Since driver tracks reserved memory, freeing a non-reserved GPMC should
not be fatal and crash the system. Printing a warning is friendlier.
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20200724182328.3348-9-krzk@kernel.org
The GPMC_SET_ONE_CD_MAX macro uses return statement and variable 'cs'
coming from called scope. This is not a good practice. Also
checkpatch complained:
WARNING: Macros with flow control statements should be avoided
ERROR: Macros starting with if should be enclosed by a do - while
loop to avoid possible if/else logic defects
Since GPMC_SET_ONE_CD_MAX macro just calls one function, it can be open
coded. The difference with original code is that function will exit on
error not after every register set, but after a group of sets.
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20200724074038.5597-22-krzk@kernel.org
The callers of gpmc_cs_set_timings() expect to receive -ERRNO on errors
and they pass further what they have received.
However gpmc_cs_set_timings() was returning -1 (equal to -EPERM) which
does not make sense in this context.
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20200724074038.5597-21-krzk@kernel.org
The file-scope variables phys_base and mem_size are assigned in
gpmc_probe() but never read.
This fixes build error when compile testing on x86_64 architecture:
drivers/memory/omap-gpmc.c:246:24: error: conflicting types for ‘phys_base’
static resource_size_t phys_base, mem_size;
In file included from arch/x86/include/asm/page.h:12:0,
from arch/x86/include/asm/thread_info.h:12,
from include/linux/thread_info.h:38,
from arch/x86/include/asm/preempt.h:7,
from include/linux/preempt.h:78,
from include/linux/spinlock.h:51,
from include/linux/irq.h:14,
from drivers/memory/omap-gpmc.c:12:
arch/x86/include/asm/page_64.h:12:22: note: previous declaration of ‘phys_base’ was here
extern unsigned long phys_base;
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20200724074038.5597-3-krzk@kernel.org
The driver does not use macros from asm/mach-types.h (neither MACH_TYPE
nor machine_is_xxx()). Removal of this include allows compile testing
on non-ARM architectures which lack this header.
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20200724074038.5597-2-krzk@kernel.org
This field is not defined for DCN3
Signed-off-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Header Changes]
- Add command for panel power seq control
Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
DSC will be disabled if DPCD 0006F[6:3] is set to a non-zero value
because bits 6:3 are not currently supported. When 6:3 is populated, an
unsupported INCREMENT OF bits_per_pixel value is read (DPCD 0006F[2:0])
[How]
Mask the INCREMENT OF bits_per_pixel field so that values in the
unsupported field are ignored.
Signed-off-by: Jaehyun Chung <jaehyun.chung@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Programming is missing the sequence where for eDP the scrambling is
reset when testing for eye diagram test pattern.
[How]
Include the required register in the definition
Signed-off-by: Chris Park <Chris.Park@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
For some special panel, it will have an invalid HPD high cause driver
DPCD read/write fail.
[How]
Add extra T3 delay as a monitor patch in dce110_edp_wait_for_hpd_ready
Signed-off-by: Lewis Huang <Lewis.Huang@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Add the devicetree properties to enable instruction and data prefetch
on exynos4210 and exynos4412 which use the L2C-310 cache. No other
Exynos chip appears to be using this L2 cache hardware.
This follows the default bits being set in the l2c_aux_val register
for the Exynos platform, which can now be cleared as a result.
Signed-off-by: Guillaume Tucker <guillaume.tucker@collabora.com>
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
[why]
Previously when force enabling DSC on SST display we unknowingly
supressed lane count, which caused DSC to be enabled automatically.
[how]
By adding an additional flag to force enable DSC in dc_dsc.c DSC can
always be enabled with debugfs dsc_clock_en forced to 1
Cc: stable@vger.kernel.org
Signed-off-by: Eryk Brol <eryk.brol@amd.com>
Signed-off-by: Mikita Lipski <mikita.lipski@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[why]
Current pipe merge and split logic only supports cases where new
dc_state is allocated and relies on dc->current_state to gather
information from previous dc_state.
Calls to validate_bandwidth on UPDATE_TYPE_MED would cause an issue
because there is no new dc_state allocated, and data in
dc->current_state would be overwritten during pipe merge.
[how]
Only allow validate_bandwidth when new dc_state space is created.
Signed-off-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
with eDP + DP, each display use one pipe. after DP unplugged, eDP switch
from one pipe to two pipes -- pipe split. dpp1_cm_set_regamma_pwl will
be executed too. The duration from switch single pipe to dual pipes is a
little long which could let eDP enter PSR mode. upon two pipes for eDP
are setup, eDP PHY is disabled. front pipe is not really running to
fetch data from frame buffer. i.e., dchubp is not in normal working
status. execution of hubbub1_wm_change_req_wa may cause p-state warning.
[How]
disable eDP PSR before dc_commit_state. psr is disabled when execute
hubbub1_wm_change_req_wa.
Signed-off-by: Hersen Wu <hersenxs.wu@amd.com>
Reviewed-by: Roman Li <Roman.Li@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Place the cursor in the center of screen between two pipes then
adjusting the viewport but cursour doesn't update cause DFPstate hang.
[How]
If viewport changed, update cursor as well.
Cc: stable@vger.kernel.org
Signed-off-by: Paul Hsieh <paul.hsieh@amd.com>
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why & How]
change abm config init interface to support multiple ABMs.
Signed-off-by: Yongqiang Sun <yongqiang.sun@amd.com>
Reviewed-by: Chris Park <Chris.Park@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The L310_AUX_CTRL_NS_LOCKDOWN flag is set during the L2C enable
sequence. There is no need to set it in the default register value,
this was done before support for it was implemented in the code. It
is not set in the hardware initial value either.
Clean this up by removing this flag from the default l2c_aux_val, and
add it to the l2c_aux_mask to print an alert message if it was already
set before the kernel initialisation.
Signed-off-by: Guillaume Tucker <guillaume.tucker@collabora.com>
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
[why]
create a dedicated function to make mst link settings decision, so that
the policy's decision is made in a unified place.
Signed-off-by: Wenjing Liu <wenjing.liu@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
Reading for DP alt registers return incorrect values due to LE_SF
definition missing.
[How]
Define correct LE_SF or DP alt registers.
Signed-off-by: Chris Park <Chris.Park@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[Why]
If DMUB is used, LVTMA VBIOS call can be used to control eDP instead of
tranditional transmitter control. Interface is agreed with VBIOS for
eDP to use this new path to program LVTMA registers.
[How]
Create DAL interface to send DMUB command for LVTMA as currently
implemented in VBIOS.
Signed-off-by: Chris Park <Chris.Park@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
v1:
add trace event enabled check to avoid nop loop when submit multi ibs
in amdgpu_cs_ioctl() function.
v2:
add a new wrapper function to trace all amdgpu cs ibs.
Signed-off-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
fix amdgpu_bo_release_notify() comment error.
Signed-off-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Dennis Li <Dennis.Li@amd.com>
Acked-by: Nirmoy Das <nirmoy.das@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Renoir only has one sdma instance, it will get failed once query the
sdma1 registers. So use switch-case instead of static register array.
Signed-off-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Some devices have broken extension unit where getting current value
doesn't work. Attempt that once when creating mixer control for it. If
it fails, just ignore it, so that it won't cripple the device entirely
(and/or make the error floods).
Signed-off-by: Tom Yan <tom.ty89@gmail.com>
Link: https://lore.kernel.org/r/5f3abc52.1c69fb81.9cf2.fe91@mx.google.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Previously the driver would use devm_* related functions at
the codec level probe() to allocate clock resources for MCLK
and the DAI clocks exposed by the device. This caused issues
when registering clocks on a re-probe (no device level
remove/prove involved) as the devm_* resources were never
freed up so the clocks were still registered from the previous
codec level probe().
This commit updates the clock handling for MCLK usage and DAI
clock provision to fix this discrepancy and allow the codec level
probe/remove functionality to operate as intended.
Signed-off-by: Adam Thomson <Adam.Thomson.Opensource@diasemi.com>
Link: https://lore.kernel.org/r/b92c461baeed27a6cd92e59e36a55c2547218683.1597164865.git.Adam.Thomson.Opensource@diasemi.com
Signed-off-by: Mark Brown <broonie@kernel.org>
As part of the reorganisation of the device level and codec
level probe functionlity, the soft reset handling should really
reside at the codec level and after the instantiation of supplies.
This commit makes the relevant changes to support this change of
scope including the remove of devm_* functions being called for
regulator instantiation at the codec level.
Signed-off-by: Adam Thomson <Adam.Thomson.Opensource@diasemi.com>
Link: https://lore.kernel.org/r/f7603a4855647429b754ce76f887ec441622015c.1597164865.git.Adam.Thomson.Opensource@diasemi.com
Signed-off-by: Mark Brown <broonie@kernel.org>
If debug_regs.c is built with newer binutils, the resulting binary is "optimized"
by the assembler:
asm volatile("ss_start: "
"xor %%rax,%%rax\n\t"
"cpuid\n\t"
"movl $0x1a0,%%ecx\n\t"
"rdmsr\n\t"
: : : "rax", "ecx");
is translated to :
000000000040194e <ss_start>:
40194e: 31 c0 xor %eax,%eax <----- rax->eax?
401950: 0f a2 cpuid
401952: b9 a0 01 00 00 mov $0x1a0,%ecx
401957: 0f 32 rdmsr
As you can see rax is replaced with eax in target binary code.
This causes a difference is the length of xor instruction (2 Byte vs 3 Byte),
and makes the hard-coded instruction length check fail:
/* Instruction lengths starting at ss_start */
int ss_size[4] = {
3, /* xor */ <-------- 2 or 3?
2, /* cpuid */
5, /* mov */
2, /* rdmsr */
};
Encode the shorter version directly and, while at it, fix the "clobbers"
of the asm.
Cc: stable@vger.kernel.org
Signed-off-by: Yang Weijiang <weijiang.yang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This fell off in the conversion in
commit 9bcaa3fe58
Author: Michal Orzel <michalorzel.eng@gmail.com>
Date: Tue Apr 28 19:10:04 2020 +0200
drm: Replace drm_modeset_lock/unlock_all with DRM_MODESET_LOCK_ALL_* helpers
but it's caught by the drm_warn_on_modeset_not_all_locked() that the
legacy modeset code uses. Since this is the bkl and it's unclear
what's all protected, play it safe and grab it again for legacy
drivers.
Unfortunately this means we need to sprinkle a few more #includes
around.
Also we need to add the drm_device as a parameter to the _END macro.
Finally remove the mute_lock() from setcrtc, since that's now done by
the macro.
Cc: Alex Deucher <alexdeucher@gmail.com>
References: https://gitlab.freedesktop.org/drm/amd/-/issues/1224
Fixes: 9bcaa3fe58 ("drm: Replace drm_modeset_lock/unlock_all with DRM_MODESET_LOCK_ALL_* helpers")
Cc: Michal Orzel <michalorzel.eng@gmail.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: dri-devel@lists.freedesktop.org
Cc: <stable@vger.kernel.org> # v5.8+
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200814093842.3048472-1-daniel.vetter@ffwll.ch
During gapless playback, its possible for previous track to
end at unaligned boundary, starting next track on the same
boundary can lead to unaligned address exception in dsp.
So implement copy callback for finer control on the buffer offsets.
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Tested-by: Vinod Koul <vkoul@kernel.org>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Vinod Koul <vkoul@kernel.org>
Link: https://lore.kernel.org/r/20200727093806.17089-11-srinivas.kandagatla@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Add support to gapless playback by implementing metadata,
next_track, drain and partial drain support.
Gapless on Q6ASM is implemented by opening 2 streams in a single
q6asm stream and toggling them on next track.
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Tested-by: Vinod Koul <vkoul@kernel.org>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Vinod Koul <vkoul@kernel.org>
Link: https://lore.kernel.org/r/20200727093806.17089-10-srinivas.kandagatla@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
rearrange code so that it will be easy to change the codec
profile at runtime. This means moving exiting set_params
to an internal wrapper which can be called when codec
profile changes.
This is also preparing the code for easy to use in gapless cases.
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Tested-by: Vinod Koul <vkoul@kernel.org>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Vinod Koul <vkoul@kernel.org>
Link: https://lore.kernel.org/r/20200727093806.17089-9-srinivas.kandagatla@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
This patch adds support to metadata required to do a gapless playback.
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Tested-by: Vinod Koul <vkoul@kernel.org>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Vinod Koul <vkoul@kernel.org>
Link: https://lore.kernel.org/r/20200727093806.17089-8-srinivas.kandagatla@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
This patch adds support to gapless flag to q6asm_open_write().
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Tested-by: Vinod Koul <vkoul@kernel.org>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Vinod Koul <vkoul@kernel.org>
Link: https://lore.kernel.org/r/20200727093806.17089-7-srinivas.kandagatla@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
This patch adds support to ASM_DATA_CMD_REMOVE_INITIAL_SILENCE
and ASM_DATA_CMD_REMOVE_TRAILING_SILENCE q6asm command to support
compressed metadata for gapless playback.
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Tested-by: Vinod Koul <vkoul@kernel.org>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Vinod Koul <vkoul@kernel.org>
Link: https://lore.kernel.org/r/20200727093806.17089-6-srinivas.kandagatla@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Add length to write command packet token so that we can track exactly
how many bytes are consumed by DSP in the command reply.
This is useful in some use-cases where the end of the file/stream
is not aligned with period size.
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Tested-by: Vinod Koul <vkoul@kernel.org>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Vinod Koul <vkoul@kernel.org>
Link: https://lore.kernel.org/r/20200727093806.17089-5-srinivas.kandagatla@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
use flags set by q6asm-dais directly!
This will be useful gapless case where write needs a special flag to indicate
that last buffer.
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Tested-by: Vinod Koul <vkoul@kernel.org>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Vinod Koul <vkoul@kernel.org>
Link: https://lore.kernel.org/r/20200727093806.17089-4-srinivas.kandagatla@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Each ASM session can have multiple streams attached to it,
current design was to allow only one static stream id 1 per each session.
However for use-case like gapless, we would need 2 streams to open per session.
This patch converts all the q6asm apis to take stream id as argument
to allow multiple streams to open on a single session, This is useful
for gapless playback cases.
Now the dai driver can specify which stream id for each command.
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Tested-by: Vinod Koul <vkoul@kernel.org>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Vinod Koul <vkoul@kernel.org>
Link: https://lore.kernel.org/r/20200727093806.17089-3-srinivas.kandagatla@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Each q6asm session can have multiple streams, mixing usage of these
names in variable are bit misleading to reader, so rename them accordingly.
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Tested-by: Vinod Koul <vkoul@kernel.org>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Vinod Koul <vkoul@kernel.org>
Link: https://lore.kernel.org/r/20200727093806.17089-2-srinivas.kandagatla@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
The vfio_iommu_replay() function does not currently unwind on error,
yet it does pin pages, perform IOMMU mapping, and modify the vfio_dma
structure to indicate IOMMU mapping. The IOMMU mappings are torn down
when the domain is destroyed, but the other actions go on to cause
trouble later. For example, the iommu->domain_list can be empty if we
only have a non-IOMMU backed mdev attached. We don't currently check
if the list is empty before getting the first entry in the list, which
leads to a bogus domain pointer. If a vfio_dma entry is erroneously
marked as iommu_mapped, we'll attempt to use that bogus pointer to
retrieve the existing physical page addresses.
This is the scenario that uncovered this issue, attempting to hot-add
a vfio-pci device to a container with an existing mdev device and DMA
mappings, one of which could not be pinned, causing a failure adding
the new group to the existing container and setting the conditions
for a subsequent attempt to explode.
To resolve this, we can first check if the domain_list is empty so
that we can reject replay of a bogus domain, should we ever encounter
this inconsistent state again in the future. The real fix though is
to add the necessary unwind support, which means cleaning up the
current pinning if an IOMMU mapping fails, then walking back through
the r-b tree of DMA entries, reading from the IOMMU which ranges are
mapped, and unmapping and unpinning those ranges. To be able to do
this, we also defer marking the DMA entry as IOMMU mapped until all
entries are processed, in order to allow the unwind to know the
disposition of each entry.
Fixes: a54eb55045 ("vfio iommu type1: Add support for mediated devices")
Reported-by: Zhiyi Guo <zhguo@redhat.com>
Tested-by: Zhiyi Guo <zhguo@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
A down_read on memory_lock is held when performing read/write accesses
to MMIO BAR space, including across the copy_to/from_user() callouts
which may fault. If the user buffer for these copies resides in an
mmap of device MMIO space, the mmap fault handler will acquire a
recursive read-lock on memory_lock. Avoid this by reducing the lock
granularity. Sequential accesses requiring multiple ioread/iowrite
cycles are expected to be rare, therefore typical accesses should not
see additional overhead.
VGA MMIO accesses are expected to be non-fatal regardless of the PCI
memory enable bit to allow legacy probing, this behavior remains with
a comment added. ioeventfds are now included in memory access testing,
with writes dropped while memory space is disabled.
Fixes: abafbc551f ("vfio-pci: Invalidate mmaps and block MMIO access on disabled memory")
Reported-by: Zhiyi Guo <zhguo@redhat.com>
Tested-by: Zhiyi Guo <zhguo@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Fix W=1 compile warnings (invalid kerneldoc):
arch/arm/plat-samsung/pm-common.c:68: warning: Function parameter or member 'ptr' not described in 's3c_pm_do_restore_core'
arch/arm/plat-samsung/pm-common.c:68: warning: Function parameter or member 'count' not described in 's3c_pm_do_restore_core'
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
The s3cmci driver does not use machine header mach/dma.h.
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Acked-by: Ulf Hansson <ulf.hansson@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
