In case we're running on a MediaTek SoC that is not declared in the
socinfo_data_table, print a message stating that and, with it, also
print out the read eFuse data: this allows to easily add a new SoC
in the table, granted that its actual soc/segment/marketing names
are already previously known.
This is especially useful when booting an already known SoC that
already has socinfo support, but has a different silicon version
and/or revision, hence different model information in the eFuses.
Link: https://lore.kernel.org/r/20240314113015.121940-1-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Most of the callers to this function do not require CMDQ support, it is
optional, so the missing property shouldn't cause an error message.
However, it could result on degraded performance, so the fact that it's
missing should still be alerted. Furthermore, the callers that do
require CMDQ support already log at the error level when an error is
returned.
Change the log message in this helper to be printed at the warning level
instead.
Signed-off-by: Nícolas F. R. A. Prado <nfraprado@collabora.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20240229-gce-client-reg-log-dbg-v2-1-4975077173d0@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Add cmdq_pkt_acquire_event() function to support CMDQ user making
an instruction for acquiring event.
CMDQ users can use cmdq_pkt_acquire_event() as `mutex_lock`
and cmdq_pkt_clear_event() as `mutex_unlock` to protect the global
resource modified instructions between them.
cmdq_pkt_acquire_event() would wait for event to be cleared.
After event is cleared by cmdq_pkt_clear_event() in other GCE threads,
cmdq_pkt_acquire_event() would set event and keep executing next
instruction. So the mutex would work like this:
cmdq_pkt_acquire_event() /* mutex lock */
/* critical secton instructions that modified global resource */
cmdq_pkt_clear_event() /* mutex unlock */
Prevent the critical section instructions from being affected by other
GCE threads.
Signed-off-by: Jason-JH.Lin <jason-jh.lin@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20240307013458.23550-5-jason-jh.lin@mediatek.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Add cmdq_pkt_poll_addr function to support CMDQ user making
an instruction for polling a specific address of hardware rigster
to check the value with or without mask.
POLL is a legacy operation in GCE, so it does not support SPR and
CMDQ_CODE_LOGIC. To support polling the register address which doesn't
have the subsys id, CMDQ users need to make an instruction with GPR and
CMDQ_CODE_MASK operation to move the register address to be poll into GPR.
Then users can make an POLL instruction with GPR to poll the register
address assigned in previous instruction.
Signed-off-by: Jason-JH.Lin <jason-jh.lin@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20240307013458.23550-4-jason-jh.lin@mediatek.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Add cmdq_pkt_mem_move() function to support CMDQ user making
an instruction for moving a value from a source address to a
destination address.
Signed-off-by: Jason-JH.Lin <jason-jh.lin@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20240307013458.23550-3-jason-jh.lin@mediatek.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Add specific purpose register definitions for GCE, so CMDQ users can
use them as a buffer to store data.
Signed-off-by: Jason-JH.Lin <jason-jh.lin@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20240307013458.23550-2-jason-jh.lin@mediatek.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
cmdq_pkt_create() and cmdq_pkt_destroy() is not suitable for
client drivers so each client driver has implement its own
function. This refinement would pass struct cmdq_pkt pointer into
cmdq_pkt_create(). In addition, client driver has the struct
cmdq_client information, so it's not necessary to store this
information in struct cmdq_pkt. After this refinement, client
drivers could use these helper funciton instead of implementing
its own version.
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Link: https://lore.kernel.org/r/20240222154120.16959-8-chunkuang.hu@kernel.org
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
cmdq_pkt_flush_async() is not used by all client drivers (MediaTek
drm driver and MediaTek mdp3 driver), so remove it.
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Link: https://lore.kernel.org/r/20240222154120.16959-7-chunkuang.hu@kernel.org
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
cmdq_pkt_eoc() append eoc command to CMDQ packet. eoc command
would ask GCE to generate IRQ. It's usually appended to the end
of packet to notify all command in the packet is done.
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Link: https://lore.kernel.org/r/20240222154120.16959-6-chunkuang.hu@kernel.org
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
cmdq_pkt_jump_rel() append relative jump command to the packet.
Relative jump change PC to the target address with offset from
current PC.
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20240222154120.16959-5-chunkuang.hu@kernel.org
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
In order to distinguish absolute jump and relative jump,
cmdq_pkt_jump() append absolute jump command, so rename it to
cmdq_pkt_jump_abs().
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20240222154120.16959-4-chunkuang.hu@kernel.org
[Angelo: Added temporary wrapper to avoid build breakage]
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
In original design, cmdq_pkt_jump() call cmdq_get_shift_pa() every
time to get shift_pa. But the shift_pa is constant value for each
SoC, so client driver just need to call cmdq_get_shift_pa() once
and pass shift_pa to cmdq_pkt_jump() to prevent frequent function
call.
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Link: https://lore.kernel.org/r/20240222154120.16959-3-chunkuang.hu@kernel.org
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
For cmdq jump command, offset 0 means relative jump and offset 1
means absolute jump. cmdq_pkt_jump() is absolute jump, so fix the
typo of CMDQ_JUMP_RELATIVE in cmdq_pkt_jump().
Fixes: 946f1792d3 ("soc: mediatek: cmdq: add jump function")
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20240222154120.16959-2-chunkuang.hu@kernel.org
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
The mtk-socinfo driver uses symbols 'soc_device_register' and
'soc_device_unregister' which are part of the bus driver for
System-on-Chip devices.
Select SOC_BUS to make sure that driver is built and the symbols are
available.
Fixes: 423a54da3c ("soc: mediatek: mtk-socinfo: Add driver for getting chip information")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Reviewed-by: Chen-Yu Tsai <wenst@chromium.org>
Reviewed-by: Matthias Brugger <matthias.bgg@gmail.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/cc8f7f7da5bdccce514a320e0ae7468659cf7346.1707327680.git.daniel@makrotopia.org
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
The thermal framework registers thermal zones as specified in DT and
including the "-thermal" suffix: append that to the driver specified
tzone_name to actually match the thermal zone name as registered by
the thermal API.
Fixes: 2bfbf82956 ("soc: mediatek: mtk-svs: Constify runtime-immutable members of svs_bank")
Link: https://lore.kernel.org/r/20240318113237.125802-1-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Chintan Vankar says:
====================
Enable RX HW timestamp for PTP packets using CPTS FIFO
The CPSW offers two mechanisms for communicating packet ingress timestamp
information to the host.
The first mechanism is via the CPTS Event FIFO which records timestamp
when triggered by certain events. One such event is the reception of an
Ethernet packet with a specified EtherType field. This is used to capture
ingress timestamps for PTP packets. With this mechanism the host must
read the timestamp (from the CPTS FIFO) separately from the packet payload
which is delivered via DMA.
In the second mechanism of timestamping, CPSW driver enables hardware
timestamping for all received packets by setting the TSTAMP_EN bit in
CPTS_CONTROL register, which directs the CPTS module to timestamp all
received packets, followed by passing timestamp via DMA descriptors.
This mechanism is responsible for triggering errata i2401:
"CPSW: Host Timestamps Cause CPSW Port to Lock up."
The errata affects all K3 SoCs. Link to errata for AM64x:
https://www.ti.com/lit/er/sprz457h/sprz457h.pdf
As a workaround we can use first mechanism to timestamp received
packets.
====================
Link: https://lore.kernel.org/r/20240419082626.57225-1-c-vankar@ti.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
In the current mechanism of timestamping, am65-cpsw-nuss driver
enables hardware timestamping for all received packets by setting
the TSTAMP_EN bit in CPTS_CONTROL register, which directs the CPTS
module to timestamp all received packets, followed by passing
timestamp via DMA descriptors. This mechanism causes CPSW Port to
Lock up.
To prevent port lock up, don't enable rx packet timestamping by
setting TSTAMP_EN bit in CPTS_CONTROL register. The workaround for
timestamping received packets is to utilize the CPTS Event FIFO
that records timestamps corresponding to certain events. The CPTS
module is configured to generate timestamps for Multicast Ethernet,
UDP/IPv4 and UDP/IPv6 PTP packets.
Update supported hwtstamp_rx_filters values for CPSW's timestamping
capability.
Fixes: b1f66a5bee ("net: ethernet: ti: am65-cpsw-nuss: enable packet timestamping support")
Signed-off-by: Chintan Vankar <c-vankar@ti.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Add a new function "am65_cpts_rx_timestamp()" which checks for PTP
packets from header and timestamps them.
Add another function "am65_cpts_find_rx_ts()" which finds CPTS FIFO
Event to get the timestamp of received PTP packet.
Signed-off-by: Chintan Vankar <c-vankar@ti.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
In dpu_core_irq_callback_handler() callback function pointer is compared to NULL,
but then callback function is unconditionally called by this pointer.
Fix this bug by adding conditional return.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Fixes: c929ac60b3 ("drm/msm/dpu: allow just single IRQ callback")
Signed-off-by: Aleksandr Mishin <amishin@t-argos.ru>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/588237/
Link: https://lore.kernel.org/r/20240408085523.12231-1-amishin@t-argos.ru
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
MDP4 and MDP5 drivers enumerate supported formats each time the plane is
created. In preparation to merger of MDP DPU format databases, define
precise formats list, so that changes to the database do not cause the
driver to add unsupported format to the list.
Reviewed-by: Abhinav Kumar <quic_abhinavk@quicinc.com>
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/590421/
Link: https://lore.kernel.org/r/20240420-dpu-format-v2-2-9e93226cbffd@linaro.org
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Make use of the __free() cleanup handler to automatically free nodes
when they get out of scope.
Signed-off-by: Javier Carrasco <javier.carrasco.cruz@gmail.com>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Make use of the __free() cleanup handler to automatically free nodes
when they get out of scope.
Only find_supply_name() is affected, and the new mechanism removes the
need for a 'goto' and the 'name' local variable.
Signed-off-by: Javier Carrasco <javier.carrasco.cruz@gmail.com>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Currently, ath12k does not have the ability to set band edge channel power
for WCN7850. In order to support this, ath12k gets band edge channel power
table in ath12k_acpi_dsm_get_data() function and sets pdev_id and
param_type_id, then finally sends these data and
WMI_PDEV_SET_BIOS_INTERFACE_CMDID to firmware to set band edge channel
power.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Signed-off-by: Lingbo Kong <quic_lingbok@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240422033054.979-5-quic_lingbok@quicinc.com
Currently, ath12k does not have the ability to adjust Clear Channel Assessment
(CCA) threshold values to meet the regulatory requirements. Get the values from
ACPI and send them to the firmware using ath12k_wmi_set_bios_cmd() function.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Signed-off-by: Lingbo Kong <quic_lingbok@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240422033054.979-4-quic_lingbok@quicinc.com
In order to enable ACPI SAR (Specific Absorption Rate), ath12k gets SAR and GEO
offset tables from ACPI and sends the data to firmware using
WMI_PDEV_SET_BIOS_SAR_TABLE_CMDID and WMI_PDEV_SET_BIOS_GEO_TABLE_CMDID
commands.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Signed-off-by: Lingbo Kong <quic_lingbok@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240422033054.979-3-quic_lingbok@quicinc.com
Currently, ath12k does not support Time-Average-SAR (TAS). In order to enable
TAS read the tables from ACPI and send them to the firmware using
WMI_PDEV_SET_BIOS_INTERFACE_CMDID command. Besides, ath12k registers an ACPI
event callback so that ACPI can notify ath12k to get the updated SAR power
table and sends it to the firmware when the device state is changed.
ACPI is only enabled for WCN7850 using struct ath12k_hw_params::acpi_guid
field. Most likely QCN9274 will never support ACPI as the chip is not used in
laptops.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Signed-off-by: Lingbo Kong <quic_lingbok@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240422033054.979-2-quic_lingbok@quicinc.com
Now that all things are ready, enable supports_suspend to
make suspend/resume work for WCN7850. Don't touch other
chips because they don't support suspend.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240412060620.27519-11-quic_bqiang@quicinc.com
Now that all infrastructure is in place and ath12k is fixed to handle all the
corner cases, power down the ath12k firmware during suspend and power it back
up during resume.
For suspend, two conditions needs to be satisfied:
1. since MHI channel unprepare would be done in late suspend stage,
ath12k needs to get all QMI-dependent things done before that stage.
2. and because unprepare MHI channels requires a working MHI stack,
ath12k is not allowed to call mhi_power_down() until that finishes.
So the original suspend callback is separated into two parts: the first part
handles all QMI-dependent things in suspend callback; while the second part
powers down MHI in suspend_late callback. This is valid because kernel calls
ath12k's suspend callback before calling all suspend_late callbacks, making
the first condition satisfied. And because MHI devices are children of ath12k
device (ab->dev), kernel guarantees that ath12k's suspend_late callback is
called after QRTR's suspend_late callback, this satisfies the second condition.
Above analysis also applies to resume process. so the original resume
callback is separated into two parts: the first part powers up MHI stack
in resume_early callback, this guarantees MHI stack is working when
QRTR tries to prepare MHI channels (kernel calls QRTR's resume_early callback
after ath12k's resume_early callback, due to the child-father relationship);
the second part waits for the completion of restart, which would succeed
since MHI channels are ready for use by QMI.
Another notable change is in power down path, we tell mhi_power_down() to not
to destroy MHI devices, making it possible for QRTR to help unprepare/prepare
MHI channels, and finally get us rid of the potential probe-defer issue when
resume.
Also change related code due to interface changes.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240412060620.27519-10-quic_bqiang@quicinc.com
Currently when resume ath12k_core_restart() calls
ath12k_core_pre_reconfigure_recovery() where mac80211 queues are
stopped by calling ieee80211_stop_queues(). Then in
ath12k_mac_op_reconfig_complete() those queues are not started
because ieee80211_wake_queues() is skipped due to the check on
reconfig_type. The result is that mac80211
could not deliver any frame to ath12k to send out, finally making
connection fail.
[84473.104249] PM: suspend exit
[84479.372397] wlan0: no VHT 160 MHz capability on 5 GHz, limiting to 80 MHz
[84479.372401] wlan0: determined local STA to be EHT, BW limited to 80 MHz
[84479.372416] wlan0: determined AP 00:03:7f:12:b7:b7 to be HE
[84479.372420] wlan0: connecting with HE mode, max bandwidth 80 MHz
[84479.580348] wlan0: authenticate with 00:03:7f:12:b7:b7 (local address=00:03:7f:37:11:53)
[84479.580351] wlan0: send auth to 00:03:7f:12:b7:b7 (try 1/3)
[84480.698993] wlan0: send auth to 00:03:7f:12:b7:b7 (try 2/3)
[84481.816505] wlan0: send auth to 00:03:7f:12:b7:b7 (try 3/3)
[84482.810966] wlan0: authentication with 00:03:7f:12:b7:b7 timed out
Actually we don't need to stop/start queues during suspend/resume,
so remove ath12k_core_pre_reconfigure_recovery() from ath12k_core_restart().
This won't cause any regression because currently the only chance
ath12k_core_restart() gets called is in reset case, where ab->is_reset
is set so that function will never be executed.
Also remove ath12k_core_post_reconfigure_recovery() because it is
not needed in suspend/resume case. This is also valid due to above
analysis.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240412060620.27519-9-quic_bqiang@quicinc.com
Currently pktlog is stopped in suspend callback and started in
resume callback, and in either scenarios it's basically to
delete/modify ab->mon_reap_timer and to purge related rings.
For WCN7850 it's pointless because pktlog is not enabled: both
ab->mon_reap_timer and those rings are not initialized.
So remove pktlog handling in suspend/resume callbacks. And
further, remove these two functions and related callee because
no one is calling them.
Other chips are not affected because now only WCN7850 supports
suspend/resume.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240412060620.27519-8-quic_bqiang@quicinc.com
In order to send out all packets before going to suspend, current code
adds a 500ms delay as a workaround. It is a rough estimate and may not
work.
Fix this by checking packet counters, if counters become zero, then all
packets are sent out or dropped.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240412060620.27519-7-quic_bqiang@quicinc.com
Currently buf_len field of ath12k_mhi_config_wcn7850 is assigned
with 0, making MHI use a default size, 64KB, to allocate channel
buffers. This is likely to fail in some scenarios where system
memory is highly fragmented and memory compaction or reclaim is
not allowed.
For now we haven't get any failure report on this in ath12k, but
there indeed is one such case in ath11k [1].
Actually those buffers are used only by QMI target -> host communication.
And for WCN7850, the largest packet size for that is less than 6KB. So
change buf_len field to 8KB, which results in order 1 allocation if page
size is 4KB. In this way, we can at least save some memory, and as well
as decrease the possibility of allocation failure in those scenarios.
[1] https://lore.kernel.org/ath11k/96481a45-3547-4d23-ad34-3a8f1d90c1cd@suse.cz/
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240412060620.27519-6-quic_bqiang@quicinc.com
We are seeing below warning in both reset and suspend/resume scenarios:
[ 4153.776040] ath12k_pci 0000:04:00.0: Already processed, so ignoring dma ring caps
This is because ab->num_db_cap is not cleared in
ath12k_wmi_free_dbring_caps(), so clear it to avoid such
warnings.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240412060620.27519-5-quic_bqiang@quicinc.com
Both the firmware reset feature and the power management
suspend/resume feature share common power-down and power-up
functionality. One aspect of the power-up functionality is
the handling of the ATH12K_QMI_EVENT_FW_INIT_DONE event.
When this event is received, a call is made to
ath12k_hal_dump_srng_stats(), with the purpose to collect
information that may be useful in debugging the cause of a
firmware reset.
Unfortunately, since this functionality is shared between
both the firmware reset path and the power management
resume path, the kernel log is flooded with messages during
resume. Since these messages are not useful during resume,
and in fact can be confusing and can increase the time it
takes to resume, update the logic to only call
ath12k_hal_dump_srng_stats() during firmware reset.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240412060620.27519-4-quic_bqiang@quicinc.com
There is no driver to match these two channels, so
remove them. This fixes warnings from MHI subsystem during suspend:
mhi mhi0_LOOPBACK: 1: Failed to reset channel, still resetting
mhi mhi0_LOOPBACK: 0: Failed to reset channel, still resetting
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240412060620.27519-3-quic_bqiang@quicinc.com
For non-WoW suspend/resume, ath12k host powers down whole hardware
when suspend and powers up it when resume, the code path it goes
through is very like the ath12k reset logic.
In order to reuse that logic, rearrange IRQ handling in the reset
path.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240412060620.27519-2-quic_bqiang@quicinc.com
Currently during resume, QMI target memory is not properly handled, resulting
in kernel crash in case DMA remap is not supported:
BUG: Bad page state in process kworker/u16:54 pfn:36e80
page: refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x36e80
page dumped because: nonzero _refcount
Call Trace:
bad_page
free_page_is_bad_report
__free_pages_ok
__free_pages
dma_direct_free
dma_free_attrs
ath12k_qmi_free_target_mem_chunk
ath12k_qmi_msg_mem_request_cb
The reason is:
Once ath12k module is loaded, firmware sends memory request to host. In case
DMA remap not supported, ath12k refuses the first request due to failure in
allocating with large segment size:
ath12k_pci 0000:04:00.0: qmi firmware request memory request
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 7077888
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 8454144
ath12k_pci 0000:04:00.0: qmi dma allocation failed (7077888 B type 1), will try later with small size
ath12k_pci 0000:04:00.0: qmi delays mem_request 2
ath12k_pci 0000:04:00.0: qmi firmware request memory request
Later firmware comes back with more but small segments and allocation
succeeds:
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 262144
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 524288
ath12k_pci 0000:04:00.0: qmi mem seg type 4 size 65536
ath12k_pci 0000:04:00.0: qmi mem seg type 1 size 524288
Now ath12k is working. If suspend is triggered, firmware will be reloaded
during resume. As same as before, firmware requests two large segments at
first. In ath12k_qmi_msg_mem_request_cb() segment count and size are
assigned:
ab->qmi.mem_seg_count == 2
ab->qmi.target_mem[0].size == 7077888
ab->qmi.target_mem[1].size == 8454144
Then allocation failed like before and ath12k_qmi_free_target_mem_chunk()
is called to free all allocated segments. Note the first segment is skipped
because its v.addr is cleared due to allocation failure:
chunk->v.addr = dma_alloc_coherent()
Also note that this leaks that segment because it has not been freed.
While freeing the second segment, a size of 8454144 is passed to
dma_free_coherent(). However remember that this segment is allocated at
the first time firmware is loaded, before suspend. So its real size is
524288, much smaller than 8454144. As a result kernel found we are freeing
some memory which is in use and thus crashed.
So one possible fix would be to free those segments during suspend. This
works because with them freed, ath12k_qmi_free_target_mem_chunk() does
nothing: all segment addresses are NULL so dma_free_coherent() is not called.
But note that ath11k has similar logic but never hits this issue. Reviewing
code there shows the luck comes from QMI memory reuse logic. So the decision
is to port it to ath12k. Like in ath11k, the crash is avoided by adding
prev_size to target_mem_chunk structure and caching real segment size in it,
then prev_size instead of current size is passed to dma_free_coherent(),
no unexpected memory is freed now.
Also reuse m3 buffer.
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4
Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0.c5-00481-QCAHMTSWPL_V1.0_V2.0_SILICONZ-3
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://msgid.link/20240419034034.2842-1-quic_bqiang@quicinc.com
