linux-stable.git/arch/xtensa, branch v5.10.78

xtensa: xtfpga: Try software restart before simulating CPU reset

2021-10-27T07:56:47+00:00

[ Upstream commit 012e974501a270d8dfd4ee2039e1fdf7579c907e ]

Rebooting xtensa images loaded with the '-kernel' option in qemu does
not work. When executing a reboot command, the qemu session either hangs
or experiences an endless sequence of error messages.

  Kernel panic - not syncing: Unrecoverable error in exception handler

Reset code jumps to the CPU restart address, but Linux can not recover
from there because code and data in the kernel init sections have been
discarded and overwritten at this point.

XTFPGA platforms have a means to reset the CPU by writing 0xdead into a
specific FPGA IO address. When used in QEMU the kernel image loaded with
the '-kernel' option gets restored to its original state allowing the
machine to boot successfully.

Use that mechanism to attempt a platform reset. If it does not work,
fall back to the existing mechanism.

Signed-off-by: Guenter Roeck 
Signed-off-by: Max Filippov 
Signed-off-by: Sasha Levin

xtensa: xtfpga: use CONFIG_USE_OF instead of CONFIG_OF

2021-10-27T07:56:47+00:00

[ Upstream commit f3d7c2cdf6dc0d5402ec29c3673893b3542c5ad1 ]

Use platform data to initialize xtfpga device drivers when CONFIG_USE_OF
is not selected. This fixes xtfpga networking when CONFIG_USE_OF is not
selected but CONFIG_OF is.

Signed-off-by: Max Filippov 
Signed-off-by: Sasha Levin

xtensa: call irqchip_init only when CONFIG_USE_OF is selected

2021-10-13T08:04:26+00:00

[ Upstream commit 6489f8d0e1d93a3603d8dad8125797559e4cf2a2 ]

During boot time kernel configured with OF=y but USE_OF=n displays the
following warnings and hangs shortly after starting userspace:

------------[ cut here ]------------
WARNING: CPU: 0 PID: 0 at kernel/irq/irqdomain.c:695 irq_create_mapping_affinity+0x29/0xc0
irq_create_mapping_affinity(, 6) called with NULL domain
CPU: 0 PID: 0 Comm: swapper Not tainted 5.15.0-rc3-00001-gd67ed2510d28 #30
Call Trace:
  __warn+0x69/0xc4
  warn_slowpath_fmt+0x6c/0x94
  irq_create_mapping_affinity+0x29/0xc0
  local_timer_setup+0x40/0x88
  time_init+0xb1/0xe8
  start_kernel+0x31d/0x3f4
  _startup+0x13b/0x13b
---[ end trace 1e6630e1c5eda35b ]---
------------[ cut here ]------------
WARNING: CPU: 0 PID: 0 at arch/xtensa/kernel/time.c:141 local_timer_setup+0x58/0x88
error: can't map timer irq
CPU: 0 PID: 0 Comm: swapper Tainted: G        W         5.15.0-rc3-00001-gd67ed2510d28 #30
Call Trace:
  __warn+0x69/0xc4
  warn_slowpath_fmt+0x6c/0x94
  local_timer_setup+0x58/0x88
  time_init+0xb1/0xe8
  start_kernel+0x31d/0x3f4
  _startup+0x13b/0x13b
---[ end trace 1e6630e1c5eda35c ]---
Failed to request irq 0 (timer)

Fix that by calling irqchip_init only when CONFIG_USE_OF is selected and
calling legacy interrupt controller init otherwise.

Fixes: da844a81779e ("xtensa: add device trees support")
Signed-off-by: Max Filippov 
Signed-off-by: Sasha Levin

xtensa: use CONFIG_USE_OF instead of CONFIG_OF

2021-10-13T08:04:26+00:00

[ Upstream commit d67ed2510d28a1eb33171010d35cf52178cfcbdd ]

CONFIG_OF can be set by a randconfig or by a user -- without setting the
early flattree option (OF_EARLY_FLATTREE).  This causes build errors.
However, if randconfig or a user sets USE_OF in the Xtensa config,
the right kconfig symbols are set to fix the build.

Fixes these build errors:

../arch/xtensa/kernel/setup.c:67:19: error: ‘__dtb_start’ undeclared here (not in a function); did you mean ‘dtb_start’?
   67 | void *dtb_start = __dtb_start;
      |                   ^~~~~~~~~~~
../arch/xtensa/kernel/setup.c: In function 'xtensa_dt_io_area':
../arch/xtensa/kernel/setup.c:201:14: error: implicit declaration of function 'of_flat_dt_is_compatible'; did you mean 'of_machine_is_compatible'? [-Werror=implicit-function-declaration]
  201 |         if (!of_flat_dt_is_compatible(node, "simple-bus"))
../arch/xtensa/kernel/setup.c:204:18: error: implicit declaration of function 'of_get_flat_dt_prop' [-Werror=implicit-function-declaration]
  204 |         ranges = of_get_flat_dt_prop(node, "ranges", &len);
../arch/xtensa/kernel/setup.c:204:16: error: assignment to 'const __be32 *' {aka 'const unsigned int *'} from 'int' makes pointer from integer without a cast [-Werror=int-conversion]
  204 |         ranges = of_get_flat_dt_prop(node, "ranges", &len);
      |                ^
../arch/xtensa/kernel/setup.c: In function 'early_init_devtree':
../arch/xtensa/kernel/setup.c:228:9: error: implicit declaration of function 'early_init_dt_scan'; did you mean 'early_init_devtree'? [-Werror=implicit-function-declaration]
  228 |         early_init_dt_scan(params);
../arch/xtensa/kernel/setup.c:229:9: error: implicit declaration of function 'of_scan_flat_dt' [-Werror=implicit-function-declaration]
  229 |         of_scan_flat_dt(xtensa_dt_io_area, NULL);

xtensa-elf-ld: arch/xtensa/mm/mmu.o:(.text+0x0): undefined reference to `xtensa_kio_paddr'

Fixes: da844a81779e ("xtensa: add device trees support")
Fixes: 6cb971114f63 ("xtensa: remap io area defined in device tree")
Signed-off-by: Randy Dunlap 
Signed-off-by: Max Filippov 
Signed-off-by: Sasha Levin

xtensa: ISS: don't panic in rs_init

2021-09-18T11:40:22+00:00

[ Upstream commit 23411c720052ad860b3e579ee4873511e367130a ]

While alloc_tty_driver failure in rs_init would mean we have much bigger
problem, there is no reason to panic when tty_register_driver fails
there. It can fail for various reasons.

So handle the failure gracefully. Actually handle them both while at it.
This will make at least the console functional as it was enabled earlier
by console_initcall in iss_console_init. Instead of shooting down the
whole system.

We move tty_port_init() after alloc_tty_driver(), so that we don't need
to destroy the port in case the latter function fails.

Cc: Chris Zankel 
Cc: Max Filippov 
Cc: linux-xtensa@linux-xtensa.org
Acked-by: Max Filippov 
Signed-off-by: Jiri Slaby 
Link: https://lore.kernel.org/r/20210723074317.32690-2-jslaby@suse.cz
Signed-off-by: Greg Kroah-Hartman 
Signed-off-by: Sasha Levin

xtensa: fix kconfig unmet dependency warning for HAVE_FUTEX_CMPXCHG

2021-09-08T06:49:00+00:00

commit ed5aacc81cd41efc4d561e14af408d1003f7b855 upstream.

XTENSA should only select HAVE_FUTEX_CMPXCHG when FUTEX is
set/enabled. This prevents a kconfig warning.

WARNING: unmet direct dependencies detected for HAVE_FUTEX_CMPXCHG
  Depends on [n]: FUTEX [=n]
  Selected by [y]:
  - XTENSA [=y] && !MMU [=n]

Fixes: d951ba21b959 ("xtensa: nommu: select HAVE_FUTEX_CMPXCHG")
Signed-off-by: Randy Dunlap 
Cc: Max Filippov 
Cc: Chris Zankel 
Cc: linux-xtensa@linux-xtensa.org
Message-Id: <20210526070337.28130-1-rdunlap@infradead.org>
Signed-off-by: Max Filippov 
Signed-off-by: Greg Kroah-Hartman

sched/core: Initialize the idle task with preemption disabled

2021-07-14T14:55:50+00:00

[ Upstream commit f1a0a376ca0c4ef1fc3d24e3e502acbb5b795674 ]

As pointed out by commit

  de9b8f5dcbd9 ("sched: Fix crash trying to dequeue/enqueue the idle thread")

init_idle() can and will be invoked more than once on the same idle
task. At boot time, it is invoked for the boot CPU thread by
sched_init(). Then smp_init() creates the threads for all the secondary
CPUs and invokes init_idle() on them.

As the hotplug machinery brings the secondaries to life, it will issue
calls to idle_thread_get(), which itself invokes init_idle() yet again.
In this case it's invoked twice more per secondary: at _cpu_up(), and at
bringup_cpu().

Given smp_init() already initializes the idle tasks for all *possible*
CPUs, no further initialization should be required. Now, removing
init_idle() from idle_thread_get() exposes some interesting expectations
with regards to the idle task's preempt_count: the secondary startup always
issues a preempt_disable(), requiring some reset of the preempt count to 0
between hot-unplug and hotplug, which is currently served by
idle_thread_get() -> idle_init().

Given the idle task is supposed to have preemption disabled once and never
see it re-enabled, it seems that what we actually want is to initialize its
preempt_count to PREEMPT_DISABLED and leave it there. Do that, and remove
init_idle() from idle_thread_get().

Secondary startups were patched via coccinelle:

  @begone@
  @@

  -preempt_disable();
  ...
  cpu_startup_entry(CPUHP_AP_ONLINE_IDLE);

Signed-off-by: Valentin Schneider 
Signed-off-by: Ingo Molnar 
Acked-by: Peter Zijlstra 
Link: https://lore.kernel.org/r/20210512094636.2958515-1-valentin.schneider@arm.com
Signed-off-by: Sasha Levin

xtensa: move coprocessor_flush to the .text section

2021-04-07T13:00:09+00:00

commit ab5eb336411f18fd449a1fb37d36a55ec422603f upstream.

coprocessor_flush is not a part of fast exception handlers, but it uses
parts of fast coprocessor handling code that's why it's in the same
source file. It uses call0 opcode to invoke those parts so there are no
limitations on their relative location, but the rest of the code calls
coprocessor_flush with call8 and that doesn't work when vectors are
placed in a different gigabyte-aligned area than the rest of the kernel.

Move coprocessor_flush from the .exception.text section to the .text so
that it's reachable from the rest of the kernel with call8.

Cc: stable@vger.kernel.org
Signed-off-by: Max Filippov 
Signed-off-by: Greg Kroah-Hartman

xtensa: fix uaccess-related livelock in do_page_fault

2021-04-07T13:00:09+00:00

commit 7b9acbb6aad4f54623dcd4bd4b1a60fe0c727b09 upstream.

If a uaccess (e.g. get_user()) triggers a fault and there's a
fault signal pending, the handler will return to the uaccess without
having performed a uaccess fault fixup, and so the CPU will immediately
execute the uaccess instruction again, whereupon it will livelock
bouncing between that instruction and the fault handler.

https://lore.kernel.org/lkml/20210121123140.GD48431@C02TD0UTHF1T.local/

Cc: stable@vger.kernel.org
Reported-by: Mark Rutland 
Signed-off-by: Max Filippov 
Signed-off-by: Greg Kroah-Hartman

local64.h: make mandatory

2021-01-12T19:18:16+00:00

[ Upstream commit 87dbc209ea04645fd2351981f09eff5d23f8e2e9 ]

Make  mandatory in include/asm-generic/Kbuild and
remove all arch/*/include/asm/local64.h arch-specific files since they
only #include .

This fixes build errors on arch/c6x/ and arch/nios2/ for
block/blk-iocost.c.

Build-tested on 21 of 25 arch-es.  (tools problems on the others)

Yes, we could even rename  to
 and change all #includes to use
 instead.

Link: https://lkml.kernel.org/r/20201227024446.17018-1-rdunlap@infradead.org
Signed-off-by: Randy Dunlap 
Suggested-by: Christoph Hellwig 
Reviewed-by: Masahiro Yamada 
Cc: Jens Axboe 
Cc: Ley Foon Tan 
Cc: Mark Salter 
Cc: Aurelien Jacquiot 
Cc: Peter Zijlstra 
Cc: Arnd Bergmann 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Sasha Levin