Use vb2_video_unregister_device() to automatically stop streaming
at unregister time.
This avoids the use of vb2_queue_release() which should not be
called by drivers that set vdev->queue.
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
This is only needed for drivers that do not use vb2_fop_release().
Note that vb2_queue_release() is *not* the counterpart of vb2_queue_init()
as some drivers here seem to think.
Also use vb2_video_unregister_device() to automatically stop streaming
at unregister time for those drivers that set vdev->queue.
Note that sun4i-csi didn't unregister the video device at all. That's
now fixed.
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Use vb2_video_unregister_device() to automatically stop streaming
at unregister time.
This avoids the use of vb2_queue_release() which should not be
called by drivers that set vdev->queue.
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Use vb2_video_unregister_device() to automatically stop streaming
at unregister time.
This avoids the use of vb2_queue_release() which should not be
called by drivers that set vdev->queue.
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Tested-by: Andrey Konovalov <andrey.konovalov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
If a driver calls (_)vb2_fop_release(), then such a driver should also
call vb2_video_unregister_device() instead of video_unregister_device().
This helper will call vb2_queue_release() if a filehandle is marked as
owner of the queue. This ensures that at unregister time any streaming
is cancelled and all buffers are returned to userspace.
This is very useful for complex drivers since this stops all streaming
in all subdevs in the pipeline controlled by this video device. Otherwise
this would be delayed until the owner filehandle is closed, which can be
quite some time later.
Bonus points for ordering the includes :-)
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
bpf_link_info.iter is used by link_query to return bpf_iter_link_info
to user space. Fields may be different, e.g., map_fd vs. map_id, so
we cannot reuse the exact structure. But make them similar, e.g.,
struct bpf_link_info {
/* common fields */
union {
struct { ... } raw_tracepoint;
struct { ... } tracing;
...
struct {
/* common fields for iter */
union {
struct {
__u32 map_id;
} map;
/* other structs for other targets */
};
};
};
};
so the structure is extensible the same way as bpf_iter_link_info.
Fixes: 6b0a249a30 ("bpf: Implement link_query for bpf iterators")
Signed-off-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Link: https://lore.kernel.org/bpf/20200828051922.758950-1-yhs@fb.com
vdso32 should only be installed if CONFIG_COMPAT_VDSO is enabled,
since it's not even supposed to be compiled otherwise, and arm64
builds without a 32bit crosscompiler will fail.
Fixes: 8d75785a81 ("ARM64: vdso32: Install vdso32 from vdso_install")
Signed-off-by: Frank van der Linden <fllinden@amazon.com>
Cc: stable@vger.kernel.org [5.4+]
Link: https://lore.kernel.org/r/20200827234012.19757-1-fllinden@amazon.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
There is no point in using GFP_ATOMIC here.
It is a probe function, and GFP_KERNEL is already used the line before
and the line after.
Use GFP_KERNEL instead.
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
During bit-banging the IR on a gpio pin, we cannot be scheduled or have
anything interrupt us, else the generated signal will be incorrect.
Therefore, we need to disable interrupts on the local cpu. This also
disables preemption.
local_irq_disable() does exactly what we need and does not require a
spinlock.
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Device drivers do not expect to have change_protocol or wakeup
re-programming to be accesed after rc_unregister_device(). This can
cause the device driver to access deallocated resources.
Cc: <stable@vger.kernel.org> # 4.16+
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The system for "Auto-detecting system features" located under
tools/build/ are (currently) used by perf, libbpf and bpftool. It can
contain stalled feature detection files, which are not cleaned up by
libbpf and bpftool on make clean (side-note: perf tool is correct).
Fix this by making the users invoke the make clean target.
Some details about the changes. The libbpf Makefile already had a
clean-config target (which seems to be copy-pasted from perf), but this
target was not "connected" (a make dependency) to clean target. Choose
not to rename target as someone might be using it. Did change the output
from "CLEAN config" to "CLEAN feature-detect", to make it more clear
what happens.
This is related to the complaint and troubleshooting in the following
link: https://lore.kernel.org/lkml/20200818122007.2d1cfe2d@carbon/
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Link: https://lore.kernel.org/lkml/20200818122007.2d1cfe2d@carbon/
Link: https://lore.kernel.org/bpf/159851841661.1072907.13770213104521805592.stgit@firesoul
When stdout output from the selftests tool 'test_maps' gets redirected
into e.g file or pipe, then the output lines increase a lot (from 21
to 33949 lines). This is caused by the printf that happens before the
fork() call, and there are user-space buffered printf data that seems
to be duplicated into the forked process.
To fix this fflush() stdout before the fork loop in __run_parallel().
Fixes: 1a97cf1fe5 ("selftests/bpf: speedup test_maps")
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/159842985651.1050885.2154399297503372406.stgit@firesoul
The files maintained as part of the RDACM20 were incorrectly sorted
while they were added.
Correct the sort-order.
Fixes: 34009bffc1 ("media: i2c: Add RDACM20 driver")
Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Acked-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
The Kinetic KTD253 backlight driver is controlled with a
single GPIO line, but still supports a range of brightness
settings by sending fast pulses on the line.
This is based off the source code release for the Samsung
GT-S7710 mobile phone.
Cc: Sam Ravnborg <sam@ravnborg.org>
Reviewed-by: Daniel Thompson <daniel.thompson@linaro.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
This adds device tree bindings for the Kinetic KTD253
white LED backlight driver.
Cc: devicetree@vger.kernel.org
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Daniel Thompson <daniel.thompson@linaro.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Let's use a common.yaml include for the backlight like we do with
the LEDs. The LEDs are inherently incompatible so their bindings
cannot be reused for backlight.
Cc: devicetree@vger.kernel.org
Suggested-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Daniel Thompson <daniel.thompson@linaro.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
commit 25612477d2 ("ASoC: soc-dai: set dai_link dpcm_ flags with a helper")
added snd_soc_dai_link_set_capabilities().
But it is using snd_soc_find_dai() (A) which is required client_mutex (B).
And client_mutex is soc-core.c local.
struct snd_soc_dai *snd_soc_find_dai(xxx)
{
...
(B) lockdep_assert_held(&client_mutex);
...
}
void snd_soc_dai_link_set_capabilities(xxx)
{
...
for_each_pcm_streams(direction) {
...
for_each_link_cpus(dai_link, i, cpu) {
(A) dai = snd_soc_find_dai(cpu);
...
}
...
for_each_link_codecs(dai_link, i, codec) {
(A) dai = snd_soc_find_dai(codec);
...
}
}
...
}
Because of these background, we will get WARNING if .config has CONFIG_LOCKDEP.
WARNING: CPU: 2 PID: 53 at sound/soc/soc-core.c:814 snd_soc_find_dai+0xf8/0x100
CPU: 2 PID: 53 Comm: kworker/2:1 Not tainted 5.7.0-rc1+ #328
Hardware name: Renesas H3ULCB Kingfisher board based on r8a77951 (DT)
Workqueue: events deferred_probe_work_func
pstate: 60000005 (nZCv daif -PAN -UAO)
pc : snd_soc_find_dai+0xf8/0x100
lr : snd_soc_find_dai+0xf4/0x100
...
Call trace:
snd_soc_find_dai+0xf8/0x100
snd_soc_dai_link_set_capabilities+0xa0/0x16c
graph_dai_link_of_dpcm+0x390/0x3c0
graph_for_each_link+0x134/0x200
graph_probe+0x144/0x230
platform_drv_probe+0x5c/0xb0
really_probe+0xe4/0x430
driver_probe_device+0x60/0xf4
snd_soc_find_dai() will be used from (X) CPU/Codec/Platform driver with
mutex lock, and (Y) Card driver without mutex lock.
This snd_soc_dai_link_set_capabilities() is for Card driver,
this means called without mutex.
This patch adds snd_soc_find_dai_with_mutex() to solve it.
Fixes: 25612477d2 ("ASoC: soc-dai: set dai_link dpcm_ flags with a helper")
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/87blixvuab.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Improve the clarity and grammar of descriptive comment on top of the
minor number assignments.
Fix a typo within 2 comments for macros.
s/This helps in eleminating of boilerplate code.
/This helps to eliminate boilerplate code./
Signed-off-by: Sebastian Fricke <sebastian.fricke.linux@gmail.com>
Link: https://lore.kernel.org/r/20200803054346.4285-1-sebastian.fricke.linux@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In a couple of places in qp_host_get_user_memory(),
get_user_pages_fast() is called without properly checking for errors. If
e.g. -EFAULT is returned, this negative value will then be passed on to
qp_release_pages(), which expects a u64 as input.
Fix this by only calling qp_release_pages() when we have a positive
number returned.
Fixes: 06164d2b72 ("VMCI: queue pairs implementation.")
Signed-off-by: Alex Dewar <alex.dewar90@gmail.com>
Link: https://lore.kernel.org/r/20200825164522.412392-1-alex.dewar90@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Use kvzalloc() in udf_sb_alloc_bitmap() instead of open-coding it.
Size computation wrapped in struct_size() macro to prevent potential
integer overflows.
Link: https://lore.kernel.org/r/20200827221652.64660-1-efremov@linux.com
Signed-off-by: Denis Efremov <efremov@linux.com>
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Jan Kara <jack@suse.cz>
Remove linux/fiemap.h which is included more than once
Link: https://lore.kernel.org/r/20200819025434.65763-1-wanghai38@huawei.com
Reported-by: Hulk Robot <hulkci@huawei.com>
Reviewed-by: Ritesh Harjani <riteshh@linux.ibm.com>
Signed-off-by: Wang Hai <wanghai38@huawei.com>
Signed-off-by: Jan Kara <jack@suse.cz>
In nbu2ss_eq_queue() memory is allocated with dma_alloc_coherent(),
though, strangely, NULL is passed as the struct device* argument. Pass
the UDC's device instead. Fix up the corresponding call to
dma_free_coherent() in the same way.
Build-tested on x86 only.
Fixes: 33aa8d45a4 ("staging: emxx_udc: Add Emma Mobile USB Gadget driver")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Alex Dewar <alex.dewar90@gmail.com>
Link: https://lore.kernel.org/r/20200825091928.55794-1-alex.dewar90@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Convert the atomisp LM3554 driver to use GPIO descriptors
fully. It was already retrieveing the GPIO lines as descriptors
but for some reason converting them back into global GPIO
numbers. There is no reason to do this, just deal with the
descriptors as-is.
Cc: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20200827081445.186882-1-linus.walleij@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When devm_clk_get() returns -EPROBE_DEFER, spi_priv
should be freed just like when wilc_cfg80211_init()
fails.
Fixes: 854d66df74 ("staging: wilc1000: look for rtc_clk clock in spi mode")
Signed-off-by: Dinghao Liu <dinghao.liu@zju.edu.cn>
Link: https://lore.kernel.org/r/20200819115426.29852-1-dinghao.liu@zju.edu.cn
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The host and the device can be connected with a called Wake-Up GPIO.
When the host fall down this GPIO, it allows the device to enter in deep
sleep and no communication with the device is no more possible (the
device wakes up automatically on DTIM and fetch data if necessary).
So, before to communicate with the device, the driver have to raise the
Wake-up GPIO and then wait for an IRQ from the device.
Unfortunately, old firmwares have a race in sleep/wake-up process and
the device may never wake up. In this case, the IRQ is not sent and
driver complains with "timeout while wake up chip". Then, the driver
tries anyway to access the bus and an other error is raised by the bus.
Fortunately, when the bug occurs, it is possible to fall down the IRQ
and the device will eventually finish the sleep process. Then the driver
can wake it up normally.
The patch implements that workaround and add a retry limit in case
something goes very wrong.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200825085828.399505-12-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The initial developer has feared msecs_to_jiffies() could round down the
result. However, the documentation of msecs_to_jiffies() says that the
result is rounded upward. So the increment of the result of
msecs_to_jiffies() is not necessary.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200825085828.399505-11-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In the old days, ieee80211 powersave has some impact on the Rx speed.
These problems are solved for a long time now. There is no more reason
to not enabling it.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200825085828.399505-10-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When mac80211 debug is enabled, the trace below appears:
[60744.340037] wlan0: Rx A-MPDU request on aa:bb:cc:97:60:24 tid 0 result -524
This imply that ___ieee80211_start_rx_ba_session will prematurely exit
and frame reordering won't be enabled.
Fixes: e5da5fbd77 ("staging: wfx: fix CCMP/TKIP replay protection")
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200825085828.399505-7-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The device need to receive a skb with necessary space for the ICV. So,
the driver adds this space before to send the frame.
Currently, once the frame is sent, the driver restore the original
content of the skb. However, this step is useless. Mac80211 don't do it
when software encryption is enabled.
Once we have removed this step, it appears that it is no more necessary
to keep hw_key in tx_priv. Then, it is possible to simplify a bunch of
code in the Tx path.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200825085828.399505-6-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When MFP is enabled, the multicast management frames are not protected,
in fact. Instead, but they should include an IE containing the MMIC of
the frames (i.e. a cryptographic signature).
Until now, the driver didn't correctly detect this kind of frames (they
are not marked protected but they are associated to a key) and didn't
ask to the device to encrypt them.
In add, the device is not able to generate the IE itself. Mac80211 has
to generate the IE and let the device compute the MMIC.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200825085828.399505-5-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The protection of the management frames is mainly done by mac80211.
However, frames for the management of the BlockAck sessions are directly
sent by the device. These frames have to be protected if MFP is in use.
So the driver has to pass the MFP configuration to the device.
Until now, the driver directly read the RSN IE of the BSS. However, it
didn't work when the BSS was MFP capable (ieee80211w=1) and the local
device has disabled MFP (ieee80211w=0).
This patch read the MFP information directly from the struct
ieee80211_sta. This information take into account the MFP negotiated
during the association. In addition, the code is far simpler.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200825085828.399505-3-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Until now, hif_map_link() get as argument the raw value for
map_link_flags when map_link_flags is defined as a bitfield. It was
error prone.
Now hif_map_link() takes explicit value for every flags of the
struct map_link_flags.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200825085828.399505-2-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The protection of the management frames is mainly done by mac80211.
However, frames for the management of the BlockAck sessions are directly
sent by the device. These frames have to be protected if MFP is in use.
So the driver has to pass the MFP configuration to the device.
Until now, the BlockAck management frames were completely unprotected
whatever the status of the MFP negotiation. So, some devices dropped
these frames.
The device has two knobs to control the MFP. One global and one per
station. Normally, the driver should always enable global MFP. Then it
should enable MFP on every station with which MFP was successfully
negotiated. Unfortunately, the older firmwares only provide the
global control.
So, this patch enable global MFP as it is exposed in the beacon. Then it
marks every station with which the MFP is effective.
Thus, the support for the old firmwares is not so bad. It may only
encounter some difficulties to negotiate BA sessions when the local
device (the AP) is MFP capable (ieee80211w=1) but the station is not.
The only solution for this case is to upgrade the firmware.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200825085828.399505-1-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Indent a number of arguments so that they align with the opening
parentheses of the function calls. Issues found by checkpatch.
Signed-off-by: Steff Richards <steff.richards.the.third@gmail.com>
Link: https://lore.kernel.org/r/20200824165356.GA5204@steffpad
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The word "enought" was changed to read as "enough".
In addition to this, there was a missing new-line after varaible
declarations within a function.
Signed-off-by: Timo A. Aranjo <timo.aranjo@aol.com>
Link: https://lore.kernel.org/r/20200826012101.12162-1-timo.aranjo@aol.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There is a spelling mistake in the MODULE_ALIAS, fix it.
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/20200820075136.186199-1-colin.king@canonical.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 7c78f67e9b ("arm64: enable tlbi range instructions") breaks
LLVM's integrated assembler, because -Wa,-march is only passed to
external assemblers and therefore, the new instructions are not enabled
when IAS is used.
This change adds a common architecture version preamble, which can be
used in inline assembly blocks that contain instructions that require
a newer architecture version, and uses it to fix __TLBI_0 and __TLBI_1
with ARM64_TLB_RANGE.
Fixes: 7c78f67e9b ("arm64: enable tlbi range instructions")
Signed-off-by: Sami Tolvanen <samitolvanen@google.com>
Tested-by: Nathan Chancellor <natechancellor@gmail.com>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Link: https://github.com/ClangBuiltLinux/linux/issues/1106
Link: https://lore.kernel.org/r/20200827203608.1225689-1-samitolvanen@google.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
If the hypervisor doesn't support hugepages, the kernel ends up allocating a large
number of page table pages. The early page table allocation was wrongly
setting the max memblock limit to ppc64_rma_size with radix translation
which resulted in boot failure as shown below.
Kernel panic - not syncing:
early_alloc_pgtable: Failed to allocate 16777216 bytes align=0x1000000 nid=-1 from=0x0000000000000000 max_addr=0xffffffffffffffff
CPU: 0 PID: 0 Comm: swapper Not tainted 5.8.0-24.9-default+ #2
Call Trace:
[c0000000016f3d00] [c0000000007c6470] dump_stack+0xc4/0x114 (unreliable)
[c0000000016f3d40] [c00000000014c78c] panic+0x164/0x418
[c0000000016f3dd0] [c000000000098890] early_alloc_pgtable+0xe0/0xec
[c0000000016f3e60] [c0000000010a5440] radix__early_init_mmu+0x360/0x4b4
[c0000000016f3ef0] [c000000001099bac] early_init_mmu+0x1c/0x3c
[c0000000016f3f10] [c00000000109a320] early_setup+0x134/0x170
This was because the kernel was checking for the radix feature before we enable the
feature via mmu_features. This resulted in the kernel using hash restrictions on
radix.
Rework the early init code such that the kernel boot with memblock restrictions
as imposed by hash. At that point, the kernel still hasn't finalized the
translation the kernel will end up using.
We have three different ways of detecting radix.
1. dt_cpu_ftrs_scan -> used only in case of PowerNV
2. ibm,pa-features -> Used when we don't use cpu_dt_ftr_scan
3. CAS -> Where we negotiate with hypervisor about the supported translation.
We look at 1 or 2 early in the boot and after that, we look at the CAS vector to
finalize the translation the kernel will use. We also support a kernel command
line option (disable_radix) to switch to hash.
Update the memblock limit after mmu_early_init_devtree() if the kernel is going
to use radix translation. This forces some of the memblock allocations we do before
mmu_early_init_devtree() to be within the RMA limit.
Fixes: 2bfd65e45e ("powerpc/mm/radix: Add radix callbacks for early init routines")
Reported-by: Shirisha Ganta <shiganta@in.ibm.com>
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Reviewed-by: Hari Bathini <hbathini@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200828100852.426575-1-aneesh.kumar@linux.ibm.com
