in userspace then kill the task.
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Now that we are ready to handle machine check directly in linux, do not
register with firmware to handle machine check exception.
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
When machine check real mode handler can not continue into host kernel
in V mode, it returns from the interrupt and we loose MCE event which
never gets logged. In such a situation queue up the MCE event so that
we can log it later when we get back into host kernel with r1 pointing to
kernel stack e.g. during syscall exit.
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Now that we handle machine check in linux, the MCE decoding should also
take place in linux host. This info is crucial to log before we go down
in case we can not handle the machine check errors. This patch decodes
and populates a machine check event which contain high level meaning full
MCE information.
We do this in real mode C code with ME bit on. The MCE information is still
available on emergency stack (in pt_regs structure format). Even if we take
another exception at this point the MCE early handler will allocate a new
stack frame on top of current one. So when we return back here we still have
our MCE information safe on current stack.
We use per cpu buffer to save high level MCE information. Each per cpu buffer
is an array of machine check event structure indexed by per cpu counter
mce_nest_count. The mce_nest_count is incremented every time we enter
machine check early handler in real mode to get the current free slot
(index = mce_nest_count - 1). The mce_nest_count is decremented once the
MCE info is consumed by virtual mode machine exception handler.
This patch provides save_mce_event(), get_mce_event() and release_mce_event()
generic routines that can be used by machine check handlers to populate and
retrieve the event. The routine release_mce_event() will free the event slot so
that it can be reused. Caller can invoke get_mce_event() with a release flag
either to release the event slot immediately OR keep it so that it can be
fetched again. The event slot can be also released anytime by invoking
release_mce_event().
This patch also updates kvm code to invoke get_mce_event to retrieve generic
mce event rather than paca->opal_mce_evt.
The KVM code always calls get_mce_event() with release flags set to false so
that event is available for linus host machine
If machine check occurs while we are in guest, KVM tries to handle the error.
If KVM is able to handle MC error successfully, it enters the guest and
delivers the machine check to guest. If KVM is not able to handle MC error, it
exists the guest and passes the control to linux host machine check handler
which then logs MC event and decides how to handle it in linux host. In failure
case, KVM needs to make sure that the MC event is available for linux host to
consume. Hence KVM always calls get_mce_event() with release flags set to false
and later it invokes release_mce_event() only if it succeeds to handle error.
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This patch handles the memory errors on power8. If we get a machine check
exception due to SLB or TLB errors, then flush SLBs/TLBs and reload SLBs to
recover.
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
If we get a machine check exception due to SLB or TLB errors, then flush
SLBs/TLBs and reload SLBs to recover. We do this in real mode before turning
on MMU. Otherwise we would run into nested machine checks.
If we get a machine check when we are in guest, then just flush the
SLBs and continue. This patch handles errors for power7. The next
patch will handle errors for power8
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This patch introduces flush_tlb operation in cpu_spec structure. This will
help us to invoke appropriate CPU-side flush tlb routine. This patch
adds the foundation to invoke CPU specific flush routine for respective
architectures. Currently this patch introduce flush_tlb for p7 and p8.
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This patch adds the early machine check function pointer in cputable for
CPU specific early machine check handling. The early machine handle routine
will be called in real mode to handle SLB and TLB errors. We can not reuse
the existing machine_check hook because it is always invoked in kernel
virtual mode and we would already be in trouble if we get SLB or TLB errors.
This patch just sets up a mechanism to invoke CPU specific handler. The
subsequent patches will populate the function pointer.
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We can get machine checks from any context. We need to make sure that
we handle all of them correctly. If we are coming from hypervisor user-space,
we can continue in host kernel in virtual mode to deliver the MC event.
If we got woken up from power-saving mode then we may come in with one of
the following state:
a. No state loss
b. Supervisor state loss
c. Hypervisor state loss
For (a) and (b), we go back to nap again. State (c) is fatal, keep spinning.
For all other context which we not sure of queue up the MCE event and return
from the interrupt.
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Move machine check entry point into Linux. So far we were dependent on
firmware to decode MCE error details and handover the high level info to OS.
This patch introduces early machine check routine that saves the MCE
information (srr1, srr0, dar and dsisr) to the emergency stack. We allocate
stack frame on emergency stack and set the r1 accordingly. This allows us to be
prepared to take another exception without loosing context. One thing to note
here that, if we get another machine check while ME bit is off then we risk a
checkstop. Hence we restrict ourselves to save only MCE information and
register saved on PACA_EXMC save are before we turn the ME bit on. We use
paca->in_mce flag to differentiate between first entry and nested machine check
entry which helps proper use of emergency stack. We increment paca->in_mce
every time we enter in early machine check handler and decrement it while
leaving. When we enter machine check early handler first time (paca->in_mce ==
0), we are sure nobody is using MC emergency stack and allocate a stack frame
at the start of the emergency stack. During subsequent entry (paca->in_mce >
0), we know that r1 points inside emergency stack and we allocate separate
stack frame accordingly. This prevents us from clobbering MCE information
during nested machine checks.
The early machine check handler changes are placed under CPU_FTR_HVMODE
section. This makes sure that the early machine check handler will get executed
only in hypervisor kernel.
This is the code flow:
Machine Check Interrupt
|
V
0x200 vector ME=0, IR=0, DR=0
|
V
+-----------------------------------------------+
|machine_check_pSeries_early: | ME=0, IR=0, DR=0
| Alloc frame on emergency stack |
| Save srr1, srr0, dar and dsisr on stack |
+-----------------------------------------------+
|
(ME=1, IR=0, DR=0, RFID)
|
V
machine_check_handle_early ME=1, IR=0, DR=0
|
V
+-----------------------------------------------+
| machine_check_early (r3=pt_regs) | ME=1, IR=0, DR=0
| Things to do: (in next patches) |
| Flush SLB for SLB errors |
| Flush TLB for TLB errors |
| Decode and save MCE info |
+-----------------------------------------------+
|
(Fall through existing exception handler routine.)
|
V
machine_check_pSerie ME=1, IR=0, DR=0
|
(ME=1, IR=1, DR=1, RFID)
|
V
machine_check_common ME=1, IR=1, DR=1
.
.
.
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
