openwrt/staging/blogic.git
5 years agoMerge tag 'drm-intel-fixes-2019-05-23' of git://anongit.freedesktop.org/drm/drm-intel...
Dave Airlie [Fri, 24 May 2019 04:01:00 +0000 (14:01 +1000)]
Merge tag 'drm-intel-fixes-2019-05-23' of git://anongit.freedesktop.org/drm/drm-intel into drm-fixes

- Fix boosting of new client to be non-preemptive
- Fix to actually bump ready tasks ahead of busywaits
- Includes gvt-fixes-2019-05-21

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190523094221.GA26026@jlahtine-desk.ger.corp.intel.com
5 years agoMerge branch 'drm-fixes-5.2' of git://people.freedesktop.org/~agd5f/linux into drm...
Dave Airlie [Thu, 23 May 2019 02:04:00 +0000 (12:04 +1000)]
Merge branch 'drm-fixes-5.2' of git://people.freedesktop.org/~agd5f/linux into drm-fixes

Fixes for 5.2:
- Fix for DMCU firmware issues for stable
- Add missing polaris10 pci id to kfd
- Screen corruption fix on picasso
- Fix for driver reload on vega10
- SR-IOV fixes
- Locking fix in new SMU code
- Compute profile switching fix for KFD

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190522205425.3657-1-alexander.deucher@amd.com
5 years agoMerge tag 'drm-misc-fixes-2019-05-22' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Thu, 23 May 2019 02:03:35 +0000 (12:03 +1000)]
Merge tag 'drm-misc-fixes-2019-05-22' of git://anongit.freedesktop.org/drm/drm-misc into drm-fixes

- sun4i fixes to hdmi phy as well as u16 overflow in dsi (left from -next-fixes)
- gma500 fix to make lvds detection more reliable
- select devfreq for panfrost since it can't probe without it

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Sean Paul <sean@poorly.run>
Link: https://patchwork.freedesktop.org/patch/msgid/20190522194440.GA22359@art_vandelay
5 years agoMerge branch 'vmwgfx-fixes-5.2' of git://people.freedesktop.org/~thomash/linux into...
Dave Airlie [Thu, 23 May 2019 02:02:30 +0000 (12:02 +1000)]
Merge branch 'vmwgfx-fixes-5.2' of git://people.freedesktop.org/~thomash/linux into drm-fixes

A set of misc fixes for various issues that have surfaced recently.
All Cc'd stable except the dma iterator fix which shouldn't really cause
any real issues on older kernels.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: "Thomas Hellstrom (VMware)" <thomas@shipmail.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20190522115408.33185-1-thomas@shipmail.org
5 years agodrm/panfrost: Select devfreq
Ezequiel Garcia [Fri, 17 May 2019 15:00:42 +0000 (12:00 -0300)]
drm/panfrost: Select devfreq

Currently, there is some logic for the driver to work without devfreq.
However, the driver actually fails to probe if !CONFIG_PM_DEVFREQ.

Fix this by selecting devfreq, and drop the additional checks
for devfreq.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Rob Herring <robh@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20190517150042.776-1-ezequiel@collabora.com
5 years agoMerge tag 'gvt-fixes-2019-05-21' of https://github.com/intel/gvt-linux into drm-intel...
Joonas Lahtinen [Tue, 21 May 2019 11:37:37 +0000 (14:37 +0300)]
Merge tag 'gvt-fixes-2019-05-21' of https://github.com/intel/gvt-linux into drm-intel-fixes

gvt-fixes-2019-05-21

- vGPU reset fix with sane init breadcrumb (Weinan)
- Fix TRTT handling to use context state (Yan)
- Fix one error return (Dan)

Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
From: Zhenyu Wang <zhenyuw@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190521062408.GH12913@zhen-hp.sh.intel.com
5 years agodrm/gma500/cdv: Check vbt config bits when detecting lvds panels
Patrik Jakobsson [Tue, 16 Apr 2019 11:46:07 +0000 (13:46 +0200)]
drm/gma500/cdv: Check vbt config bits when detecting lvds panels

Some machines have an lvds child device in vbt even though a panel is
not attached. To make detection more reliable we now also check the lvds
config bits available in the vbt.

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1665766
Cc: stable@vger.kernel.org
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190416114607.1072-1-patrik.r.jakobsson@gmail.com
5 years agodrm/vmwgfx: integer underflow in vmw_cmd_dx_set_shader() leading to an invalid read
Murray McAllister [Mon, 20 May 2019 09:57:34 +0000 (21:57 +1200)]
drm/vmwgfx: integer underflow in vmw_cmd_dx_set_shader() leading to an invalid read

If SVGA_3D_CMD_DX_SET_SHADER is called with a shader ID
of SVGA3D_INVALID_ID, and a shader type of
SVGA3D_SHADERTYPE_INVALID, the calculated binding.shader_slot
will be 4294967295, leading to an out-of-bounds read in vmw_binding_loc()
when the offset is calculated.

Cc: <stable@vger.kernel.org>
Fixes: d80efd5cb3de ("drm/vmwgfx: Initial DX support")
Signed-off-by: Murray McAllister <murray.mcallister@gmail.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
5 years agodrm/vmwgfx: NULL pointer dereference from vmw_cmd_dx_view_define()
Murray McAllister [Sat, 11 May 2019 06:01:37 +0000 (18:01 +1200)]
drm/vmwgfx: NULL pointer dereference from vmw_cmd_dx_view_define()

If SVGA_3D_CMD_DX_DEFINE_RENDERTARGET_VIEW is called with a surface
ID of SVGA3D_INVALID_ID, the srf struct will remain NULL after
vmw_cmd_res_check(), leading to a null pointer dereference in
vmw_view_add().

Cc: <stable@vger.kernel.org>
Fixes: d80efd5cb3de ("drm/vmwgfx: Initial DX support")
Signed-off-by: Murray McAllister <murray.mcallister@gmail.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
5 years agodrm/vmwgfx: Use the dma scatter-gather iterator to get dma addresses
Thomas Hellstrom [Wed, 15 May 2019 15:45:23 +0000 (17:45 +0200)]
drm/vmwgfx: Use the dma scatter-gather iterator to get dma addresses

Use struct sg_dma_page_iter in favour struct of sg_page_iter, which fairly
recently was declared useless for obtaining dma addresses.

With a struct sg_dma_page_iter we can't call sg_page_iter_page() so
when the page is needed, use the same page lookup mechanism as for the
non-sg dma modes instead of calling sg_dma_page_iter.

Note, the fixes tag doesn't really point to a commit introducing a
failure / regression, but rather to a commit that implemented a simple
workaround for this problem.

Cc: Jason Gunthorpe <jgg@mellanox.com>
Fixes: d901b2760dc6 ("lib/scatterlist: Provide a DMA page iterator")
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
5 years agodrm/vmwgfx: Fix compat mode shader operation
Thomas Hellstrom [Thu, 4 Apr 2019 13:25:43 +0000 (13:25 +0000)]
drm/vmwgfx: Fix compat mode shader operation

In compat mode, we allowed host-backed user-space with guest-backed
kernel / device. In this mode, set shader commands was broken since
no relocations were emitted. Fix this.

Cc: <stable@vger.kernel.org>
Fixes: e8c66efbfe3a ("drm/vmwgfx: Make user resource lookups reference-free during validation")
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
5 years agodrm/vmwgfx: Fix user space handle equal to zero
Thomas Hellstrom [Tue, 7 May 2019 09:10:10 +0000 (11:10 +0200)]
drm/vmwgfx: Fix user space handle equal to zero

User-space handles equal to zero are interpreted as uninitialized or
illegal by some drm systems (most notably kms). This means that a
dumb buffer or surface with a zero user-space handle can never be
used as a kms frame-buffer.

Cc: <stable@vger.kernel.org>
Fixes: c7eae62666ad ("drm/vmwgfx: Make the object handles idr-generated")
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Deepak Rawat <drawat@vmware.com>
5 years agodrm/vmwgfx: Don't send drm sysfs hotplug events on initial master set
Thomas Hellstrom [Tue, 7 May 2019 09:07:53 +0000 (11:07 +0200)]
drm/vmwgfx: Don't send drm sysfs hotplug events on initial master set

This may confuse user-space clients like plymouth that opens a drm
file descriptor as a result of a hotplug event and then generates a
new event...

Cc: <stable@vger.kernel.org>
Fixes: 5ea1734827bb ("drm/vmwgfx: Send a hotplug event at master_set")
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Deepak Rawat <drawat@vmware.com>
5 years agodrm/i915/gvt: Fix an error code in ppgtt_populate_spt_by_guest_entry()
Dan Carpenter [Mon, 13 May 2019 09:22:44 +0000 (12:22 +0300)]
drm/i915/gvt: Fix an error code in ppgtt_populate_spt_by_guest_entry()

"ret" is uninitialized on this path but it should be -EINVAL.

Fixes: 930c8dfea4b8 ("drm/i915/gvt: Check if get_next_pt_type() always returns a valid value")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
5 years agodrm/i915/gvt: do not let TRTTE and 0x4dfc write passthrough to hardware
Yan Zhao [Wed, 8 May 2019 02:16:44 +0000 (22:16 -0400)]
drm/i915/gvt: do not let TRTTE and 0x4dfc write passthrough to hardware

the vGPU write on TRTTE and 0x4dfc is now write to vreg first. their
values all be restored hardware when context switching.

Fixes: e39c5add3221 ("drm/i915/gvt: vGPU MMIO virtualization")
Acked-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Yan Zhao <yan.y.zhao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
5 years agodrm/i915/gvt: add 0x4dfc to gen9 save-restore list
Yan Zhao [Wed, 8 May 2019 02:16:33 +0000 (22:16 -0400)]
drm/i915/gvt: add 0x4dfc to gen9 save-restore list

0x4dfc is in-context mmio for gen9+, but each vm have different settings
need to add it to save-restore list along with other trtt registers

Fixes: 178657139307 ("drm/i915/gvt: vGPU context switch")
Acked-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Yan Zhao <yan.y.zhao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
5 years agodrm/i915/gvt: Tiled Resources mmios are in-context mmios for gen9+
Yan Zhao [Wed, 8 May 2019 02:15:00 +0000 (22:15 -0400)]
drm/i915/gvt: Tiled Resources mmios are in-context mmios for gen9+

TRVATTL3PTRDW(0x4de0-0x4de4), TRNULLDETCT(0x4de8), TRINVTILEDETCT(0x4dec),
TRTTE(0x4df0), TRVADR(0x4df4) are in-context mmios for gen9+

Fixes: 178657139307 ("drm/i915/gvt: vGPU context switch")
Acked-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Yan Zhao <yan.y.zhao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
5 years agodrm/i915/gvt: use cmd to restore in-context mmios to hw for gen9 platform
Yan Zhao [Wed, 8 May 2019 02:14:04 +0000 (22:14 -0400)]
drm/i915/gvt: use cmd to restore in-context mmios to hw for gen9 platform

for restore-inhibit context, hardware will not load in-context mmios
(engine context part) to hardware, but hardware will save the mmio
values in hardware back to context image. So, in order to save correct
values of vGPU back to context image, values of vGPU mmios have to be
loaded into hardware first for restore-inhibit context.

In this patch, the mechanism is applied to all gen9 platform.

The reason excluding gen8 platforms is only because of lacking of testing
on those platforms.

v3: for mocs registers, goto in-context mmios save-restore path for skl
platform as well (weinan li)
v2: update vreg when scanning indirect context for inhibit context for
gen9

Cc: Weinan Li <weinan.z.li@intel.com>
Acked-by: Weinan Li <weinan.z.li@intel.com>
Signed-off-by: Yan Zhao <yan.y.zhao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
5 years agodrm/i915/gvt: emit init breadcrumb for gvt request
Weinan [Fri, 10 May 2019 07:57:20 +0000 (15:57 +0800)]
drm/i915/gvt: emit init breadcrumb for gvt request

"To track whether a request has started on HW, we can emit a breadcrumb at
the beginning of the request and check its timeline's HWSP to see if the
breadcrumb has advanced past the start of this request." It means all the
request which timeline's has_init_breadcrumb is true, then the
emit_init_breadcrumb process must have before emitting the real commands,
otherwise, the scheduler might get a wrong state of this request during
reset. If the request is exactly the guilty one, the scheduler won't
terminate it with the wrong state. To avoid this, do emit_init_breadcrumb
for all the requests from gvt.

v2: cc to stable kernel

Fixes: 8547444137ec ("drm/i915: Identify active requests")
Cc: stable@vger.kernel.org
Acked-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Weinan <weinan.z.li@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
5 years agodrm/amdkfd: Fix compute profile switching
Harish Kasiviswanathan [Fri, 25 Jan 2019 21:35:35 +0000 (16:35 -0500)]
drm/amdkfd: Fix compute profile switching

Fix compute profile switching on process termination.

Add a dedicated reference counter to keep track of entry/exit to/from
compute profile. This enables switching compute profiles for other
reasons than process creation or termination.

Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Eric Huang <JinhuiEric.Huang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
5 years agodrm/amdgpu: skip fw pri bo alloc for SRIOV
Yintian Tao [Thu, 16 May 2019 05:07:26 +0000 (13:07 +0800)]
drm/amdgpu: skip fw pri bo alloc for SRIOV

PSP fw primary buffer is not used under SRIOV.
Under SRIOV, VBIOS or hypervisor driver will load psp
sos and psp sysdrv. Therefore, we don't need to
allocate memory for it.

v2: remove superfluous check for amdgpu_bo_free_kernel().

Signed-off-by: Yintian Tao <yttao@amd.com>
Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
5 years agodrm/amd/powerplay: fix locking in smu_feature_set_supported()
Dan Carpenter [Wed, 15 May 2019 09:51:30 +0000 (12:51 +0300)]
drm/amd/powerplay: fix locking in smu_feature_set_supported()

There is a typo so the code unlocks twice instead of taking the lock and
then releasing it.

Fixes: f14a323db5b0 ("drm/amd/powerplay: implement update enabled feature state to smc for smu11")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
5 years agodrm/amdgpu/gmc9: set vram_width properly for SR-IOV
Alex Deucher [Fri, 17 May 2019 14:31:43 +0000 (09:31 -0500)]
drm/amdgpu/gmc9: set vram_width properly for SR-IOV

For SR-IOV, vram_width can't be read from ATOM as
RAVEN, and DF related registers is not readable, so hardcord
is the only way to set the correct vram_width.

Reviewed-by: Yintian Tao <yttao@amd.com>
Signed-off-by: Trigger Huang <Trigger.Huang@amd.com>
Signed-off-by: Yintian Tao <yttao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
5 years agodrm/amdgpu/soc15: skip reset on init
Alex Deucher [Fri, 17 May 2019 14:21:13 +0000 (09:21 -0500)]
drm/amdgpu/soc15: skip reset on init

Not necessary on soc15 and breaks driver reload on server cards.

Acked-by: Amber Lin <Amber.Lin@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
5 years agodrm/amdgpu: keep stolen memory on picasso
Flora Cui [Fri, 17 May 2019 03:33:56 +0000 (11:33 +0800)]
drm/amdgpu: keep stolen memory on picasso

otherwise screen corrupts during modprobe.

Signed-off-by: Flora Cui <flora.cui@amd.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
5 years agodrm/amdkfd: Add missing Polaris10 ID
Kent Russell [Mon, 13 May 2019 13:00:05 +0000 (09:00 -0400)]
drm/amdkfd: Add missing Polaris10 ID

This was added to amdgpu but was missed in amdkfd

Signed-off-by: Kent Russell <kent.russell@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.rg
5 years agodrm/amd/display: Don't load DMCU for Raven 1
Harry Wentland [Mon, 29 Apr 2019 13:39:15 +0000 (09:39 -0400)]
drm/amd/display: Don't load DMCU for Raven 1

[WHY]
Some early Raven boards had a bad SBIOS that doesn't play nicely with
the DMCU FW. We thought the issues were fixed by ignoring errors on DMCU
load but that doesn't seem to be the case. We've still seen reports of
users unable to boot their systems at all.

[HOW]
Disable DMCU load on Raven 1. Only load it for Raven 2 and Picasso.

Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
5 years agodrm/amd/display: Add ASICREV_IS_PICASSO
Harry Wentland [Tue, 14 May 2019 13:05:37 +0000 (09:05 -0400)]
drm/amd/display: Add ASICREV_IS_PICASSO

[WHY]
We only want to load DMCU FW on Picasso and Raven 2, not on Raven 1.

Signed-off-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
5 years agodrm/i915: Truly bump ready tasks ahead of busywaits
Chris Wilson [Wed, 15 May 2019 13:00:49 +0000 (14:00 +0100)]
drm/i915: Truly bump ready tasks ahead of busywaits

In commit b7404c7ecb38 ("drm/i915: Bump ready tasks ahead of
busywaits"), I tried cutting a corner in order to not install a signal
for each of our dependencies, and only listened to requests on which we
were intending to busywait. The compromise that was made was that
instead of then being able to promote the request with a full
NOSEMAPHORE like its non-busywaiting brethren, as we had not ensured we
had cleared the semaphore chain, we settled for only using the NEWCLIENT
boost. With an over saturated system with multiple NEWCLIENTS in flight
at any time, this was found to be an inadequate promotion and left us
with a much poorer scheduling order than prior to using semaphores.

The outcome of this patch, is that all requests have NOSEMAPHORE
priority when they have no dependencies and are ready to run and not
busywait, restoring the pre-semaphore ordering on saturated systems.

We can demonstrate the effect of poor scheduling order by oversaturating
the system using gem_wsim on a system with multiple vcs engines
(i.e running the same workloads across more clients than required for
peak throughput, e.g. media_load_balance_17i7.wsim -c4 -b context):

x v5.1 (normalized)
+ tip
* fix
+------------------------------------------------------------------------+
|                                                                    x   |
|                                                                    x   |
|                                                                    x   |
|                                                                    x   |
|                                                                   %x   |
|                                                                  %%x   |
|                                                                  %%x   |
|                                                                  %%x   |
|                                                                  %%x   |
|                                                                  %%x   |
|                                                                  %%x   |
|                                                                  %%x   |
|                                                                  %%x   |
|                                                                  %%x   |
|                                                                  %%x   |
|                                                                  %#x   |
|                                                                  %#x   |
|                                                                  %#x   |
|                                                                  %#x   |
|                                                                  %#x   |
|         +                                                        %#xx  |
|         +                                                        %#xx  |
|         +                                                       %%#xx  |
|         +                                                       %%#xx  |
|         +                                                       %%#xx  |
|         +                                                       %%#xx  |
|         +                                                       %%##x  |
|         +++                                                     %%##x  |
|         +++                                                     %%##x  |
|         +++                                                     %%##x  |
|        ++++                                                     %%##x  |
|        ++++                                                     %%##x  |
|        ++++                                                     %%##xx |
|        ++++                                                     %###xx |
|        ++++                                                     %###xx |
|        ++++                                                     %###xx |
|        ++++                                                     %###xx |
|        ++++ +                                                   %#O#xx |
|        ++++ +                                                   %#O#xx |
|        ++++++ +                                                 %#O#xx |
|       ++++++++++                                                %OOOxxx|
|       ++++++++++       +                                       %#OOO#xx|
|     + ++++++++++++ ++ +++++    +                        ++    @@OOOO#xx|
|                                                                   |A_| |
||__________M_______A____________________|                               |
|                                                                 |A_|   |
+------------------------------------------------------------------------+
    N           Min           Max        Median           Avg        Stddev
x 120       0.99456       1.00628      0.999985     1.0001545  0.0024387139
+ 120      0.873021       1.00037      0.884134    0.90148752   0.039190862
Difference at 99.5% confidence
-0.098667 +/- 0.0110762
-9.86517% +/- 1.10745%
(Student's t, pooled s = 0.0277657)
% 120      0.990207       1.00165     0.9970265    0.99699748     0.0021024
Difference at 99.5% confidence
-0.003157 +/- 0.000908245
-0.315651% +/- 0.0908105%
(Student's t, pooled s = 0.00227678)

Fixes: b7404c7ecb38 ("drm/i915: Bump ready tasks ahead of busywaits")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Dmitry Rogozhkin <dmitry.v.rogozhkin@intel.com>
Cc: Dmitry Ermilov <dmitry.ermilov@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190515130052.4475-2-chris@chris-wilson.co.uk
(cherry picked from commit 17db337f5098d29415314c4a588b842fc684394b)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
5 years agodrm/i915: Downgrade NEWCLIENT to non-preemptive
Chris Wilson [Wed, 15 May 2019 13:00:51 +0000 (14:00 +0100)]
drm/i915: Downgrade NEWCLIENT to non-preemptive

Commit 1413b2bc0717 ("drm/i915: Trim NEWCLIENT boosting") had the
intended consequence of not allowing a sequence of work that merely
crossed into a new engine the privilege to be promoted to NEWCLIENT
status. It also had the unintended consequence of actually making
NEWCLIENT effective on heavily oversubscribed transcode machines and
impacting upon their throughput.

If we consider a client packet composed of (rcsA, rcsB, vcs) and 30 of
those clients, using the NEWCLIENT boost that will be scheduled as

rcsA x 30, (rcsB, vcs) x 30

where as before it would have been

(rcsA, rcsB, vcs) x 30

That is with NEWCLIENT only boosting the first request of each client,
we would execute all rcsA requests prior to running on the vcs engines;
acruing a lot of dead time as compared to the previous case where the
vcs engine would be started in parallel to processing the second client.

The previous patch has the effect of delaying submission until it is
required by a third party (either the user with an explicit wait, or by
another client/engine). We reduce the NEWCLIENT bump to a mere WAIT,
which has the effect of removing its preemptive grant and reducing it to
the same level as any other user interaction -- that it will not be
promoted above the interengine dependencies, and so preventing NEWCLIENTS
from starving other engines. This a large nerf to the rrul properties of
the current NEWCLIENT, but it still does give prioritised submission to
new requests from light workloads.

References: b16c765122f9 ("drm/i915: Priority boost for new clients")
Fixes: 1413b2bc0717 ("drm/i915: Trim NEWCLIENT boosting") # customer impact
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Dmitry Rogozhkin <dmitry.v.rogozhkin@intel.com>
Cc: Dmitry Ermilov <dmitry.ermilov@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190515130052.4475-4-chris@chris-wilson.co.uk
(cherry picked from commit 68fc728b01fcc93b26d52f6e884e738962a49a66)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
5 years agodrm/i915: Bump signaler priority on adding a waiter
Chris Wilson [Wed, 15 May 2019 13:00:50 +0000 (14:00 +0100)]
drm/i915: Bump signaler priority on adding a waiter

The handling of the no-preemption priority level imposes the restriction
that we need to maintain the implied ordering even though preemption is
disabled. Otherwise we may end up with an AB-BA deadlock across multiple
engine due to a real preemption event reordering the no-preemption
WAITs. To resolve this issue we currently promote all requests to WAIT
on unsubmission, however this interferes with the timeslicing
requirement that we do not apply any implicit promotion that will defeat
the round-robin timeslice list. (If we automatically promote the active
request it will go back to the head of the queue and not the tail!)

So we need implicit promotion to prevent reordering around semaphores
where we are not allowed to preempt, and we must avoid implicit
promotion on unsubmission. So instead of at unsubmit, if we apply that
implicit promotion on adding the dependency, we avoid the semaphore
deadlock and we also reduce the gains made by the promotion for user
space waiting. Furthermore, by keeping the earlier dependencies at a
higher level, we reduce the search space for timeslicing without
altering runtime scheduling too badly (no dependencies at all will be
assigned a higher priority for rrul).

v2: Limit the bump to external edges (as originally intended) i.e.
between contexts and out to the user.

Testcase: igt/gem_concurrent_blit
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190515130052.4475-3-chris@chris-wilson.co.uk
(cherry picked from commit 6e7eb7a80769e7250e31652b96918cf7f3e0d285)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
5 years agodrm/i915: Pass i915_sched_node around internally
Chris Wilson [Mon, 13 May 2019 12:01:01 +0000 (13:01 +0100)]
drm/i915: Pass i915_sched_node around internally

To simplify the next patch, update bump_priority and schedule to accept
the internal i915_sched_ndoe directly and not expect a request pointer.

add/remove: 0/0 grow/shrink: 2/1 up/down: 8/-15 (-7)
Function                                     old     new   delta
i915_schedule_bump_priority                  109     113      +4
i915_schedule                                 50      54      +4
__i915_schedule                              922     907     -15

v2: Adopt node for the old rq local, since it no longer is a request but
the origin node.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190513120102.29660-2-chris@chris-wilson.co.uk
(cherry picked from commit 52c76fb18a34fc08dd06f32b9fc83f1375f083ee)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
5 years agodrm/i915: Rearrange i915_scheduler.c
Chris Wilson [Mon, 13 May 2019 12:01:00 +0000 (13:01 +0100)]
drm/i915: Rearrange i915_scheduler.c

To avoid pulling in a forward declaration in the next patch, move the
i915_sched_node handling to after the main dfs of the scheduler.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190513120102.29660-1-chris@chris-wilson.co.uk
(cherry picked from commit 5ae87063c162679a61f2141041d0918cc3045daf)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
5 years agoMerge drm-misc-next-fixes-2019-05-20 into drm-misc-fixes
Sean Paul [Mon, 20 May 2019 15:02:53 +0000 (11:02 -0400)]
Merge drm-misc-next-fixes-2019-05-20 into drm-misc-fixes

Picking up 3 sun4i patches that missed the last drm-misc-next-fixes pull
request for 5.2

Signed-off-by: Sean Paul <seanpaul@chromium.org>
5 years agoLinux 5.2-rc1
Linus Torvalds [Sun, 19 May 2019 22:47:09 +0000 (15:47 -0700)]
Linux 5.2-rc1

5 years agoMerge tag 'upstream-5.2-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rw...
Linus Torvalds [Sun, 19 May 2019 22:22:03 +0000 (15:22 -0700)]
Merge tag 'upstream-5.2-rc2' of git://git./linux/kernel/git/rw/ubifs

Pull UBIFS fixes from Richard Weinberger:

 - build errors wrt xattrs

 - mismerge which lead to a wrong Kconfig ifdef

 - missing endianness conversion

* tag 'upstream-5.2-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rw/ubifs:
  ubifs: Convert xattr inum to host order
  ubifs: Use correct config name for encryption
  ubifs: Fix build error without CONFIG_UBIFS_FS_XATTR

5 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Sun, 19 May 2019 19:15:32 +0000 (12:15 -0700)]
Merge branch 'akpm' (patches from Andrew)

Merge yet more updates from Andrew Morton:
 "A few final bits:

   - large changes to vmalloc, yielding large performance benefits

   - tweak the console-flush-on-panic code

   - a few fixes"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  panic: add an option to replay all the printk message in buffer
  initramfs: don't free a non-existent initrd
  fs/writeback.c: use rcu_barrier() to wait for inflight wb switches going into workqueue when umount
  mm/compaction.c: correct zone boundary handling when isolating pages from a pageblock
  mm/vmap: add DEBUG_AUGMENT_LOWEST_MATCH_CHECK macro
  mm/vmap: add DEBUG_AUGMENT_PROPAGATE_CHECK macro
  mm/vmalloc.c: keep track of free blocks for vmap allocation

5 years agoMerge tag 'kbuild-v5.2-2' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy...
Linus Torvalds [Sun, 19 May 2019 18:53:58 +0000 (11:53 -0700)]
Merge tag 'kbuild-v5.2-2' of git://git./linux/kernel/git/masahiroy/linux-kbuild

Pull more Kbuild updates from Masahiro Yamada:

 - remove unneeded use of cc-option, cc-disable-warning, cc-ldoption

 - exclude tracked files from .gitignore

 - re-enable -Wint-in-bool-context warning

 - refactor samples/Makefile

 - stop building immediately if syncconfig fails

 - do not sprinkle error messages when $(CC) does not exist

 - move arch/alpha/defconfig to the configs subdirectory

 - remove crappy header search path manipulation

 - add comment lines to .config to clarify the end of menu blocks

 - check uniqueness of module names (adding new warnings intentionally)

* tag 'kbuild-v5.2-2' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild: (24 commits)
  kconfig: use 'else ifneq' for Makefile to improve readability
  kbuild: check uniqueness of module names
  kconfig: Terminate menu blocks with a comment in the generated config
  kbuild: add LICENSES to KBUILD_ALLDIRS
  kbuild: remove 'addtree' and 'flags' magic for header search paths
  treewide: prefix header search paths with $(srctree)/
  media: prefix header search paths with $(srctree)/
  media: remove unneeded header search paths
  alpha: move arch/alpha/defconfig to arch/alpha/configs/defconfig
  kbuild: terminate Kconfig when $(CC) or $(LD) is missing
  kbuild: turn auto.conf.cmd into a mandatory include file
  .gitignore: exclude .get_maintainer.ignore and .gitattributes
  kbuild: add all Clang-specific flags unconditionally
  kbuild: Don't try to add '-fcatch-undefined-behavior' flag
  kbuild: add some extra warning flags unconditionally
  kbuild: add -Wvla flag unconditionally
  arch: remove dangling asm-generic wrappers
  samples: guard sub-directories with CONFIG options
  kbuild: re-enable int-in-bool-context warning
  MAINTAINERS: kbuild: Add pattern for scripts/*vmlinux*
  ...

5 years agoMerge branch 'i2c/for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Sun, 19 May 2019 18:47:03 +0000 (11:47 -0700)]
Merge branch 'i2c/for-next' of git://git./linux/kernel/git/wsa/linux

Pull i2c updates from Wolfram Sang:
 "Some I2C core API additions which are kind of simple but enhance error
  checking for users a lot, especially by returning errno now.

  There are wrappers to still support the old API but it will be removed
  once all users are converted"

* 'i2c/for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: core: add device-managed version of i2c_new_dummy
  i2c: core: improve return value handling of i2c_new_device and i2c_new_dummy

5 years agoMerge tag 'ext4_for_linus_stable' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 19 May 2019 18:43:16 +0000 (11:43 -0700)]
Merge tag 'ext4_for_linus_stable' of git://git./linux/kernel/git/tytso/ext4

Pull ext4 fixes from Ted Ts'o:
 "Some bug fixes, and an update to the URL's for the final version of
  Unicode 12.1.0"

* tag 'ext4_for_linus_stable' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  ext4: avoid panic during forced reboot due to aborted journal
  ext4: fix block validity checks for journal inodes using indirect blocks
  unicode: update to Unicode 12.1.0 final
  unicode: add missing check for an error return from utf8lookup()
  ext4: fix miscellaneous sparse warnings
  ext4: unsigned int compared against zero
  ext4: fix use-after-free in dx_release()
  ext4: fix data corruption caused by overlapping unaligned and aligned IO
  jbd2: fix potential double free
  ext4: zero out the unused memory region in the extent tree block

5 years agoMerge tag '5.2-rc-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Sun, 19 May 2019 18:38:18 +0000 (11:38 -0700)]
Merge tag '5.2-rc-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6

Pull cifs fixes from Steve French:
 "Minor cleanup and fixes, one for stable, four rdma (smbdirect)
  related. Also adds SEEK_HOLE support"

* tag '5.2-rc-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: add support for SEEK_DATA and SEEK_HOLE
  Fixed https://bugzilla.kernel.org/show_bug.cgi?id=202935 allow write on the same file
  cifs: Allocate memory for all iovs in smb2_ioctl
  cifs: Don't match port on SMBDirect transport
  cifs:smbd Use the correct DMA direction when sending data
  cifs:smbd When reconnecting to server, call smbd_destroy() after all MIDs have been called
  cifs: use the right include for signal_pending()
  smb3: trivial cleanup to smb2ops.c
  cifs: cleanup smb2ops.c and normalize strings
  smb3: display session id in debug data

5 years agoMerge branch 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 19 May 2019 18:20:22 +0000 (11:20 -0700)]
Merge branch 'perf-core-for-linus' of git://git./linux/kernel/git/tip/tip

Pull perf tooling updates from Ingo Molnar:
 "perf.data:

   - Streaming compression of perf ring buffer into
     PERF_RECORD_COMPRESSED user space records, resulting in ~3-5x
     perf.data file size reduction on variety of tested workloads what
     saves storage space on larger server systems where perf.data size
     can easily reach several tens or even hundreds of GiBs, especially
     when profiling with DWARF-based stacks and tracing of context
     switches.

  perf record:

   - Improve -user-regs/intr-regs suggestions to overcome errors

  perf annotate:

   - Remove hist__account_cycles() from callback, speeding up branch
     processing (perf record -b)

  perf stat:

   - Add a 'percore' event qualifier, e.g.: -e
     cpu/event=0,umask=0x3,percore=1/, that sums up the event counts for
     both hardware threads in a core.

     We can already do this with --per-core, but it's often useful to do
     this together with other metrics that are collected per hardware
     thread.

     I.e. now its possible to do this per-event, and have it mixed with
     other events not aggregated by core.

  arm64:

   - Map Brahma-B53 CPUID to cortex-a53 events.

   - Add Cortex-A57 and Cortex-A72 events.

  csky:

   - Add DWARF register mappings for libdw, allowing --call-graph=dwarf
     to work on the C-SKY arch.

  x86:

   - Add support for recording and printing XMM registers, available,
     for instance, on Icelake.

   - Add uncore_upi (Intel's "Ultra Path Interconnect" events) JSON
     support. UPI replaced the Intel QuickPath Interconnect (QPI) in
     Xeon Skylake-SP.

  Intel PT:

   - Fix instructions sampling rate.

   - Timestamp fixes.

   - Improve exported-sql-viewer GUI, allowing, for instance, to
     copy'n'paste the trees, useful for e-mailing"

* 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (73 commits)
  perf stat: Support 'percore' event qualifier
  perf stat: Factor out aggregate counts printing
  perf tools: Add a 'percore' event qualifier
  perf docs: Add description for stderr
  perf intel-pt: Fix sample timestamp wrt non-taken branches
  perf intel-pt: Fix improved sample timestamp
  perf intel-pt: Fix instructions sampling rate
  perf regs x86: Add X86 specific arch__intr_reg_mask()
  perf parse-regs: Add generic support for arch__intr/user_reg_mask()
  perf parse-regs: Split parse_regs
  perf vendor events arm64: Add Cortex-A57 and Cortex-A72 events
  perf vendor events arm64: Map Brahma-B53 CPUID to cortex-a53 events
  perf vendor events arm64: Remove [[:xdigit:]] wildcard
  perf jevents: Remove unused variable
  perf test zstd: Fixup verbose mode output
  perf tests: Implement Zstd comp/decomp integration test
  perf inject: Enable COMPRESSED record decompression
  perf report: Implement perf.data record decompression
  perf record: Implement -z,--compression_level[=<n>] option
  perf report: Add stub processing of compressed events for -D
  ...

5 years agoMerge branch 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 19 May 2019 18:11:20 +0000 (11:11 -0700)]
Merge branch 'timers-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull clocksource updates from Ingo Molnar:
 "Misc clocksource/clockevent driver updates that came in a bit late but
  are ready for v5.2"

* 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  misc: atmel_tclib: Do not probe already used TCBs
  clocksource/drivers/timer-atmel-tcb: Convert tc_clksrc_suspend|resume() to static
  clocksource/drivers/tcb_clksrc: Rename the file for consistency
  clocksource/drivers/timer-atmel-pit: Rework Kconfig option
  clocksource/drivers/tcb_clksrc: Move Kconfig option
  ARM: at91: Implement clocksource selection
  clocksource/drivers/tcb_clksrc: Use tcb as sched_clock
  clocksource/drivers/tcb_clksrc: Stop depending on atmel_tclib
  ARM: at91: move SoC specific definitions to SoC folder
  clocksource/drivers/timer-milbeaut: Cleanup common register accesses
  clocksource/drivers/timer-milbeaut: Add shutdown function
  clocksource/drivers/timer-milbeaut: Fix to enable one-shot timer
  clocksource/drivers/tegra: Rework for compensation of suspend time
  clocksource/drivers/sp804: Add COMPILE_TEST to CONFIG_ARM_TIMER_SP804
  clocksource/drivers/sun4i: Add a compatible for suniv
  dt-bindings: timer: Add Allwinner suniv timer

5 years agoMerge branch 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 19 May 2019 17:58:45 +0000 (10:58 -0700)]
Merge branch 'irq-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull IRQ chip updates from Ingo Molnar:
 "A late irqchips update:

   - New TI INTR/INTA set of drivers

   - Rewrite of the stm32mp1-exti driver as a platform driver

   - Update the IOMMU MSI mapping API to be RT friendly

   - A number of cleanups and other low impact fixes"

* 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (34 commits)
  iommu/dma-iommu: Remove iommu_dma_map_msi_msg()
  irqchip/gic-v3-mbi: Don't map the MSI page in mbi_compose_m{b, s}i_msg()
  irqchip/ls-scfg-msi: Don't map the MSI page in ls_scfg_msi_compose_msg()
  irqchip/gic-v3-its: Don't map the MSI page in its_irq_compose_msi_msg()
  irqchip/gicv2m: Don't map the MSI page in gicv2m_compose_msi_msg()
  iommu/dma-iommu: Split iommu_dma_map_msi_msg() in two parts
  genirq/msi: Add a new field in msi_desc to store an IOMMU cookie
  arm64: arch_k3: Enable interrupt controller drivers
  irqchip/ti-sci-inta: Add msi domain support
  soc: ti: Add MSI domain bus support for Interrupt Aggregator
  irqchip/ti-sci-inta: Add support for Interrupt Aggregator driver
  dt-bindings: irqchip: Introduce TISCI Interrupt Aggregator bindings
  irqchip/ti-sci-intr: Add support for Interrupt Router driver
  dt-bindings: irqchip: Introduce TISCI Interrupt router bindings
  gpio: thunderx: Use the default parent apis for {request,release}_resources
  genirq: Introduce irq_chip_{request,release}_resource_parent() apis
  firmware: ti_sci: Add helper apis to manage resources
  firmware: ti_sci: Add RM mapping table for am654
  firmware: ti_sci: Add support for IRQ management
  firmware: ti_sci: Add support for RM core ops
  ...

5 years agoMerge branch 'efi-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 19 May 2019 17:33:26 +0000 (10:33 -0700)]
Merge branch 'efi-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull EFI fix from Ingo Molnar:
 "Fix an EFI-fb regression that affects certain x86 systems"

* 'efi-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  fbdev/efifb: Ignore framebuffer memmap entries that lack any memory types

5 years agoMerge branch 'core-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 19 May 2019 17:23:24 +0000 (10:23 -0700)]
Merge branch 'core-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull core fixes from Ingo Molnar:
 "This fixes a particularly thorny munmap() bug with MPX, plus fixes a
  host build environment assumption in objtool"

* 'core-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  objtool: Allow AR to be overridden with HOSTAR
  x86/mpx, mm/core: Fix recursive munmap() corruption

5 years agoMerge tag 'armsoc-late' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Linus Torvalds [Sun, 19 May 2019 17:16:39 +0000 (10:16 -0700)]
Merge tag 'armsoc-late' of git://git./linux/kernel/git/soc/soc

Pull ARM SoC late updates from Olof Johansson:
 "This is some material that we picked up into our tree late. Most of it
  are smaller fixes and additions, some defconfig updates due to recent
  development, etc.

  Code-wise the largest portion is a series of PM updates for the at91
  platform, and those have been in linux-next a while through the at91
  tree before we picked them up"

* tag 'armsoc-late' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc: (29 commits)
  arm64: dts: sprd: Add clock properties for serial devices
  Opt out of scripts/get_maintainer.pl
  ARM: ixp4xx: Remove duplicated include from common.c
  soc: ixp4xx: qmgr: Fix an NULL vs IS_ERR() check in probe
  arm64: tegra: Disable XUSB support on Jetson TX2
  arm64: tegra: Enable SMMU translation for PCI on Tegra186
  arm64: tegra: Fix insecure SMMU users for Tegra186
  arm64: tegra: Select ARM_GIC_PM
  amba: tegra-ahb: Mark PM functions as __maybe_unused
  ARM: dts: logicpd-som-lv: Fix MMC1 card detect
  ARM: mvebu: drop return from void function
  ARM: mvebu: prefix coprocessor operand with p
  ARM: mvebu: drop unnecessary label
  ARM: mvebu: fix a leaked reference by adding missing of_node_put
  ARM: socfpga_defconfig: enable LTC2497
  ARM: mvebu: kirkwood: remove error message when retrieving mac address
  ARM: at91: sama5: make ov2640 as a module
  ARM: OMAP1: ams-delta: fix early boot crash when LED support is disabled
  ARM: at91: remove HAVE_FB_ATMEL for sama5 SoC as they use DRM
  soc/fsl/qe: Fix an error code in qe_pin_request()
  ...

5 years agoMerge tag 'powerpc-5.2-2' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc...
Linus Torvalds [Sun, 19 May 2019 17:10:15 +0000 (10:10 -0700)]
Merge tag 'powerpc-5.2-2' of git://git./linux/kernel/git/powerpc/linux

Pull powerpc fixes from Michael Ellerman:
 "One fix going back to stable, for a bug on 32-bit introduced when we
  added support for THREAD_INFO_IN_TASK.

  A fix for a typo in a recent rework of our hugetlb code that leads to
  crashes on 64-bit when using hugetlbfs with a 4K PAGE_SIZE.

  Two fixes for our recent rework of the address layout on 64-bit hash
  CPUs, both only triggered when userspace tries to access addresses
  outside the user or kernel address ranges.

  Finally a fix for a recently introduced double free in an error path
  in our cacheinfo code.

  Thanks to: Aneesh Kumar K.V, Christophe Leroy, Sachin Sant, Tobin C.
  Harding"

* tag 'powerpc-5.2-2' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/cacheinfo: Remove double free
  powerpc/mm/hash: Fix get_region_id() for invalid addresses
  powerpc/mm: Drop VM_BUG_ON in get_region_id()
  powerpc/mm: Fix crashes with hugepages & 4K pages
  powerpc/32s: fix flush_hash_pages() on SMP

5 years agoMerge tag 'mips_5.2_2' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux
Linus Torvalds [Sun, 19 May 2019 17:05:28 +0000 (10:05 -0700)]
Merge tag 'mips_5.2_2' of git://git./linux/kernel/git/mips/linux

Pull a few more MIPS updates from Paul Burton:
 "Some SGI IP27 specific PCI rework and a batch of fixes:

   - A build fix for BMIPS5000 configurations with
     CONFIG_HW_PERF_EVENTS=y, which also neatly removes some #ifdefery.

   - A fix to report supported ISAs correctly on older Ingenic SoCs
     which incorrectly indicate MIPSr2 support in their cop0 Config
     register.

   - Some PCI modernization for SGI IP27 systems as part of ongoing work
     to support some other SGI systems.

   - A fix allowing use of appended DTB files with generic kernels.

   - DMA mask fixes for SGI IP22 & Alchemy systems"

* tag 'mips_5.2_2' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux:
  MIPS: Alchemy: add DMA masks for on-chip ethernet
  MIPS: SGI-IP22: provide missing dma_mask/coherent_dma_mask
  generic: fix appended dtb support
  MIPS: SGI-IP27: abstract chipset irq from bridge
  MIPS: SGI-IP27: use generic PCI driver
  MIPS: Fix Ingenic SoCs sometimes reporting wrong ISA
  MIPS: perf: Fix build with CONFIG_CPU_BMIPS5000 enabled

5 years agoMerge tag 'riscv-for-linus-5.2-mw2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 19 May 2019 16:56:36 +0000 (09:56 -0700)]
Merge tag 'riscv-for-linus-5.2-mw2' of git://git./linux/kernel/git/palmer/riscv-linux

Pull RISC-V updates from Palmer Dabbelt:
 "This contains an assortment of RISC-V related patches that I'd like to
  target for the 5.2 merge window. Most of the patches are cleanups, but
  there are a handful of user-visible changes:

   - The nosmp and nr_cpus command-line arguments are now supported,
     which work like normal.

   - The SBI console no longer installs itself as a preferred console,
     we rely on standard mechanisms (/chosen, command-line, hueristics)
     instead.

   - sfence_remove_sfence_vma{,_asid} now pass their arguments along to
     the SBI call.

   - Modules now support BUG().

   - A missing sfence.vma during boot has been added. This bug only
     manifests during boot.

   - The arch/riscv support for SiFive's L2 cache controller has been
     merged, which should un-block the EDAC framework work.

  I've only tested this on QEMU again, as I didn't have time to get
  things running on the Unleashed. The latest master from this morning
  merges in cleanly and passes the tests as well"

* tag 'riscv-for-linus-5.2-mw2' of git://git.kernel.org/pub/scm/linux/kernel/git/palmer/riscv-linux: (31 commits)
  riscv: fix locking violation in page fault handler
  RISC-V: sifive_l2_cache: Add L2 cache controller driver for SiFive SoCs
  RISC-V: Add DT documentation for SiFive L2 Cache Controller
  RISC-V: Avoid using invalid intermediate translations
  riscv: Support BUG() in kernel module
  riscv: Add the support for c.ebreak check in is_valid_bugaddr()
  riscv: support trap-based WARN()
  riscv: fix sbi_remote_sfence_vma{,_asid}.
  riscv: move switch_mm to its own file
  riscv: move flush_icache_{all,mm} to cacheflush.c
  tty: Don't force RISCV SBI console as preferred console
  RISC-V: Access CSRs using CSR numbers
  RISC-V: Add interrupt related SCAUSE defines in asm/csr.h
  RISC-V: Use tabs to align macro values in asm/csr.h
  RISC-V: Fix minor checkpatch issues.
  RISC-V: Support nr_cpus command line option.
  RISC-V: Implement nosmp commandline option.
  RISC-V: Add RISC-V specific arch_match_cpu_phys_id
  riscv: vdso: drop unnecessary cc-ldoption
  riscv: call pm_power_off from machine_halt / machine_power_off
  ...

5 years agokconfig: use 'else ifneq' for Makefile to improve readability
Masahiro Yamada [Sat, 18 May 2019 08:07:48 +0000 (17:07 +0900)]
kconfig: use 'else ifneq' for Makefile to improve readability

'ifeq ... else ifneq ... endif' notation is supported by GNU Make 3.81
or later, which is the requirement for building the kernel since
commit 37d69ee30808 ("docs: bump minimal GNU Make version to 3.81").

Use it to improve the readability.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agopanic: add an option to replay all the printk message in buffer
Feng Tang [Fri, 17 May 2019 21:31:50 +0000 (14:31 -0700)]
panic: add an option to replay all the printk message in buffer

Currently on panic, kernel will lower the loglevel and print out pending
printk msg only with console_flush_on_panic().

Add an option for users to configure the "panic_print" to replay all
dmesg in buffer, some of which they may have never seen due to the
loglevel setting, which will help panic debugging .

[feng.tang@intel.com: keep the original console_flush_on_panic() inside panic()]
Link: http://lkml.kernel.org/r/1556199137-14163-1-git-send-email-feng.tang@intel.com
[feng.tang@intel.com: use logbuf lock to protect the console log index]
Link: http://lkml.kernel.org/r/1556269868-22654-1-git-send-email-feng.tang@intel.com
Link: http://lkml.kernel.org/r/1556095872-36838-1-git-send-email-feng.tang@intel.com
Signed-off-by: Feng Tang <feng.tang@intel.com>
Reviewed-by: Petr Mladek <pmladek@suse.com>
Cc: Aaro Koskinen <aaro.koskinen@nokia.com>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Borislav Petkov <bp@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoinitramfs: don't free a non-existent initrd
Steven Price [Fri, 17 May 2019 21:31:47 +0000 (14:31 -0700)]
initramfs: don't free a non-existent initrd

Since commit 54c7a8916a88 ("initramfs: free initrd memory if opening
/initrd.image fails"), the kernel has unconditionally attempted to free
the initrd even if it doesn't exist.

In the non-existent case this causes a boot-time splat if
CONFIG_DEBUG_VIRTUAL is enabled due to a call to virt_to_phys() with a
NULL address.

Instead we should check that the initrd actually exists and only attempt
to free it if it does.

Link: http://lkml.kernel.org/r/20190516143125.48948-1-steven.price@arm.com
Fixes: 54c7a8916a88 ("initramfs: free initrd memory if opening /initrd.image fails")
Signed-off-by: Steven Price <steven.price@arm.com>
Reported-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agofs/writeback.c: use rcu_barrier() to wait for inflight wb switches going into workque...
Jiufei Xue [Fri, 17 May 2019 21:31:44 +0000 (14:31 -0700)]
fs/writeback.c: use rcu_barrier() to wait for inflight wb switches going into workqueue when umount

synchronize_rcu() didn't wait for call_rcu() callbacks, so inode wb
switch may not go to the workqueue after synchronize_rcu().  Thus
previous scheduled switches was not finished even flushing the
workqueue, which will cause a NULL pointer dereferenced followed below.

  VFS: Busy inodes after unmount of vdd. Self-destruct in 5 seconds.  Have a nice day...
  BUG: unable to handle kernel NULL pointer dereference at 0000000000000278
    evict+0xb3/0x180
    iput+0x1b0/0x230
    inode_switch_wbs_work_fn+0x3c0/0x6a0
    worker_thread+0x4e/0x490
    ? process_one_work+0x410/0x410
    kthread+0xe6/0x100
    ret_from_fork+0x39/0x50

Replace the synchronize_rcu() call with a rcu_barrier() to wait for all
pending callbacks to finish.  And inc isw_nr_in_flight after call_rcu()
in inode_switch_wbs() to make more sense.

Link: http://lkml.kernel.org/r/20190429024108.54150-1-jiufei.xue@linux.alibaba.com
Signed-off-by: Jiufei Xue <jiufei.xue@linux.alibaba.com>
Acked-by: Tejun Heo <tj@kernel.org>
Suggested-by: Tejun Heo <tj@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/compaction.c: correct zone boundary handling when isolating pages from a pageblock
Mel Gorman [Fri, 17 May 2019 21:31:41 +0000 (14:31 -0700)]
mm/compaction.c: correct zone boundary handling when isolating pages from a pageblock

syzbot reported the following error from a tree with a head commit of
baf76f0c58ae ("slip: make slhc_free() silently accept an error pointer")

  BUG: unable to handle kernel paging request at ffffea0003348000
  #PF error: [normal kernel read fault]
  PGD 12c3f9067 P4D 12c3f9067 PUD 12c3f8067 PMD 0
  Oops: 0000 [#1] PREEMPT SMP KASAN
  CPU: 1 PID: 28916 Comm: syz-executor.2 Not tainted 5.1.0-rc6+ #89
  Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
  RIP: 0010:constant_test_bit arch/x86/include/asm/bitops.h:314 [inline]
  RIP: 0010:PageCompound include/linux/page-flags.h:186 [inline]
  RIP: 0010:isolate_freepages_block+0x1c0/0xd40 mm/compaction.c:579
  Code: 01 d8 ff 4d 85 ed 0f 84 ef 07 00 00 e8 29 00 d8 ff 4c 89 e0 83 85 38 ff
  ff ff 01 48 c1 e8 03 42 80 3c 38 00 0f 85 31 0a 00 00 <4d> 8b 2c 24 31 ff 49
  c1 ed 10 41 83 e5 01 44 89 ee e8 3a 01 d8 ff
  RSP: 0018:ffff88802b31eab8 EFLAGS: 00010246
  RAX: 1ffffd4000669000 RBX: 00000000000cd200 RCX: ffffc9000a235000
  RDX: 000000000001ca5e RSI: ffffffff81988cc7 RDI: 0000000000000001
  RBP: ffff88802b31ebd8 R08: ffff88805af700c0 R09: 0000000000000000
  R10: 0000000000000000 R11: 0000000000000000 R12: ffffea0003348000
  R13: 0000000000000000 R14: ffff88802b31f030 R15: dffffc0000000000
  FS:  00007f61648dc700(0000) GS:ffff8880ae900000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: ffffea0003348000 CR3: 0000000037c64000 CR4: 00000000001426e0
  Call Trace:
   fast_isolate_around mm/compaction.c:1243 [inline]
   fast_isolate_freepages mm/compaction.c:1418 [inline]
   isolate_freepages mm/compaction.c:1438 [inline]
   compaction_alloc+0x1aee/0x22e0 mm/compaction.c:1550

There is no reproducer and it is difficult to hit -- 1 crash every few
days.  The issue is very similar to the fix in commit 6b0868c820ff
("mm/compaction.c: correct zone boundary handling when resetting pageblock
skip hints").  When isolating free pages around a target pageblock, the
boundary handling is off by one and can stray into the next pageblock.
Triggering the syzbot error requires that the end of pageblock is section
or zone aligned, and that the next section is unpopulated.

A more subtle consequence of the bug is that pageblocks were being
improperly used as migration targets which potentially hurts fragmentation
avoidance in the long-term one page at a time.

A debugging patch revealed that it's definitely possible to stray outside
of a pageblock which is not intended.  While syzbot cannot be used to
verify this patch, it was confirmed that the debugging warning no longer
triggers with this patch applied.  It has also been confirmed that the THP
allocation stress tests are not degraded by this patch.

Link: http://lkml.kernel.org/r/20190510182124.GI18914@techsingularity.net
Fixes: e332f741a8dd ("mm, compaction: be selective about what pageblocks to clear skip hints")
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Reported-by: syzbot+d84c80f9fe26a0f7a734@syzkaller.appspotmail.com
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Qian Cai <cai@lca.pw>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org> # v5.1+
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/vmap: add DEBUG_AUGMENT_LOWEST_MATCH_CHECK macro
Uladzislau Rezki (Sony) [Fri, 17 May 2019 21:31:37 +0000 (14:31 -0700)]
mm/vmap: add DEBUG_AUGMENT_LOWEST_MATCH_CHECK macro

This macro adds some debug code to check that vmap allocations are
happened in ascending order.

By default this option is set to 0 and not active.  It requires
recompilation of the kernel to activate it.  Set to 1, compile the
kernel.

[urezki@gmail.com: v4]
Link: http://lkml.kernel.org/r/20190406183508.25273-4-urezki@gmail.com
Link: http://lkml.kernel.org/r/20190402162531.10888-4-urezki@gmail.com
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Reviewed-by: Roman Gushchin <guro@fb.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Joel Fernandes <joelaf@google.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Oleksiy Avramchenko <oleksiy.avramchenko@sonymobile.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Garnier <thgarnie@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/vmap: add DEBUG_AUGMENT_PROPAGATE_CHECK macro
Uladzislau Rezki (Sony) [Fri, 17 May 2019 21:31:34 +0000 (14:31 -0700)]
mm/vmap: add DEBUG_AUGMENT_PROPAGATE_CHECK macro

This macro adds some debug code to check that the augment tree is
maintained correctly, meaning that every node contains valid
subtree_max_size value.

By default this option is set to 0 and not active.  It requires
recompilation of the kernel to activate it.  Set to 1, compile the
kernel.

[urezki@gmail.com: v4]
Link: http://lkml.kernel.org/r/20190406183508.25273-3-urezki@gmail.com
Link: http://lkml.kernel.org/r/20190402162531.10888-3-urezki@gmail.com
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Reviewed-by: Roman Gushchin <guro@fb.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Joel Fernandes <joelaf@google.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Oleksiy Avramchenko <oleksiy.avramchenko@sonymobile.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Garnier <thgarnie@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/vmalloc.c: keep track of free blocks for vmap allocation
Uladzislau Rezki (Sony) [Fri, 17 May 2019 21:31:31 +0000 (14:31 -0700)]
mm/vmalloc.c: keep track of free blocks for vmap allocation

Patch series "improve vmap allocation", v3.

Objective
---------

Please have a look for the description at:

  https://lkml.org/lkml/2018/10/19/786

but let me also summarize it a bit here as well.

The current implementation has O(N) complexity. Requests with different
permissive parameters can lead to long allocation time. When i say
"long" i mean milliseconds.

Description
-----------

This approach organizes the KVA memory layout into free areas of the
1-ULONG_MAX range, i.e.  an allocation is done over free areas lookups,
instead of finding a hole between two busy blocks.  It allows to have
lower number of objects which represent the free space, therefore to have
less fragmented memory allocator.  Because free blocks are always as large
as possible.

It uses the augment tree where all free areas are sorted in ascending
order of va->va_start address in pair with linked list that provides
O(1) access to prev/next elements.

Since the tree is augment, we also maintain the "subtree_max_size" of VA
that reflects a maximum available free block in its left or right
sub-tree.  Knowing that, we can easily traversal toward the lowest (left
most path) free area.

Allocation: ~O(log(N)) complexity.  It is sequential allocation method
therefore tends to maximize locality.  The search is done until a first
suitable block is large enough to encompass the requested parameters.
Bigger areas are split.

I copy paste here the description of how the area is split, since i
described it in https://lkml.org/lkml/2018/10/19/786

<snip>

A free block can be split by three different ways.  Their names are
FL_FIT_TYPE, LE_FIT_TYPE/RE_FIT_TYPE and NE_FIT_TYPE, i.e.  they
correspond to how requested size and alignment fit to a free block.

FL_FIT_TYPE - in this case a free block is just removed from the free
list/tree because it fully fits.  Comparing with current design there is
an extra work with rb-tree updating.

LE_FIT_TYPE/RE_FIT_TYPE - left/right edges fit.  In this case what we do
is just cutting a free block.  It is as fast as a current design.  Most of
the vmalloc allocations just end up with this case, because the edge is
always aligned to 1.

NE_FIT_TYPE - Is much less common case.  Basically it happens when
requested size and alignment does not fit left nor right edges, i.e.  it
is between them.  In this case during splitting we have to build a
remaining left free area and place it back to the free list/tree.

Comparing with current design there are two extra steps.  First one is we
have to allocate a new vmap_area structure.  Second one we have to insert
that remaining free block to the address sorted list/tree.

In order to optimize a first case there is a cache with free_vmap objects.
Instead of allocating from slab we just take an object from the cache and
reuse it.

Second one is pretty optimized.  Since we know a start point in the tree
we do not do a search from the top.  Instead a traversal begins from a
rb-tree node we split.
<snip>

De-allocation.  ~O(log(N)) complexity.  An area is not inserted straight
away to the tree/list, instead we identify the spot first, checking if it
can be merged around neighbors.  The list provides O(1) access to
prev/next, so it is pretty fast to check it.  Summarizing.  If merged then
large coalesced areas are created, if not the area is just linked making
more fragments.

There is one more thing that i should mention here.  After modification of
VA node, its subtree_max_size is updated if it was/is the biggest area in
its left or right sub-tree.  Apart of that it can also be populated back
to upper levels to fix the tree.  For more details please have a look at
the __augment_tree_propagate_from() function and the description.

Tests and stressing
-------------------

I use the "test_vmalloc.sh" test driver available under
"tools/testing/selftests/vm/" since 5.1-rc1 kernel.  Just trigger "sudo
./test_vmalloc.sh" to find out how to deal with it.

Tested on different platforms including x86_64/i686/ARM64/x86_64_NUMA.
Regarding last one, i do not have any physical access to NUMA system,
therefore i emulated it.  The time of stressing is days.

If you run the test driver in "stress mode", you also need the patch that
is in Andrew's tree but not in Linux 5.1-rc1.  So, please apply it:

http://git.cmpxchg.org/cgit.cgi/linux-mmotm.git/commit/?id=e0cf7749bade6da318e98e934a24d8b62fab512c

After massive testing, i have not identified any problems like memory
leaks, crashes or kernel panics.  I find it stable, but more testing would
be good.

Performance analysis
--------------------

I have used two systems to test.  One is i5-3320M CPU @ 2.60GHz and
another is HiKey960(arm64) board.  i5-3320M runs on 4.20 kernel, whereas
Hikey960 uses 4.15 kernel.  I have both system which could run on 5.1-rc1
as well, but the results have not been ready by time i an writing this.

Currently it consist of 8 tests.  There are three of them which correspond
to different types of splitting(to compare with default).  We have 3
ones(see above).  Another 5 do allocations in different conditions.

a) sudo ./test_vmalloc.sh performance

When the test driver is run in "performance" mode, it runs all available
tests pinned to first online CPU with sequential execution test order.  We
do it in order to get stable and repeatable results.  Take a look at time
difference in "long_busy_list_alloc_test".  It is not surprising because
the worst case is O(N).

# i5-3320M
How many cycles all tests took:
CPU0=646919905370(default) cycles vs CPU0=193290498550(patched) cycles

# See detailed table with results here:
ftp://vps418301.ovh.net/incoming/vmap_test_results_v2/i5-3320M_performance_default.txt
ftp://vps418301.ovh.net/incoming/vmap_test_results_v2/i5-3320M_performance_patched.txt

# Hikey960 8x CPUs
How many cycles all tests took:
CPU0=3478683207 cycles vs CPU0=463767978 cycles

# See detailed table with results here:
ftp://vps418301.ovh.net/incoming/vmap_test_results_v2/HiKey960_performance_default.txt
ftp://vps418301.ovh.net/incoming/vmap_test_results_v2/HiKey960_performance_patched.txt

b) time sudo ./test_vmalloc.sh test_repeat_count=1

With this configuration, all tests are run on all available online CPUs.
Before running each CPU shuffles its tests execution order.  It gives
random allocation behaviour.  So it is rough comparison, but it puts in
the picture for sure.

# i5-3320M
<default>            vs            <patched>
real    101m22.813s                real    0m56.805s
user    0m0.011s                   user    0m0.015s
sys     0m5.076s                   sys     0m0.023s

# See detailed table with results here:
ftp://vps418301.ovh.net/incoming/vmap_test_results_v2/i5-3320M_test_repeat_count_1_default.txt
ftp://vps418301.ovh.net/incoming/vmap_test_results_v2/i5-3320M_test_repeat_count_1_patched.txt

# Hikey960 8x CPUs
<default>            vs            <patched>
real    unknown                    real    4m25.214s
user    unknown                    user    0m0.011s
sys     unknown                    sys     0m0.670s

I did not manage to complete this test on "default Hikey960" kernel
version.  After 24 hours it was still running, therefore i had to cancel
it.  That is why real/user/sys are "unknown".

This patch (of 3):

Currently an allocation of the new vmap area is done over busy list
iteration(complexity O(n)) until a suitable hole is found between two busy
areas.  Therefore each new allocation causes the list being grown.  Due to
over fragmented list and different permissive parameters an allocation can
take a long time.  For example on embedded devices it is milliseconds.

This patch organizes the KVA memory layout into free areas of the
1-ULONG_MAX range.  It uses an augment red-black tree that keeps blocks
sorted by their offsets in pair with linked list keeping the free space in
order of increasing addresses.

Nodes are augmented with the size of the maximum available free block in
its left or right sub-tree.  Thus, that allows to take a decision and
traversal toward the block that will fit and will have the lowest start
address, i.e.  it is sequential allocation.

Allocation: to allocate a new block a search is done over the tree until a
suitable lowest(left most) block is large enough to encompass: the
requested size, alignment and vstart point.  If the block is bigger than
requested size - it is split.

De-allocation: when a busy vmap area is freed it can either be merged or
inserted to the tree.  Red-black tree allows efficiently find a spot
whereas a linked list provides a constant-time access to previous and next
blocks to check if merging can be done.  In case of merging of
de-allocated memory chunk a large coalesced area is created.

Complexity: ~O(log(N))

[urezki@gmail.com: v3]
Link: http://lkml.kernel.org/r/20190402162531.10888-2-urezki@gmail.com
[urezki@gmail.com: v4]
Link: http://lkml.kernel.org/r/20190406183508.25273-2-urezki@gmail.com
Link: http://lkml.kernel.org/r/20190321190327.11813-2-urezki@gmail.com
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Reviewed-by: Roman Gushchin <guro@fb.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Thomas Garnier <thgarnie@google.com>
Cc: Oleksiy Avramchenko <oleksiy.avramchenko@sonymobile.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Joel Fernandes <joelaf@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoMerge tag 'perf-core-for-mingo-5.2-20190517' of git://git.kernel.org/pub/scm/linux...
Ingo Molnar [Sat, 18 May 2019 08:24:43 +0000 (10:24 +0200)]
Merge tag 'perf-core-for-mingo-5.2-20190517' of git://git./linux/kernel/git/acme/linux into perf/core

Pull perf/core improvements and fixes from Arnaldo Carvalho de Melo:

perf.data:

  Alexey Budankov:

  - Streaming compression of perf ring buffer into PERF_RECORD_COMPRESSED
    user space records, resulting in ~3-5x perf.data file size reduction
    on variety of tested workloads what saves storage space on larger
    server systems where perf.data size can easily reach several tens or
    even hundreds of GiBs, especially when profiling with DWARF-based
    stacks and tracing of context switches.

perf record:

  Arnaldo Carvalho de Melo

  - Improve -user-regs/intr-regs suggestions to overcome errors.

perf annotate:

  Jin Yao:

  - Remove hist__account_cycles() from callback, speeding up branch processing
    (perf record -b).

perf stat:

  - Add a 'percore' event qualifier, e.g.: -e cpu/event=0,umask=0x3,percore=1/,
    that sums up the event counts for both hardware threads in a core.

    We can already do this with --per-core, but it's often useful to do
    this together with other metrics that are collected per hardware thread.

    I.e. now its possible to do this per-event, and have it mixed with other
    events not aggregated by core.

core libraries:

  Donald Yandt:

  - Check for errors when doing fgets(/proc/version).

  Jiri Olsa:

  - Speed up report for perf compiled with linbunwind.

tools headers:

  Arnaldo Carvalho de Melo

  - Update memcpy_64.S, x86's kvm.h and pt_regs.h.

arm64:

  Florian Fainelli:

  - Map Brahma-B53 CPUID to cortex-a53 events.

  - Add Cortex-A57 and Cortex-A72 events.

csky:

  Mao Han:

  - Add DWARF register mappings for libdw, allowing --call-graph=dwarf to work
    on the C-SKY arch.

x86:

  Andi Kleen/Kan Liang:

  - Add support for recording and printing XMM registers, available, for
    instance, on Icelake.

  Kan Liang:

  - Add uncore_upi (Intel's "Ultra Path Interconnect" events) JSON support.
    UPI replaced the Intel QuickPath Interconnect (QPI) in Xeon Skylake-SP.

Intel PT:

  Adrian Hunter

  . Fix instructions sampling rate.

  . Timestamp fixes.

  . Improve exported-sql-viewer GUI, allowing, for instance, to copy'n'paste
    the trees, useful for e-mailing.

Documentation:

  Thomas Richter:

  - Add description for 'perf --debug stderr=1', which redirects stderr to stdout.

libtraceevent:

  Tzvetomir Stoyanov:

  - Add man pages for the various APIs.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
5 years agokbuild: check uniqueness of module names
Masahiro Yamada [Fri, 17 May 2019 16:07:15 +0000 (01:07 +0900)]
kbuild: check uniqueness of module names

In the recent build test of linux-next, Stephen saw a build error
caused by a broken .tmp_versions/*.mod file:

  https://lkml.org/lkml/2019/5/13/991

drivers/net/phy/asix.ko and drivers/net/usb/asix.ko have the same
basename, and there is a race in generating .tmp_versions/asix.mod

Kbuild has not checked this before, and it suddenly shows up with
obscure error messages when this kind of race occurs.

Non-unique module names cause various sort of problems, but it is
not trivial to catch them by eyes.

Hence, this script.

It checks not only real modules, but also built-in modules (i.e.
controlled by tristate CONFIG option, but currently compiled with =y).
Non-unique names for built-in modules also cause problems because
/sys/modules/ would fall over.

For the latest kernel, I tested "make allmodconfig all" (or more
quickly "make allyesconfig modules"), and it detected the following:

warning: same basename if the following are built as modules:
  drivers/regulator/88pm800.ko
  drivers/mfd/88pm800.ko
warning: same basename if the following are built as modules:
  drivers/gpu/drm/bridge/adv7511/adv7511.ko
  drivers/media/i2c/adv7511.ko
warning: same basename if the following are built as modules:
  drivers/net/phy/asix.ko
  drivers/net/usb/asix.ko
warning: same basename if the following are built as modules:
  fs/coda/coda.ko
  drivers/media/platform/coda/coda.ko
warning: same basename if the following are built as modules:
  drivers/net/phy/realtek.ko
  drivers/net/dsa/realtek.ko

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Stephen Rothwell <sfr@canb.auug.org.au>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
5 years agokconfig: Terminate menu blocks with a comment in the generated config
Alexander Popov [Fri, 17 May 2019 19:42:22 +0000 (22:42 +0300)]
kconfig: Terminate menu blocks with a comment in the generated config

Currently menu blocks start with a pretty header but end with nothing in
the generated config. So next config options stick together with the
options from the menu block.

Let's terminate menu blocks in the generated config with a comment and
a newline if needed. Example:

...
CONFIG_BPF_STREAM_PARSER=y
CONFIG_NET_FLOW_LIMIT=y

#
# Network testing
#
CONFIG_NET_PKTGEN=y
CONFIG_NET_DROP_MONITOR=y
# end of Network testing
# end of Networking options

CONFIG_HAMRADIO=y
...

Signed-off-by: Alexander Popov <alex.popov@linux.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agokbuild: add LICENSES to KBUILD_ALLDIRS
Masahiro Yamada [Wed, 15 May 2019 16:18:54 +0000 (01:18 +0900)]
kbuild: add LICENSES to KBUILD_ALLDIRS

For *-pkg targets, the LICENSES directory should be included in the
source tarball.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agokbuild: remove 'addtree' and 'flags' magic for header search paths
Masahiro Yamada [Mon, 13 May 2019 06:22:17 +0000 (15:22 +0900)]
kbuild: remove 'addtree' and 'flags' magic for header search paths

The 'addtree' and 'flags' in scripts/Kbuild.include are so compilecated
and ugly.

As I mentioned in [1], Kbuild should stop automatic prefixing of header
search path options.

I fixed up (almost) all Makefiles in the kernel. Now 'addtree' and
'flags' have been removed.

Kbuild still caters to add $(srctree)/$(src) and $(objtree)/$(obj)
to the header search path for O= building, but never touches extra
compiler options from ccflags-y etc.

[1]: https://patchwork.kernel.org/patch/9632347/

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agotreewide: prefix header search paths with $(srctree)/
Masahiro Yamada [Mon, 13 May 2019 06:22:16 +0000 (15:22 +0900)]
treewide: prefix header search paths with $(srctree)/

Currently, the Kbuild core manipulates header search paths in a crazy
way [1].

To fix this mess, I want all Makefiles to add explicit $(srctree)/ to
the search paths in the srctree. Some Makefiles are already written in
that way, but not all. The goal of this work is to make the notation
consistent, and finally get rid of the gross hacks.

Having whitespaces after -I does not matter since commit 48f6e3cf5bc6
("kbuild: do not drop -I without parameter").

[1]: https://patchwork.kernel.org/patch/9632347/

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agomedia: prefix header search paths with $(srctree)/
Masahiro Yamada [Mon, 13 May 2019 06:22:15 +0000 (15:22 +0900)]
media: prefix header search paths with $(srctree)/

Currently, the Kbuild core manipulates header search paths in a crazy
way [1].

To fix this mess, I want all Makefiles to add explicit $(srctree)/ to
the search paths in the srctree. Some Makefiles are already written in
that way, but not all. The goal of this work is to make the notation
consistent, and finally get rid of the gross hacks.

Having whitespaces after -I does not matter since commit 48f6e3cf5bc6
("kbuild: do not drop -I without parameter").

[1]: https://patchwork.kernel.org/patch/9632347/

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com>
5 years agomedia: remove unneeded header search paths
Masahiro Yamada [Mon, 13 May 2019 06:22:14 +0000 (15:22 +0900)]
media: remove unneeded header search paths

I was able to build without these extra header search paths.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agoalpha: move arch/alpha/defconfig to arch/alpha/configs/defconfig
Masahiro Yamada [Mon, 13 May 2019 02:14:05 +0000 (11:14 +0900)]
alpha: move arch/alpha/defconfig to arch/alpha/configs/defconfig

As of Linux 5.1, alpha and s390 are the last architectures that
have defconfig in arch/*/ instead of arch/*/configs/.

  $ find arch -name defconfig | sort
  arch/alpha/defconfig
  arch/arm64/configs/defconfig
  arch/csky/configs/defconfig
  arch/nds32/configs/defconfig
  arch/riscv/configs/defconfig
  arch/s390/defconfig

The arch/$(ARCH)/defconfig is the hard-coded default in Kconfig,
and I want to deprecate it after evacuating the remaining defconfig
into the standard location, arch/*/configs/.

Define KBUILD_DEFCONFIG like other architectures, and move defconfig
into the configs/ subdirectory.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Paul Walmsley <paul@pwsan.com>
5 years agokbuild: terminate Kconfig when $(CC) or $(LD) is missing
Masahiro Yamada [Thu, 9 May 2019 07:35:55 +0000 (16:35 +0900)]
kbuild: terminate Kconfig when $(CC) or $(LD) is missing

If the compiler specified by $(CC) is not present, the Kconfig stage
sprinkles 'not found' messages, then succeeds.

  $ make CROSS_COMPILE=foo defconfig
  /bin/sh: 1: foogcc: not found
  /bin/sh: 1: foogcc: not found
  *** Default configuration is based on 'x86_64_defconfig'
  ./scripts/gcc-version.sh: 17: ./scripts/gcc-version.sh: foogcc: not found
  ./scripts/gcc-version.sh: 18: ./scripts/gcc-version.sh: foogcc: not found
  ./scripts/gcc-version.sh: 19: ./scripts/gcc-version.sh: foogcc: not found
  ./scripts/gcc-version.sh: 17: ./scripts/gcc-version.sh: foogcc: not found
  ./scripts/gcc-version.sh: 18: ./scripts/gcc-version.sh: foogcc: not found
  ./scripts/gcc-version.sh: 19: ./scripts/gcc-version.sh: foogcc: not found
  ./scripts/clang-version.sh: 11: ./scripts/clang-version.sh: foogcc: not found
  ./scripts/gcc-plugin.sh: 11: ./scripts/gcc-plugin.sh: foogcc: not found
  init/Kconfig:16:warning: 'GCC_VERSION': number is invalid
  #
  # configuration written to .config
  #

Terminate parsing files immediately if $(CC) or $(LD) is not found.
"make *config" will fail more nicely.

  $ make CROSS_COMPILE=foo defconfig
  *** Default configuration is based on 'x86_64_defconfig'
  scripts/Kconfig.include:34: compiler 'foogcc' not found
  make[1]: *** [scripts/kconfig/Makefile;82: defconfig] Error 1
  make: *** [Makefile;557: defconfig] Error 2

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agokbuild: turn auto.conf.cmd into a mandatory include file
Masahiro Yamada [Sun, 12 May 2019 02:13:48 +0000 (11:13 +0900)]
kbuild: turn auto.conf.cmd into a mandatory include file

syncconfig is responsible for keeping auto.conf up-to-date, so if it
fails for any reason, the build must be terminated immediately.

However, since commit 9390dff66a52 ("kbuild: invoke syncconfig if
include/config/auto.conf.cmd is missing"), Kbuild continues running
even after syncconfig fails.

You can confirm this by intentionally making syncconfig error out:

  diff --git a/scripts/kconfig/confdata.c b/scripts/kconfig/confdata.c
  index 08ba146..307b9de 100644
  --- a/scripts/kconfig/confdata.c
  +++ b/scripts/kconfig/confdata.c
  @@ -1023,6 +1023,9 @@ int conf_write_autoconf(int overwrite)
          FILE *out, *tristate, *out_h;
          int i;

  +       if (overwrite)
  +               return 1;
  +
          if (!overwrite && is_present(autoconf_name))
                  return 0;

Then, syncconfig fails, but Make would not stop:

  $ make -s mrproper allyesconfig defconfig
  $ make
  scripts/kconfig/conf  --syncconfig Kconfig

  *** Error during sync of the configuration.

  make[2]: *** [scripts/kconfig/Makefile;69: syncconfig] Error 1
  make[1]: *** [Makefile;557: syncconfig] Error 2
  make: *** [include/config/auto.conf.cmd] Deleting file 'include/config/tristate.conf'
  make: Failed to remake makefile 'include/config/auto.conf'.
    SYSTBL  arch/x86/include/generated/asm/syscalls_32.h
    SYSHDR  arch/x86/include/generated/asm/unistd_32_ia32.h
    SYSHDR  arch/x86/include/generated/asm/unistd_64_x32.h
    SYSTBL  arch/x86/include/generated/asm/syscalls_64.h
  [ continue running ... ]

The reason is in the behavior of a pattern rule with multi-targets.

  %/auto.conf %/auto.conf.cmd %/tristate.conf: $(KCONFIG_CONFIG)
          $(Q)$(MAKE) -f $(srctree)/Makefile syncconfig

GNU Make knows this rule is responsible for making all the three files
simultaneously. As far as examined, auto.conf.cmd is the target in
question when this rule is invoked. It is probably because auto.conf.cmd
is included below the inclusion of auto.conf.

The inclusion of auto.conf is mandatory, while that of auto.conf.cmd
is optional. GNU Make does not care about the failure in the process
of updating optional include files.

I filed this issue (https://savannah.gnu.org/bugs/?56301) in case this
behavior could be improved somehow in future releases of GNU Make.
Anyway, it is quite easy to fix our Makefile.

Given that auto.conf is already a mandatory include file, there is no
reason to stick auto.conf.cmd optional. Make it mandatory as well.

Cc: linux-stable <stable@vger.kernel.org> # 5.0+
Fixes: 9390dff66a52 ("kbuild: invoke syncconfig if include/config/auto.conf.cmd is missing")
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years ago.gitignore: exclude .get_maintainer.ignore and .gitattributes
Masahiro Yamada [Sat, 11 May 2019 03:13:54 +0000 (12:13 +0900)]
.gitignore: exclude .get_maintainer.ignore and .gitattributes

Also, sort the patterns alphabetically. Update the comment since
we have non-git files here.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agokbuild: add all Clang-specific flags unconditionally
Masahiro Yamada [Fri, 10 May 2019 14:10:09 +0000 (23:10 +0900)]
kbuild: add all Clang-specific flags unconditionally

We do not support old Clang versions. Upgrade your clang version
if any of these flags is unsupported.

Let's add all flags inside ifdef CONFIG_CC_IS_CLANG unconditionally.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Sedat Dilek <sedat.dilek@gmail.com>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Tested-by: Nick Desaulniers <ndesaulniers@google.com>
5 years agokbuild: Don't try to add '-fcatch-undefined-behavior' flag
Nathan Chancellor [Thu, 9 May 2019 11:48:25 +0000 (04:48 -0700)]
kbuild: Don't try to add '-fcatch-undefined-behavior' flag

This is no longer a valid option in clang, it was removed in 3.5, which
we don't support.

https://github.com/llvm/llvm-project/commit/cb3f812b6b9fab8f3b41414f24e90222170417b4

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agokbuild: add some extra warning flags unconditionally
Masahiro Yamada [Thu, 9 May 2019 06:46:35 +0000 (15:46 +0900)]
kbuild: add some extra warning flags unconditionally

These flags are documented in the GCC 4.6 manual, and recognized by
Clang as well. Let's rip off the cc-option / cc-disable-warning switches.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Tested-by: Nick Desaulniers <ndesaulniers@google.com>
5 years agokbuild: add -Wvla flag unconditionally
Masahiro Yamada [Thu, 9 May 2019 06:45:49 +0000 (15:45 +0900)]
kbuild: add -Wvla flag unconditionally

This flag is documented in the GCC 4.6 manual, and recognized by
Clang as well. Let's rip off the cc-option switch.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Sedat Dilek <sedat.dilek@gmail.com>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Acked-by: Kees Cook <keescook@chromium.org>
Tested-by: Nick Desaulniers <ndesaulniers@google.com>
5 years agoarch: remove dangling asm-generic wrappers
Masahiro Yamada [Thu, 9 May 2019 07:59:34 +0000 (16:59 +0900)]
arch: remove dangling asm-generic wrappers

These generic-y defines do not have the corresponding generic header
in include/asm-generic/, so they are definitely invalid.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agosamples: guard sub-directories with CONFIG options
Masahiro Yamada [Thu, 9 May 2019 01:00:19 +0000 (10:00 +0900)]
samples: guard sub-directories with CONFIG options

Do not descend to sub-directories when unneeded.

I used subdir-$(CONFIG_...) for hidraw, seccomp, and vfs because
they only contain host programs.

While we are here, let's add SPDX License tag, and sort the directories
alphabetically.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agokbuild: re-enable int-in-bool-context warning
Masahiro Yamada [Thu, 9 May 2019 00:58:01 +0000 (09:58 +0900)]
kbuild: re-enable int-in-bool-context warning

This warning was disabled by commit bd664f6b3e37 ("disable new
gcc-7.1.1 warnings for now") just because it was too noisy.

Thanks to Arnd Bergmann, all warnings have been fixed. Now, we are
ready to re-enable it.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Cc: Arnd Bergmann <arnd@arndb.de>
5 years agoMAINTAINERS: kbuild: Add pattern for scripts/*vmlinux*
Krzysztof Kozlowski [Mon, 6 May 2019 12:47:00 +0000 (14:47 +0200)]
MAINTAINERS: kbuild: Add pattern for scripts/*vmlinux*

scripts/link-vmlinux.sh is part of kbuild so extend the pattern to match
any vmlinux related scripts.

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agosh: exclude vmlinux.scr from .gitignore pattern
Masahiro Yamada [Wed, 1 May 2019 13:56:01 +0000 (22:56 +0900)]
sh: exclude vmlinux.scr from .gitignore pattern

arch/sh/boot/.gitignore has the pattern "vmlinux*"; this is effective
not only for the current directory, but also for any sub-directories.

So, from the point of .gitignore grammar, the following check-in files
are also considered to be ignored:

  arch/sh/boot/compressed/vmlinux.scr
  arch/sh/boot/romimage/vmlinux.scr

As the manual gitignore(5) says "Files already tracked by Git are not
affected", this is not a problem as far as Git is concerned.

However, Git is not the only program that parses .gitignore because
.gitignore is useful to distinguish build artifacts from source files.

For example, tar(1) supports the --exclude-vcs-ignore option. As of
writing, this option does not work perfectly, but it intends to create
a tarball excluding files specified by .gitignore.

So, I believe it is better to fix this issue.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agosh: vsyscall: drop unnecessary cc-ldoption
Nick Desaulniers [Wed, 24 Apr 2019 18:02:21 +0000 (11:02 -0700)]
sh: vsyscall: drop unnecessary cc-ldoption

Towards the goal of removing cc-ldoption, it seems that --hash-style=
was added to binutils 2.17.50.0.2 in 2006. The minimal required version
of binutils for the kernel according to
Documentation/process/changes.rst is 2.20.

Link: https://gcc.gnu.org/ml/gcc/2007-01/msg01141.html
Cc: clang-built-linux@googlegroups.com
Suggested-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Acked-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agoia64: require -Wl,--hash-style=sysv
Nick Desaulniers [Tue, 23 Apr 2019 20:48:20 +0000 (13:48 -0700)]
ia64: require -Wl,--hash-style=sysv

Towards the goal of removing cc-ldoption, it seems that --hash-style=
was added to binutils 2.17.50.0.2 in 2006. The minimal required version
of binutils for the kernel according to
Documentation/process/changes.rst is 2.20.

Link: https://gcc.gnu.org/ml/gcc/2007-01/msg01141.html
Cc: clang-built-linux@googlegroups.com
Suggested-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agocsky: remove deprecated arch/csky/boot/dts/include/dt-bindings
Masahiro Yamada [Mon, 4 Feb 2019 07:56:01 +0000 (16:56 +0900)]
csky: remove deprecated arch/csky/boot/dts/include/dt-bindings

Having a symbolic link arch/*/boot/dts/include/dt-bindings was
deprecated by commit d5d332d3f7e8 ("devicetree: Move include
prefixes from arch to separate directory").

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
5 years agoext4: avoid panic during forced reboot due to aborted journal
Jan Kara [Fri, 17 May 2019 21:37:18 +0000 (17:37 -0400)]
ext4: avoid panic during forced reboot due to aborted journal

Handling of aborted journal is a special code path different from
standard ext4_error() one and it can call panic() as well. Commit
1dc1097ff60e ("ext4: avoid panic during forced reboot") forgot to update
this path so fix that omission.

Fixes: 1dc1097ff60e ("ext4: avoid panic during forced reboot")
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@kernel.org # 5.1
5 years agoMerge tag 'sound-fix-5.2-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Fri, 17 May 2019 20:57:54 +0000 (13:57 -0700)]
Merge tag 'sound-fix-5.2-rc1' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "Just a few HD-audio fixes, most of which are specific to Realtek
  codecs"

* tag 'sound-fix-5.2-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ALSA: hda/realtek - Fix for Lenovo B50-70 inverted internal microphone bug
  ALSA: hda: Fix race between creating and refreshing sysfs entries
  ALSA: hda/realtek - Corrected fixup for System76 Gazelle (gaze14)
  ALSA: hda/realtek - Avoid superfluous COEF EAPD setups
  ALSA: hda/realtek - Fixup headphone noise via runtime suspend

5 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Fri, 17 May 2019 17:33:30 +0000 (10:33 -0700)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull KVM updates from Paolo Bonzini:
 "ARM:
   - support for SVE and Pointer Authentication in guests
   - PMU improvements

  POWER:
   - support for direct access to the POWER9 XIVE interrupt controller
   - memory and performance optimizations

  x86:
   - support for accessing memory not backed by struct page
   - fixes and refactoring

  Generic:
   - dirty page tracking improvements"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (155 commits)
  kvm: fix compilation on aarch64
  Revert "KVM: nVMX: Expose RDPMC-exiting only when guest supports PMU"
  kvm: x86: Fix L1TF mitigation for shadow MMU
  KVM: nVMX: Disable intercept for FS/GS base MSRs in vmcs02 when possible
  KVM: PPC: Book3S: Remove useless checks in 'release' method of KVM device
  KVM: PPC: Book3S HV: XIVE: Fix spelling mistake "acessing" -> "accessing"
  KVM: PPC: Book3S HV: Make sure to load LPID for radix VCPUs
  kvm: nVMX: Set nested_run_pending in vmx_set_nested_state after checks complete
  tests: kvm: Add tests for KVM_SET_NESTED_STATE
  KVM: nVMX: KVM_SET_NESTED_STATE - Tear down old EVMCS state before setting new state
  tests: kvm: Add tests for KVM_CAP_MAX_VCPUS and KVM_CAP_MAX_CPU_ID
  tests: kvm: Add tests to .gitignore
  KVM: Introduce KVM_CAP_MANUAL_DIRTY_LOG_PROTECT2
  KVM: Fix kvm_clear_dirty_log_protect off-by-(minus-)one
  KVM: Fix the bitmap range to copy during clear dirty
  KVM: arm64: Fix ptrauth ID register masking logic
  KVM: x86: use direct accessors for RIP and RSP
  KVM: VMX: Use accessors for GPRs outside of dedicated caching logic
  KVM: x86: Omit caching logic for always-available GPRs
  kvm, x86: Properly check whether a pfn is an MMIO or not
  ...

5 years agoi2c: core: add device-managed version of i2c_new_dummy
Heiner Kallweit [Thu, 16 May 2019 21:13:09 +0000 (23:13 +0200)]
i2c: core: add device-managed version of i2c_new_dummy

i2c_new_dummy is typically called from the probe function of the
driver for the primary i2c client. It requires calls to
i2c_unregister_device in the error path of the probe function and
in the remove function.
This can be simplified by introducing a device-managed version.

Note the changed error case return value type: i2c_new_dummy returns
NULL whilst devm_i2c_new_dummy_device returns an ERR_PTR.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
[wsa: rename new functions and fix minor kdoc issues]
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: Peter Rosin <peda@axentia.se>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
5 years agoi2c: core: improve return value handling of i2c_new_device and i2c_new_dummy
Heiner Kallweit [Thu, 16 May 2019 21:13:08 +0000 (23:13 +0200)]
i2c: core: improve return value handling of i2c_new_device and i2c_new_dummy

Currently i2c_new_device and i2c_new_dummy return just NULL in error
case although they have more error details internally. Therefore move
the functionality into new functions returning detailed errors and
add wrappers for compatibility with the current API.

This allows to use these functions with detailed error codes within
the i2c core or for API extensions.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
[wsa: rename new functions and fix minor kdoc issues]
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: Peter Rosin <peda@axentia.se>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
5 years agoMerge tag 'nds32-for-linus-5.2-rc1' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 17 May 2019 17:17:29 +0000 (10:17 -0700)]
Merge tag 'nds32-for-linus-5.2-rc1' of git://git./linux/kernel/git/greentime/linux

Pull nds32 updates from Greentime Hu:

 - Clean up codes and Makefile

 - Fix a vDSO bug

 - Remove useless functions/header files

 - Update git repo path in MAINTAINERS

* tag 'nds32-for-linus-5.2-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/greentime/linux:
  nds32: Fix vDSO clock_getres()
  MAINTAINERS: update nds32 git repo path
  nds32: don't export low-level cache flushing routines
  arch: nds32: Kconfig: pedantic formatting
  nds32: fix semicolon code style issue
  nds32: vdso: drop unnecessary cc-ldoption
  nds32: remove unused generic-y += cmpxchg-local.h
  nds32: Use the correct style for SPDX License Identifier
  nds32: remove __virt_to_bus and __bus_to_virt
  nds32: vdso: fix and clean-up Makefile
  nds32: add vmlinux.lds and vdso.so to .gitignore
  nds32: ex-exit: Remove unneeded need_resched() loop
  nds32/io: Remove useless definition of mmiowb()
  nds32: Removed unused thread flag TIF_USEDFPU

5 years agoMerge tag 's390-5.2-2' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Linus Torvalds [Fri, 17 May 2019 17:08:59 +0000 (10:08 -0700)]
Merge tag 's390-5.2-2' of git://git./linux/kernel/git/s390/linux

Pull more s390 updates from Martin Schwidefsky:

 - Enhancements for the QDIO layer

 - Remove the RCP trace event

 - Avoid three build issues

 - Move the defconfig to the configs directory

* tag 's390-5.2-2' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390: move arch/s390/defconfig to arch/s390/configs/defconfig
  s390/qdio: optimize state inspection of HW-owned SBALs
  s390/qdio: use get_buf_state() in debug_get_buf_state()
  s390/qdio: allow to scan all Output SBALs in one go
  s390/cio: Remove tracing for rchp instruction
  s390/kasan: adapt disabled_wait usage to avoid build error
  latent_entropy: avoid build error when plugin cflags are not set
  s390/boot: fix compiler error due to missing awk strtonum

5 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Fri, 17 May 2019 16:46:31 +0000 (09:46 -0700)]
Merge branch 'fixes' of git://git./linux/kernel/git/viro/vfs

Pull more vfs mount updates from Al Viro:
 "Propagation of new syscalls to other architectures + cosmetic change
  from Christian (fscontext didn't follow the convention for anon inode
  names)"

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  uapi: Wire up the mount API syscalls on non-x86 arches [ver #2]
  uapi, x86: Fix the syscall numbering of the mount API syscalls [ver #2]
  uapi, fsopen: use square brackets around "fscontext" [ver #2]

5 years agoMerge tag 'timers-v5.2' of http://git.linaro.org/people/daniel.lezcano/linux into...
Thomas Gleixner [Fri, 17 May 2019 13:48:06 +0000 (15:48 +0200)]
Merge tag 'timers-v5.2' of git.linaro.org/people/daniel.lezcano/linux into timers/core

Pull clockevent updates from Daniel Lezcano:

 - Add compatible string for suniv for sun4i (Mesih Kilinc)

 - Add COMPILE_TEST option for sp804 (David Abdurachmanov)

 - Replace the compensation time when suspend happens on tegra with the one
   provided by the generic framework (Joseph Lo)

 - Cleanup, shutdown and oneshot mode fix on milbeaut timer (Sugaya Taichi)

 - Atmel TCB rework to fix boot failure on boards without PIT or misfunction
   on system using a preempt-rt kernel (Alexandre Belloni)

5 years agopowerpc/cacheinfo: Remove double free
Tobin C. Harding [Wed, 15 May 2019 09:07:50 +0000 (19:07 +1000)]
powerpc/cacheinfo: Remove double free

kfree() after kobject_put(). Who ever wrote this was on crack.

Fixes: 7e8039795a80 ("powerpc/cacheinfo: Fix kobject memleak")
Signed-off-by: Tobin C. Harding <tobin@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
5 years agopowerpc/mm/hash: Fix get_region_id() for invalid addresses
Aneesh Kumar K.V [Thu, 16 May 2019 11:50:54 +0000 (17:20 +0530)]
powerpc/mm/hash: Fix get_region_id() for invalid addresses

Accesses by userspace to random addresses outside the user or kernel
address range will generate an SLB fault. When we handle that fault we
classify the effective address into several classes, eg. user, kernel
linear, kernel virtual etc.

For addresses that are completely outside of any valid range, we
should not insert an SLB entry at all, and instead immediately an
exception.

In the past this was handled in two ways. Firstly we would check the
top nibble of the address (using REGION_ID(ea)) and that would tell us
if the address was user (0), kernel linear (c), kernel virtual (d), or
vmemmap (f). If the address didn't match any of these it was invalid.

Then for each type of address we would do a secondary check. For the
user region we check against H_PGTABLE_RANGE, for kernel linear we
would mask the top nibble of the address and then check the address
against MAX_PHYSMEM_BITS.

As part of commit 0034d395f89d ("powerpc/mm/hash64: Map all the kernel
regions in the same 0xc range") we replaced REGION_ID() with
get_region_id() and changed the masking of the top nibble to only mask
the top two bits, which introduced a bug.

Addresses less than (4 << 60) are still handled correctly, they are
either less than (1 << 60) in which case they are subject to the
H_PGTABLE_RANGE check, or they are correctly checked against
MAX_PHYSMEM_BITS.

However addresses from (4 << 60) to ((0xc << 60) - 1), are incorrectly
treated as kernel linear addresses in get_region_id(). Then the top
two bits are cleared by EA_MASK in slb_allocate_kernel() and the
address is checked against MAX_PHYSMEM_BITS, which it passes due to
the masking. The end result is we incorrectly insert SLB entries for
those addresses.

That is not actually catastrophic, having inserted the SLB entry we
will then go on to take a page fault for the address and at that point
we detect the problem and report it as a bad fault.

Still we should not be inserting those entries, or treating them as
kernel linear addresses in the first place. So fix get_region_id() to
detect addresses in that range and return an invalid region id, which
we cause use to not insert an SLB entry and directly report an
exception.

Fixes: 0034d395f89d ("powerpc/mm/hash64: Map all the kernel regions in the same 0xc range")
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
[mpe: Drop change to EA_MASK for now, rewrite change log]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
5 years agokvm: fix compilation on aarch64
Paolo Bonzini [Fri, 17 May 2019 12:08:53 +0000 (14:08 +0200)]
kvm: fix compilation on aarch64

Commit e45adf665a53 ("KVM: Introduce a new guest mapping API", 2019-01-31)
introduced a build failure on aarch64 defconfig:

$ make -j$(nproc) ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu- O=out defconfig \
                Image.gz
...
../arch/arm64/kvm/../../../virt/kvm/kvm_main.c:
    In function '__kvm_map_gfn':
../arch/arm64/kvm/../../../virt/kvm/kvm_main.c:1763:9: error:
    implicit declaration of function 'memremap'; did you mean 'memset_p'?
../arch/arm64/kvm/../../../virt/kvm/kvm_main.c:1763:46: error:
    'MEMREMAP_WB' undeclared (first use in this function)
../arch/arm64/kvm/../../../virt/kvm/kvm_main.c:
    In function 'kvm_vcpu_unmap':
../arch/arm64/kvm/../../../virt/kvm/kvm_main.c:1795:3: error:
    implicit declaration of function 'memunmap'; did you mean 'vm_munmap'?

because these functions are declared in <linux/io.h> rather than <asm/io.h>,
and the former was being pulled in already on x86 but not on aarch64.

Reported-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 years agoobjtool: Allow AR to be overridden with HOSTAR
Nathan Chancellor [Thu, 16 May 2019 17:49:42 +0000 (12:49 -0500)]
objtool: Allow AR to be overridden with HOSTAR

Currently, this Makefile hardcodes GNU ar, meaning that if it is not
available, there is no way to supply a different one and the build will
fail.

  $ make AR=llvm-ar CC=clang LD=ld.lld HOSTAR=llvm-ar HOSTCC=clang \
         HOSTLD=ld.lld HOSTLDFLAGS=-fuse-ld=lld defconfig modules_prepare
  ...
    AR       /out/tools/objtool/libsubcmd.a
  /bin/sh: 1: ar: not found
  ...

Follow the logic of HOST{CC,LD} and allow the user to specify a
different ar tool via HOSTAR (which is used elsewhere in other
tools/ Makefiles).

Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Reviewed-by: Mukesh Ojha <mojha@codeaurora.org>
Cc: <stable@vger.kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/80822a9353926c38fd7a152991c6292491a9d0e8.1558028966.git.jpoimboe@redhat.com
Link: https://github.com/ClangBuiltLinux/linux/issues/481
Signed-off-by: Ingo Molnar <mingo@kernel.org>
5 years agofbdev/efifb: Ignore framebuffer memmap entries that lack any memory types
Ard Biesheuvel [Thu, 16 May 2019 21:31:59 +0000 (23:31 +0200)]
fbdev/efifb: Ignore framebuffer memmap entries that lack any memory types

The following commit:

  38ac0287b7f4 ("fbdev/efifb: Honour UEFI memory map attributes when mapping the FB")

updated the EFI framebuffer code to use memory mappings for the linear
framebuffer that are permitted by the memory attributes described by the
EFI memory map for the particular region, if the framebuffer happens to
be covered by the EFI memory map (which is typically only the case for
framebuffers in shared memory). This is required since non-x86 systems
may require cacheable attributes for memory mappings that are shared
with other masters (such as GPUs), and this information cannot be
described by the Graphics Output Protocol (GOP) EFI protocol itself,
and so we rely on the EFI memory map for this.

As reported by James, this breaks some x86 systems:

  [ 1.173368] efifb: probing for efifb
  [ 1.173386] efifb: abort, cannot remap video memory 0x1d5000 @ 0xcf800000
  [ 1.173395] Trying to free nonexistent resource <00000000cf800000-00000000cf9d4bff>
  [ 1.173413] efi-framebuffer: probe of efi-framebuffer.0 failed with error -5

The problem turns out to be that the memory map entry that describes the
framebuffer has no memory attributes listed at all, and so we end up with
a mem_flags value of 0x0.

So work around this by ensuring that the memory map entry's attribute field
has a sane value before using it to mask the set of usable attributes.

Reported-by: James Hilliard <james.hilliard1@gmail.com>
Tested-by: James Hilliard <james.hilliard1@gmail.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: <stable@vger.kernel.org> # v4.19+
Cc: Borislav Petkov <bp@alien8.de>
Cc: James Morse <james.morse@arm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Peter Jones <pjones@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-efi@vger.kernel.org
Fixes: 38ac0287b7f4 ("fbdev/efifb: Honour UEFI memory map attributes when ...")
Link: http://lkml.kernel.org/r/20190516213159.3530-2-ard.biesheuvel@linaro.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
5 years agoriscv: fix locking violation in page fault handler
Andreas Schwab [Tue, 7 May 2019 07:36:46 +0000 (09:36 +0200)]
riscv: fix locking violation in page fault handler

When a user mode process accesses an address in the vmalloc area
do_page_fault tries to unlock the mmap semaphore when it isn't locked.

Signed-off-by: Andreas Schwab <schwab@suse.de>
[Palmer: Duplicated code instead of a goto]
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
5 years agoRISC-V: sifive_l2_cache: Add L2 cache controller driver for SiFive SoCs
Yash Shah [Mon, 6 May 2019 10:48:40 +0000 (16:18 +0530)]
RISC-V: sifive_l2_cache: Add L2 cache controller driver for SiFive SoCs

The driver currently supports only SiFive FU540-C000 platform.

The initial version of L2 cache controller driver includes:
- Initial configuration reporting at boot up.
- Support for ECC related functionality.

Signed-off-by: Yash Shah <yash.shah@sifive.com>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
5 years agoRISC-V: Add DT documentation for SiFive L2 Cache Controller
Yash Shah [Mon, 6 May 2019 10:48:39 +0000 (16:18 +0530)]
RISC-V: Add DT documentation for SiFive L2 Cache Controller

Add device tree bindings for SiFive FU540 L2 cache controller driver

Signed-off-by: Yash Shah <yash.shah@sifive.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
5 years agoRISC-V: Avoid using invalid intermediate translations
Palmer Dabbelt [Wed, 27 Mar 2019 00:40:24 +0000 (17:40 -0700)]
RISC-V: Avoid using invalid intermediate translations

This is almost entirely a comment.

Signed-off-by: Palmer Dabbelt <palmer@sifive.com>
Reviewed-by: Anup Patel <anup@brainfault.org>