The XIVE driver deals with CPU IPIs in a peculiar way. Each CPU has
its own XIVE IPI interrupt allocated at the HW level, for PowerNV, or
at the hypervisor level for pSeries. In practice, these interrupts are
not always used. pSeries/PowerVM prefers local doorbells for local
threads since they are faster. On PowerNV, global doorbells are also
preferred for the same reason.
The mapping in the Linux is reduced to a single interrupt using HW
interrupt number 0 and a custom irq_chip to handle EOI. This can cause
performance issues in some benchmark (ipistorm) on multichip systems.
Clarify the use of the 0 value, it will help in improving multichip
support.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20201210171450.1933725-4-clg@kaod.org
This is a simple cleanup to identify easily all flags of the XIVE
interrupt structure. The interrupts flagged with XIVE_IRQ_FLAG_NO_EOI
are the escalations used to wake up vCPUs in KVM. They are handled
very differently from the rest.
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20201210171450.1933725-3-clg@kaod.org
This is useful to track allocation of the HW resources on per guest
basis. Making sure IPIs are local to the chip of the vCPUs reduces
rerouting between interrupt controllers and gives better performance
in case of pinning. Checking the distribution of VP structures on the
chips also helps in reducing PowerBUS traffic.
[ clg: resurrected show_sources and reworked ouput ]
Signed-off-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20201210171450.1933725-2-clg@kaod.org
On POWER platforms where only some groups of threads within a core
share the L2-cache (indicated by the ibm,thread-groups device-tree
property), we currently print the incorrect shared_cpu_map/list for
L2-cache in the sysfs.
This patch reports the correct shared_cpu_map/list on such platforms.
Example:
On a platform with "ibm,thread-groups" set to
00000001 00000002 00000004 00000000
00000002 00000004 00000006 00000001
00000003 00000005 00000007 00000002
00000002 00000004 00000000 00000002
00000004 00000006 00000001 00000003
00000005 00000007
This indicates that threads {0,2,4,6} in the core share the L2-cache
and threads {1,3,5,7} in the core share the L2 cache.
However, without the patch, the shared_cpu_map/list for L2 for CPUs 0,
1 is reported in the sysfs as follows:
/sys/devices/system/cpu/cpu0/cache/index2/shared_cpu_list:0-7
/sys/devices/system/cpu/cpu0/cache/index2/shared_cpu_map:000000,000000ff
/sys/devices/system/cpu/cpu1/cache/index2/shared_cpu_list:0-7
/sys/devices/system/cpu/cpu1/cache/index2/shared_cpu_map:000000,000000ff
With the patch, the shared_cpu_map/list for L2 cache for CPUs 0, 1 is
correctly reported as follows:
/sys/devices/system/cpu/cpu0/cache/index2/shared_cpu_list:0,2,4,6
/sys/devices/system/cpu/cpu0/cache/index2/shared_cpu_map:000000,00000055
/sys/devices/system/cpu/cpu1/cache/index2/shared_cpu_list:1,3,5,7
/sys/devices/system/cpu/cpu1/cache/index2/shared_cpu_map:000000,000000aa
This patch also defines cpu_l2_cache_mask() for !CONFIG_SMP case.
Signed-off-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/1607596739-32439-6-git-send-email-ego@linux.vnet.ibm.com
On POWER systems, groups of threads within a core sharing the L2-cache
can be indicated by the "ibm,thread-groups" property array with the
identifier "2".
This patch adds support for detecting this, and when present, populate
the populating the cpu_l2_cache_mask of every CPU to the core-siblings
which share L2 with the CPU as specified in the by the
"ibm,thread-groups" property array.
On a platform with the following "ibm,thread-group" configuration
00000001 00000002 00000004 00000000
00000002 00000004 00000006 00000001
00000003 00000005 00000007 00000002
00000002 00000004 00000000 00000002
00000004 00000006 00000001 00000003
00000005 00000007
Without this patch, the sched-domain hierarchy for CPUs 0,1 would be
CPU0 attaching sched-domain(s):
domain-0: span=0,2,4,6 level=SMT
domain-1: span=0-7 level=CACHE
domain-2: span=0-15,24-39,48-55 level=MC
domain-3: span=0-55 level=DIE
CPU1 attaching sched-domain(s):
domain-0: span=1,3,5,7 level=SMT
domain-1: span=0-7 level=CACHE
domain-2: span=0-15,24-39,48-55 level=MC
domain-3: span=0-55 level=DIE
The CACHE domain at 0-7 is incorrect since the ibm,thread-groups
sub-array
[00000002 00000002 00000004
00000000 00000002 00000004 00000006
00000001 00000003 00000005 00000007]
indicates that L2 (Property "2") is shared only between the threads of a single
group. There are "2" groups of threads where each group contains "4"
threads each. The groups being {0,2,4,6} and {1,3,5,7}.
With this patch, the sched-domain hierarchy for CPUs 0,1 would be
CPU0 attaching sched-domain(s):
domain-0: span=0,2,4,6 level=SMT
domain-1: span=0-15,24-39,48-55 level=MC
domain-2: span=0-55 level=DIE
CPU1 attaching sched-domain(s):
domain-0: span=1,3,5,7 level=SMT
domain-1: span=0-15,24-39,48-55 level=MC
domain-2: span=0-55 level=DIE
The CACHE domain with span=0,2,4,6 for CPU 0 (span=1,3,5,7 for CPU 1
resp.) gets degenerated into the SMT domain. Furthermore, the
last-level-cache domain gets correctly set to the SMT sched-domain.
Signed-off-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/1607596739-32439-5-git-send-email-ego@linux.vnet.ibm.com
init_thread_group_l1_cache_map() initializes the per-cpu cpumask
thread_group_l1_cache_map with the core-siblings which share L1 cache
with the CPU. Make this function generic to the cache-property (L1 or
L2) and update a suitable mask. This is a preparatory patch for the
next patch where we will introduce discovery of thread-groups that
share L2-cache.
No functional change.
Signed-off-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/1607596739-32439-4-git-send-email-ego@linux.vnet.ibm.com
On platforms which have the "ibm,thread-groups" property, the per-cpu
variable cpu_l1_cache_map keeps a track of which group of threads
within the same core share the L1 cache, Instruction and Data flow.
This patch renames the variable to "thread_group_l1_cache_map" to make
it consistent with a subsequent patch which will introduce
thread_group_l2_cache_map.
This patch introduces no functional change.
Signed-off-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/1607596739-32439-3-git-send-email-ego@linux.vnet.ibm.com
The "ibm,thread-groups" device-tree property is an array that is used
to indicate if groups of threads within a core share certain
properties. It provides details of which property is being shared by
which groups of threads. This array can encode information about
multiple properties being shared by different thread-groups within the
core.
Example: Suppose,
"ibm,thread-groups" = [1,2,4,8,10,12,14,9,11,13,15,2,2,4,8,10,12,14,9,11,13,15]
This can be decomposed up into two consecutive arrays:
a) [1,2,4,8,10,12,14,9,11,13,15]
b) [2,2,4,8,10,12,14,9,11,13,15]
where in,
a) provides information of Property "1" being shared by "2" groups,
each with "4" threads each. The "ibm,ppc-interrupt-server#s" of the
first group is {8,10,12,14} and the "ibm,ppc-interrupt-server#s" of
the second group is {9,11,13,15}. Property "1" is indicative of
the thread in the group sharing L1 cache, translation cache and
Instruction Data flow.
b) provides information of Property "2" being shared by "2" groups,
each group with "4" threads. The "ibm,ppc-interrupt-server#s" of
the first group is {8,10,12,14} and the
"ibm,ppc-interrupt-server#s" of the second group is
{9,11,13,15}. Property "2" indicates that the threads in each group
share the L2-cache.
The existing code assumes that the "ibm,thread-groups" encodes
information about only one property. Hence even on platforms which
encode information about multiple properties being shared by the
corresponding groups of threads, the current code will only pick the
first one. (In the above example, it will only consider
[1,2,4,8,10,12,14,9,11,13,15] but not [2,2,4,8,10,12,14,9,11,13,15]).
This patch extends the parsing support on platforms which encode
information about multiple properties being shared by the
corresponding groups of threads.
Signed-off-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/1607596739-32439-2-git-send-email-ego@linux.vnet.ibm.com
POWER10 DD1 has an issue where it generates watchpoint exceptions when
it shouldn't. The conditions where this occur are:
- octword op
- ending address of DAWR range is less than starting address of op
- those addresses need to be in the same or in two consecutive 512B
blocks
- 'op address + 64B' generates an address that has a carry into bit
52 (crosses 2K boundary)
Handle such spurious exception by considering them as extraneous and
emulating/single-steeping instruction without generating an event.
[ravi: Fixed build warning reported by lkp@intel.com]
Signed-off-by: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20201106045650.278987-1-ravi.bangoria@linux.ibm.com
Perf event attritube supports exclude_kernel flag to avoid
sampling/profiling in supervisor state (kernel). Based on this event
attr flag, Monitor Mode Control Register bit is set to freeze on
supervisor state. But sometimes (due to hardware limitation), Sampled
Instruction Address Register (SIAR) locks on to kernel address even
when freeze on supervisor is set. Patch here adds a check to drop
those samples.
Cc: stable@vger.kernel.org
Signed-off-by: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/1606289215-1433-1-git-send-email-atrajeev@linux.vnet.ibm.com
This is way to catch some cases of decrementer overflow, when the
decrementer has underflowed an odd number of times, while MSR[EE] was
disabled.
With a typical small decrementer, a timer that fires when MSR[EE] is
disabled will be "lost" if MSR[EE] remains disabled for between 4.3 and
8.6 seconds after the timer expires. In any case, the decrementer
interrupt would be taken at 8.6 seconds and the timer would be found at
that point.
So this check is for catching extreme latency events, and it prevents
those latencies from being a further few seconds long. It's not obvious
this is a good tradeoff. This is already a watchdog magnitude event and
that situation is not improved a significantly with this check. For
large decrementers, it's useless.
Therefore remove this check, which avoids a mftb when enabling hard
disabled interrupts (e.g., when enabling after coming from hardware
interrupt handlers). Perhaps more importantly, it also removes the
clunky MSR[EE] vs PACA_IRQ_HARD_DIS incoherency in soft-interrupt replay
which simplifies the code.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20201107014336.2337337-1-npiggin@gmail.com
All hugetlb range freeing functions have a verification like the following,
which only differs by the mask used, depending on the page table level.
start &= MASK;
if (start < floor)
return;
if (ceiling) {
ceiling &= MASK;
if (! ceiling)
return;
}
if (end - 1 > ceiling - 1)
return;
Refactor that into a helper function which takes the mask as
an argument, returning true when [start;end[ is not fully
contained inside [floor;ceiling[
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/16a571bb32eb6e8cd44bda484c8d81cd8a25e6d7.1604668827.git.christophe.leroy@csgroup.eu
Exception fixup doesn't require the heady full regs saving,
do it from do_page_fault() directly.
For that, split bad_page_fault() in two parts.
As bad_page_fault() can also be called from other places than
handle_page_fault(), it will still perform exception fixup and
fallback on __bad_page_fault().
handle_page_fault() directly calls __bad_page_fault() as the
exception fixup will now be done by do_page_fault()
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Reviewed-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/bd07d6fef9237614cd6d318d8f19faeeadaa816b.1607491748.git.christophe.leroy@csgroup.eu
search_exception_tables() is an heavy operation, we have to avoid it.
When KUAP is selected, we'll know the fault has been blocked by KUAP.
When it is blocked by KUAP, check whether we are in an expected
userspace access place. If so, emit a warning to spot something is
going work. Otherwise, just remain silent, it will likely Oops soon.
When KUAP is not selected, it behaves just as if the address was
already in the TLBs and no fault was generated.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Reviewed-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/9870f01e293a5a76c4f4e4ddd4a6b0f63038c591.1607491748.git.christophe.leroy@csgroup.eu
On hash 32 bits, handling minor protection faults like unsetting
dirty flag is heavy if done from the normal page_fault processing,
because it implies hash table software lookup for flushing the entry
and then a DSI is taken anyway to add the entry back.
When KUAP was implemented, as explained in commit a68c31fc01
("powerpc/32s: Implement Kernel Userspace Access Protection"),
protection faults has been diverted from hash_page() because
hash_page() was not able to identify a KUAP fault.
Implement KUAP verification in hash_page(), by clearing write
permission when the access is a kernel access and Ks is 1.
This works regardless of the address because kernel segments always
have Ks set to 0 while user segments have Ks set to 0 only
when kernel write to userspace is granted.
Then protection faults can be handled by hash_page() even for KUAP.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/8a4ffe4798e9ea32aaaccdf85e411bb1beed3500.1605542955.git.christophe.leroy@csgroup.eu
book3s/32 has two main families:
- CPU with 603 cores that don't have HASH PTE table and
perform SW TLB loading.
- Other CPUs based on 604+ cores that have HASH PTE table.
This leads to some complex logic and additionnal code to
support both. This makes sense for distribution kernels
that aim at running on any CPU, but when you are fine
tuning a kernel for an embedded 603 based board you
don't need all the HASH logic.
Allow selection of support for each family, in order to opt
out unneeded parts of code. At least one must be selected.
Note that some of the CPU supporting HASH also support SW TLB
loading, however it is not supported by Linux kernel at the
time being, because they do not have alternate registers in
the TLB miss exception handlers.
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/8dde0cdb629a71abc29b0d85a52a86e920376cb6.1603348103.git.christophe.leroy@csgroup.eu
