firefly-linux-kernel-4.4.55.git
10 years agopowerpc/powernv: Set memory_block_size_bytes to 256MB
Anton Blanchard [Wed, 4 Jun 2014 07:52:42 +0000 (17:52 +1000)]
powerpc/powernv: Set memory_block_size_bytes to 256MB

powerpc sets a low SECTION_SIZE_BITS to accomodate small pseries
boxes. We default to 16MB memory blocks, and boxes with a lot
of memory end up with enormous numbers of sysfs memory nodes.

Set a more reasonable default for powernv of 256MB.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Allow ppc_md platform hook to override memory_block_size_bytes
Anton Blanchard [Wed, 4 Jun 2014 07:50:47 +0000 (17:50 +1000)]
powerpc: Allow ppc_md platform hook to override memory_block_size_bytes

The pseries platform code unconditionally overrides
memory_block_size_bytes regardless of the running platform.

Create a ppc_md hook that so each platform can choose to
do what it wants.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/powernv: Fix endian issues in memory error handling code
Anton Blanchard [Wed, 4 Jun 2014 04:48:48 +0000 (14:48 +1000)]
powerpc/powernv: Fix endian issues in memory error handling code

struct OpalMemoryErrorData is passed to us from firmware, so we
have to byteswap it.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/eeh: Skip eeh sysfs when eeh is disabled
Wei Yang [Wed, 4 Jun 2014 01:49:17 +0000 (09:49 +0800)]
powerpc/eeh: Skip eeh sysfs when eeh is disabled

When eeh is not enabled, and hotplug two pci devices on the same bus, eeh
related sysfs would be added twice for the first added pci device. Since the
eeh_dev is not created when eeh is not enabled.

This patch adds the check, if eeh is not enabled, eeh sysfs will not be
created.

After applying this patch, following warnings are reduced:

sysfs: cannot create duplicate filename '/devices/pci0000:00/0000:00:00.0/eeh_mode'
sysfs: cannot create duplicate filename '/devices/pci0000:00/0000:00:00.0/eeh_config_addr'
sysfs: cannot create duplicate filename '/devices/pci0000:00/0000:00:00.0/eeh_pe_config_addr'

Signed-off-by: Wei Yang <weiyang@linux.vnet.ibm.com>
Acked-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: 64bit sendfile is capped at 2GB
Anton Blanchard [Wed, 4 Jun 2014 00:48:48 +0000 (10:48 +1000)]
powerpc: 64bit sendfile is capped at 2GB

commit 8f9c0119d7ba (compat: fs: Generic compat_sys_sendfile
implementation) changed the PowerPC 64bit sendfile call from
sys_sendile64 to sys_sendfile.

Unfortunately this broke sendfile of lengths greater than 2G because
sys_sendfile caps at MAX_NON_LFS. Restore what we had previously which
fixes the bug.

Cc: stable@vger.kernel.org
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/powernv: Provide debugfs access to the LPC bus via OPAL
Benjamin Herrenschmidt [Tue, 3 Jun 2014 10:07:39 +0000 (20:07 +1000)]
powerpc/powernv: Provide debugfs access to the LPC bus via OPAL

This provides debugfs files to access the LPC bus on Power8
non-virtualized using the appropriate OPAL firmware calls.

The usage is simple: one file per space (IO, MEM and FW),
lseek to the address and read/write the data. IO and MEM always
generate series of byte accesses. FW can generate word and dword
accesses if aligned properly.

Based on an original patch from Rob Lippert and reworked.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/serial: Use saner flags when creating legacy ports
Benjamin Herrenschmidt [Tue, 3 Jun 2014 07:33:41 +0000 (17:33 +1000)]
powerpc/serial: Use saner flags when creating legacy ports

We had a mix & match of flags used when creating legacy ports
depending on where we found them in the device-tree. Among others
we were missing UPF_SKIP_TEST for some kind of ISA ports which is
a problem as quite a few UARTs out there don't support the loopback
test (such as a lot of BMCs).

Let's pick the set of flags used by the SoC code and generalize it
which means autoconf, no loopback test, irq maybe shared and fixed
port.

Sending to stable as the lack of UPF_SKIP_TEST is breaking
serial on some machines so I want this back into distros

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: stable@vger.kernel.org
10 years agopowerpc: Add cpu family documentation
Michael Ellerman [Mon, 26 May 2014 11:44:25 +0000 (21:44 +1000)]
powerpc: Add cpu family documentation

This patch adds some documentation on the different cpu families
supported by arch/powerpc.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/xmon: Fix up xmon format strings
Michael Ellerman [Mon, 26 May 2014 11:02:14 +0000 (21:02 +1000)]
powerpc/xmon: Fix up xmon format strings

There are a couple of places where xmon is using %x to print values that
are unsigned long.

I found this out the hard way recently:

 0:mon> p c000000000d0e7c8 c00000033dc90000 00000000a0000089 c000000000000000
 return value is 0x96300500

Which is calling find_linux_pte_or_hugepte(), the result should be a
kernel pointer. After decoding the page tables by hand I discovered the
correct value was c000000396300500.

So fix up that case and a few others.

We also use a mix of 0x%x, %x and %u to print cpu numbers. So
standardise on 0x%x.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/powernv: Add calls to support little endian host
Benjamin Herrenschmidt [Tue, 20 May 2014 01:01:28 +0000 (11:01 +1000)]
powerpc/powernv: Add calls to support little endian host

When running as a powernv "host" system on P8, we need to switch
the endianness of interrupt handlers. This does it via the appropriate
call to the OPAL firmware which may result in just switching HID0:HILE
but depending on the processor version might need to do a few more
things. This call must be done early before any other processor has
been brought out of firmware.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andy Whitcroft <apw@canonical.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Document sysfs DSCR interface
Sam bobroff [Wed, 21 May 2014 06:32:39 +0000 (16:32 +1000)]
powerpc: Document sysfs DSCR interface

Add some documentation about ...

/sys/devices/system/cpu/dscr_default
/sys/devices/system/cpu/cpuN/dscr

... to Documentation/ABI/stable.

Signed-off-by: Sam Bobroff <sam.bobroff@au1.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Fix regression of per-CPU DSCR setting
Sam bobroff [Wed, 21 May 2014 06:32:38 +0000 (16:32 +1000)]
powerpc: Fix regression of per-CPU DSCR setting

Since commit "efcac65 powerpc: Per process DSCR + some fixes (try#4)"
it is no longer possible to set the DSCR on a per-CPU basis.

The old behaviour was to minipulate the DSCR SPR directly but this is no
longer sufficient: the value is quickly overwritten by context switching.

This patch stores the per-CPU DSCR value in a kernel variable rather than
directly in the SPR and it is used whenever a process has not set the DSCR
itself. The sysfs interface (/sys/devices/system/cpu/cpuN/dscr) is unchanged.

Writes to the old global default (/sys/devices/system/cpu/dscr_default)
now set all of the per-CPU values and reads return the last written value.

The new per-CPU default is added to the paca_struct and is used everywhere
outside of sysfs.c instead of the old global default.

Signed-off-by: Sam Bobroff <sam.bobroff@au1.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Split __SYSFS_SPRSETUP macro
Sam bobroff [Wed, 21 May 2014 06:32:37 +0000 (16:32 +1000)]
powerpc: Split __SYSFS_SPRSETUP macro

Split the __SYSFS_SPRSETUP macro into two parts so that registers requiring
custom read and write functions can use common code for their show and store
functions.

Signed-off-by: Sam Bobroff <sam.bobroff@au1.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agoarch: powerpc/fadump: Cleaning up inconsistent NULL checks
Rickard Strandqvist [Thu, 22 May 2014 22:03:16 +0000 (00:03 +0200)]
arch: powerpc/fadump: Cleaning up inconsistent NULL checks

Cleaning up inconsistent NULL checks.
There is otherwise a risk of a possible null pointer dereference.

Was largely found by using a static code analysis program called cppcheck.

Signed-off-by: Rickard Strandqvist <rickard_strandqvist@spectrumdigital.se>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Fix comment around arch specific definition of RECLAIM_DISTANCE
Preeti U Murthy [Mon, 5 May 2014 05:17:14 +0000 (10:47 +0530)]
powerpc: Fix comment around arch specific definition of RECLAIM_DISTANCE

Commit 32e45ff43eaf5c17f changed the default value of
RECLAIM_DISTANCE to 30. However the comment around arch
specifc definition of RECLAIM_DISTANCE is not updated to
reflect the same. Correct the value mentioned in the comment.

Signed-off-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Anton Blanchard <anton@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: KOSAKI Motohiro <Kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/powernv: Add support for POWER8 split core on powernv
Michael Ellerman [Fri, 23 May 2014 08:15:30 +0000 (18:15 +1000)]
powerpc/powernv: Add support for POWER8 split core on powernv

Upcoming POWER8 chips support a concept called split core. This is where the
core can be split into subcores that although not full cores, are able to
appear as full cores to a guest.

The splitting & unsplitting procedure is mildly complicated, and explained at
length in the comments within the patch.

One notable detail is that when splitting or unsplitting we need to pull
offline cpus out of their offline state to do work as part of the procedure.

The interface for changing the split mode is via a sysfs file, eg:

 $ echo 2 > /sys/devices/system/cpu/subcores_per_core

Currently supported values are '1', '2' and '4'. And indicate respectively that
the core should be unsplit, split in half, and split in quarters. These modes
correspond to threads_per_subcore of 8, 4 and 2.

We do not allow changing the split mode while KVM VMs are active. This is to
prevent the value changing while userspace is configuring the VM, and also to
prevent the mode being changed in such a way that existing guests are unable to
be run.

CPU hotplug fixes by Srivatsa.  max_cpus fixes by Mahesh.  cpuset fixes by
benh.  Fix for irq race by paulus.  The rest by mikey and mpe.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/kvm/book3s_hv: Use threads_per_subcore in KVM
Michael Ellerman [Fri, 23 May 2014 08:15:29 +0000 (18:15 +1000)]
powerpc/kvm/book3s_hv: Use threads_per_subcore in KVM

To support split core on POWER8 we need to modify various parts of the
KVM code to use threads_per_subcore instead of threads_per_core. On
systems that do not support split core threads_per_subcore ==
threads_per_core and these changes are a nop.

We use threads_per_subcore as the value reported by KVM_CAP_PPC_SMT.
This communicates to userspace that guests can only be created with
a value of threads_per_core that is less than or equal to the current
threads_per_subcore. This ensures that guests can only be created with a
thread configuration that we are able to run given the current split
core mode.

Although threads_per_subcore can change during the life of the system,
the commit that enables that will ensure that threads_per_subcore does
not change during the life of a KVM VM.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Michael Neuling <mikey@neuling.org>
Acked-by: Alexander Graf <agraf@suse.de>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Check cpu_thread_in_subcore() in __cpu_up()
Michael Ellerman [Fri, 23 May 2014 08:15:28 +0000 (18:15 +1000)]
powerpc: Check cpu_thread_in_subcore() in __cpu_up()

To support split core we need to change the check in __cpu_up() that
determines if a cpu is allowed to come online.

Currently we refuse to online cpus which are not the primary thread
within their core.

On POWER8 with split core support this check needs to instead refuse to
online cpus which are not the primary thread within their *sub* core.

On POWER7 and other systems that do not support split core,
threads_per_subcore == threads_per_core and so the check is equivalent.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Add threads_per_subcore
Michael Ellerman [Fri, 23 May 2014 08:15:27 +0000 (18:15 +1000)]
powerpc: Add threads_per_subcore

On POWER8 we have a new concept of a subcore. This is what happens when
you take a regular core and split it. A subcore is a grouping of two or
four SMT threads, as well as a handfull of SPRs which allows the subcore
to appear as if it were a core from the point of view of a guest.

Unlike threads_per_core which is fixed at boot, threads_per_subcore can
change while the system is running. Most code will not want to use
threads_per_subcore.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/powernv: Make it possible to skip the IRQHAPPENED check in power7_nap()
Michael Ellerman [Fri, 23 May 2014 08:15:26 +0000 (18:15 +1000)]
powerpc/powernv: Make it possible to skip the IRQHAPPENED check in power7_nap()

To support split core we need to be able to force all secondaries into
nap, so the core can detect they are idle and do an unsplit.

Currently power7_nap() will return without napping if there is an irq
pending. We want to ignore the pending irq and nap anyway, we will deal
with the interrupt later.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/kvm/book3s_hv: Rework the secondary inhibit code
Michael Ellerman [Fri, 23 May 2014 08:15:25 +0000 (18:15 +1000)]
powerpc/kvm/book3s_hv: Rework the secondary inhibit code

As part of the support for split core on POWER8, we want to be able to
block splitting of the core while KVM VMs are active.

The logic to do that would be exactly the same as the code we currently
have for inhibiting onlining of secondaries.

Instead of adding an identical mechanism to block split core, rework the
secondary inhibit code to be a "HV KVM is active" check. We can then use
that in both the cpu hotplug code and the upcoming split core code.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Michael Neuling <mikey@neuling.org>
Acked-by: Alexander Graf <agraf@suse.de>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/numa: Enable CONFIG_HAVE_MEMORYLESS_NODES
Nishanth Aravamudan [Fri, 16 May 2014 23:41:20 +0000 (16:41 -0700)]
powerpc/numa: Enable CONFIG_HAVE_MEMORYLESS_NODES

Based off fd1197f1 for ia64, enable CONFIG_HAVE_MEMORYLESS_NODES if
NUMA. Initialize the local memory node in start_secondary.

With this commit and the preceding to enable
CONFIG_USER_PERCPU_NUMA_NODE_ID, which is a prerequisite, in a PowerKVM
guest with the following topology:

numactl --hardware
available: 3 nodes (0-2)
node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46
47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70
71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94
95 96 97 98 99
node 0 size: 1998 MB
node 0 free: 521 MB
node 1 cpus: 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114
115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132
133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150
151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168
169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186
187 188 189 190 191 192 193 194 195 196 197 198 199
node 1 size: 0 MB
node 1 free: 0 MB
node 2 cpus:
node 2 size: 2039 MB
node 2 free: 1739 MB
node distances:
node   0   1   2
  0:  10  40  40
  1:  40  10  40
  2:  40  40  10

the unreclaimable slab is reduced by close to 130M:

Before:
        Slab:             418176 kB
        SReclaimable:      26624 kB
        SUnreclaim:       391552 kB

After:
        Slab:             298944 kB
        SReclaimable:      31744 kB
        SUnreclaim:       267200 kB

Signed-off-by: Nishanth Aravamudan <nacc@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/numa: Enable USE_PERCPU_NUMA_NODE_ID
Nishanth Aravamudan [Mon, 19 May 2014 18:14:23 +0000 (11:14 -0700)]
powerpc/numa: Enable USE_PERCPU_NUMA_NODE_ID

Based off 3bccd996 for ia64, convert powerpc to use the generic per-CPU
topology tracking, specifically:

    initialize per cpu numa_node entry in start_secondary
    remove the powerpc cpu_to_node()
    define CONFIG_USE_PERCPU_NUMA_NODE_ID if NUMA

Signed-off-by: Nishanth Aravamudan <nacc@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agoMerge branch 'merge' into next
Benjamin Herrenschmidt [Wed, 28 May 2014 03:30:12 +0000 (13:30 +1000)]
Merge branch 'merge' into next

Merge the binutils and kexec fixes.

10 years agopowerpc, kexec: Fix "Processor X is stuck" issue during kexec from ST mode
Srivatsa S. Bhat [Tue, 27 May 2014 10:55:34 +0000 (16:25 +0530)]
powerpc, kexec: Fix "Processor X is stuck" issue during kexec from ST mode

If we try to perform a kexec when the machine is in ST (Single-Threaded) mode
(ppc64_cpu --smt=off), the kexec operation doesn't succeed properly, and we
get the following messages during boot:

[    0.089866] POWER8 performance monitor hardware support registered
[    0.089985] power8-pmu: PMAO restore workaround active.
[    5.095419] Processor 1 is stuck.
[   10.097933] Processor 2 is stuck.
[   15.100480] Processor 3 is stuck.
[   20.102982] Processor 4 is stuck.
[   25.105489] Processor 5 is stuck.
[   30.108005] Processor 6 is stuck.
[   35.110518] Processor 7 is stuck.
[   40.113369] Processor 9 is stuck.
[   45.115879] Processor 10 is stuck.
[   50.118389] Processor 11 is stuck.
[   55.120904] Processor 12 is stuck.
[   60.123425] Processor 13 is stuck.
[   65.125970] Processor 14 is stuck.
[   70.128495] Processor 15 is stuck.
[   75.131316] Processor 17 is stuck.

Note that only the sibling threads are stuck, while the primary threads (0, 8,
16 etc) boot just fine. Looking closer at the previous step of kexec, we observe
that kexec tries to wakeup (bring online) the sibling threads of all the cores,
before performing kexec:

[ 9464.131231] Starting new kernel
[ 9464.148507] kexec: Waking offline cpu 1.
[ 9464.148552] kexec: Waking offline cpu 2.
[ 9464.148600] kexec: Waking offline cpu 3.
[ 9464.148636] kexec: Waking offline cpu 4.
[ 9464.148671] kexec: Waking offline cpu 5.
[ 9464.148708] kexec: Waking offline cpu 6.
[ 9464.148743] kexec: Waking offline cpu 7.
[ 9464.148779] kexec: Waking offline cpu 9.
[ 9464.148815] kexec: Waking offline cpu 10.
[ 9464.148851] kexec: Waking offline cpu 11.
[ 9464.148887] kexec: Waking offline cpu 12.
[ 9464.148922] kexec: Waking offline cpu 13.
[ 9464.148958] kexec: Waking offline cpu 14.
[ 9464.148994] kexec: Waking offline cpu 15.
[ 9464.149030] kexec: Waking offline cpu 17.

Instrumenting this piece of code revealed that the cpu_up() operation actually
fails with -EBUSY. Thus, only the primary threads of all the cores are online
during kexec, and hence this is a sure-shot receipe for disaster, as explained
in commit e8e5c2155b (powerpc/kexec: Fix orphaned offline CPUs across kexec),
as well as in the comment above wake_offline_cpus().

It turns out that cpu_up() was returning -EBUSY because the variable
'cpu_hotplug_disabled' was set to 1; and this disabling of CPU hotplug was done
by migrate_to_reboot_cpu() inside kernel_kexec().

Now, migrate_to_reboot_cpu() was originally written with the assumption that
any further code will not need to perform CPU hotplug, since we are anyway in
the reboot path. However, kexec is clearly not such a case, since we depend on
onlining CPUs, atleast on powerpc.

So re-enable cpu-hotplug after returning from migrate_to_reboot_cpu() in the
kexec path, to fix this regression in kexec on powerpc.

Also, wrap the cpu_up() in powerpc kexec code within a WARN_ON(), so that we
can catch such issues more easily in the future.

Fixes: c97102ba963 (kexec: migrate to reboot cpu)
Cc: stable@vger.kernel.org
Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Fix 64 bit builds with binutils 2.24
Guenter Roeck [Thu, 15 May 2014 16:33:42 +0000 (09:33 -0700)]
powerpc: Fix 64 bit builds with binutils 2.24

With binutils 2.24, various 64 bit builds fail with relocation errors
such as

arch/powerpc/kernel/built-in.o: In function `exc_debug_crit_book3e':
(.text+0x165ee): relocation truncated to fit: R_PPC64_ADDR16_HI
against symbol `interrupt_base_book3e' defined in .text section
in arch/powerpc/kernel/built-in.o
arch/powerpc/kernel/built-in.o: In function `exc_debug_crit_book3e':
(.text+0x16602): relocation truncated to fit: R_PPC64_ADDR16_HI
against symbol `interrupt_end_book3e' defined in .text section
in arch/powerpc/kernel/built-in.o

The assembler maintainer says:

 I changed the ABI, something that had to be done but unfortunately
 happens to break the booke kernel code.  When building up a 64-bit
 value with lis, ori, shl, oris, ori or similar sequences, you now
 should use @high and @higha in place of @h and @ha.  @h and @ha
 (and their associated relocs R_PPC64_ADDR16_HI and R_PPC64_ADDR16_HA)
 now report overflow if the value is out of 32-bit signed range.
 ie. @h and @ha assume you're building a 32-bit value. This is needed
 to report out-of-range -mcmodel=medium toc pointer offsets in @toc@h
 and @toc@ha expressions, and for consistency I did the same for all
 other @h and @ha relocs.

Replacing @h with @high in one strategic location fixes the relocation
errors. This has to be done conditionally since the assembler either
supports @h or @high but not both.

Cc: <stable@vger.kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agoMerge remote-tracking branch 'scott/next' into next
Benjamin Herrenschmidt [Wed, 28 May 2014 00:02:58 +0000 (10:02 +1000)]
Merge remote-tracking branch 'scott/next' into next

<<
Highlights include a few new boards, a device tree binding for CCF
(including backwards-compatible device tree updates to distinguish
incompatible versions), and some fixes.
>>

10 years agopowerpc/fsl: Add fsl,portid-mapping to corenet1-cf chips
Scott Wood [Tue, 6 May 2014 01:35:10 +0000 (20:35 -0500)]
powerpc/fsl: Add fsl,portid-mapping to corenet1-cf chips

Signed-off-by: Scott Wood <scottwood@freescale.com>
Cc: Diana Craciun <diana.craciun@freescale.com>
10 years agoPPC: ePAPR: Fix hypercall on LE guest
Alexander Graf [Thu, 15 May 2014 12:35:19 +0000 (14:35 +0200)]
PPC: ePAPR: Fix hypercall on LE guest

We get an array of instructions from the hypervisor via device tree that
we write into a buffer that gets executed whenever we want to make an
ePAPR compliant hypercall.

However, the hypervisor passes us these instructions in BE order which
we have to manually convert to LE when we want to run them in LE mode.

With this fixup in place, I can successfully run LE kernels with KVM
PV enabled on PR KVM.

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc/mpc85xx: Add BSC9132 QDS Support
harninder rai [Thu, 15 May 2014 07:45:33 +0000 (13:15 +0530)]
powerpc/mpc85xx: Add BSC9132 QDS Support

- BSC9132 is an integrated device that targets Femto base station market.
  It combines Power Architecture e500v2 and DSP StarCore SC3850 technologies
  with MAPLE-B2F baseband acceleration processing elements

- BSC9132QDS Overview
     2Gbyte DDR3 (on board DDR)
     32Mbyte 16bit NOR flash
     128Mbyte 2K page size NAND Flash
     256 Kbit M24256 I2C EEPROM
     128 Mbit SPI Flash memory
     SD slot
     eTSEC1: Connected to SGMII PHY
     eTSEC2: Connected to SGMII PHY
     DUART interface: supports one UARTs up to 115200 bps for console display

Signed-off-by: Harninder Rai <harninder.rai@freescale.com>
Signed-off-by: Ruchika Gupta <ruchika.gupta@freescale.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc/mpc85xx: Remove P1023 RDS support
Lijun Pan [Mon, 5 May 2014 18:23:15 +0000 (13:23 -0500)]
powerpc/mpc85xx: Remove P1023 RDS support

P1023RDS is no longer supported/manufactured by Freescale while P1023RDB is.

Signed-off-by: Lijun Pan <Lijun.Pan@freescale.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc/mpic: Don't init the fsl error int until after mpic init
Scott Wood [Wed, 21 May 2014 01:26:01 +0000 (20:26 -0500)]
powerpc/mpic: Don't init the fsl error int until after mpic init

Besides other potential problems, if MPIC_NO_RESET is  not set,
the error interrupt will be masked after it is requested.

Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc/fsl-booke: Add initial T104x_QDS board support
Prabhakar Kushwaha [Mon, 21 Apr 2014 11:34:45 +0000 (17:04 +0530)]
powerpc/fsl-booke: Add initial T104x_QDS board support

Add support for T104x board in board file t104x_qds.c, It is common for
 both T1040 and T1042 as they share same QDS board.

 T1040QDS board Overview
 -----------------------
 - SERDES Connections, 8 lanes supporting:
      — PCI Express: supporting Gen 1 and Gen 2;
      — SGMII
      — QSGMII
      — SATA 2.0
      — Aurora debug with dedicated connectors (T1040 only)
 - DDR Controller
     - Supports rates of up to 1600 MHz data-rate
     - Supports one DDR3LP UDIMM/RDIMMs, of single-, dual- or quad-rank types.
 -IFC/Local Bus
     - NAND flash: 8-bit, async, up to 2GB.
     - NOR: 8-bit or 16-bit, non-multiplexed, up to 512MB
     - GASIC: Simple (minimal) target within Qixis FPGA
     - PromJET rapid memory download support
 - Ethernet
     - Two on-board RGMII 10/100/1G ethernet ports.
     - PHY #0 remains powered up during deep-sleep (T1040 only)
 - QIXIS System Logic FPGA
 - Clocks
     - System and DDR clock (SYSCLK, “DDRCLK”)
     - SERDES clocks
 - Power Supplies
 - Video
     - DIU supports video at up to 1280x1024x32bpp
 - USB
     - Supports two USB 2.0 ports with integrated PHYs
     — Two type A ports with 5V@1.5A per port.
     — Second port can be converted to OTG mini-AB
 - SDHC
     - SDHC port connects directly to an adapter card slot, featuring:
     - Supporting SD slots for: SD, SDHC (1x, 4x, 8x) and/or MMC
     — Supporting eMMC memory devices
 - SPI
    -  On-board support of 3 different devices and sizes
 - Other IO
    - Two Serial ports
    - ProfiBus port
    - Four I2C ports

Add T104xQDS support in Kconfig and Makefile. Also create device tree.
Following features are currently not implmented.
  - SerDes: Aurora
  - IFC: GASIC, Promjet
  - QIXIS
  - Ethernet
  - DIU
  - power supplies management
  - ProfiBus

Signed-off-by: Priyanka Jain <Priyanka.Jain@freescale.com>
Signed-off-by: Poonam Aggrwal <poonam.aggrwal@freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc/mpc85xx:Add initial device tree support of T104x
Prabhakar Kushwaha [Mon, 21 Apr 2014 11:34:28 +0000 (17:04 +0530)]
powerpc/mpc85xx:Add initial device tree support of T104x

The QorIQ T1040/T1042 processor support four integrated 64-bit e5500 PA
processor cores with high-performance data path acceleration architecture
and network peripheral interfaces required for networking & telecommunications.

T1042 personality is a reduced personality of T1040 without Integrated 8-port
Gigabit Ethernet switch.

The T1040/T1042 SoC includes the following function and features:

 - Four e5500 cores, each with a private 256 KB L2 cache
 - 256 KB shared L3 CoreNet platform cache (CPC)
 - Interconnect CoreNet platform
 - 32-/64-bit DDR3L/DDR4 SDRAM memory controller with ECC and interleaving
   support
 - Data Path Acceleration Architecture (DPAA) incorporating acceleration
 for the following functions:
    -  Packet parsing, classification, and distribution
    -  Queue management for scheduling, packet sequencing, and congestion
     management
    -  Cryptography Acceleration (SEC 5.0)
    - RegEx Pattern Matching Acceleration (PME 2.2)
    - IEEE Std 1588 support
    - Hardware buffer management for buffer allocation and deallocation
 - Ethernet interfaces
    - Integrated 8-port Gigabit Ethernet switch (T1040 only)
    - Four 1 Gbps Ethernet controllers
 - Two RGMII interfaces or one RGMII and one MII interfaces
 - High speed peripheral interfaces
   - Four PCI Express 2.0 controllers running at up to 5 GHz
   - Two SATA controllers supporting 1.5 and 3.0 Gb/s operation
   - Upto two QSGMII interface
   - Upto six SGMII interface supporting 1000 Mbps
   - One SGMII interface supporting upto 2500 Mbps
 - Additional peripheral interfaces
   - Two USB 2.0 controllers with integrated PHY
   - SD/eSDHC/eMMC
   -  eSPI controller
   - Four I2C controllers
   - Four UARTs
   - Four GPIO controllers
   - Integrated flash controller (IFC)
   - Change this to  LCD/ HDMI interface (DIU) with 12 bit dual data rate
   - TDM interface
 - Multicore programmable interrupt controller (PIC)
 - Two 8-channel DMA engines
 - Single source clocking implementation
 - Deep Sleep power implementaion (wakeup from GPIO/Timer/Ethernet/USB)

Signed-off-by: Poonam Aggrwal <poonam.aggrwal@freescale.com>
Signed-off-by: Priyanka Jain <Priyanka.Jain@freescale.com>
Signed-off-by: Varun Sethi <Varun.Sethi@freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc/fsl: Updated corenet-cf compatible string for corenet1-cf chips
Diana Craciun [Wed, 7 May 2014 06:29:17 +0000 (09:29 +0300)]
powerpc/fsl: Updated corenet-cf compatible string for corenet1-cf chips

Updated the device trees according to the corenet-cf
binding definition.

Signed-off-by: Diana Craciun <Diana.Craciun@freescale.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc/fsl: Updated device trees for platforms with corenet version 2
Diana Craciun [Mon, 5 May 2014 16:04:27 +0000 (19:04 +0300)]
powerpc/fsl: Updated device trees for platforms with corenet version 2

Updated the device trees according to the corenet-cf
binding definition.

Signed-off-by: Diana Craciun <Diana.Craciun@freescale.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc/fsl: Added binding for Freescale CoreNet coherency fabric (CCF)
Diana Craciun [Mon, 5 May 2014 15:58:19 +0000 (18:58 +0300)]
powerpc/fsl: Added binding for Freescale CoreNet coherency fabric (CCF)

The CoreNet coherency fabric is a fabric-oriented, conectivity
infrastructure that enables the implementation of coherent, multicore
systems. The CCF acts as a central interconnect for cores,
platform-level caches, memory subsystem, peripheral devices and I/O host
bridges in the system.

Signed-off-by: Diana Craciun <Diana.Craciun@freescale.com>
[scottwood@freescale.com: formatting and minor changes]
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc: Fix unused variable warning for epapr_has_idle
Scott Wood [Thu, 15 May 2014 16:28:26 +0000 (11:28 -0500)]
powerpc: Fix unused variable warning for epapr_has_idle

This warning can be seen in allyesconfig, and was introduced by commit
f9eb581c63b2acce827570e105205c0789360650 "powerpc: fix build of
epapr_paravirt on 64-bit book3s".

Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc: fix build of epapr_paravirt on 64-bit book3s
Scott Wood [Mon, 12 May 2014 15:05:19 +0000 (10:05 -0500)]
powerpc: fix build of epapr_paravirt on 64-bit book3s

This fixes an allyesconfig build break introduced by commit
7762b1ed7aaee223230793fcee80672e2e3aa7a8 "powerpc: move epapr paravirt
init of power_save to an initcall".

Signed-off-by: Scott Wood <scottwood@freescale.com>
Cc: Stuart Yoder <stuart.yoder@freescale.com>
10 years agoclk: qoriq: Update the clock bindings
Tang Yuantian [Thu, 8 May 2014 03:12:10 +0000 (11:12 +0800)]
clk: qoriq: Update the clock bindings

Main changs include:
- Clarified the clock nodes' version number
- Fixed a issue in example

Singed-off-by: Tang Yuantian <Yuantian.Tang@freescale.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc: move epapr paravirt init of power_save to an initcall
Stuart Yoder [Wed, 30 Apr 2014 23:34:23 +0000 (18:34 -0500)]
powerpc: move epapr paravirt init of power_save to an initcall

some restructuring of epapr paravirt init resulted in
ppc_md.power_save being set, and then overwritten to
NULL during machine_init.  This patch splits the
initialization of ppc_md.power_save out into a postcore
init call.

Signed-off-by: Stuart Yoder <stuart.yoder@freescale.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc/85xx: Add OCA4080 board support
Martijn de Gouw [Tue, 15 Apr 2014 17:51:46 +0000 (19:51 +0200)]
powerpc/85xx: Add OCA4080 board support

OCA4080 overview:
- 1.466 GHz Freescale QorIQ P4080E Processor
- 4Gbyte DDR3 on board
- 8Mbyte Nor flash
- Serial RapidIO 1.2
- 1 x 10/100/1000 BASE-T front ethernet
- 1 x 1000 BASE-BX ethernet on AMC connector

Signed-off-by: Martijn de Gouw <martijn.de.gouw@prodrive.nl>
[scottwood@freescale.com: minor conflict-related changes]
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc/mpc85xx: add support for Keymile's kmcoge4 board
Valentin Longchamp [Tue, 22 Apr 2014 09:30:31 +0000 (11:30 +0200)]
powerpc/mpc85xx: add support for Keymile's kmcoge4 board

This patch introduces the support for Keymile's kmcoge4 board which is
the internal reference design for boards based on Freescale's
P2040/P2041 SoCs. This internal reference design is named kmp204x.

The peripherals used on this board are:
- SPI NOR Flash as bootloader medium
- NAND Flash with a ubi partition
- 2 PCIe busses (hosts 1 and 3)
- 3 FMAN Ethernet devices (FMAN1 DTSEC1/2/5)
- 4 Local Bus windows, with one dedicated to the QRIO reset/power mgmt
  CPLD
- 2 I2C busses
- last but not least, the mandatory serial port

The patch also adds a defconfig file for this reference design that is
necessary because of the lowmem option that must be set higher due to
the number of PCIe devices with big ioremapped mem ranges on the boad.

Signed-off-by: Valentin Longchamp <valentin.longchamp@keymile.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agodevcietree: bindings: add some MFD Keymile FPGAs
Valentin Longchamp [Tue, 22 Apr 2014 09:30:30 +0000 (11:30 +0200)]
devcietree: bindings: add some MFD Keymile FPGAs

These are the bindings for 2 MFD devices used on some of the Keymile boards.
The first one is the chassis managmenet bfticu FPGA.
The second one is the board controller (reset, LEDs, GPIOs) QRIO CPDL.
These FPGAs are used in the kmcoge4 board.

Signed-off-by: Valentin Longchamp <valentin.longchamp@keymile.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agodevicetree: bindings: add Zarlink to the vendor prefixes
Valentin Longchamp [Tue, 22 Apr 2014 09:30:29 +0000 (11:30 +0200)]
devicetree: bindings: add Zarlink to the vendor prefixes

Even though the company belongs to Microsemi, many chips are still
labeled as Zarlink. Among them is the family of network clock generators,
the zl3034x.

Signed-off-by: Valentin Longchamp <valentin.longchamp@keymile.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agofsl/pci: fix RC cannot detect PME message coming
Wang Dongsheng [Tue, 15 Apr 2014 07:43:18 +0000 (15:43 +0800)]
fsl/pci: fix RC cannot detect PME message coming

PCI controller disable PME message report feature, that shouldn't
have happened. Fix it and enable PME message report feature.

Signed-off-by: Wang Dongsheng <dongsheng.wang@freescale.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc/rmu: Fix the error memory free parameters
Liu Gang [Wed, 9 Apr 2014 10:04:45 +0000 (18:04 +0800)]
powerpc/rmu: Fix the error memory free parameters

There are error parameters should be corrected when
calling dma_free_coherent to free rmu rx-ring buffers
in fsl_open_inb_mbox() function.

Signed-off-by: Liu Gang <Gang.Liu@freescale.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc/fsl-booke64: Set vmemmap_psize to 4K
Scott Wood [Tue, 6 May 2014 00:41:10 +0000 (19:41 -0500)]
powerpc/fsl-booke64: Set vmemmap_psize to 4K

The only way Freescale booke chips support mappings larger than 4K
is via TLB1.  The only way we support (direct) TLB1 entries is via
hugetlb, which is not what map_kernel_page() does when given a large
page size.

Without this, a kernel with CONFIG_SPARSEMEM_VMEMMAP enabled crashes on
boot with messages such as:

PID hash table entries: 4096 (order: 3, 32768 bytes)
Sorting __ex_table...
BUG: Bad page state in process swapper  pfn:00a2f
page:8000040000023a48 count:0 mapcount:0 mapping:0000040000ffce48 index:0x40000ffbe50
page flags: 0x40000ffda40(active|arch_1|private|private_2|head|tail|swapcache|mappedtodisk|reclaim|swapbacked|unevictable|mlocked)
page dumped because: PAGE_FLAGS_CHECK_AT_FREE flag(s) set
bad because of flags:
page flags: 0x311840(active|private|private_2|swapcache|unevictable|mlocked)
Modules linked in:
CPU: 0 PID: 0 Comm: swapper Not tainted 3.15.0-rc1-00003-g7fa250c #299
Call Trace:
[c00000000098ba20] [c000000000008b3c] .show_stack+0x7c/0x1cc (unreliable)
[c00000000098baf0] [c00000000060aa50] .dump_stack+0x88/0xb4
[c00000000098bb70] [c0000000000c0468] .bad_page+0x144/0x1a0
[c00000000098bc10] [c0000000000c0628] .free_pages_prepare+0x164/0x17c
[c00000000098bcc0] [c0000000000c24cc] .free_hot_cold_page+0x48/0x214
[c00000000098bd60] [c00000000086c318] .free_all_bootmem+0x1fc/0x354
[c00000000098be70] [c00000000085da84] .mem_init+0xac/0xdc
[c00000000098bef0] [c0000000008547b0] .start_kernel+0x21c/0x4d4
[c00000000098bf90] [c000000000000448] .start_here_common+0x20/0x58

Signed-off-by: Scott Wood <scottwood@freescale.com>
10 years agopowerpc/module: Fix stubs for BE
Rusty Russell [Mon, 12 May 2014 07:45:02 +0000 (17:15 +0930)]
powerpc/module: Fix stubs for BE

A simple patch which was supposed to swap r12 and r11 also
inexplicably changed the offset by two bytes.  This instruction
(to load r2) isn't used in LE, so it wasn't noticed.

Fixes: b1ce369e82 ("powerpc: modules: use r12 for stub jump address.)
Reported-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Tested-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agomacintosh/windfarm_pm121.c: Fix for possible null pointer dereference
Rickard Strandqvist [Sat, 17 May 2014 17:20:57 +0000 (19:20 +0200)]
macintosh/windfarm_pm121.c: Fix for possible null pointer dereference

There is otherwise a risk of a possible null pointer dereference.

Was largely found by using a static code analysis program called cppcheck.

Signed-off-by: Rickard Strandqvist <rickard_strandqvist@spectrumdigital.se>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Clear ELF personality flag if ELFv2 is not requested.
Jeff Bailey [Sat, 17 May 2014 15:05:43 +0000 (15:05 +0000)]
powerpc: Clear ELF personality flag if ELFv2 is not requested.

powerpc: Clear ELF personality flag if ELFv2 is not requested.

The POWER kernel uses a personality flag to determine whether it should
be setting up function descriptors or not (per the updated ABI).  This
flag wasn't being cleared on a new process but instead was being
inherited.  The visible effect was that an ELFv2 binary could not execve
to an ELFv1 binary.

Signed-off-by: Jeff Bailey <jeffbailey@google.com>
 arch/powerpc/include/asm/elf.h | 2 ++
 1 file changed, 2 insertions(+)
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Fix smp_processor_id() in preemptible splat in set_breakpoint
Paul Gortmaker [Tue, 29 Apr 2014 19:25:17 +0000 (15:25 -0400)]
powerpc: Fix smp_processor_id() in preemptible splat in set_breakpoint

Currently, on 8641D, which doesn't set CONFIG_HAVE_HW_BREAKPOINT
we get the following splat:

BUG: using smp_processor_id() in preemptible [00000000] code: login/1382
caller is set_breakpoint+0x1c/0xa0
CPU: 0 PID: 1382 Comm: login Not tainted 3.15.0-rc3-00041-g2aafe1a4d451 #1
Call Trace:
[decd5d80] [c0008dc4] show_stack+0x50/0x158 (unreliable)
[decd5dc0] [c03c6fa0] dump_stack+0x7c/0xdc
[decd5de0] [c01f8818] check_preemption_disabled+0xf4/0x104
[decd5e00] [c00086b8] set_breakpoint+0x1c/0xa0
[decd5e10] [c00d4530] flush_old_exec+0x2bc/0x588
[decd5e40] [c011c468] load_elf_binary+0x2ac/0x1164
[decd5ec0] [c00d35f8] search_binary_handler+0xc4/0x1f8
[decd5ef0] [c00d4ee8] do_execve+0x3d8/0x4b8
[decd5f40] [c001185c] ret_from_syscall+0x0/0x38
 --- Exception: c01 at 0xfeee554
    LR = 0xfeee7d4

The call path in this case is:

flush_thread
   --> set_debug_reg_defaults
     --> set_breakpoint
       --> __get_cpu_var

Since preemption is enabled in the cleanup of flush thread, and
there is no need to disable it, introduce the distinction between
set_breakpoint and __set_breakpoint, leaving only the flush_thread
instance as the current user of set_breakpoint.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Drop return value from set_breakpoint as it is unused
Paul Gortmaker [Tue, 29 Apr 2014 19:25:16 +0000 (15:25 -0400)]
powerpc: Drop return value from set_breakpoint as it is unused

None of the callers check the return value, so it might as
well not have one at all.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Remove non-uapi linkage.h export
James Hogan [Fri, 2 May 2014 10:20:41 +0000 (11:20 +0100)]
powerpc: Remove non-uapi linkage.h export

The arch/powerpc/include/asm/linkage.h is being unintentionally exported
in the kernel headers since commit e1b5bb6d1236 (consolidate
cond_syscall and SYSCALL_ALIAS declarations) when
arch/powerpc/include/uapi/asm/linkage.h was deleted but the header-y not
removed from the Kbuild file. This happens because Makefile.headersinst
still checks the old asm/ directory if the specified header doesn't
exist in the uapi directory.

The asm/linkage.h shouldn't ever have been exported anyway. No other
arch does and it doesn't contain anything useful to userland, so remove
the header-y line from the Kbuild file which triggers the export.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: linuxppc-dev@lists.ozlabs.org
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: David Howells <dhowells@redhat.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/eeh: Fix build error for celleb
Gavin Shan [Mon, 5 May 2014 02:09:05 +0000 (12:09 +1000)]
powerpc/eeh: Fix build error for celleb

Commit 7f52a526f ("powerpc/eeh: Allow to disable EEH") caused
following build error with "celleb_defconfig" as being catched
by Mikey on linux-next.

arch/powerpc/kernel/eeh.c: In function 'eeh_init_proc':
arch/powerpc/kernel/eeh.c:1173:37: error: 'powerpc_debugfs_root' \
undeclared (first use in this function)
arch/powerpc/kernel/eeh.c:1173:37: note: each undeclared identifier \
is reported only once for each function it appears in

Reported-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agoMerge branch 'merge' into next
Benjamin Herrenschmidt [Tue, 20 May 2014 00:21:48 +0000 (10:21 +1000)]
Merge branch 'merge' into next

Merge "merge" branch to get two fairly important bug fixes:

powerpc/powernv: Reset root port in firmware
powerpc: irq work racing with timer interrupt can result in timer interrupt

10 years agoRevert "powerpc/powernv: Fundamental reset on PLX ports"
Benjamin Herrenschmidt [Tue, 20 May 2014 00:20:49 +0000 (10:20 +1000)]
Revert "powerpc/powernv: Fundamental reset on PLX ports"

This reverts commit b2b5efcf208ddc9444aca77336627428782a39f4.

This code was way too board specific, there are quirks as to how
the PERST line is wired on different boards, we'll have to revisit
this using/creating appropriate firmware interfaces.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: irq work racing with timer interrupt can result in timer interrupt hang
Anton Blanchard [Fri, 9 May 2014 07:47:12 +0000 (17:47 +1000)]
powerpc: irq work racing with timer interrupt can result in timer interrupt hang

I am seeing an issue where a CPU running perf eventually hangs.
Traces show timer interrupts happening every 4 seconds even
when a userspace task is running on the CPU. /proc/timer_list
also shows pending hrtimers have not run in over an hour,
including the scheduler.

Looking closer, decrementers_next_tb is getting set to
0xffffffffffffffff, and at that point we will never take
a timer interrupt again.

In __timer_interrupt() we set decrementers_next_tb to
0xffffffffffffffff and rely on ->event_handler to update it:

        *next_tb = ~(u64)0;
        if (evt->event_handler)
                evt->event_handler(evt);

In this case ->event_handler is hrtimer_interrupt. This will eventually
call back through the clockevents code with the next event to be
programmed:

static int decrementer_set_next_event(unsigned long evt,
                                      struct clock_event_device *dev)
{
        /* Don't adjust the decrementer if some irq work is pending */
        if (test_irq_work_pending())
                return 0;
        __get_cpu_var(decrementers_next_tb) = get_tb_or_rtc() + evt;

If irq work came in between these two points, we will return
before updating decrementers_next_tb and we never process a timer
interrupt again.

This looks to have been introduced by 0215f7d8c53f (powerpc: Fix races
with irq_work). Fix it by removing the early exit and relying on
code later on in the function to force an early decrementer:

       /* We may have raced with new irq work */
       if (test_irq_work_pending())
               set_dec(1);

Signed-off-by: Anton Blanchard <anton@samba.org>
Cc: stable@vger.kernel.org # 3.14+
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/powernv: Reset root port in firmware
Gavin Shan [Thu, 24 Apr 2014 08:00:22 +0000 (18:00 +1000)]
powerpc/powernv: Reset root port in firmware

Resetting root port has more stuff to do than that for PCIe switch
ports and we should have resetting root port done in firmware instead
of the kernel itself. The problem was introduced by commit 5b2e198e
("powerpc/powernv: Rework EEH reset").

Cc: linux-stable <stable@vger.kernel.org>
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/fsl-rio: Fix fsl_rio_setup error paths and use-after-unmap
Scott Wood [Mon, 28 Apr 2014 23:20:09 +0000 (18:20 -0500)]
powerpc/fsl-rio: Fix fsl_rio_setup error paths and use-after-unmap

Several of the error paths from fsl_rio_setup are missing error
messages.

Worse, fsl_rio_setup initializes several global pointers and does not
NULL them out after freeing/unmapping on error.  This caused
fsl_rio_mcheck_exception() to crash when accessing rio_regs_win which
was non-NULL but had been unmapped.

Signed-off-by: Scott Wood <scottwood@freescale.com>
Cc: Liu Gang <Gang.Liu@freescale.com>
---
Liu Gang, are you sure all of these error conditions are fatal?  Why
does the rio driver fail if rmu is not present (e.g.  on t4240)?

10 years agoMerge remote-tracking branch 'anton/abiv2' into next
Benjamin Herrenschmidt [Mon, 5 May 2014 10:33:10 +0000 (20:33 +1000)]
Merge remote-tracking branch 'anton/abiv2' into next

This series adds support for building the powerpc 64-bit
LE kernel using the new ABI v2. We already supported
running ABI v2 userspace programs but this adds support
for building the kernel itself using the new ABI.

10 years agopowerpc/mm: use macro PGTABLE_EADDR_SIZE instead of digital
Liu Ping Fan [Thu, 21 Nov 2013 02:17:54 +0000 (10:17 +0800)]
powerpc/mm: use macro PGTABLE_EADDR_SIZE instead of digital

In case of extending the eaddr in future, use this macro
PGTABLE_EADDR_SIZE to ease the maintenance of the code.

Signed-off-by: Liu Ping Fan <pingfank@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agoppc476: Enable a linker work around for IBM errata #46
Alistair Popple [Mon, 24 Feb 2014 07:00:56 +0000 (18:00 +1100)]
ppc476: Enable a linker work around for IBM errata #46

This patch adds an option to enable a work around for an icache bug on
476 that can cause execution of stale instructions when falling
through pages (IBM errata #46). It requires a recent version of
binutils which supports the --ppc476-workaround option.

The work around enables the appropriate linker options and ensures
that all module output sections are aligned to 4K page boundaries. The
work around is only required when building modules.

Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Added PCI MSI support using the HSTA module
Alistair Popple [Thu, 6 Mar 2014 03:52:28 +0000 (14:52 +1100)]
powerpc: Added PCI MSI support using the HSTA module

The PPC476GTR SoC supports message signalled interrupts (MSI) by writing
to special addresses within the High Speed Transfer Assist (HSTA) module.

This patch adds support for PCI MSI with a new system device. The DMA
window is also updated to allow access to the entire 42-bit address range
to allow PCI devices write access to the HSTA module.

Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agoIBM Akebono: Add the Akebono platform
Alistair Popple [Thu, 6 Mar 2014 03:52:27 +0000 (14:52 +1100)]
IBM Akebono: Add the Akebono platform

This patch adds support for the IBM Akebono board.

Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agoIBM Currituck: Clean up board specific code before adding Akebono code
Alistair Popple [Thu, 6 Mar 2014 03:52:26 +0000 (14:52 +1100)]
IBM Currituck: Clean up board specific code before adding Akebono code

The IBM Akebono code uses the same initialisation functions as the
earlier Currituck board. Rather than create a copy of this code for
Akebono we will instead integrate support for it into the same file as
the Currituck code.

This patch just renames the board support file and updates the Makefile.

Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/le: Show the endianess of the LPAR under PowerVM.
Tony Breeds [Wed, 12 Mar 2014 08:17:07 +0000 (19:17 +1100)]
powerpc/le: Show the endianess of the LPAR under PowerVM.

Signed-off-by: Tony Breeds <tony@bakeyournoodle.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: Use 64k io pages when we never see an HEA
Alexander Graf [Tue, 1 Apr 2014 13:46:05 +0000 (15:46 +0200)]
powerpc: Use 64k io pages when we never see an HEA

When we never get around to seeing an HEA ethernet adapter, there's
no point in restricting ourselves to 4k IO page size.

This speeds up IO maps when CONFIG_IBMEBUS is disabled.

[ Updated the test to also lift the restriction on arch 2.07
  (Power 8) which cannot have an HEA
 -- BenH ]

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
foo

10 years agopowerpc: memcpy optimization for 64bit LE
Philippe Bergheaud [Tue, 29 Apr 2014 23:12:01 +0000 (09:12 +1000)]
powerpc: memcpy optimization for 64bit LE

Unaligned stores take alignment exceptions on POWER7 running in little-endian.
This is a dumb little-endian base memcpy that prevents unaligned stores.
Once booted the feature fixup code switches over to the VMX copy loops
(which are already endian safe).

The question is what we do before that switch over. The base 64bit
memcpy takes alignment exceptions on POWER7 so we can't use it as is.
Fixing the causes of alignment exception would slow it down, because
we'd need to ensure all loads and stores are aligned either through
rotate tricks or bytewise loads and stores. Either would be bad for
all other 64bit platforms.

[ I simplified the loop a bit - Anton ]

Signed-off-by: Philippe Bergheaud <felix@linux.vnet.ibm.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agoMerge commit 'f3cae355a962784101478504ef7f6a389ad62979' into next
Benjamin Herrenschmidt [Tue, 29 Apr 2014 23:36:34 +0000 (09:36 +1000)]
Merge commit 'f3cae355a962784101478504ef7f6a389ad62979' into next

Merge Linus tree to get "cpufreq, powernv: Fix build failure on UP"
to avoid build breakages in some of my test configs.

10 years agopowerpc/tm: Add checking to treclaim/trechkpt
Michael Neuling [Fri, 28 Mar 2014 05:40:34 +0000 (16:40 +1100)]
powerpc/tm: Add checking to treclaim/trechkpt

If we do a treclaim and we are not in TM suspend mode, it results in a TM bad
thing (ie. a 0x700 program check).  Similarly if we do a trechkpt and we have
an active transaction or TEXASR Failure Summary (FS) is not set, we also take a
TM bad thing.

This should never happen, but if it does (ie. a kernel bug), the cause is
almost impossible to debug as the GPR state is mostly userspace and hence we
don't get a call chain.

This adds some checks in these cases case a BUG_ON() (in asm) in case we ever
hit these cases.  It moves the register saving around to preserve r1 till later
also.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/tm: Remove unnecessary r1 save
Michael Neuling [Fri, 28 Mar 2014 05:40:33 +0000 (16:40 +1100)]
powerpc/tm: Remove unnecessary r1 save

We save r1 to the scratch SPR and restore it from there after the trechkpt so
saving r1 to the paca is not needed.

Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: powernv: Implement ppc_md.get_proc_freq()
Gautham R. Shenoy [Tue, 11 Mar 2014 11:31:19 +0000 (17:01 +0530)]
powerpc: powernv: Implement ppc_md.get_proc_freq()

Implement a method named pnv_get_proc_freq(unsigned int cpu) which
returns the current clock rate on the 'cpu' in Hz to be reported in
/proc/cpuinfo. This method uses the value reported by cpufreq when
such a value is sane. Otherwise it falls back to old way of reporting
the clockrate, i.e. ppc_proc_freq.

Set the ppc_md.get_proc_freq() hook to pnv_get_proc_freq() on the
PowerNV platform.

Signed-off-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc: powernv: Framework to show the correct clock in /proc/cpuinfo
Gautham R. Shenoy [Tue, 11 Mar 2014 11:31:18 +0000 (17:01 +0530)]
powerpc: powernv: Framework to show the correct clock in /proc/cpuinfo

Currently, the code in setup-common.c for powerpc assumes that all
clock rates are same in a smp system. This value is cached in the
variable named ppc_proc_freq and is the value that is reported in
/proc/cpuinfo.

However on the PowerNV platform, the clock rate is same only across
the threads of the same core. Hence the value that is reported in
/proc/cpuinfo is incorrect on PowerNV platforms. We need a better way
to query and report the correct value of the processor clock in
/proc/cpuinfo.

The patch achieves this by creating a machdep_call named
get_proc_freq() which is expected to returns the frequency in Hz. The
code in show_cpuinfo() can invoke this method to display the correct
clock rate on platforms that have implemented this method. On the
other powerpc platforms it can use the value cached in ppc_proc_freq.

Signed-off-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/powernv: Return secondary CPUs to firmware before FW update
Vasant Hegde [Wed, 9 Apr 2014 17:18:55 +0000 (22:48 +0530)]
powerpc/powernv: Return secondary CPUs to firmware before FW update

Firmware update on PowerNV platform takes several minutes. During
this time one CPU is stuck in FW and the kernel complains about "soft
lockups".

This patch returns all secondary CPUs to firmware before starting
firmware update process.

[ Reworked a bit and cleaned up -- BenH ]

Signed-off-by: Vasant Hegde <hegdevasant@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/pci: Use of_pci_range_parser helper in pci_process_bridge_OF_ranges
Andrew Murray [Tue, 25 Feb 2014 06:32:11 +0000 (06:32 +0000)]
powerpc/pci: Use of_pci_range_parser helper in pci_process_bridge_OF_ranges

This patch updates the implementation of pci_process_bridge_OF_ranges to use
the of_pci_range_parser helpers.

Signed-off-by: Andrew Murray <amurray@embedded-bits.co.uk>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/legacy_serial: Support MVME5100 UARTS with shifted registers
Stephen Chivers [Sat, 19 Apr 2014 23:43:10 +0000 (09:43 +1000)]
powerpc/legacy_serial: Support MVME5100 UARTS with shifted registers

This patch adds support to legacy serial for
UARTS with shifted registers.

The MVME5100 Single Board Computer is a PowerPC platform
that has 16550 style UARTS with register addresses that are
16 bytes apart (shifted by 4).

Commit  309257484cc1a592e8ac5fbdd8cd661be2b80bf8
"powerpc: Cleanup udbg_16550 and add support for LPC PIO-only UARTs"
added support to udbg_16550 for shifted registers by adding a "stride"
parameter to the initialisation operations for Programmed IO and
Memory Mapped IO.

As a consequence it is now possible to use the services of legacy serial
to provide early serial console messages for the MVME5100.

An added benefit of this is that the serial console will always be
"ttyS0" irrespective of whether the computer is fitted with extra
PCI 8250 interface boards or not.

I have tested this patch using the four PowerPC platforms available to me:

MVME5100 - shifted registers,
SAM440EP - unshifted registers,
MPC8349 - unshifted registers,
MVME4100 - unshifted registers.

Signed-off-by: Stephen Chivers <schivers@csc.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Add support for 64bit little endian wrapper
Cédric Le Goater [Thu, 24 Apr 2014 07:23:39 +0000 (09:23 +0200)]
powerpc/boot: Add support for 64bit little endian wrapper

The code is only slightly modified : entry points now use the
FIXUP_ENDIAN trampoline to switch endian order. The 32bit wrapper
is kept for big endian kernels and 64bit is enforced for little
endian kernels with a PPC64_BOOT_WRAPPER config option.

The linker script is generated using the kernel preprocessor flags
to make use of the CONFIG_* definitions and the wrapper script is
modified to take into account the new elf64ppc format.

Finally, the zImage file is compiled as a position independent
executable (-pie) which makes it loadable at any address by the
firmware.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Add a global entry point for pseries
Cédric Le Goater [Thu, 24 Apr 2014 07:23:38 +0000 (09:23 +0200)]
powerpc/boot: Add a global entry point for pseries

When entering the boot wrapper in little endian, we will need to fix
the endian order using a fixup trampoline like in the kernel. This
patch overrides the _zimage_start entry point for this purpose.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Modify entry point for 64bit
Cédric Le Goater [Thu, 24 Apr 2014 07:23:37 +0000 (09:23 +0200)]
powerpc/boot: Modify entry point for 64bit

This patch adds support a 64bit wrapper entry point. As in 32bit, the
entry point does its own relocation and can be loaded at any address
by the firmware.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Define a routine to enter prom
Cédric Le Goater [Thu, 24 Apr 2014 07:23:36 +0000 (09:23 +0200)]
powerpc/boot: Define a routine to enter prom

This patch defines a 'prom' routine similar to 'enter_prom' in the
kernel.

The difference is in the MSR which is built before entering prom. Big
endian order is enforced as in the kernel but 32bit mode is not. It
prepares ground for the next patches which will introduce Little endian
order.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Add little endian support to elf utils
Cédric Le Goater [Thu, 24 Apr 2014 07:23:35 +0000 (09:23 +0200)]
powerpc/boot: Add little endian support to elf utils

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Add 64bit and little endian support to addnote
Cédric Le Goater [Thu, 24 Apr 2014 07:23:34 +0000 (09:23 +0200)]
powerpc/boot: Add 64bit and little endian support to addnote

It could certainly be improved using Elf macros and byteswapping
routines, but the initial version of the code is organised to be a
single file program with limited dependencies. yaboot is the same.

Please scream if you want a total rewrite.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Define byteswapping routines for little endian
Cédric Le Goater [Thu, 24 Apr 2014 07:23:33 +0000 (09:23 +0200)]
powerpc/boot: Define byteswapping routines for little endian

These are not the most efficient versions of swab but the wrapper does
not do much byte swapping. On a big endian cpu, these routines are
a no-op.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Fix compile warning in 64bit
Cédric Le Goater [Thu, 24 Apr 2014 07:23:32 +0000 (09:23 +0200)]
powerpc/boot: Fix compile warning in 64bit

arch/powerpc/boot/oflib.c:211:9: warning: cast to pointer from integer of \
  different size [-Wint-to-pointer-cast]
  return (phandle) of_call_prom("finddevice", 1, 1, name);

This is a work around. The definite solution would be to define the
phandle typedef as a u32, as in the kernel, but this would break the
device tree ops API.

Let it be for the moment.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Define typedef ihandle as u32
Cédric Le Goater [Thu, 24 Apr 2014 07:23:31 +0000 (09:23 +0200)]
powerpc/boot: Define typedef ihandle as u32

This makes ihandle 64bit friendly.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Rework of_claim() to make it 64bit friendly
Cédric Le Goater [Thu, 24 Apr 2014 07:23:30 +0000 (09:23 +0200)]
powerpc/boot: Rework of_claim() to make it 64bit friendly

This patch fixes 64bit compile warnings and updates the wrapper code
to converge the kernel code in prom_init.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Add PROM_ERROR define in oflib
Cédric Le Goater [Thu, 24 Apr 2014 07:23:29 +0000 (09:23 +0200)]
powerpc/boot: Add PROM_ERROR define in oflib

This is mostly useful to make to the boot wrapper code closer with
the kernel code in prom_init.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Add byteswapping routines in oflib
Cédric Le Goater [Thu, 24 Apr 2014 07:23:28 +0000 (09:23 +0200)]
powerpc/boot: Add byteswapping routines in oflib

Values will need to be byte-swapped when calling prom (big endian) from
a little endian boot wrapper.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Use prom_arg_t in oflib
Cédric Le Goater [Thu, 24 Apr 2014 07:23:27 +0000 (09:23 +0200)]
powerpc/boot: Use prom_arg_t in oflib

This patch updates the wrapper code to converge with the kernel code in
prom_init.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Use a common prom_args struct in oflib
Cédric Le Goater [Thu, 24 Apr 2014 07:23:26 +0000 (09:23 +0200)]
powerpc/boot: Use a common prom_args struct in oflib

This patch fixes warnings when the wrapper is compiled in 64bit and
updates the boot wrapper code related to prom to converge with the
kernel code in prom_init. This should make the review of changes easier.

The kernel has a different number of possible arguments (10) when
entering prom. There does not seem to be any good reason to have
12 in the wrapper, so the patch changes this value to args[10] in
the prom_args struct.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/boot: Fix do_div for 64bit wrapper
Cédric Le Goater [Thu, 24 Apr 2014 07:23:25 +0000 (09:23 +0200)]
powerpc/boot: Fix do_div for 64bit wrapper

When the boot wrapper is compiled in 64bit, there is no need to
use __div64_32.

Signed-off-by: Cédric Le Goater <clg@fr.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/prom: Stop scanning dev-tree for fdump early
Gavin Shan [Thu, 24 Apr 2014 08:00:31 +0000 (18:00 +1000)]
powerpc/prom: Stop scanning dev-tree for fdump early

Function early_init_dt_scan_fw_dump() is called to scan the device
tree for fdump properties under node "rtas". Any one of them is
invalid, we can stop scanning the device tree early by returning
"1". It would save a bit time during boot.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/powernv: Don't use pe->pbus to get the domain number
Gavin Shan [Thu, 24 Apr 2014 08:00:29 +0000 (18:00 +1000)]
powerpc/powernv: Don't use pe->pbus to get the domain number

If the PE contains single PCI function, "pe->pbus" would be NULL.
It's not reliable to be used by pci_domain_nr().  We just grab the
PCI domain number from the PCI host controller (struct pci_controller)
instance.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/powernv: Missed IOMMU table type
Gavin Shan [Thu, 24 Apr 2014 08:00:28 +0000 (18:00 +1000)]
powerpc/powernv: Missed IOMMU table type

In function pnv_pci_ioda2_setup_dma_pe(), the IOMMU table type is
set to (TCE_PCI_SWINV_CREATE | TCE_PCI_SWINV_FREE) unconditionally.
It was just set to TCE_PCI by pnv_pci_setup_iommu_table(). So the
primary IOMMU table type (TCE_PCI) is lost. The patch fixes it.

Also, pnv_pci_setup_iommu_table() already set "tbl->it_busno" to
zero and we needn't do it again. The patch removes the redundant
assignment.

The patch also fixes similar issues in pnv_pci_ioda_setup_dma_pe().

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/powernv: Fundamental reset on PLX ports
Gavin Shan [Thu, 24 Apr 2014 08:00:27 +0000 (18:00 +1000)]
powerpc/powernv: Fundamental reset on PLX ports

The patch intends to support fundamental reset on PLX downstream
ports. If the PCI device matches any one of the internal table,
which includes PLX vendor ID, bridge device ID, register offset
for fundamental reset and bit, fundamental reset will be done
accordingly. Otherwise, it will fail back to hot reset.

Additional flag (EEH_DEV_FRESET) is introduced to record the last
reset type on the PCI bridge.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/eeh: Can't recover from non-PE-reset case
Gavin Shan [Thu, 24 Apr 2014 08:00:26 +0000 (18:00 +1000)]
powerpc/eeh: Can't recover from non-PE-reset case

When PCI_ERS_RESULT_CAN_RECOVER returned from device drivers, the
EEH core should enable I/O and DMA for the affected PE. However,
it was missed to have DMA enabled in eeh_handle_normal_event().
Besides, the frozen state of the affected PE should be cleared
after successful recovery, but we didn't.

The patch fixes both of the issues as above.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowrpc/powernv: Reset PHB in kdump kernel
Gavin Shan [Thu, 24 Apr 2014 08:00:25 +0000 (18:00 +1000)]
powrpc/powernv: Reset PHB in kdump kernel

In the kdump scenario, the first kerenl doesn't shutdown PCI devices
and the kdump kerenl clean PHB IODA table at the early probe time.
That means the kdump kerenl can't support PCI transactions piled
by the first kerenl. Otherwise, lots of EEH errors and frozen PEs
will be detected.

In order to avoid the EEH errors, the PHB is resetted to drop all
PCI transaction from the first kerenl.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/pci: Mask linkDown on resetting PCI bus
Gavin Shan [Thu, 24 Apr 2014 08:00:24 +0000 (18:00 +1000)]
powerpc/pci: Mask linkDown on resetting PCI bus

The problem was initially reported by Wendy who tried pass through
IPR adapter, which was connected to PHB root port directly, to KVM
based guest. When doing that, pci_reset_bridge_secondary_bus() was
called by VFIO driver and linkDown was detected by the root port.
That caused all PEs to be frozen.

The patch fixes the issue by routing the reset for the secondary bus
of root port to underly firmware. For that, one more weak function
pci_reset_secondary_bus() is introduced so that the individual platforms
can override that and do specific reset for bridge's secondary bus.

Reported-by: Wendy Xiong <wenxiong@linux.vnet.ibm.com>
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
10 years agopowerpc/eeh: Make the delay for PE reset unified
Gavin Shan [Thu, 24 Apr 2014 08:00:23 +0000 (18:00 +1000)]
powerpc/eeh: Make the delay for PE reset unified

Basically, we have 3 types of resets to fulfil PE reset: fundamental,
hot and PHB reset. For the later 2 cases, we need PCI bus reset hold
and settlement delay as specified by PCI spec. PowerNV and pSeries
platforms are running on top of different firmware and some of the
delays have been covered by underly firmware (PowerNV).

The patch makes the delays unified to be done in backend, instead of
EEH core.

Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>