Kevin Hao [Thu, 26 Sep 2013 01:42:28 +0000 (09:42 +0800)]
powerpc/85xx: use one kernel option for all the CoreNet_Generic boards
Currently all these boards use the same machine struct and also select
the same kernel options, so it seems a bit of redundant to keep one
separate kernel option for each board. Also update the defconfigs
according to this change.
Signed-off-by: Kevin Hao <haokexin@gmail.com> Signed-off-by: Scott Wood <scottwood@freescale.com>
Kevin Hao [Thu, 26 Sep 2013 01:42:27 +0000 (09:42 +0800)]
powerpc/85xx: rename the corenet_ds.c to corenet_generic.c
This file is also used by some RDB and QDS boards. So the name seems
not so accurate. Rename it to corenet_generic.c. Also update the
function names in this file according to the change.
Kevin Hao [Thu, 26 Sep 2013 01:42:26 +0000 (09:42 +0800)]
powerpc/85xx: introduce corenet_generic machine
In the current kernel, the board files for p2041rdb, p3041ds, p4080ds,
p5020ds, p5040ds, t4240qds and b4qds are almost the same except the
machine name. So this introduces a cornet_generic machine to support
all these boards to avoid the code duplication.
With these changes the file corenet_ds.h becomes useless. Just delete
it.
Signed-off-by: Kevin Hao <haokexin@gmail.com> Signed-off-by: Scott Wood <scottwood@freescale.com>
York Sun [Mon, 16 Sep 2013 23:38:07 +0000 (16:38 -0700)]
powerpc/b4860emu: Add device tree file for b4860emu
B4860EMU is a emualtor target with minimum peripherals. It is based on
B4860QDS and trimmed down most peripherals due to either not modeled or
lack of board level connections. The main purpose of this minimum dts is
to speed up booting on emulator.
Signed-off-by: York Sun <yorksun@freescale.com>
[scottwood@freescale.com: whitespace fix] Signed-off-by: Scott Wood <scottwood@freescale.com>
York Sun [Mon, 16 Sep 2013 23:38:06 +0000 (16:38 -0700)]
powerpc/t4240emu: Add device tree file for t4240emu
T4240EMU is an emulator target with minimum peripherals. It is based on
T4240QDS and trimmed down most peripherals due to either not modeled or
lack of board level connections. The main purpose of this minimum dts is
to speed up booting on emulator.
Signed-off-by: York Sun <yorksun@freescale.com>
[scottwood@freescale.com: whitespace fixes] Signed-off-by: Scott Wood <scottwood@freescale.com>
Scott Wood [Wed, 11 Sep 2013 23:36:52 +0000 (18:36 -0500)]
powerpc/b4qds: enable coreint
Commit 9837b43c5f3514e5d28f65f1513f4dc6759d2810 ("powerpc/85xx: enable
coreint for all the 64bit boards") removed the ifdef that avoided
coreint on 64-bit, but it missed b4_qds.c.
Signed-off-by: Scott Wood <scottwood@freescale.com> Cc: Kevin Hao <haokexin@gmail.com> Cc: Shaveta Leekha <shaveta@freescale.com>
The commit e0908085fc2391c85b85fb814ae1df377c8e0dcb ("powerpc/8xx: Fix
regression introduced by cache coherency rewrite") is not needed
anymore. The issue was because dcbst wrongly sets the store bit when
causing a DTLB error, but this is now fixed by commit 0a2ab51ffb8dfdf51402dcfb446629648c96bc78 ("powerpc/8xx: Fixup DAR from
buggy dcbX instructions.") which handles the buggy dcbx instructions on
data page faults on the 8xx.
Haijun.Zhang [Thu, 29 Aug 2013 03:31:28 +0000 (11:31 +0800)]
powerpc/eSDCH: Specify voltage for T4240QDS
Freescale T4240QDS reference board has extra voltage shifters added
to allow 3.3V operation, so add 3.3v voltage support for T4240QDS.
1.8v and 3.3v is recommand for eMMC and SDHC card.
Signed-off-by: Haijun Zhang <haijun.zhang@freescale.com> Signed-off-by: Scott Wood <scottwood@freescale.com>
LEROY Christophe [Wed, 28 Aug 2013 14:19:17 +0000 (16:19 +0200)]
powerpc/mpc8xx: Clearer Oops message for Software Emulation Exception
This patch modifies the Oops message in case of Software Emulation Exception.
The existing message is quite confusing because it refers to FPU Emulation
while most often the issue is due to either a non supported instruction
(not necessarily FPU related) or a stale instruction due to HW issues.
The new message tries to be more generic in order to make the user understand
that the Oops is due to something wrong with an instruction, not necessarily
due to an FPU instruction.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Signed-off-by: Scott Wood <scottwood@freescale.com>
Chunhe Lan [Fri, 2 Aug 2013 08:46:25 +0000 (16:46 +0800)]
powerpc/pci: Change the DECLARE_PCI_FIXUP_{HEADER => EARLY} macro of pci quirk
Freescale platform has class code = 0x0b2000, when it boots. This makes
kernel PCI bus code to setup these devices resulting into the following
notice information when trying to enable them:
pci 0000:00:00.0: ignoring class 0x0b2000 (doesn't match header type 01)
The above information is outputted by judging value of dev->class before
pci_setup_device() function, and the DECLARE_PCI_FIXUP_HEADER quirk runs
after pci_setup_device() function. But the DECLARE_PCI_FIXUP_EARLY quirk
runs before judging value of dev->class and pci_setup_device() function.
So we use the DECLARE_PCI_FIXUP_EARLY macro to fix this issue.
Signed-off-by: Chunhe Lan <Chunhe.Lan@freescale.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Bjorn Helgaas <bhelgaas@google.com> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Scott Wood <scottwood@freescale.com>
Minghuan Lian [Wed, 31 Jul 2013 02:59:07 +0000 (10:59 +0800)]
powerpc/dts: fix sRIO error interrupt for b4860
For B4 platform, MPIC EISR register is in reversed bitmap order,
instead of "Error interrupt source 0-31. Bit 0 represents SRC0."
the correct ordering is "Error interrupt source 0-31. Bit 0
represents SRC31." This patch is to fix sRIO EISR bit value
of error interrupt in dts node.
Signed-off-by: Minghuan Lian <Minghuan.Lian@freescale.com> Signed-off-by: Scott Wood <scottwood@freescale.com>
James Yang [Fri, 5 Jul 2013 19:49:43 +0000 (14:49 -0500)]
powerpc/booke: clear DBCR0_BT in user_disable_single_step()
BookE version of user_disable_single_step() clears DBCR0_IC for the
instruction completion debug, but did not also clear DBCR0_BT for the
branch taken exception. This behavior was lost by the 2/2010 patch.
Signed-off-by: James Yang <James.Yang@freescale.com> Signed-off-by: Scott Wood <scottwood@freescale.com>
James Yang [Wed, 3 Jul 2013 21:26:47 +0000 (16:26 -0500)]
powerpc: Emulate sync instruction variants
Reserved fields of the sync instruction have been used for other
instructions (e.g. lwsync). On processors that do not support variants
of the sync instruction, emulate it by executing a sync to subsume the
effect of the intended instruction.
Signed-off-by: James Yang <James.Yang@freescale.com>
[scottwood@freescale.com: whitespace and subject line fix] Signed-off-by: Scott Wood <scottwood@freescale.com>
Paul Bolle [Wed, 15 May 2013 09:21:01 +0000 (11:21 +0200)]
powerpc: remove dependency on MV64360
The Kconfig entry that allows to "Distribute interrupts on all CPUs by
default" has a (negative) dependency on MV64360. But that Kconfig symbol
was removed in v2.6.27, which means that this dependency has evaluated
to true ever since. It can be removed too.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl> Signed-off-by: Scott Wood <scottwood@freescale.com>
Paul Mackerras [Tue, 10 Sep 2013 10:21:10 +0000 (20:21 +1000)]
powerpc: Provide for giveup_fpu/altivec to save state in alternate location
This provides a facility which is intended for use by KVM, where the
contents of the FP/VSX and VMX (Altivec) registers can be saved away
to somewhere other than the thread_struct when kernel code wants to
use floating point or VMX instructions. This is done by providing a
pointer in the thread_struct to indicate where the state should be
saved to. The giveup_fpu() and giveup_altivec() functions test these
pointers and save state to the indicated location if they are non-NULL.
Note that the MSR_FP/VEC bits in task->thread.regs->msr are still used
to indicate whether the CPU register state is live, even when an
alternate save location is being used.
This also provides load_fp_state() and load_vr_state() functions, which
load up FP/VSX and VMX state from memory into the CPU registers, and
corresponding store_fp_state() and store_vr_state() functions, which
store FP/VSX and VMX state into memory from the CPU registers.
Signed-off-by: Paul Mackerras <paulus@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Paul Mackerras [Tue, 10 Sep 2013 10:20:42 +0000 (20:20 +1000)]
powerpc: Put FP/VSX and VR state into structures
This creates new 'thread_fp_state' and 'thread_vr_state' structures
to store FP/VSX state (including FPSCR) and Altivec/VSX state
(including VSCR), and uses them in the thread_struct. In the
thread_fp_state, the FPRs and VSRs are represented as u64 rather
than double, since we rarely perform floating-point computations
on the values, and this will enable the structures to be used
in KVM code as well. Similarly FPSCR is now a u64 rather than
a structure of two 32-bit values.
This takes the offsets out of the macros such as SAVE_32FPRS,
REST_32FPRS, etc. This enables the same macros to be used for normal
and transactional state, enabling us to delete the transactional
versions of the macros. This also removes the unused do_load_up_fpu
and do_load_up_altivec, which were in fact buggy since they didn't
create large enough stack frames to account for the fact that
load_up_fpu and load_up_altivec are not designed to be called from C
and assume that their caller's stack frame is an interrupt frame.
Signed-off-by: Paul Mackerras <paulus@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
powerpc: add real mode support for dma operations on powernv
The existing TCE machine calls (tce_build and tce_free) only support
virtual mode as they call __raw_writeq for TCE invalidation what
fails in real mode.
This introduces tce_build_rm and tce_free_rm real mode versions
which do mostly the same but use "Store Doubleword Caching Inhibited
Indexed" instruction for TCE invalidation.
This new feature is going to be utilized by real mode support of VFIO.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
powerpc: Prepare to support kernel handling of IOMMU map/unmap
The current VFIO-on-POWER implementation supports only user mode
driven mapping, i.e. QEMU is sending requests to map/unmap pages.
However this approach is really slow, so we want to move that to KVM.
Since H_PUT_TCE can be extremely performance sensitive (especially with
network adapters where each packet needs to be mapped/unmapped) we chose
to implement that as a "fast" hypercall directly in "real
mode" (processor still in the guest context but MMU off).
To be able to do that, we need to provide some facilities to
access the struct page count within that real mode environment as things
like the sparsemem vmemmap mappings aren't accessible.
This adds an API function realmode_pfn_to_page() to get page struct when
MMU is off.
This adds to MM a new function put_page_unless_one() which drops a page
if counter is bigger than 1. It is going to be used when MMU is off
(for example, real mode on PPC64) and we want to make sure that page
release will not happen in real mode as it may crash the kernel in
a horrible way.
CONFIG_SPARSEMEM_VMEMMAP and CONFIG_FLATMEM are supported.
Cc: linux-mm@kvack.org Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Paul Mackerras <paulus@samba.org> Signed-off-by: Paul Mackerras <paulus@samba.org> Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This adds hash_for_each_possible_rcu_notrace() which is basically
a notrace clone of hash_for_each_possible_rcu() which cannot be
used in real mode due to its tracing/debugging capability.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We already had some output messages from EEH core. Occasionally,
we can see the output messages from EEH core before the stack
dump. That's not what we expected. The patch fixes that and shows
the stack dump prior to output messages from EEH core.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The patch adds function ioda_eeh_phb3_phb_diag() to dump PHB3
PHB diag-data. That's called while detecting informative errors
or frozen PE on the specific PHB.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Each PHB instance (struct pnv_phb) has its corresponding log blob,
which is used to hold the retrieved error log from firmware. The
current size of that (4096) isn't enough for PHB3 case and the patch
makes that double to 8192.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
For now, we only support outbound error injection. Actually, the
hardware supports injecting inbound errors as well. The patch enables
to inject inbound errors.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
powerpc: prom_init exception when updating core value
Since the CPU is generating an exception when accessing unaligned word, and
as this exception is not yet handled when running prom_init, data should be
copied from the architecture vector byte per byte.
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Michael Ellerman [Fri, 11 Oct 2013 03:07:59 +0000 (14:07 +1100)]
powerpc/pseries: Implement arch_get_random_long() based on H_RANDOM
Add support for the arch_get_random_long() hook based on the H_RANDOM
hypervisor call. We trust the hypervisor to provide us with random data,
ie. we don't whiten it in anyway.
Signed-off-by: Michael Ellerman <michael@ellerman.id.au> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Michael Ellerman [Fri, 11 Oct 2013 03:07:58 +0000 (14:07 +1100)]
hwrng: Add a driver for the hwrng found in power7+ systems
Add a driver for the hwrng found in power7+ systems, based on the
existing code for the arch_get_random_long() hook.
We only register a single instance of the driver, not one per device,
because we use the existing per_cpu array of devices in the arch code.
This means we always read from the "closest" device, avoiding inter-chip
memory traffic.
Signed-off-by: Guo Chao <yan@linux.vnet.ibm.com> Signed-off-by: Michael Ellerman <michael@ellerman.id.au> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Michael Ellerman [Fri, 11 Oct 2013 03:07:57 +0000 (14:07 +1100)]
powerpc: Implement arch_get_random_long/int() for powernv
Add the plumbing to implement arch_get_random_long/int(). It didn't seem
worth adding an extra ppc_md hook for int, so we reuse the one for long.
Add an implementation for powernv based on the hwrng found in power7+
systems. We whiten the output of the hwrng, and the result passes all
the dieharder tests.
Signed-off-by: Michael Ellerman <michael@ellerman.id.au> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Michael Ellerman [Wed, 25 Sep 2013 09:24:17 +0000 (19:24 +1000)]
hwrng: Return errors to upper levels in pseries-rng.c
We don't expect to get errors from the hypervisor when reading the rng,
but if we do we should pass the error up to the hwrng driver. Otherwise
the hwrng driver will continue calling us forever.
Signed-off-by: Michael Ellerman <michael@ellerman.id.au> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Bharat Bhushan [Wed, 9 Oct 2013 05:11:17 +0000 (10:41 +0530)]
powerpc: Added __cmpdi2 for signed 64bit comparision
This was missing on powerpc and I am getting compilation error
drivers/vfio/pci/vfio_pci_rdwr.c:193: undefined reference to `__cmpdi2'
drivers/vfio/pci/vfio_pci_rdwr.c:193: undefined reference to `__cmpdi2'
Signed-off-by: Bharat Bhushan <bharat.bhushan@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Vladimir Murzin [Sun, 29 Sep 2013 12:41:18 +0000 (14:41 +0200)]
powerpc: Fix section mismatch warning in free_lppacas
While cross-building for PPC64 I've got bunch of
WARNING: arch/powerpc/kernel/built-in.o(.text.unlikely+0x2d2): Section
mismatch in reference from the function .free_lppacas() to the variable
.init.data:lppaca_size The function .free_lppacas() references the variable
__initdata lppaca_size. This is often because .free_lppacas lacks a __initdata
annotation or the annotation of lppaca_size is wrong.
Fix it by using proper annotation for free_lppacas. Additionally, annotate
{allocate,new}_llpcas properly.
Signed-off-by: Vladimir Murzin <murzin.v@gmail.com> Acked-by: Michael Ellerman <michael@ellerman.id.au> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Kevin Hao [Thu, 26 Sep 2013 08:41:34 +0000 (16:41 +0800)]
powerpc/ppc64: Remove the unneeded load of ti_flags in resume_kernel
We already got the value of current_thread_info and ti_flags and store
them into r9 and r4 respectively before jumping to resume_kernel. So
there is no reason to reload them again.
Signed-off-by: Kevin Hao <haokexin@gmail.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Scott Wood [Fri, 27 Sep 2013 00:18:18 +0000 (19:18 -0500)]
powerpc/mpic: Disable preemption when calling mpic_processor_id()
Otherwise, we get a debug traceback due to the use of
smp_processor_id() (or get_paca()) inside hard_smp_processor_id().
mpic_host_map() is just looking for a default CPU, so it doesn't matter
if we migrate after getting the CPU ID.
Signed-off-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Anton Blanchard [Mon, 23 Sep 2013 02:05:12 +0000 (12:05 +1000)]
powerpc: Don't set HAVE_EFFICIENT_UNALIGNED_ACCESS on little endian builds
POWER7 takes alignment exceptions on some unaligned addresses, so
disable HAVE_EFFICIENT_UNALIGNED_ACCESS. This fixes an early boot
issue in the printk code.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Ian Munsie [Mon, 23 Sep 2013 02:05:11 +0000 (12:05 +1000)]
powerpc: Add ability to build little endian kernels
This patch allows the kbuild system to successfully compile a kernel
for the little endian PowerPC64 architecture. A subsequent patch
will add the CONFIG_CPU_LITTLE_ENDIAN kernel config option which
must be set to build such a kernel.
If cross compiling, CROSS_COMPILE must point to a suitable toolchain
(compiled for the powerpc64le-linux and powerpcle-linux targets).
Signed-off-by: Ian Munsie <imunsie@au1.ibm.com> Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Anton Blanchard [Mon, 23 Sep 2013 02:04:50 +0000 (12:04 +1000)]
powerpc: Handle VSX alignment faults in little endian mode
Things are complicated by the fact that VSX elements are big
endian ordered even in little endian mode. 8 byte loads and
stores also write to the top 8 bytes of the register.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Anton Blanchard [Mon, 23 Sep 2013 02:04:48 +0000 (12:04 +1000)]
powerpc: Alignment handler shouldn't access VSX registers with TS_FPR
The TS_FPR macro selects the FPR component of a VSX register (the
high doubleword). emulate_vsx is using this macro to get the
address of the associated VSX register. This happens to work on big
endian, but fails on little endian.
Replace it with an explicit array access.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Anton Blanchard [Mon, 23 Sep 2013 02:04:47 +0000 (12:04 +1000)]
powerpc: Remove hard coded FP offsets in alignment handler
The alignment handler assumes big endian ordering when selecting
the low word of a 64bit floating point value. Use the existing
union which works in both little and big endian.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Create a trampoline that works in either endian and flips to
the expected endian. Use it for primary and secondary thread
entry as well as RTAS and OF call return.
Credit for finding the magic instruction goes to Paul Mackerras
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Ian Munsie [Mon, 23 Sep 2013 02:04:44 +0000 (12:04 +1000)]
powerpc: Include the appropriate endianness header
This patch will have powerpc include the appropriate generic endianness
header depending on what the compiler reports.
Signed-off-by: Ian Munsie <imunsie@au1.ibm.com> Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Ian Munsie [Mon, 23 Sep 2013 02:04:40 +0000 (12:04 +1000)]
powerpc: Support endian agnostic MMIO
This patch maps the MMIO functions for 32bit PowerPC to their
appropriate instructions depending on CPU endianness.
The macros used to create the corresponding inline functions are also
renamed by this patch. Previously they had BE or LE in their names which
was misleading - they had nothing to do with endianness, but actually
created different instruction forms so their new names reflect the
instruction form they are creating (D-Form and X-Form).
Little endian 64bit PowerPC is not supported, so the lack of mappings
(and corresponding breakage) for that case is intentional to bring the
attention of anyone doing a 64bit little endian port. 64bit big endian
is unaffected.
[ Added 64 bit versions - Anton ]
Signed-off-by: Ian Munsie <imunsie@au1.ibm.com> Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Anton Blanchard [Mon, 23 Sep 2013 02:04:39 +0000 (12:04 +1000)]
powerpc: Little endian builds double word swap VSX state during context save/restore
The elements within VSX loads and stores are big endian ordered
regardless of endianness. Our VSX context save/restore code uses
lxvd2x and stxvd2x which is a 2x doubleword operation. This means
the two doublewords will be swapped and we have to perform another
swap to undo it.
We need to do this on save and restore.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Anton Blanchard [Mon, 23 Sep 2013 02:04:37 +0000 (12:04 +1000)]
powerpc: Fix offset of FPRs in VSX registers in little endian builds
The FPRs overlap the high doublewords of the first 32 VSX registers.
Fix TS_FPROFFSET and TS_VSRLOWOFFSET so we access the correct fields
in little endian mode.
If VSX is disabled the FPRs are only one doubleword in length so
TS_FPROFFSET needs adjusting in little endian.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
powerpc/irq: Don't switch to irq stack from softirq stack
irq_exit() is now called on the irq stack, which can trigger a switch to
the softirq stack from the irq stack. If an interrupt happens at that
point, we will not properly detect the re-entrancy and clobber the
original return context on the irq stack.
This fixes it. The side effect is to prevent all nesting from softirq
stack to irq stack even in the "safe" case but it's simpler that way and
matches what x86_64 does.
Reported-by: Cédric Le Goater <clg@fr.ibm.com> Tested-by: Cédric Le Goater <clg@fr.ibm.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>