drm/msm/dpu: Fix for non-visible planes

[ Upstream commit cb77085b1f0a86ef9dfba86b5f3ed6c3340c2ea3 ]

Fixes `kms_cursor_crc --run-subtest cursor-offscreen`.. when the cursor
moves offscreen the plane becomes non-visible, so we need to skip over
it in crtc atomic test and mixer setup.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Fixes: 25fdd5933e4c ("drm/msm: Add SDM845 DPU support")
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Abhinav Kumar <quic_abhinavk@quicinc.com>
Patchwork: https://patchwork.freedesktop.org/patch/492819/
Link: https://lore.kernel.org/r/20220707212003.1710163-1-robdclark@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/msm: Avoid dirtyfb stalls on video mode displays (v2)

[ Upstream commit 9e4dde28e9cd34ee13a6b7247f0857fb49fd3f19 ]

Someone on IRC once asked an innocent enough sounding question:  Why
with xf86-video-modesetting is es2gears limited at 120fps.

So I broke out the perfetto tracing mesa MR and took a look.  It turns
out the problem was drm_atomic_helper_dirtyfb(), which would end up
waiting for vblank.. es2gears would rapidly push two frames to Xorg,
which would blit them to screen and in idle hook (I assume) call the
DIRTYFB ioctl.  Which in turn would do an atomic update to flush the
dirty rects, which would stall until the next vblank.  And then the
whole process would repeat.

But this is a bit silly, we only need dirtyfb for command mode DSI
panels.  So track in plane state whether dirtyfb is required, and
track in the fb how many attached planes require dirtyfb so that we
can skip it when not required.  (Note, mdp4 does not actually have
cmd mode support.)

Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20220223191118.881321-1-robdclark@gmail.com
Signed-off-by: Rob Clark <robdclark@chromium.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: platform: mtk-mdp: Fix mdp_ipi_comm structure alignment

[ Upstream commit ab14c99c035da7156a3b66fa171171295bc4b89a ]

The mdp_ipi_comm structure defines a command that is either
PROCESS (start processing) or DEINIT (destroy instance); we
are using this one to send PROCESS or DEINIT commands from Linux
to an MDP instance through a VPU write but, while the first wants
us to stay 4-bytes aligned, the VPU instead requires an 8-bytes
data alignment.

Keeping in mind that these commands are executed immediately
after sending them (hence not chained with others before the
VPU/MDP "actually" start executing), it is fine to simply add
a padding of 4 bytes to this structure: this keeps the same
performance as before, as we're still stack-allocating it,
while avoiding hackery inside of mtk-vpu to ensure alignment
bringing a definitely bigger performance impact.

Fixes: c8eb2d7e8202 ("[media] media: Add Mediatek MDP Driver")
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Houlong Wei <houlong.wei@mediatek.com>
Reviewed-by: Irui Wang <irui.wang@mediatek.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: hisilicon - Kunpeng916 crypto driver don't sleep when in softirq

[ Upstream commit 68740ab505431f268dc1ee26a54b871e75f0ddaa ]

When kunpeng916 encryption driver is used to deencrypt and decrypt
packets during the softirq, it is not allowed to use mutex lock.

Fixes: 915e4e8413da ("crypto: hisilicon - SEC security accelerator driver")
Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: hisilicon/sec - don't sleep when in softirq

[ Upstream commit 02884a4f12de11f54d4ca67a07dd1f111d96fdbd ]

When kunpeng920 encryption driver is used to deencrypt and decrypt
packets during the softirq, it is not allowed to use mutex lock. The
kernel will report the following error:

BUG: scheduling while atomic: swapper/57/0/0x00000300
Call trace:
dump_backtrace+0x0/0x1e4
show_stack+0x20/0x2c
dump_stack+0xd8/0x140
__schedule_bug+0x68/0x80
__schedule+0x728/0x840
schedule+0x50/0xe0
schedule_preempt_disabled+0x18/0x24
__mutex_lock.constprop.0+0x594/0x5dc
__mutex_lock_slowpath+0x1c/0x30
mutex_lock+0x50/0x60
sec_request_init+0x8c/0x1a0 [hisi_sec2]
sec_process+0x28/0x1ac [hisi_sec2]
sec_skcipher_crypto+0xf4/0x1d4 [hisi_sec2]
sec_skcipher_encrypt+0x1c/0x30 [hisi_sec2]
crypto_skcipher_encrypt+0x2c/0x40
crypto_authenc_encrypt+0xc8/0xfc [authenc]
crypto_aead_encrypt+0x2c/0x40
echainiv_encrypt+0x144/0x1a0 [echainiv]
crypto_aead_encrypt+0x2c/0x40
esp_output_tail+0x348/0x5c0 [esp4]
esp_output+0x120/0x19c [esp4]
xfrm_output_one+0x25c/0x4d4
xfrm_output_resume+0x6c/0x1fc
xfrm_output+0xac/0x3c0
xfrm4_output+0x64/0x130
ip_build_and_send_pkt+0x158/0x20c
tcp_v4_send_synack+0xdc/0x1f0
tcp_conn_request+0x7d0/0x994
tcp_v4_conn_request+0x58/0x6c
tcp_v6_conn_request+0xf0/0x100
tcp_rcv_state_process+0x1cc/0xd60
tcp_v4_do_rcv+0x10c/0x250
tcp_v4_rcv+0xfc4/0x10a4
ip_protocol_deliver_rcu+0xf4/0x200
ip_local_deliver_finish+0x58/0x70
ip_local_deliver+0x68/0x120
ip_sublist_rcv_finish+0x70/0x94
ip_list_rcv_finish.constprop.0+0x17c/0x1d0
ip_sublist_rcv+0x40/0xb0
ip_list_rcv+0x140/0x1dc
__netif_receive_skb_list_core+0x154/0x28c
__netif_receive_skb_list+0x120/0x1a0
netif_receive_skb_list_internal+0xe4/0x1f0
napi_complete_done+0x70/0x1f0
gro_cell_poll+0x9c/0xb0
napi_poll+0xcc/0x264
net_rx_action+0xd4/0x21c
__do_softirq+0x130/0x358
irq_exit+0x11c/0x13c
__handle_domain_irq+0x88/0xf0
gic_handle_irq+0x78/0x2c0
el1_irq+0xb8/0x140
arch_cpu_idle+0x18/0x40
default_idle_call+0x5c/0x1c0
cpuidle_idle_call+0x174/0x1b0
do_idle+0xc8/0x160
cpu_startup_entry+0x30/0x11c
secondary_start_kernel+0x158/0x1e4
softirq: huh, entered softirq 3 NET_RX 0000000093774ee4 with
preempt_count 00000100, exited with fffffe00?

Fixes: 416d82204df4 ("crypto: hisilicon - add HiSilicon SEC V2 driver")
Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/msm/mdp5: Fix global state lock backoff

[ Upstream commit 92ef86ab513593c6329d04146e61f9a670e72fc5 ]

We need to grab the lock after the early return for !hwpipe case.
Otherwise, we could have hit contention yet still returned 0.

Fixes an issue that the new CONFIG_DRM_DEBUG_MODESET_LOCK stuff flagged
in CI:

   WARNING: CPU: 0 PID: 282 at drivers/gpu/drm/drm_modeset_lock.c:296 drm_modeset_lock+0xf8/0x154
   Modules linked in:
   CPU: 0 PID: 282 Comm: kms_cursor_lega Tainted: G        W         5.19.0-rc2-15930-g875cc8bc536a #1
   Hardware name: Qualcomm Technologies, Inc. DB820c (DT)
   pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
   pc : drm_modeset_lock+0xf8/0x154
   lr : drm_atomic_get_private_obj_state+0x84/0x170
   sp : ffff80000cfab6a0
   x29: ffff80000cfab6a0 x28: 0000000000000000 x27: ffff000083bc4d00
   x26: 0000000000000038 x25: 0000000000000000 x24: ffff80000957ca58
   x23: 0000000000000000 x22: ffff000081ace080 x21: 0000000000000001
   x20: ffff000081acec18 x19: ffff80000cfabb80 x18: 0000000000000038
   x17: 0000000000000000 x16: 0000000000000000 x15: fffffffffffea0d0
   x14: 0000000000000000 x13: 284e4f5f4e524157 x12: 5f534b434f4c5f47
   x11: ffff80000a386aa8 x10: 0000000000000029 x9 : ffff80000cfab610
   x8 : 0000000000000029 x7 : 0000000000000014 x6 : 0000000000000000
   x5 : 0000000000000001 x4 : ffff8000081ad904 x3 : 0000000000000029
   x2 : ffff0000801db4c0 x1 : ffff80000cfabb80 x0 : ffff000081aceb58
   Call trace:
    drm_modeset_lock+0xf8/0x154
    drm_atomic_get_private_obj_state+0x84/0x170
    mdp5_get_global_state+0x54/0x6c
    mdp5_pipe_release+0x2c/0xd4
    mdp5_plane_atomic_check+0x2ec/0x414
    drm_atomic_helper_check_planes+0xd8/0x210
    drm_atomic_helper_check+0x54/0xb0
    ...
   ---[ end trace 0000000000000000 ]---
   drm_modeset_lock attempting to lock a contended lock without backoff:
      drm_modeset_lock+0x148/0x154
      mdp5_get_global_state+0x30/0x6c
      mdp5_pipe_release+0x2c/0xd4
      mdp5_plane_atomic_check+0x290/0x414
      drm_atomic_helper_check_planes+0xd8/0x210
      drm_atomic_helper_check+0x54/0xb0
      drm_atomic_check_only+0x4b0/0x8f4
      drm_atomic_commit+0x68/0xe0

Fixes: d59be579fa93 ("drm/msm/mdp5: Return error code in mdp5_pipe_release when deadlock is detected")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Abhinav Kumar <quic_abhinavk@quicinc.com>
Patchwork: https://patchwork.freedesktop.org/patch/492701/
Link: https://lore.kernel.org/r/20220707162040.1594855-1-robdclark@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: hinic: avoid kernel hung in hinic_get_stats64()

[ Upstream commit 98f9fcdee35add80505b6c73f72de5f750d5c03c ]

When using hinic device as a bond slave device, and reading device stats
of master bond device, the kernel may hung.

The kernel panic calltrace as follows:
Kernel panic - not syncing: softlockup: hung tasks
Call trace:
  native_queued_spin_lock_slowpath+0x1ec/0x31c
  dev_get_stats+0x60/0xcc
  dev_seq_printf_stats+0x40/0x120
  dev_seq_show+0x1c/0x40
  seq_read_iter+0x3c8/0x4dc
  seq_read+0xe0/0x130
  proc_reg_read+0xa8/0xe0
  vfs_read+0xb0/0x1d4
  ksys_read+0x70/0xfc
  __arm64_sys_read+0x20/0x30
  el0_svc_common+0x88/0x234
  do_el0_svc+0x2c/0x90
  el0_svc+0x1c/0x30
  el0_sync_handler+0xa8/0xb0
  el0_sync+0x148/0x180

And the calltrace of task that actually caused kernel hungs as follows:
  __switch_to+124
  __schedule+548
  schedule+72
  schedule_timeout+348
  __down_common+188
  __down+24
  down+104
  hinic_get_stats64+44 [hinic]
  dev_get_stats+92
  bond_get_stats+172 [bonding]
  dev_get_stats+92
  dev_seq_printf_stats+60
  dev_seq_show+24
  seq_read_iter+964
  seq_read+220
  proc_reg_read+164
  vfs_read+172
  ksys_read+108
  __arm64_sys_read+28
  el0_svc_common+132
  do_el0_svc+40
  el0_svc+24
  el0_sync_handler+164
  el0_sync+324

When getting device stats from bond, kernel will call bond_get_stats().
It first holds the spinlock bond->stats_lock, and then call
hinic_get_stats64() to collect hinic device's stats.
However, hinic_get_stats64() calls `down(&nic_dev->mgmt_lock)` to
protect its critical section, which may schedule current task out.
And if system is under high pressure, the task cannot be woken up
immediately, which eventually triggers kernel hung panic.

Since previous patch has replaced hinic_dev.tx_stats/rx_stats with local
variable in hinic_get_stats64(), there is nothing need to be protected
by lock, so just removing down()/up() is ok.

Fixes: edd384f682cc ("net-next/hinic: Add ethtool and stats")
Signed-off-by: Qiao Ma <mqaio@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: hinic: fix bug that ethtool get wrong stats

[ Upstream commit 67dffd3db98570af8ff54c934f7d14664c0d182a ]

Function hinic_get_stats64() will do two operations:
1. reads stats from every hinic_rxq/txq and accumulates them
2. calls hinic_rxq/txq_clean_stats() to clean every rxq/txq's stats

For hinic_get_stats64(), it could get right data, because it sums all
data to nic_dev->rx_stats/tx_stats.
But it is wrong for get_drv_queue_stats(), this function will read
hinic_rxq's stats, which have been cleared to zero by hinic_get_stats64().

I have observed hinic's cleanup operation by using such command:
> watch -n 1 "cat ethtool -S eth4 | tail -40"

Result before:
     ...
     rxq7_pkts: 1
     rxq7_bytes: 90
     rxq7_errors: 0
     rxq7_csum_errors: 0
     rxq7_other_errors: 0
     ...
     rxq9_pkts: 11
     rxq9_bytes: 726
     rxq9_errors: 0
     rxq9_csum_errors: 0
     rxq9_other_errors: 0
     ...
     rxq11_pkts: 0
     rxq11_bytes: 0
     rxq11_errors: 0
     rxq11_csum_errors: 0
     rxq11_other_errors: 0

Result after a few seconds:
     ...
     rxq7_pkts: 0
     rxq7_bytes: 0
     rxq7_errors: 0
     rxq7_csum_errors: 0
     rxq7_other_errors: 0
     ...
     rxq9_pkts: 2
     rxq9_bytes: 132
     rxq9_errors: 0
     rxq9_csum_errors: 0
     rxq9_other_errors: 0
     ...
     rxq11_pkts: 1
     rxq11_bytes: 170
     rxq11_errors: 0
     rxq11_csum_errors: 0
     rxq11_other_errors: 0

To solve this problem, we just keep every queue's total stats in their own
queue (aka hinic_{rxq|txq}), and simply sum all per-queue stats every time
calling hinic_get_stats64().
With that solution, there is no need to clean per-queue stats now,
and there is no need to maintain global hinic_dev.{tx|rx}_stats, too.

Fixes: edd384f682cc ("net-next/hinic: Add ethtool and stats")
Signed-off-by: Qiao Ma <mqaio@linux.alibaba.com>
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

hinic: Use the bitmap API when applicable

[ Upstream commit 7c2c57263af41cfd8b5022274e6801542831bb69 ]

'vlan_bitmap' is a bitmap and is used as such. So allocate it with
devm_bitmap_zalloc() and its explicit bit size (i.e. VLAN_N_VID).

This avoids the need of the VLAN_BITMAP_SIZE macro which:
   - needlessly has a 'nic_dev' parameter
   - should be "long" (and not byte) aligned, so that the bitmap semantic
     is respected

This is in fact not an issue because VLAN_N_VID is 4096 at the time
being, but devm_bitmap_zalloc() is less verbose and easier to understand.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/6ff7b7d21414240794a77dc2456914412718a145.1656260842.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm: bridge: sii8620: fix possible off-by-one

[ Upstream commit 21779cc21c732c5eff8ea1624be6590450baa30f ]

The next call to sii8620_burst_get_tx_buf will result in off-by-one
When ctx->burst.tx_count + size == ARRAY_SIZE(ctx->burst.tx_buf). The same
thing happens in sii8620_burst_get_rx_buf.

This patch also change tx_count and tx_buf to rx_count and rx_buf in
sii8620_burst_get_rx_buf. It is unreasonable to check tx_buf's size and
use rx_buf.

Fixes: e19e9c692f81 ("drm/bridge/sii8620: add support for burst eMSC transmissions")
Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220518065856.18936-1-hbh25y@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/mediatek: dpi: Only enable dpi after the bridge is enabled

[ Upstream commit aed61ef6beb911cc043af0f2f291167663995065 ]

Enabling the dpi too early causes glitches on screen.

Move the call to mtk_dpi_enable() at the end of the bridge_enable
callback to ensure everything is setup properly before enabling dpi.

Fixes: 9e629c17aa8d ("drm/mediatek: Add DPI sub driver")
Signed-off-by: Guillaume Ranquet <granquet@baylibre.com>
Signed-off-by: Bo-Chen Chen <rex-bc.chen@mediatek.com>
Tested-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://patchwork.kernel.org/project/linux-mediatek/patch/20220701035845.16458-16-rex-bc.chen@mediatek.com/
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/mediatek: dpi: Remove output format of YUV

[ Upstream commit c9ed0713b3c35fc45677707ba47f432cad95da56 ]

DPI is not support output format as YUV, but there is the setting of
configuring output YUV. Therefore, remove them in this patch.

Fixes: 9e629c17aa8d ("drm/mediatek: Add DPI sub driver")
Signed-off-by: Bo-Chen Chen <rex-bc.chen@mediatek.com>
Link: https://patchwork.kernel.org/project/linux-mediatek/patch/20220701035845.16458-5-rex-bc.chen@mediatek.com/
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/rockchip: Fix an error handling path rockchip_dp_probe()

[ Upstream commit 5074376822fe99fa4ce344b851c5016d00c0444f ]

Should component_add() fail, we should call analogix_dp_remove() in the
error handling path, as already done in the remove function.

Fixes: 152cce0006ab ("drm/bridge: analogix_dp: Split bind() into probe() and real bind()")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/b719d9061bb97eb85145fbd3c5e63f4549f2e13e.1655572071.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/rockchip: vop: Don't crash for invalid duplicate_state()

[ Upstream commit 1449110b0dade8b638d2c17ab7c5b0ff696bfccb ]

It's possible for users to try to duplicate the CRTC state even when the
state doesn't exist. drm_atomic_helper_crtc_duplicate_state() (and other
users of __drm_atomic_helper_crtc_duplicate_state()) already guard this
with a WARN_ON() instead of crashing, so let's do that here too.

Fixes: 4e257d9eee23 ("drm/rockchip: get rid of rockchip_drm_crtc_mode_config")
Signed-off-by: Brian Norris <briannorris@chromium.org>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20220617172623.1.I62db228170b1559ada60b8d3e1637e1688424926@changeid
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftests/xsk: Destroy BPF resources only when ctx refcount drops to 0

[ Upstream commit 39e940d4abfabb08b6937a315546b24d10be67e3 ]

Currently, xsk_socket__delete frees BPF resources regardless of ctx
refcount. Xdpxceiver has a test to verify whether underlying BPF
resources would not be wiped out after closing XSK socket that was
bound to interface with other active sockets. From library's xsk part
perspective it also means that the internal xsk context is shared and
its refcount is bumped accordingly.

After a switch to loading XDP prog based on previously opened XSK
socket, mentioned xdpxceiver test fails with:

  not ok 16 [xdpxceiver.c:swap_xsk_resources:1334]: ERROR: 9/"Bad file descriptor

which means that in swap_xsk_resources(), xsk_socket__delete() released
xskmap which in turn caused a failure of xsk_socket__update_xskmap().

To fix this, when deleting socket, decrement ctx refcount before
releasing BPF resources and do so only when refcount dropped to 0 which
means there are no more active sockets for this ctx so BPF resources can
be freed safely.

Fixes: 2f6324a3937f ("libbpf: Support shared umems between queues and devices")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Magnus Karlsson <magnus.karlsson@intel.com>
Link: https://lore.kernel.org/bpf/20220629143458.934337-5-maciej.fijalkowski@intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: arm64/gcm - Select AEAD for GHASH_ARM64_CE

[ Upstream commit fac76f2260893dde5aa05bb693b4c13e8ed0454b ]

Otherwise, we could fail to compile.

ld: arch/arm64/crypto/ghash-ce-glue.o: in function 'ghash_ce_mod_exit':
ghash-ce-glue.c:(.exit.text+0x24): undefined reference to 'crypto_unregister_aead'
ld: arch/arm64/crypto/ghash-ce-glue.o: in function 'ghash_ce_mod_init':
ghash-ce-glue.c:(.init.text+0x34): undefined reference to 'crypto_register_aead'

Fixes: 537c1445ab0b ("crypto: arm64/gcm - implement native driver using v8 Crypto Extensions")
Signed-off-by: Qian Cai <quic_qiancai@quicinc.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: hdmi: Correct HDMI timing registers for interlaced modes

[ Upstream commit fb10dc451c0f15e3c19798a2f41d357f3f7576f5 ]

For interlaced modes the timings were not being correctly
programmed into the HDMI block, so correct them.

Fixes: 8323989140f3 ("drm/vc4: hdmi: Support the BCM2711 HDMI controllers")
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://lore.kernel.org/r/20220613144800.326124-33-maxime@cerno.tech
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: hdmi: Fix timings for interlaced modes

[ Upstream commit 0ee5a40152b15f200ed3a0d51e8aa782ea979c6a ]

Increase the number of post-sync blanking lines on odd fields instead of
decreasing it on even fields. This makes the total number of lines
properly match the modelines.

Additionally fix the value of PV_VCONTROL_ODD_DELAY, which did not take
pixels_per_clock into account, causing some displays to invert the
fields when driven by bcm2711.

Fixes: 682e62c45406 ("drm/vc4: Fix support for interlaced modes on HDMI.")
Signed-off-by: Mateusz Kwiatkowski <kfyatek+publicgit@gmail.com>
Link: https://lore.kernel.org/r/20220613144800.326124-31-maxime@cerno.tech
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: hdmi: Reset HDMI MISC_CONTROL register

[ Upstream commit 35dc00c12a72700a9c4592afee7d136ecb280cbd ]

The HDMI block can repeat pixels for double clocked modes,
and the firmware is now configuring the block to do this as
the PV is doing it incorrectly when at 2pixels/clock.
If the kernel doesn't reset it then we end up with strange
modes.

Reset MISC_CONTROL.

Fixes: 8323989140f3 ("drm/vc4: hdmi: Support the BCM2711 HDMI controllers")
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://lore.kernel.org/r/20220613144800.326124-22-maxime@cerno.tech
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: hdmi: Avoid full hdmi audio fifo writes

[ Upstream commit 1c594eeccf92368177c2e22f1d3ee4933dfb8567 ]

We are getting occasional VC4_HD_MAI_CTL_ERRORF in
HDMI_MAI_CTL which seem to correspond with audio dropouts.

Reduce the threshold where we deassert DREQ to avoid the fifo
overfilling

Fixes: bb7d78568814 ("drm/vc4: Add HDMI audio support")
Signed-off-by: Dom Cobley <popcornmix@gmail.com>
Link: https://lore.kernel.org/r/20220613144800.326124-21-maxime@cerno.tech
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: hdmi: Fix HPD GPIO detection

[ Upstream commit e32e5723256a99c5324824503572f743377dd0fe ]

Prior to commit 6800234ceee0 ("drm/vc4: hdmi: Convert to gpiod"), in the
detect hook, if we had an HPD GPIO we would only rely on it and return
whatever state it was in.

However, that commit changed that by mistake to only consider the case
where we have a GPIO and it returns a logical high, and would fall back
to the other methods otherwise.

Since we can read the EDIDs when the HPD signal is low on some displays,
we changed the detection status from disconnected to connected, and we
would ignore an HPD pulse.

Fixes: 6800234ceee0 ("drm/vc4: hdmi: Convert to gpiod")
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://lore.kernel.org/r/20211025152903.1088803-3-maxime@cerno.tech
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: dsi: Add correct stop condition to vc4_dsi_encoder_disable iteration

[ Upstream commit 7bcb9c8d0bc9f3cab8ac2634b056c2e6b63945ca ]

vc4_dsi_encoder_disable is partially an open coded version of
drm_bridge_chain_disable, but it missed a termination condition
in the loop for ->disable which meant that no post_disable
calls were made.

Add in the termination clause.

Fixes: 033bfe7538a1 ("drm/vc4: dsi: Fix bridge chain handling")
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://lore.kernel.org/r/20220613144800.326124-17-maxime@cerno.tech
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: dsi: Fix dsi0 interrupt support

[ Upstream commit bc5b815e06f90cccdb6461aba1e49fdc2f3c8cd1 ]

DSI0 seemingly had very little or no testing as a load of
the register mappings were incorrect/missing, so host
transfers always timed out due to enabling/checking incorrect
bits in the interrupt enable and status registers.

Fixes: 4078f5757144 ("drm/vc4: Add DSI driver")
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://lore.kernel.org/r/20220613144800.326124-16-maxime@cerno.tech
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: dsi: Register dsi0 as the correct vc4 encoder type

[ Upstream commit 4d9273c978d4c1af15d7874c10c732ec83d444d0 ]

vc4_dsi was registering both dsi0 and dsi1 as VC4_ENCODER_TYPE_DSI1
which seemed to work OK for a single DSI display, but fails
if there are two DSI displays connected.

Update to register the correct type.

Fixes: 4078f5757144 ("drm/vc4: Add DSI driver")
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://lore.kernel.org/r/20220613144800.326124-15-maxime@cerno.tech
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: dsi: Correct pixel order for DSI0

[ Upstream commit edfe84ae0df16be1251b5a8e840d95f1f3827500 ]

For slightly unknown reasons, dsi0 takes a different pixel format
to dsi1, and that has to be set in the pixel valve.

Amend the setup accordingly.

Fixes: a86773d120d7 ("drm/vc4: Add support for feeding DSI encoders from the pixel valve.")
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://lore.kernel.org/r/20220613144800.326124-14-maxime@cerno.tech
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: dsi: Correct DSI divider calculations

[ Upstream commit 3b45eee87da171caa28f61240ddb5c21170cda53 ]

The divider calculations tried to find the divider just faster than the
clock requested. However if it required a divider of 7 then the for loop
aborted without handling the "error" case, and could end up with a clock
lower than requested.

The integer divider from parent PLL to DSI clock is also capable of
going up to /255, not just /7 that the driver was trying.  This allows
for slower link frequencies on the DSI bus where the resolution permits.

Correct the loop so that we always have a clock greater than requested,
and covering the whole range of dividers.

Fixes: 86c1b9eff3f2 ("drm/vc4: Adjust modes in DSI to work around the integer PLL divider.")
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://lore.kernel.org/r/20220613144800.326124-13-maxime@cerno.tech
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: dsi: Release workaround buffer and DMA

[ Upstream commit 89c4bbe2a01ea401c2b0fabc104720809084b77f ]

On Pi0-3 the driver allocates a buffer and requests a DMA channel
because the ARM can't write to DSI1's registers directly.

However, we never release that buffer or channel. Let's add a
device-managed action to release each.

Fixes: 4078f5757144 ("drm/vc4: Add DSI driver")
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://lore.kernel.org/r/20220613144800.326124-12-maxime@cerno.tech
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: Use of_device_get_match_data()

[ Upstream commit 9cbe89ede58294d23af06ec12c20f2ce6acc1892 ]

Use of_device_get_match_data() to simplify the code.

Reported-by: Zeal Robot <zealci@zte.com.cn>
Signed-off-by: Minghao Chi (CGEL ZTE) <chi.minghao@zte.com.cn>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20220214020530.1714631-1-chi.minghao@zte.com.cn
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: dsi: Switch to devm_drm_of_get_bridge

[ Upstream commit a43dd76bacd0d5441a4c84f60d64bdfaedc95bac ]

The new devm_drm_of_get_bridge removes most of the boilerplate we
have to deal with. Let's switch to it.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210910130941.1740182-4-maxime@cerno.tech
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/bridge: Add a function to abstract away panels

[ Upstream commit 87ea95808d53e56b03e620e8f8f3add48899a88d ]

Display drivers so far need to have a lot of boilerplate to first
retrieve either the panel or bridge that they are connected to using
drm_of_find_panel_or_bridge(), and then either deal with each with ad-hoc
functions or create a drm panel bridge through drm_panel_bridge_add.

In order to reduce the boilerplate and hopefully create a path of least
resistance towards using the DRM panel bridge layer, let's create the
function devm_drm_of_get_bridge() to reduce that boilerplate.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210910130941.1740182-2-maxime@cerno.tech
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: plane: Fix margin calculations for the right/bottom edges

[ Upstream commit b7c3d6821627861f4ea3e1f2b595d0ed9e80aac8 ]

The current plane margin calculation code clips the right and bottom
edges of the range based using the left and top margins.

This is obviously wrong, so let's fix it.

Fixes: 666e73587f90 ("drm/vc4: Take margin setup into account when updating planes")
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Link: https://lore.kernel.org/r/20220613144800.326124-6-maxime@cerno.tech
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: plane: Remove subpixel positioning check

[ Upstream commit 517db1ab1566dba3093dbdb8de4263ba4aa66416 ]

There is little harm in ignoring fractional coordinates
(they just get truncated).

Without this:
modetest -M vc4 -F tiles,gradient -s 32:1920x1080-60 -P89@74:1920x1080*.1.1@XR24

is rejected. We have the same issue in Kodi when trying to
use zoom options on video.

Note: even if all coordinates are fully integer. e.g.
src:[0,0,1920,1080] dest:[-10,-10,1940,1100]

it will still get rejected as drm_atomic_helper_check_plane_state
uses drm_rect_clip_scaled which transforms this to fractional src coords

Fixes: 21af94cf1a4c ("drm/vc4: Add support for scaling of display planes.")
Signed-off-by: Dom Cobley <popcornmix@gmail.com>
Link: https://lore.kernel.org/r/20220613144800.326124-5-maxime@cerno.tech
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: tw686x: Fix memory leak in tw686x_video_init

[ Upstream commit e0b212ec9d8177d6f7c404315293f6a085d6ee42 ]

video_device_alloc() allocates memory for vdev,
when video_register_device() fails, it doesn't release the memory and
leads to memory leak, call video_device_release() to fix this.

Fixes: 704a84ccdbf1 ("[media] media: Support Intersil/Techwell TW686x-based video capture cards")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: driver/nxp/imx-jpeg: fix a unexpected return value problem

[ Upstream commit 5b304046a81eda221b5d06a9c62f7b5e45530fa5 ]

In function mxc_jpeg_probe(), when devm_clk_get() fail, the return value
will be unexpected, and it should be the devm_clk_get's error code.

Fixes: 4c2e5156d9fa6 ("media: imx-jpeg: Add pm-runtime support for imx-jpeg")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Jian Zhang <zhangjian210@huawei.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: v4l2-mem2mem: prevent pollerr when last_buffer_dequeued is set

[ Upstream commit d4de27a9b1eadd33a2e40de87a646d1bf5fef756 ]

If the last buffer was dequeued from the capture queue,
signal userspace. DQBUF(CAPTURE) will return -EPIPE.

But if output queue is empty and capture queue is empty,
v4l2_m2m_poll_for_data will return EPOLLERR,
This is very easy to happen in drain.

When last_buffer_dequeued is set, we shouldn't return EPOLLERR,
but return EPOLLIN | EPOLLRDNORM.

Fixes: 1698a7f151126 ("media: v4l2-mem2mem: simplify poll logic")
Signed-off-by: Ming Qian <ming.qian@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: hdpvr: fix error value returns in hdpvr_read

[ Upstream commit 359c27c6ddbde404f44a9c0d3ec88ccd1e2042f2 ]

Error return values are supposed to be negative in hdpvr_read. Most
error returns are currently handled via an unsigned integer "ret". When
setting a negative error value to "ret", the value actually becomes a
large positive value, because "ret" is unsigned. Later on, the "ret"
value is returned. But as ssize_t is a 64-bit signed number, the error
return value stays a large positive integer instead of a negative
integer. This can cause an error value to be interpreted as the read
size, which can cause a buffer overread for applications relying on the
returned size.

Fixes: 9aba42efe85b ("V4L/DVB (11096): V4L2 Driver for the Hauppauge HD PVR usb capture device")
Signed-off-by: Niels Dossche <dossche.niels@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/mcde: Fix refcount leak in mcde_dsi_bind

[ Upstream commit 3a149169e4a2f9127022fec6ef5d71b4e804b3b9 ]

Every iteration of for_each_available_child_of_node() decrements
the reference counter of the previous node. There is no decrement
when break out from the loop and results in refcount leak.
Add missing of_node_put() to fix this.

Fixes: 5fc537bfd000 ("drm/mcde: Add new driver for ST-Ericsson MCDE")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220525115411.65455-1-linmq006@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: imx-jpeg: Disable slot interrupt when frame done

[ Upstream commit 22a2bc88c139dc9757bdb1d0a3665ac27edc79a5 ]

The interrupt STMBUF_HALF may be triggered after frame done.
It may led to system hang if driver try to access the register after
power off.

Disable the slot interrupt when frame done.

Fixes: 2db16c6ed72ce ("media: imx-jpeg: Add V4L2 driver for i.MX8 JPEG Encoder/Decoder")
Signed-off-by: Ming Qian <ming.qian@nxp.com>
Reviewed-by: Mirela Rabulea <mirela.rabulea@nxp.com>
Tested-by: Mirela Rabulea <mirela.rabulea@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm: bridge: adv7511: Add check for mipi_dsi_driver_register

[ Upstream commit 831463667b5f4f1e5bce9c3b94e9e794d2bc8923 ]

As mipi_dsi_driver_register could return error if fails,
it should be better to check the return value and return error
if fails.
Moreover, if i2c_add_driver fails,  mipi_dsi_driver_register
should be reverted.

Fixes: 1e4d58cd7f88 ("drm/bridge: adv7533: Create a MIPI DSI device")
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220602103401.2980938-1-jiasheng@iscas.ac.cn
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: ccp - During shutdown, check SEV data pointer before using

[ Upstream commit 1b05ece0c931536c0a38a9385e243a7962e933f6 ]

On shutdown, each CCP device instance performs shutdown processing.
However, __sev_platform_shutdown_locked() uses the controlling psp
structure to obtain the pointer to the sev_device structure. However,
during driver initialization, it is possible that an error can be received
from the firmware that results in the sev_data pointer being cleared from
the controlling psp structure. The __sev_platform_shutdown_locked()
function does not check for this situation and will segfault.

While not common, this scenario should be accounted for. Add a check for a
NULL sev_device structure before attempting to use it.

Fixes: 5441a07a127f ("crypto: ccp - shutdown SEV firmware on kexec")
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

test_bpf: fix incorrect netdev features

[ Upstream commit 9676feccacdb0571791c88b23e3b7ac4e7c9c457 ]

The prototype of .features is netdev_features_t, it should use
NETIF_F_LLTX and NETIF_F_HW_VLAN_STAG_TX, not NETIF_F_LLTX_BIT
and NETIF_F_HW_VLAN_STAG_TX_BIT.

Fixes: cf204a718357 ("bpf, testing: Introduce 'gso_linear_no_head_frag' skb_segment test")
Signed-off-by: Jian Shen <shenjian15@huawei.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/r/20220622135002.8263-1-shenjian15@huawei.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rcutorture: Fix ksoftirqd boosting timing and iteration

[ Upstream commit 3002153a91a9732a6d1d0bb95138593c7da15743 ]

The RCU priority boosting can fail in two situations:

1) If (nr_cpus= > maxcpus=), which means if the total number of CPUs
is higher than those brought online at boot, then torture_onoff() may
later bring up CPUs that weren't online on boot. Now since rcutorture
initialization only boosts the ksoftirqds of the CPUs that have been
set online on boot, the CPUs later set online by torture_onoff won't
benefit from the boost, making RCU priority boosting fail.

2) The ksoftirqd kthreads are boosted after the creation of
rcu_torture_boost() kthreads, which opens a window large enough for these
rcu_torture_boost() kthreads to wait (despite running at FIFO priority)
for ksoftirqds that are still running at SCHED_NORMAL priority.

The issues can trigger for example with:

./kvm.sh --configs TREE01 --kconfig "CONFIG_RCU_BOOST=y"

[   34.968561] rcu-torture: !!!
[   34.968627] ------------[ cut here ]------------
[   35.014054] WARNING: CPU: 4 PID: 114 at kernel/rcu/rcutorture.c:1979 rcu_torture_stats_print+0x5ad/0x610
[   35.052043] Modules linked in:
[   35.069138] CPU: 4 PID: 114 Comm: rcu_torture_sta Not tainted 5.18.0-rc1 #1
[   35.096424] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.14.0-0-g155821a-rebuilt.opensuse.org 04/01/2014
[   35.154570] RIP: 0010:rcu_torture_stats_print+0x5ad/0x610
[   35.198527] Code: 63 1b 02 00 74 02 0f 0b 48 83 3d 35 63 1b 02 00 74 02 0f 0b 48 83 3d 21 63 1b 02 00 74 02 0f 0b 48 83 3d 0d 63 1b 02 00 74 02 <0f> 0b 83 eb 01 0f 8e ba fc ff ff 0f 0b e9 b3 fc ff f82
[   37.251049] RSP: 0000:ffffa92a0050bdf8 EFLAGS: 00010202
[   37.277320] rcu: De-offloading 8
[   37.290367] RAX: 0000000000000000 RBX: 0000000000000001 RCX: 0000000000000001
[   37.290387] RDX: 0000000000000000 RSI: 00000000ffffbfff RDI: 00000000ffffffff
[   37.290398] RBP: 000000000000007b R08: 0000000000000000 R09: c0000000ffffbfff
[   37.290407] R10: 000000000000002a R11: ffffa92a0050bc18 R12: ffffa92a0050be20
[   37.290417] R13: ffffa92a0050be78 R14: 0000000000000000 R15: 000000000001bea0
[   37.290427] FS:  0000000000000000(0000) GS:ffff96045eb00000(0000) knlGS:0000000000000000
[   37.290448] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   37.290460] CR2: 0000000000000000 CR3: 000000001dc0c000 CR4: 00000000000006e0
[   37.290470] Call Trace:
[   37.295049]  <TASK>
[   37.295065]  ? preempt_count_add+0x63/0x90
[   37.295095]  ? _raw_spin_lock_irqsave+0x12/0x40
[   37.295125]  ? rcu_torture_stats_print+0x610/0x610
[   37.295143]  rcu_torture_stats+0x29/0x70
[   37.295160]  kthread+0xe3/0x110
[   37.295176]  ? kthread_complete_and_exit+0x20/0x20
[   37.295193]  ret_from_fork+0x22/0x30
[   37.295218]  </TASK>

Fix this with boosting the ksoftirqds kthreads from the boosting
hotplug callback itself and before the boosting kthreads are created.

Fixes: ea6d962e80b6 ("rcutorture: Judge RCU priority boosting on grace periods, not callbacks")
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rcutorture: Don't cpuhp_remove_state() if cpuhp_setup_state() failed

[ Upstream commit fd13fe16db0d82612b260640f4e26f6d9d1e11fd ]

Currently, in CONFIG_RCU_BOOST kernels, if the rcu_torture_init()
function's call to cpuhp_setup_state() fails, rcu_torture_cleanup()
gamely passes nonsense to cpuhp_remove_state().  This results in
strange and misleading splats.  This commit therefore ensures that if
the rcu_torture_init() function's call to cpuhp_setup_state() fails,
rcu_torture_cleanup() avoids invoking cpuhp_remove_state().

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rcutorture: Warn on individual rcu_torture_init() error conditions

[ Upstream commit efeff6b39b9de4480572c7b0c5eb77204795cb57 ]

When running rcutorture as a module, any rcu_torture_init() issues will be
reflected in the error code from modprobe or insmod, as the case may be.
However, these error codes are not available when running rcutorture
built-in, for example, when using the kvm.sh script.  This commit
therefore adds WARN_ON_ONCE() to allow distinguishing rcu_torture_init()
errors when running rcutorture built-in.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/radeon: fix incorrrect SPDX-License-Identifiers

[ Upstream commit 1f43b8903f3aae4a26a603c36f6d5dd25d6edb51 ]

radeon is MIT.  This were incorrectly changed in
commit b24413180f56 ("License cleanup: add SPDX GPL-2.0 license identifier to files with no license")
and
commit d198b34f3855 (".gitignore: add SPDX License Identifier")
and:
commit ec8f24b7faaf ("treewide: Add SPDX license identifier - Makefile/Kconfig")

Fixes: d198b34f3855 (".gitignore: add SPDX License Identifier")
Fixes: ec8f24b7faaf ("treewide: Add SPDX license identifier - Makefile/Kconfig")
Fixes: b24413180f56 ("License cleanup: add SPDX GPL-2.0 license identifier to files with no license")
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2053
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: iwlegacy: 4965: fix potential off-by-one overflow in il4965_rs_fill_link_cmd()

[ Upstream commit a8eb8e6f7159c7c20c0ddac428bde3d110890aa7 ]

As a result of the execution of the inner while loop, the value
of 'idx' can be equal to LINK_QUAL_MAX_RETRY_NUM. However, this
is not checked after the loop and 'idx' is used to write the
LINK_QUAL_MAX_RETRY_NUM size array 'lq_cmd->rs_table[idx]' below
in the outer loop.

The fix is to check the new value of 'idx' inside the nested loop,
and break both loops if index equals the size. Checking it at the
start is now pointless, so let's remove it.

Detected using the static analysis tool - Svace.

Fixes: be663ab67077 ("iwlwifi: split the drivers for agn and legacy devices 3945/4965")
Signed-off-by: Alexey Kodanev <aleksei.kodanev@bell-sw.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220608171614.28891-1-aleksei.kodanev@bell-sw.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

ath9k: fix use-after-free in ath9k_hif_usb_rx_cb

[ Upstream commit 0ac4827f78c7ffe8eef074bc010e7e34bc22f533 ]

Syzbot reported use-after-free Read in ath9k_hif_usb_rx_cb() [0]. The
problem was in incorrect htc_handle->drv_priv initialization.

Probable call trace which can trigger use-after-free:

ath9k_htc_probe_device()
  /* htc_handle->drv_priv = priv; */
  ath9k_htc_wait_for_target()      <--- Failed
  ieee80211_free_hw()    <--- priv pointer is freed

<IRQ>
...
ath9k_hif_usb_rx_cb()
  ath9k_hif_usb_rx_stream()
   RX_STAT_INC() <--- htc_handle->drv_priv access

In order to not add fancy protection for drv_priv we can move
htc_handle->drv_priv initialization at the end of the
ath9k_htc_probe_device() and add helper macro to make
all *_STAT_* macros NULL safe, since syzbot has reported related NULL
deref in that macros [1]

Link: https://syzkaller.appspot.com/bug?id=6ead44e37afb6866ac0c7dd121b4ce07cb665f60
Link: https://syzkaller.appspot.com/bug?id=b8101ffcec107c0567a0cd8acbbacec91e9ee8de
Fixes: fb9987d0f748 ("ath9k_htc: Support for AR9271 chipset.")
Reported-and-tested-by: syzbot+03110230a11411024147@syzkaller.appspotmail.com
Reported-and-tested-by: syzbot+c6dde1f690b60e0b9fbe@syzkaller.appspotmail.com
Signed-off-by: Pavel Skripkin <paskripkin@gmail.com>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/d57bbedc857950659bfacac0ab48790c1eda00c8.1655145743.git.paskripkin@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: imx-jpeg: Implement drain using v4l2-mem2mem helpers

[ Upstream commit 4911c5acf9351c4caf692895c7cf6a4fa46c26b0 ]

v4l2 m2m has supplied some helper function to handle drain,
so the driver can use the helper function directly.

Fixes: d8ebe298d008c ("media: imx-jpeg: Set V4L2_BUF_FLAG_LAST at eos")
Signed-off-by: Ming Qian <ming.qian@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: imx-jpeg: Align upwards buffer size

[ Upstream commit 9e7aa76cdb02923ee23a0ddd48f38bdc3512f92b ]

The hardware can support any image size WxH,
with arbitrary W (image width) and H (image height) dimensions.

Align upwards buffer size for both encoder and decoder.
and leave the picture resolution unchanged.

For decoder, the risk of memory out of bounds can be avoided.
For both encoder and decoder, the driver will lift the limitation of
resolution alignment.

For example, the decoder can support jpeg whose resolution is 227x149
the encoder can support nv12 1080P, won't change it to 1920x1072.

Fixes: 2db16c6ed72ce ("media: imx-jpeg: Add V4L2 driver for i.MX8 JPEG Encoder/Decoder")
Signed-off-by: Ming Qian <ming.qian@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: imx-jpeg: Support dynamic resolution change

[ Upstream commit b4e1fb8643daabba850e97df532191acffc23e6a ]

To support dynamic resolution change,
driver should meet the following conditions:
1. the previous pictures are all decoded before source change event.
2. prevent decoding new resolution pictures with incorrect capture
   buffer, until user handle source change event and setup capture.
3. report correct fmt and resolution during source change.

Signed-off-by: Ming Qian <ming.qian@nxp.com>
Reviewed-by: Mirela Rabulea <mirela.rabulea@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: imx-jpeg: Handle source change in a function

[ Upstream commit 831f87424dd3973612782983ef7352789795b4df ]

Refine code to support dynamic resolution change

Signed-off-by: Ming Qian <ming.qian@nxp.com>
Reviewed-by: Mirela Rabulea <mirela.rabulea@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: imx-jpeg: Identify and handle precision correctly

[ Upstream commit bec0a3a67389ede106d0661a007edf832878d8b2 ]

The decoder will save the precision that was detected from jpeg header
and use it later, when choosing the pixel format and also calculate
bytesperline according to precision.

The 12bit jpeg is not supported yet,
but driver shouldn't led to serious problem if user enqueue a 12 bit jpeg.
And the 12bit jpeg is supported by hardware, driver may support it later.

[hverkuil: document the new precision field]

Signed-off-by: Ming Qian <ming.qian@nxp.com>
Reviewed-by: Mirela Rabulea <mirela.rabulea@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: imx-jpeg: Refactor function mxc_jpeg_parse

[ Upstream commit 8dd504a3a0a5f73b4c137ce3afc35936a4ecd871 ]

Refine code to support dynamic resolution change

Signed-off-by: Ming Qian <ming.qian@nxp.com>
Reviewed-by: Mirela Rabulea <mirela.rabulea@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: imx-jpeg: Set V4L2_BUF_FLAG_LAST at eos

[ Upstream commit d8ebe298d008ccbae3011cbeb139707f01a730c8 ]

The V4L2_EVENT_EOS event is a deprecated behavior,
the V4L2_BUF_FLAG_LAST buffer flag should be used instead.

Signed-off-by: Ming Qian <ming.qian@nxp.com>
Reviewed-by: Mirela Rabulea <mirela.rabulea@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: imx-jpeg: use NV12M to represent non contiguous NV12

[ Upstream commit 784a1883cff07e7510a81ad3041d6ec443d51944 ]

V4L2_PIX_FMT_NV12 requires num_planes equals to 1,
V4L2_PIX_FMT_NV12M requires num_planes equals to 2.
and mxc-jpeg supports 2 planes for nv12,
so we should use 4L2_PIX_FMT_NV12M instead of V4L2_PIX_FMT_NV12,
otherwise it will confuses gstreamer and prevent encoding and decoding.

Signed-off-by: Ming Qian <ming.qian@nxp.com>
Signed-off-by: Shijie Qin <shijie.qin@nxp.com>
Signed-off-by: Zhou Peng <eagle.zhou@nxp.com>
Reviewed-by: Mirela Rabulea <mirela.rabulea@oss.nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: imx-jpeg: Add pm-runtime support for imx-jpeg

[ Upstream commit 4c2e5156d9fa63a3f41c2bf56b694ad42df825d7 ]

Save some power by disabling/enabling the jpeg clocks with
every stream stop/start.
Do not use DL_FLAG_RPM_ACTIVE in mxc_jpeg_attach_pm_domains,
to ensure power domains are off after probe.

Signed-off-by: Mirela Rabulea <mirela.rabulea@oss.nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: imx-jpeg: Leave a blank space before the configuration data

[ Upstream commit 6285cdea19daf764bf00f662a59fc83ef67345cf ]

There is a hardware bug that it will load
the first 128 bytes of configuration data twice,
it will led to some configure error.
so shift the configuration data 128 bytes,
and make the first 128 bytes all zero,
then hardware will load the 128 zero twice,
and ignore them as garbage.
then the configuration data can be loaded correctly

Fixes: 2db16c6ed72ce ("media: imx-jpeg: Add V4L2 driver for i.MX8 JPEG Encoder/Decoder")
Signed-off-by: Ming Qian <ming.qian@nxp.com>
Reviewed-by: Mirela Rabulea <mirela.rabulea@nxp.com>
Reviewed-by: Tommaso Merciai <tommaso.merciai@amarulasolutions.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: imx-jpeg: Correct some definition according specification

[ Upstream commit 5a601f89e846c1b6005ab274d039e5036fc22015 ]

the register CAST_NOMFRSIZE_LO should be equal to CAST_STATUS16
the register CAST_NOMFRSIZE_HI should be equal to CAST_STATUS17
the register CAST_OFBSIZE_LO should be equal to CAST_STATUS18
the register CAST_OFBSIZE_HI should be equal to CAST_STATUS19

Fixes: 2db16c6ed72ce ("media: imx-jpeg: Add V4L2 driver for i.MX8 JPEG Encoder/Decoder")
Signed-off-by: Ming Qian <ming.qian@nxp.com>
Reviewed-by: Mirela Rabulea <mirela.rabulea@nxp.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: tw686x: Register the irq at the end of probe

[ Upstream commit fb730334e0f759d00f72168fbc555e5a95e35210 ]

We got the following warning when booting the kernel:

[    3.243674] INFO: trying to register non-static key.
[    3.243922] The code is fine but needs lockdep annotation, or maybe
[    3.244230] you didn't initialize this object before use?
[    3.245642] Call Trace:
[    3.247836]  lock_acquire+0xff/0x2d0
[    3.248727]  tw686x_audio_irq+0x1a5/0xcc0 [tw686x]
[    3.249211]  tw686x_irq+0x1f9/0x480 [tw686x]

The lock 'vc->qlock' will be initialized in tw686x_video_init(), but the
driver registers the irq before calling the tw686x_video_init(), and we
got the warning.

Fix this by registering the irq at the end of probe

Fixes: 704a84ccdbf1 ("[media] media: Support Intersil/Techwell TW686x-based video capture cards")
Signed-off-by: Zheyu Ma <zheyuma97@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: atmel: atmel-sama7g5-isc: fix warning in configs without OF

[ Upstream commit b2bae4b8e637dd751d27918a6b27bd5abcd08859 ]

All warnings (new ones prefixed by >>):

>> drivers/media/platform/atmel/atmel-sama7g5-isc.c:610:34: warning: unused variable 'microchip_xisc_of_match' [-Wunused-const-variable]
   static const struct of_device_id microchip_xisc_of_match[] = {
                                    ^
   13 warnings generated.

vim +/microchip_xisc_of_match +610 drivers/media/platform/atmel/atmel-sama7g5-isc.c

   609
 > 610  static const struct of_device_id microchip_xisc_of_match[] = {
   611          { .compatible = "microchip,sama7g5-isc" },
   612          { }
   613  };
   614  MODULE_DEVICE_TABLE(of, microchip_xisc_of_match);
   615

Fixed warning by guarding the atmel_isc_of_match by CONFIG_OF.

Reported-by: kernel test robot <lkp@intel.com>
Fixes: c9aa973884a1 ("media: atmel: atmel-isc: add microchip-xisc driver")
Signed-off-by: Eugen Hristev <eugen.hristev@microchip.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: sun8i-ss - fix infinite loop in sun8i_ss_setup_ivs()

[ Upstream commit d61a7b3decf7f0cf4121a7204303deefd2c7151b ]

There is no i decrement in while (i >= 0) loop.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Fixes: 359e893e8af4 ("crypto: sun8i-ss - rework handling of IV")
Acked-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Tested-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i2c: Fix a potential use after free

[ Upstream commit e4c72c06c367758a14f227c847f9d623f1994ecf ]

Free the adap structure only after we are done using it.
This patch just moves the put_device() down a bit to avoid the
use after free.

Fixes: 611e12ea0f12 ("i2c: core: manage i2c bus device refcount in i2c_[get|put]_adapter")
Signed-off-by: Xu Wang <vulab@iscas.ac.cn>
[wsa: added comment to the code, added Fixes tag]
Signed-off-by: Wolfram Sang <wsa@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

can: netlink: allow configuring of fixed data bit rates without need for do_set_data_bittiming callback

[ Upstream commit ec30c109391c5eac9b1d689a61e4bfed88148947 ]

This patch is similar to 7e193a42c37c ("can: netlink: allow
configuring of fixed bit rates without need for do_set_bittiming
callback") but for data bit rates instead of bit rates.

Usually CAN devices support configurable data bit rates. The limits
are defined by struct can_priv::data_bittiming_const. Another way is
to implement the struct can_priv::do_set_data_bittiming callback.

If the bit rate is configured via netlink, the can_changelink()
function checks that either can_priv::data_bittiming_const or struct
can_priv::do_set_data_bittiming is implemented.

In commit 431af779256c ("can: dev: add CAN interface API for fixed
bitrates") an API for configuring bit rates on CAN interfaces that
only support fixed bit rates was added. The supported bit rates are
defined by struct can_priv::bitrate_const.

However the above mentioned commit forgot to add the struct
can_priv::data_bitrate_const to the check in can_changelink().

In order to avoid to implement a no-op can_priv::do_set_data_bittiming
callback on devices with fixed data bit rates, extend the check in
can_changelink() accordingly.

Link: https://lore.kernel.org/all/20220613143633.4151884-1-mkl@pengutronix.de
Fixes: 431af779256c ("can: dev: add CAN interface API for fixed bitrates")
Acked-by: Max Staudt <max@enpas.org>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

can: netlink: allow configuring of fixed bit rates without need for do_set_bittiming callback

[ Upstream commit 7e193a42c37cf40eba8ac5af2d5e8eeb8b9506f9 ]

Usually CAN devices support configurable bit rates. The limits are
defined by struct can_priv::bittiming_const. Another way is to
implement the struct can_priv::do_set_bittiming callback.

If the bit rate is configured via netlink, the can_changelink()
function checks that either can_priv::bittiming_const or struct
can_priv::do_set_bittiming is implemented.

In commit 431af779256c ("can: dev: add CAN interface API for fixed
bitrates") an API for configuring bit rates on CAN interfaces that
only support fixed bit rates was added. The supported bit rates are
defined by struct can_priv::bitrate_const.

However the above mentioned commit forgot to add the struct
can_priv::bitrate_const to the check in can_changelink().

In order to avoid to implement a no-op can_priv::do_set_bittiming
callback on devices with fixed bit rates, extend the check in
can_changelink() accordingly.

Link: https://lore.kernel.org/all/20220611144248.3924903-1-mkl@pengutronix.de
Fixes: 431af779256c ("can: dev: add CAN interface API for fixed bitrates")
Reported-by: Max Staudt <max@enpas.org>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: fix sk_wmem_schedule() and sk_rmem_schedule() errors

[ Upstream commit 7c80b038d23e1f4c7fcc311f43f83b8c60e7fb80 ]

If sk->sk_forward_alloc is 150000, and we need to schedule 150001 bytes,
we want to allocate 1 byte more (rounded up to one page),
instead of 150001 :/

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: sun8i-ss - fix error codes in allocate_flows()

[ Upstream commit d2765e1b9ac4b2d5a5d5bf17f468c9b3566c3770 ]

These failure paths should return -ENOMEM.  Currently they return
success.

Fixes: 359e893e8af4 ("crypto: sun8i-ss - rework handling of IV")
Fixes: 8eec4563f152 ("crypto: sun8i-ss - do not allocate memory when handling hash requests")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Tested-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: sun8i-ss - do not allocate memory when handling hash requests

[ Upstream commit 8eec4563f152981a441693fc97c5459843dc5e6e ]

Instead of allocate memory on each requests, it is easier to
pre-allocate buffers.
This made error path easier.

Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm: adv7511: override i2c address of cec before accessing it

[ Upstream commit 9cc4853e4781bf0dd0f35355dc92d97c9da02f5d ]

Commit 680532c50bca ("drm: adv7511: Add support for
i2c_new_secondary_device") allows a device tree node to override
the default addresses of the secondary i2c devices. This is useful
for solving address conflicts on the i2c bus.

In adv7511_init_cec_regmap() the new i2c address of cec device is
read from device tree and immediately accessed, well before it is
written in the proper register to override the default address.
This can cause an i2c error during probe and a consequent probe
failure.

Once the new i2c address is read from the device tree, override
the default address before any attempt to access the cec.

Tested with adv7533 and stm32mp157f.

Signed-off-by: Antonio Borneo <antonio.borneo@foss.st.com>
Fixes: 680532c50bca ("drm: adv7511: Add support for i2c_new_secondary_device")
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220607213144.427177-1-antonio.borneo@foss.st.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/virtio: Fix NULL vs IS_ERR checking in virtio_gpu_object_shmem_init

[ Upstream commit c24968734abfed81c8f93dc5f44a7b7a9aecadfa ]

Since drm_prime_pages_to_sg() function return error pointers.
The drm_gem_shmem_get_sg_table() function returns error pointers too.
Using IS_ERR() to check the return value to fix this.

Fixes: 2f2aa13724d5 ("drm/virtio: move virtio_gpu_mem_entry initialization to new function")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20220602104223.54527-1-linmq006@gmail.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/shmem-helper: Pass GEM shmem object in public interfaces

[ Upstream commit a193f3b4e050e35c506a34d0870c838d8e0b0449 ]

Change all GEM SHMEM object functions that receive a GEM object
of type struct drm_gem_object to expect an object of type
struct drm_gem_shmem_object instead.

This change reduces the number of upcasts from struct drm_gem_object
by moving them into callers. The C compiler can now verify that the
GEM SHMEM functions are called with the correct type.

For consistency, the patch also renames drm_gem_shmem_free_object to
drm_gem_shmem_free. It further updates documentation for a number of
functions.

v3:
* fix docs for drm_gem_shmem_object_free()
v2:
* mention _object_ callbacks in docs (Daniel)

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20211108093149.7226-4-tzimmermann@suse.de
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/shmem-helper: Export dedicated wrappers for GEM object functions

[ Upstream commit c7fbcb7149ff9321bbbcc93c9920de534ea8102c ]

Wrap GEM SHMEM functions for struct drm_gem_object_funcs and update
all callers. This will allow for an update of the public interfaces
of the GEM SHMEM helper library.

v2:
* fix docs for drm_gem_shmem_object_print_info()

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20211108093149.7226-3-tzimmermann@suse.de
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/shmem-helper: Unexport drm_gem_shmem_create_with_handle()

[ Upstream commit 5a363c20673308e968b6640deb73d7bf77e8b463 ]

Turn drm_gem_shmem_create_with_handle() into an internal helper
function. It's not used outside of the compilation unit.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20211108093149.7226-2-tzimmermann@suse.de
Signed-off-by: Sasha Levin <sashal@kernel.org>

virtio-gpu: fix a missing check to avoid NULL dereference

[ Upstream commit bd63f11f4c3c46afec07d821f74736161ff6e526 ]

'cache_ent' could be set NULL inside virtio_gpu_cmd_get_capset()
and it will lead to a NULL dereference by a lately use of it
(i.e., ptr = cache_ent->caps_cache). Fix it with a NULL check.

Fixes: 62fb7a5e10962 ("virtio-gpu: add 3d/virgl support")
Signed-off-by: Xiaomeng Tong <xiam0nd.tong@gmail.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20220327050945.1614-1-xiam0nd.tong@gmail.com
[ kraxel: minor codestyle fixup ]

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i2c: mxs: Silence a clang warning

[ Upstream commit 3d43273d7d1e1a5374d531e901d3c537b4c97bbf ]

Change the of_device_get_match_data() cast to (uintptr_t)
to silence the following clang warning:

drivers/i2c/busses/i2c-mxs.c:802:18: warning: cast to smaller integer type 'enum mxs_i2c_devtype' from 'const void *' [-Wvoid-pointer-to-enum-cast]

Reported-by: kernel test robot <lkp@intel.com>
Fixes: c32abd8b5691 ("i2c: mxs: Remove unneeded platform_device_id")
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Wolfram Sang <wsa@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i2c: npcm: Correct slave role behavior

[ Upstream commit d7aa1b149b8fc04d802879cf4662010aa4a42deb ]

Correct the slave transaction logic to be compatible with the generic
slave backend driver.

Fixes: 56a1485b102e ("i2c: npcm7xx: Add Nuvoton NPCM I2C controller driver")
Signed-off-by: Tali Perry <tali.perry1@gmail.com>
Signed-off-by: Tyrone Ting <kfting@nuvoton.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Wolfram Sang <wsa@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i2c: npcm: Remove own slave addresses 2:10

[ Upstream commit 47d506d1a28fd10a9fb1f33df5622d88fae72095 ]

NPCM can support up to 10 own slave addresses. In practice, only one
address is actually being used. In order to access addresses 2 and above,
need to switch register banks. The switch needs spinlock.
To avoid using spinlock for this useless feature removed support of SA >=
2. Also fix returned slave event enum.

Remove some comment since the bank selection is not required. The bank
selection is not required since the supported slave addresses are reduced.

Fixes: 56a1485b102e ("i2c: npcm7xx: Add Nuvoton NPCM I2C controller driver")
Signed-off-by: Tali Perry <tali.perry1@gmail.com>
Signed-off-by: Tyrone Ting <kfting@nuvoton.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Wolfram Sang <wsa@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/bridge: lt9611uxc: Cancel only driver's work

[ Upstream commit dfa687bffc8a4a21ed929c7dececf01b8f1f52ee ]

During device remove care needs to be taken that no work is pending
before it removes the underlying DRM bridge etc, but this can be done on
the specific work rather than waiting for the flush of the system-wide
workqueue.

Fixes: bc6fa8676ebb ("drm/bridge/lontium-lt9611uxc: move HPD notification out of IRQ handler")
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Reviewed-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220601233818.1877963-1-bjorn.andersson@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/meson: encoder_hdmi: Fix refcount leak in meson_encoder_hdmi_init

[ Upstream commit d82a5a4aae9d0203234737caed1bf470aa317568 ]

of_graph_get_remote_node() returns remote device nodepointer with
refcount incremented, we should use of_node_put() on it when done.
Add missing of_node_put() to avoid refcount leak.

Fixes: e67f6037ae1b ("drm/meson: split out encoder from meson_dw_hdmi")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Reviewed-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220601033927.47814-3-linmq006@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/meson: encoder_hdmi: switch to bridge DRM_BRIDGE_ATTACH_NO_CONNECTOR

[ Upstream commit 0af5e0b41110e2da872030395231ab19c45be931 ]

This implements the necessary change to no more use the embedded
connector in dw-hdmi and use the dedicated bridge connector driver
by passing DRM_BRIDGE_ATTACH_NO_CONNECTOR to the bridge attach call.

The necessary connector properties are added to handle the same
functionalities as the embedded dw-hdmi connector, i.e. the HDR
metadata, the CEC notifier & other flags.

The dw-hdmi output_port is set to 1 in order to look for a connector
next bridge in order to get DRM_BRIDGE_ATTACH_NO_CONNECTOR working.

Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211020123947.2585572-5-narmstrong@baylibre.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/mediatek: Add pull-down MIPI operation in mtk_dsi_poweroff function

[ Upstream commit fa5d0a0205c34734c5b8daa77e39ac2817f63a10 ]

In the dsi_enable function, mtk_dsi_rxtx_control is to
pull up the MIPI signal operation. Before dsi_disable,
MIPI should also be pulled down by writing a register
instead of disabling dsi.

If disable dsi without pulling the mipi signal low, the value of
the register will still maintain the setting of the mipi signal being
pulled high.
After resume, even if the mipi signal is not pulled high, it will still
be in the high state.

Fixes: 2e54c14e310f ("drm/mediatek: Add DSI sub driver")

Link: https://patchwork.kernel.org/project/linux-mediatek/patch/1653012007-11854-5-git-send-email-xinlei.lee@mediatek.com/
Signed-off-by: Jitao Shi <jitao.shi@mediatek.com>
Signed-off-by: Xinlei Lee <xinlei.lee@mediatek.com>
Reviewed-by: Rex-BC Chen <rex-bc.chen@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/mediatek: Separate poweron/poweroff from enable/disable and define new funcs

[ Upstream commit cde7e2e35c2866d22a3a012e72a41052dfcc255d ]

In order to match the changes of "Use the drm_panel_bridge API",
the poweron/poweroff of dsi is extracted from enable/disable and
defined as new funcs (atomic_pre_enable/atomic_post_disable).

Since dsi_poweron is moved from dsi_enable to pre_enable function, in
order to avoid poweron failure, the operation of dsi register fails to
cause bus hang. Therefore, the protection mechanism is added to the
dsi_enable function.

Fixes: 2dd8075d2185 ("drm/mediatek: mtk_dsi: Use the drm_panel_bridge API")

Link: https://patchwork.kernel.org/project/linux-mediatek/patch/1653012007-11854-3-git-send-email-xinlei.lee@mediatek.com/
Signed-off-by: Jitao Shi <jitao.shi@mediatek.com>
Signed-off-by: Xinlei Lee <xinlei.lee@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Rex-BC Chen <rex-bc.chen@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/mediatek: Modify dsi funcs to atomic operations

[ Upstream commit 7f6335c6a258edf4d5ff1b904bc033188dc7b48b ]

Because .enable & .disable are deprecated.
Use .atomic_enable & .atomic_disable instead.

Link: https://patchwork.kernel.org/project/linux-mediatek/patch/1653012007-11854-2-git-send-email-xinlei.lee@mediatek.com/
Signed-off-by: Jitao Shi <jitao.shi@mediatek.com>
Signed-off-by: Xinlei Lee <xinlei.lee@mediatek.com>
Reviewed-by: Rex-BC Chen <rex-bc.chen@mediatek.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/radeon: fix potential buffer overflow in ni_set_mc_special_registers()

[ Upstream commit 136f614931a2bb73616b292cf542da3a18daefd5 ]

The last case label can write two buffers 'mc_reg_address[j]' and
'mc_data[j]' with 'j' offset equal to SMC_NISLANDS_MC_REGISTER_ARRAY_SIZE
since there are no checks for this value in both case labels after the
last 'j++'.

Instead of changing '>' to '>=' there, add the bounds check at the start
of the second 'case' (the first one already has it).

Also, remove redundant last checks for 'j' index bigger than array size.
The expression is always false. Moreover, before or after the patch
'table->last' can be equal to SMC_NISLANDS_MC_REGISTER_ARRAY_SIZE and it
seems it can be a valid value.

Detected using the static analysis tool - Svace.
Fixes: 69e0b57a91ad ("drm/radeon/kms: add dpm support for cayman (v5)")
Signed-off-by: Alexey Kodanev <aleksei.kodanev@bell-sw.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ath11k: Fix incorrect debug_mask mappings

[ Upstream commit 9331f7d3c54a263bede5055e106e40b28d0bd937 ]

Currently a couple of debug_mask entries are mapped to the same value,
this could enable unintended driver logging. If enabling DP_TX logs was
the intention, then this could also enable PCI logs flooding the dmesg
buffer or vice versa. Fix this by correctly assigning the debug masks.

Found during code review.

Tested-on: WCN6750 hw1.0 AHB WLAN.MSL.1.0.1-00887-QCAMSLSWPLZ-1

Fixes: aa2092a9bab3f ("ath11k: add raw mode and software crypto support")
Signed-off-by: Manikanta Pubbisetty <quic_mpubbise@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220602115621.15339-1-quic_mpubbise@quicinc.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/mipi-dbi: align max_chunk to 2 in spi_transfer

[ Upstream commit 435c249008cba04ed6a7975e9411f3b934620204 ]

In __spi_validate, there's a validation that no partial transfers
are accepted (xfer->len % w_size must be zero). When
max_chunk is not a multiple of bpw (e.g. max_chunk = 65535,
bpw = 16), the transfer will be rejected.

This patch aligns max_chunk to 2 bytes (the maximum value of bpw is 16),
so that no partial transfer will occur.

Fixes: d23d4d4dac01 ("drm/tinydrm: Move tinydrm_spi_transfer()")

Signed-off-by: Yunhao Tian <t123yh.xyz@gmail.com>
Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220510030219.2486687-1-t123yh.xyz@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

ath11k: fix netdev open race

[ Upstream commit d4ba1ff87b17e81686ada8f429300876f55f95ad ]

Make sure to allocate resources needed before registering the device.

This specifically avoids having a racing open() trigger a BUG_ON() in
mod_timer() when ath11k_mac_op_start() is called before the
mon_reap_timer as been set up.

I did not see this issue with next-20220310, but I hit it on every probe
with next-20220511. Perhaps some timing changed in between.

Here's the backtrace:

[   51.346947] kernel BUG at kernel/time/timer.c:990!
[   51.346958] Internal error: Oops - BUG: 0 [#1] PREEMPT SMP
...
[   51.578225] Call trace:
[   51.583293]  __mod_timer+0x298/0x390
[   51.589518]  mod_timer+0x14/0x20
[   51.595368]  ath11k_mac_op_start+0x41c/0x4a0 [ath11k]
[   51.603165]  drv_start+0x38/0x60 [mac80211]
[   51.610110]  ieee80211_do_open+0x29c/0x7d0 [mac80211]
[   51.617945]  ieee80211_open+0x60/0xb0 [mac80211]
[   51.625311]  __dev_open+0x100/0x1c0
[   51.631420]  __dev_change_flags+0x194/0x210
[   51.638214]  dev_change_flags+0x24/0x70
[   51.644646]  do_setlink+0x228/0xdb0
[   51.650723]  __rtnl_newlink+0x460/0x830
[   51.657162]  rtnl_newlink+0x4c/0x80
[   51.663229]  rtnetlink_rcv_msg+0x124/0x390
[   51.669917]  netlink_rcv_skb+0x58/0x130
[   51.676314]  rtnetlink_rcv+0x18/0x30
[   51.682460]  netlink_unicast+0x250/0x310
[   51.688960]  netlink_sendmsg+0x19c/0x3e0
[   51.695458]  ____sys_sendmsg+0x220/0x290
[   51.701938]  ___sys_sendmsg+0x7c/0xc0
[   51.708148]  __sys_sendmsg+0x68/0xd0
[   51.714254]  __arm64_sys_sendmsg+0x28/0x40
[   51.720900]  invoke_syscall+0x48/0x120

Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3

Fixes: d5c65159f289 ("ath11k: driver for Qualcomm IEEE 802.11ax devices")
Fixes: 840c36fa727a ("ath11k: dp: stop rx pktlog before suspend")
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220517103436.15867-1-johan+linaro@kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rtlwifi: fix error codes in rtl_debugfs_set_write_h2c()

[ Upstream commit b88d28146c30a8e14f0f012d56ebf19b68a348f4 ]

If the copy_from_user() fails or the user gives invalid date then the
correct thing to do is to return a negative error code.  (Currently it
returns success).

I made a copy additional related cleanups:
1) There is no need to check "buffer" for NULL.  That's handled by
copy_from_user().
2) The "h2c_len" variable cannot be negative because it is unsigned
and because sscanf() does not return negative error codes.

Fixes: 610247f46feb ("rtlwifi: Improve debugging by using debugfs")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/YoOLnDkHgVltyXK7@kili
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/panel: Fix build error when CONFIG_DRM_PANEL_SAMSUNG_ATNA33XC20=y && CONFIG_DRM_DISPLAY_HELPER=m

[ Upstream commit a67664860f7833015a683ea295f7c79ac2901332 ]

If CONFIG_DRM_PANEL_SAMSUNG_ATNA33XC20=y && CONFIG_DRM_DISPLAY_HELPER=m,
bulding fails:

drivers/gpu/drm/panel/panel-samsung-atna33xc20.o: In function `atana33xc20_probe':
panel-samsung-atna33xc20.c:(.text+0x744): undefined reference to
 `drm_panel_dp_aux_backlight'
make: *** [vmlinux] Error 1

Let CONFIG_DRM_PANEL_SAMSUNG_ATNA33XC20 select DRM_DISPLAY_DP_HELPER and
CONFIG_DRM_DISPLAY_HELPER to fix this error.

Fixes: 32ce3b320343 ("drm/panel: atna33xc20: Introduce the Samsung ATNA33XC20 panel")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Gao Chao <gaochao49@huawei.com>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220524024551.539-1-gaochao49@huawei.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/st7735r: Fix module autoloading for Okaya RH128128T

[ Upstream commit 9ad6f181ad9a19a26bda73a7b199df44ccfcdaba ]

The SPI core always reports a "MODALIAS=spi:<foo>", even if the device was
registered via OF. This means that the st7735r.ko module won't autoload if
a DT has a node with a compatible "okaya,rh128128t" string.

In that case, kmod expects a "MODALIAS=of:N*T*Cokaya,rh128128t" uevent but
instead will get a "MODALIAS=spi:rh128128t", which is not present in the
list of aliases:

  $ modinfo drivers/gpu/drm/tiny/st7735r.ko | grep alias
  alias:          of:N*T*Cokaya,rh128128tC*
  alias:          of:N*T*Cokaya,rh128128t
  alias:          of:N*T*Cjianda,jd-t18003-t01C*
  alias:          of:N*T*Cjianda,jd-t18003-t01
  alias:          spi:jd-t18003-t01

To workaround this issue, add in the SPI table an entry for that device.

Fixes: d1d511d516f7 ("drm: tiny: st7735r: Add support for Okaya RH128128T")
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Acked-by: David Lechner <david@lechnology.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220520091602.179078-1-javierm@redhat.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

ath10k: do not enforce interrupt trigger type

[ Upstream commit 1ee6c5abebd3cacf2ac4378d0ed4f57fd4850421 ]

Interrupt line can be configured on different hardware in different way,
even inverted.  Therefore driver should not enforce specific trigger
type - edge rising - but instead rely on Devicetree to configure it.

All Qualcomm DTSI with WCN3990 define the interrupt type as level high,
so the mismatch between DTSI and driver causes rebind issues:

  $ echo 18800000.wifi > /sys/bus/platform/drivers/ath10k_snoc/unbind
  $ echo 18800000.wifi > /sys/bus/platform/drivers/ath10k_snoc/bind
  [   44.763114] irq: type mismatch, failed to map hwirq-446 for interrupt-controller@17a00000!
  [   44.763130] ath10k_snoc 18800000.wifi: error -ENXIO: IRQ index 0 not found
  [   44.763140] ath10k_snoc 18800000.wifi: failed to initialize resource: -6

Tested-on: WCN3990 hw1.0 SNOC WLAN.HL.3.2.0.c8-00009-QCAHLSWSC8180XMTPLZ-1
Tested-on: WCN3990 hw1.0 SNOC WLAN.HL.2.0-01387-QCAHLSWMTPLZ-1

Fixes: c963a683e701 ("ath10k: add resource init and deinit for WCN3990")
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Tested-by: Steev Klimaszewski <steev@kali.org>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220513151516.357549-1-krzysztof.kozlowski@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/bridge: tc358767: Move (e)DP bridge endpoint parsing into dedicated function

[ Upstream commit 8478095a8c4bcea3c83b0767d6c9127434160761 ]

The TC358767/TC358867/TC9595 are all capable of operating in multiple
modes, DPI-to-(e)DP, DSI-to-(e)DP, DSI-to-DPI. Only the first mode is
currently supported. In order to support the rest of the modes without
making the tc_probe() overly long, split the bridge endpoint parsing
into dedicated function, where the necessary logic to detect the bridge
mode based on which endpoints are connected, can be implemented.

Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Tested-by: Lucas Stach <l.stach@pengutronix.de> # In both DPI to eDP and DSI to DPI mode.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Jonas Karlman <jonas@kwiboo.se>
Cc: Laurent Pinchart <Laurent.pinchart@ideasonboard.com>
Cc: Maxime Ripard <maxime@cerno.tech>
Cc: Neil Armstrong <narmstrong@baylibre.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220329085015.39159-7-marex@denx.de
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/dp: Export symbol / kerneldoc fixes for DP AUX bus

[ Upstream commit 39c28cdfb719f0e306b447f0827dfd712f81858b ]

While working on the DP AUX bus code I found a few small things that
should be fixed. Namely the non-devm version of
of_dp_aux_populate_ep_devices() was missing an export. There was also
an extra blank line in a kerneldoc and a kerneldoc that incorrectly
documented a return value. Fix these.

Fixes: aeb33699fc2c ("drm: Introduce the DP AUX bus")
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20220510122726.v3.1.Ia91f4849adfc5eb9da1eb37ba79aa65fb3c95a0f@changeid
Signed-off-by: Sasha Levin <sashal@kernel.org>

pwm: lpc18xx: Fix period handling

[ Upstream commit 8933d30c5f468d6cc1e4bf9bb535149da35f202e ]

The calculation:

val = (u64)NSEC_PER_SEC * LPC18XX_PWM_TIMER_MAX;
do_div(val, lpc18xx_pwm->clk_rate);
lpc18xx_pwm->max_period_ns = val;

is bogus because with NSEC_PER_SEC = 1000000000,
LPC18XX_PWM_TIMER_MAX = 0xffffffff and clk_rate < NSEC_PER_SEC this
overflows the (on lpc18xx (i.e. ARM32) 32 bit wide) unsigned int
.max_period_ns. This results (dependant of the actual clk rate) in an
arbitrary limitation of the maximal period.  E.g. for clkrate =
333333333 (Hz) we get max_period_ns = 9 instead of 12884901897.

So make .max_period_ns an u64 and pass period and duty as u64 to not
discard relevant digits. And also make use of mul_u64_u64_div_u64()
which prevents all overflows assuming clk_rate < NSEC_PER_SEC.

Fixes: 841e6f90bb78 ("pwm: NXP LPC18xx PWM/SCT driver")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

pwm: lpc18xx-sct: Simplify driver by not using pwm_[gs]et_chip_data()

[ Upstream commit 9136a39e6cf69e49803ac6123a4ac4431bc915a2 ]

The per-channel data is available directly in the driver data struct. So
use it without making use of pwm_[gs]et_chip_data().

The relevant change introduced by this patch to lpc18xx_pwm_disable() at
the assembler level (for an arm lpc18xx_defconfig build) is:

push    {r3, r4, r5, lr}
mov     r4, r0
mov     r0, r1
mov     r5, r1
bl      0 <pwm_get_chip_data>
ldr     r3, [r0, #0]

changes to

ldr     r3, [r1, #8]
push    {r4, lr}
add.w   r3, r0, r3, lsl #2
ldr     r3, [r3, #92]   ; 0x5c

So this reduces stack usage, has an improved runtime behavior because of
better pipeline usage, doesn't branch to an external function and the
generated code is a bit smaller occupying less memory.

The codesize of lpc18xx_pwm_probe() is reduced by 32 bytes.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

pwm: lpc18xx-sct: Reduce number of devm memory allocations

[ Upstream commit 20d9de9c4d6642bb40c935233697723b56573cbc ]

Each devm allocations has an overhead of 24 bytes to store the related
struct devres_node additionally to the fragmentation of the allocator.
So allocating 16 struct lpc18xx_pwm_data (which only hold a single int)
adds quite some overhead. Instead put the per-channel data into the
driver data struct and allocate it in one go.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

pwm: sifive: Shut down hardware only after pwmchip_remove() completed

[ Upstream commit 2375e964d541bb09158cd2dff67b5d74e8de61cd ]

The PWMs are expected to be functional until pwmchip_remove() is called.
So disable the clks only afterwards.

Fixes: 9e37a53eb051 ("pwm: sifive: Add a driver for SiFive SoC PWM")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Tested-by: Emil Renner Berthing <emil.renner.berthing@canonical.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

pwm: sifive: Ensure the clk is enabled exactly once per running PWM

[ Upstream commit ace41d7564e655c39f709a78c035188a460c7cbd ]

.apply() assumes the clk to be for a given PWM iff the PWM is enabled.
So make sure this is the case when .probe() completes. And in .remove()
disable the according number of times.

This fixes a clk enable/disable imbalance, if some PWMs are already running
at probe time.

Fixes: 9e37a53eb051 (pwm: sifive: Add a driver for SiFive SoC PWM)
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Tested-by: Emil Renner Berthing <emil.renner.berthing@canonical.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

pwm: sifive: Simplify offset calculation for PWMCMP registers

[ Upstream commit 20550a61880fc55e68a0d290ad195b74729c0e7b ]

Instead of explicitly using PWM_SIFIVE_PWMCMP0 + pwm->hwpwm *
PWM_SIFIVE_SIZE_PWMCMP for each access to one of the PWMCMP registers,
introduce a macro that takes the hwpwm id as parameter.

For the register definition using a plain 4 instead of the cpp constant
PWM_SIFIVE_SIZE_PWMCMP is easier to read, so define the offset macro
without the constant. The latter can then be dropped as there are no
users left.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Tested-by: Emil Renner Berthing <emil.renner.berthing@canonical.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

dm: return early from dm_pr_call() if DM device is suspended

[ Upstream commit e120a5f1e78fab6223544e425015f393d90d6f0d ]

Otherwise PR ops may be issued while the broader DM device is being
reconfigured, etc.

Fixes: 9c72bad1f31a ("dm: call PR reserve/unreserve on each underlying device")
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

thermal/tools/tmon: Include pthread and time headers in tmon.h

[ Upstream commit 0cf51bfe999524377fbb71becb583b4ca6d07cfc ]

Include sys/time.h and pthread.h in tmon.h, so that types
"pthread_mutex_t" and "struct timeval tv" are known when tmon.h
references them.

Without these headers, compiling tmon against musl-libc will fail with
these errors:

In file included from sysfs.c:31:0:
tmon.h:47:8: error: unknown type name 'pthread_mutex_t'
 extern pthread_mutex_t input_lock;
        ^~~~~~~~~~~~~~~
make[3]: *** [<builtin>: sysfs.o] Error 1
make[3]: *** Waiting for unfinished jobs....
In file included from tui.c:31:0:
tmon.h:54:17: error: field 'tv' has incomplete type
  struct timeval tv;
                 ^~
make[3]: *** [<builtin>: tui.o] Error 1
make[2]: *** [Makefile:83: tmon] Error 2

Signed-off-by: Markus Mayer <mmayer@broadcom.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Sumeet Pawnikar <sumeet.r.pawnikar@intel.com>
Acked-by: Alejandro González <alejandro.gonzalez.correo@gmail.com>
Tested-by: Alejandro González <alejandro.gonzalez.correo@gmail.com>
Fixes: 94f69966faf8 ("tools/thermal: Introduce tmon, a tool for thermal  subsystem")
Link: https://lore.kernel.org/r/20220718031040.44714-1-f.fainelli@gmail.com
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>