Dragonboard845c doesn't have board-specific board-id programmed, it uses
generic 0xff. Thus add the property with the 'variant' of the
calibration data.
Note: the driver will check for the calibration data for the following
IDs, so older board-2.bin files that were distributed as a part of
Linaro releases will continue to work.
- 'bus=snoc,qmi-board-id=ff,qmi-chip-id=30214,variant=Thundercomm_DB845C'
- 'bus=snoc,qmi-board-id=ff,qmi-chip-id=30214'
- 'bus=snoc,qmi-board-id=ff'
For the reference, the board is identified by the driver in the
following way:
ath10k_snoc 18800000.wifi: qmi chip_id 0x30214 chip_family 0x4001 board_id 0xff soc_id 0x40030001
ath10k_snoc 18800000.wifi: qmi fw_version 0x2009856b fw_build_timestamp 2018-07-19 12:28 fw_build_id QC_IMAGE_VERSION_STRING=WLAN.HL.2.0-01387-QCAHLSWMTPLZ-1
Fixes: 3f72e2d3e6 ("arm64: dts: qcom: Add Dragonboard 845c")
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220403105711.1173161-1-dmitry.baryshkov@linaro.org
The property vddp-ref-clk-max-microamp (for VDDP ref clk supply which is
l25 regulator) is not documented in MSM8996 UFS PHY bindings
(qcom,msm8996-qmp-ufs-phy). It is mentioned in the other UFS PHY
bindings for qcom,msm8996-ufs-phy-qmp-14nm.
The MSM8996-based Xiaomi devices configure l25 regulator in a
conflicting way:
1. with maximum 100 uAmp for VDDP ref clk supply of UFS PHY,
2. with maximum 450 mAmp for VCCQ supply of UFS.
Since the vddp-ref-clk-max-microamp property is basically not
documented for that UFS PHY and has a conflicting values, drop it
entirely as it looks like not tested and not used ever.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220407092725.232463-1-krzysztof.kozlowski@linaro.org
The referenced commit generates a reference counting error if the rkey has
the same index but the wrong key. In this case the reference taken by
rxe_pool_get_index() is not dropped.
Drop the reference if the keys don't match in rxe_recheck_mr(). Check
that the mw and mr are still valid.
Fixes: 8a1a0be894 ("RDMA/rxe: Replace mr by rkey in responder resources")
Link: https://lore.kernel.org/r/20220411030647.20011-1-rpearsonhpe@gmail.com
Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Current rxe_requester() doesn't generate a completion when processing an
unsupported/invalid opcode. If rxe driver doesn't support a new opcode
(e.g. RDMA Atomic Write) and RDMA library supports it, an application
using the new opcode can reproduce this issue. Fix the issue by calling
"goto err;".
Fixes: 8700e3e7c4 ("Soft RoCE driver")
Link: https://lore.kernel.org/r/20220410113513.27537-1-yangx.jy@fujitsu.com
Signed-off-by: Xiao Yang <yangx.jy@fujitsu.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Tasklets have long been deprecated as being too heavy on the system
by running in irq context - and this is not a performance critical
path. If a higher priority process wants to run, it must wait for
the tasklet to finish before doing so.
rtllib_sta_ps() and ieee80211_sta_ps() will now run in process context
and have further concurrency (tasklets being serialized among themselves),
but this is done holding the ieee->lock, so it should be fine.
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
Link: https://lore.kernel.org/r/20220411151620.129178-7-dave@stgolabs.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Tasklets have long been deprecated as being too heavy on the system
by running in irq context - and this is not a performance critical
path. If a higher priority process wants to run, it must wait for
the tasklet to finish before doing so.
The execution of the SME event will now occur in task context. There
are, however, changes in concurrency. Workqueues, unlike tasklets,
are not serialized among themselves and can run concurrently
updating sme_i.qhead. However, the current code is already exposed
in same ways, regardless of the deferral mechanism, in that
hostif_sme_enqueue() does unserialized enqueues updating sme_i.qtail.
Also get rid of the bogus (power save) tasklet enabling, as it
is never disabled to begin with.
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
Link: https://lore.kernel.org/r/20220411151620.129178-5-dave@stgolabs.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Tasklets have long been deprecated as being too heavy on the system
by running in irq context - and this is not a performance critical
path. If a higher priority process wants to run, it must wait for
the tasklet to finish before doing so.
The completion_bh tasklet will now run in process context and have
further concurrency (tasklets being serialized among themselves),
but this is done holding the ctlxq.lock, so it should be fine.
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
Link: https://lore.kernel.org/r/20220411151620.129178-4-dave@stgolabs.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Tasklets have long been deprecated as being too heavy on the system
by running in irq context - and this is not a performance critical
path. If a higher priority process wants to run, it must wait for
the tasklet to finish before doing so.
The reaper_bh tasklet will now run in process context and have further
concurrency (tasklets being serialized among themselves), but this
is done holding the ctlxq.lock, so it should be fine.
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
Link: https://lore.kernel.org/r/20220411151620.129178-3-dave@stgolabs.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Tasklets have long been deprecated as being too heavy on the system
by running in irq context - and this is not a performance critical
path. If a higher priority process wants to run, it must wait for
the tasklet to finish before doing so. A more suitable equivalent
is to converted to threaded irq instead and service channels in
regular task context.
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
Link: https://lore.kernel.org/r/20220411151620.129178-2-dave@stgolabs.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To support stations in power saving mode, AP should notify stations
that there are frames buffered at the AP via TIM during beacons.
Driver used to transmit identical beacons that were downloaded to
hardware during the initiation phase. This beacon will become
obsolete over time.
If the beacon does not contain sufficient information, station would
not be able to percept that there is data to receive. Hence it won't
wake up and start the PS-poll procedure, this could lead to timeout
and/or lost data segments. In order to resolve this issue, driver will
now download beacon to hardware whenever the content is updated.
Enable hardware to update dtim_count for more efficiency, this reduces
the overhead of downloading beacon at every beacon interval since most
of the time only the dtim_count needs to be updated.
Change queue mapping for broadcast/multicast frames to high queue, so
these frames can be prioritized and sent when dtim_count is zero.
Signed-off-by: Po-Hao Huang <phhuang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220407095858.46807-4-pkshih@realtek.com
