The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is ignored (apart
from emitting a warning) and this typically results in resource leaks.
To improve here there is a quest to make the remove callback return
void. In the first step of this quest all drivers are converted to
.remove_new(), which already returns void. Eventually after all drivers
are converted, .remove_new() will be renamed to .remove().
Several interconnect/qcom drivers use qnoc_remove() as remove callback.
Make this function return void (instead of unconditionally zero) and
adapt the drivers using this function accordingly.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Link: https://lore.kernel.org/r/20231031222851.3126434-12-u.kleine-koenig@pengutronix.de
Signed-off-by: Georgi Djakov <djakov@kernel.org>
This single node has its own clock which seems to be responsible for
transactions between CPUSS (CPU + some stuff) and the GNOC. See [1]
for reference.
Define it and hook it up.
[1] 02f8c342b2%5E%21/#F0
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Link: https://lore.kernel.org/r/20230726-topic-icc_coeff-v4-5-c04b60caa467@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
If this hardware couldn't get messier, some nodes are supposed to drive
their own bus clock.. Presumably to connect to some intermediate
interface between the node itself and the bus it's (supposed to be)
connected to.
Expand the node struct with the necessary data and hook up the
allocations & calculations.
Note that the node-specific AB/IB coefficients contribute (by design)
to both the node-level and the bus-level aggregation.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Link: https://lore.kernel.org/r/20230726-topic-icc_coeff-v4-3-c04b60caa467@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Presumably due to the hardware being so complex, some nodes (or busses)
have different (usually higher) requirements for bandwidth than what
the usual calculations would suggest.
Looking at the available downstream files, it seems like AB values are
adjusted per-bus and IB values are adjusted per-node.
With that in mind, introduce percentage-based coefficient struct members
and use them in the calculations.
One thing to note is that the IB coefficient is inverse (100/ib_percent)
which feels a bit backwards, but it's necessary for precision..
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Link: https://lore.kernel.org/r/20230726-topic-icc_coeff-v4-1-c04b60caa467@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
SMD RPM only provides two buckets, one each for the active-only and
active-sleep RPM contexts. Use the correct constant to allocate and
operate on them.
This will make the qcom,icc.h header no longer work with this driver,
mostly because.. it was never meant to! The commit that introduced
bucket support to SMD RPM was trying to shove a square into a round
hole and it did not work out very well. That said, there are no
active users of SMD RPM ICC + qcom,icc.h, so that doesn't hurt.
Fixes: dcbce7b0a7 ("interconnect: qcom: icc-rpm: Support multiple buckets")
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Stephan Gerhold <stephan@gerhold.net>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Acked-by: Georgi Djakov <djakov@kernel.org>
Link: https://lore.kernel.org/r/20230526-topic-smd_icc-v7-19-09c78c175546@linaro.org
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
The sole purpose of bus clocks that were previously registered with
rpmcc was to convey the aggregated bandwidth to RPM. There's no good
reason to keep them outside the interconnect framework, as it only
adds to the plentiful complexity.
Add the required code to handle these clocks from within SMD RPM ICC.
RPM-owned bus clocks are no longer considered a thing, but sadly we
have to allow for the existence of HLOS-owned bus clocks, as some
(mostly older) SoCs (ab)use these for bus scaling (e.g. MSM8998 and
&mmcc AHB_CLK_SRC).
This in turn is trivially solved with a single *clk, which is filled
and used iff qp.bus_clk_desc is absent and we have a "bus" clock-names
entry in the DT node.
This change should(tm) be fully compatible with all sorts of old
Device Trees as far as the interconnect functionality goes (modulo
abusing bus clock handles or wrongly using the qcom,icc.h binding,
but that's a mistake in and of itself).
Reviewed-by: Stephan Gerhold <stephan@gerhold.net>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Acked-by: Georgi Djakov <djakov@kernel.org>
Link: https://lore.kernel.org/r/20230526-topic-smd_icc-v7-17-09c78c175546@linaro.org
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Add the definitions for RPM bus clocks that will be used by many
different platforms.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Acked-by: Georgi Djakov <djakov@kernel.org>
Link: https://lore.kernel.org/r/20230526-topic-smd_icc-v7-10-09c78c175546@linaro.org
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Introduce qcom_icc_rpm_set_bus_rate() in preparation for handling RPM
clock resources within the interconnect framework. This lets us greatly
simplify all of the code handling, as setting the rate comes down to:
u32 rate_khz = max(clk.sleep_rate, clk.active_rate, clk_a.active_rate)
write_to_rpm(clock.description, rate_khz);
Reviewed-by: Stephan Gerhold <stephan@gerhold.net>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Acked-by: Georgi Djakov <djakov@kernel.org>
Link: https://lore.kernel.org/r/20230526-topic-smd_icc-v7-9-09c78c175546@linaro.org
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
smd-rpm.h is not very useful as-is and both files are always included
anyway.. Combine them.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Acked-by: Georgi Djakov <djakov@kernel.org>
Link: https://lore.kernel.org/r/20230526-topic-smd_icc-v7-8-09c78c175546@linaro.org
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Currently the header does not provide all the required dependencies.
Fix it.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Acked-by: Georgi Djakov <djakov@kernel.org>
Link: https://lore.kernel.org/r/20230526-topic-smd_icc-v7-7-09c78c175546@linaro.org
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
The downstream kernel employs the concept of "keeping the bus alive"
by voting for the minimum (XO/19.2MHz) rate at all times on certain
(well, most) buses. This is a very important thing to have, as if we
either have a lackluster/wrong DT that doesn't specify a (high enough)
vote on a certain bus, we may lose access to the entire bus altogether.
This is very apparent when we only start introducing interconnect
support on a given platform and haven't yet introduced voting on all
peripherals.
The same can happen if we only have a single driver casting a vote on
a certain bus and that driver exits/crashes/suspends.
The keepalive vote is limited to the ACTIVE bucket, as keeping a
permanent vote on the SLEEP one could prevent the platform from properly
entering low power mode states.
Introduce the very same concept, with a slight twist: the vendor
kernel checks whether the rate is zero before setting the minimum
vote, but that's rather silly, as in doing so we're at the mercy
of CCF. Instead, explicitly clamp the rates to always be >= 19.2 MHz
for providers with keep_alive=true.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Acked-by: Georgi Djakov <djakov@kernel.org>
Link: https://lore.kernel.org/r/20230526-topic-smd_icc-v7-6-09c78c175546@linaro.org
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
For SMD RPM bus scaling to work, we need a pair of sleep-wake clocks.
The variable number of them we previously supported was only a hack
to keep the clocks required for QoS register access, but now that
these are separated, we can leave bus_clks to the actual bus clocks.
In cases where there is no actual bus scaling (such as A0NoC on MSM8996
and GNoC on SDM660 where the HLOS is only supposed to program the QoS
registers and the bus is either static or controlled remotely), allow
for no clock scaling with a boolean property.
Remove all the code related to allowing an arbitrary number of bus_clks,
replace the number by BUS_CLK_MAX (= 2) and guard the bus clock paths
to ensure they are not taken on non-scaling buses.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Link: https://lore.kernel.org/r/20230228-topic-qos-v8-6-ee696a2c15a9@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Some (but not all) providers (or their specific nodes) require
specific clocks to be turned on before they can be accessed. Failure
to ensure that results in a seemingly random system crash (which
would usually happen at boot with the interconnect driver built-in),
resulting in the platform not booting up properly.
Limit the number of bus_clocks to 2 (which is the maximum that SMD
RPM interconnect supports anyway) and handle non-scaling clocks
separately. Update MSM8996 and SDM660 drivers to make sure they do
not regress with this change.
This unfortunately has to be done in one patch to prevent either
compile errors or broken bisect.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Link: https://lore.kernel.org/r/20230518195801.2556998-1-konrad.dybcio@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
In preparation for handling non-scaling clocks that we still have to
enable, rename num_clocks to more descriptive num_bus_clocks.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Link: https://lore.kernel.org/r/20230228-topic-qos-v8-2-ee696a2c15a9@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Rename the "clocks" (and _names) fields of qcom_icc_desc to
"bus_clocks" in preparation for introducing handling of clocks that
need to be enabled but not voted on with aggregate frequency.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Link: https://lore.kernel.org/r/20230228-topic-qos-v8-1-ee696a2c15a9@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Sort the kerneldoc entries the same way the struct members are
sorted.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20230228-topic-qos-v7-3-815606092fff@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Some nodes, like EBI0 (DDR) or L3/LLCC, may be connected over more than
one channel. This should be taken into account in bandwidth calcualtion,
as we're supposed to feed msmbus with the per-channel bandwidth. Add
support for specifying that and use it during bandwidth aggregation.
Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20230228-topic-qos-v7-2-815606092fff@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Currently NOC_QOS_MODE_FIXED is defined as 0x0 which makes it the
default option (partial struct initialization). The default option
however should be NOC_QOS_MODE_INVALID.
That results in bogus QoS configurations being sent for port 0 (which
is used for the DRAM endpoint on BIMC, for example) coming from all nodes
with .qos.ap_owned = true and uninitialized .qos.qos_mode. It's also an
issue for newer SoCs where all nodes are treated as if they were ap_owned,
but not all of them have QoS configuration.
The NOC_QOS_MODEs are defined as preprocessor constants and are not used
anywhere outside qcom_icc_set_noc_qos(), which is easily worked around.
Separate the desc->type values from the values sent to msmbus in the
aforementioned function. Make the former an enum for better mainainability.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20230228-topic-qos-v7-1-815606092fff@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Any power domain would already have been attached by the platform bus
code so drop the bogus power domain attach which always succeeds from
probe.
This effectively reverts commit 7de109c0ab ("interconnect: icc-rpm:
Add support for bus power domain").
Fixes: 7de109c0ab ("interconnect: icc-rpm: Add support for bus power domain")
Cc: Yassine Oudjana <y.oudjana@protonmail.com>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Tested-by: Konrad Dybcio <konrad.dybcio@linaro.org> # MSM8996 Sony Kagura
Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Link: https://lore.kernel.org/r/20230313084953.24088-3-johan+linaro@kernel.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
The current interconnect rpm driver uses a single aggregate bandwidth to
calculate the clock rates for both active and sleep clocks; therefore,
it has no chance to separate bandwidth requests for these two kinds of
clocks.
This patch studies the implementation from interconnect rpmh driver to
support multiple buckets. The rpmh driver provides three buckets for
AMC, WAKE, and SLEEP; this driver only needs to use WAKE and SLEEP
buckets, but we keep the same way with rpmh driver, this can allow us to
reuse the DT binding and avoid to define duplicated data structures.
This patch introduces two callbacks: qcom_icc_pre_bw_aggregate() is used
to clean up bucket values before aggregate bandwidth requests, and
qcom_icc_bw_aggregate() is to aggregate bandwidth for buckets.
Signed-off-by: Leo Yan <leo.yan@linaro.org>
Link: https://lore.kernel.org/r/20220712015929.2789881-5-leo.yan@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
This patch set is to address two clock rate setting issues.
The first patch is to fix a potential cached clock rate mismatching
issue, the issue can lead to the clock rate is missed to be set. Note,
since this potential issue requires specific time window and certain
condition (consumers need to request the same bandwidth) to produce,
the patch is based on analysis but not a real trace log.
The second patch is an extension to cache clock rates for active and
sleep clocks separately, with this change it gives us possibility to set
active and sleep clock with different clock rates.
* icc-rpm
interconnect: qcom: icc-rpm: Fix for cached clock rate
interconnect: qcom: icc-rpm: Cache every clock rate
Link: https://lore.kernel.org/r/20220416031029.693211-1-leo.yan@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
The cached clock rate is used for all bus clocks, thus it has the
assumption that all interconnect clock rates are always same, this
causes trouble if we want to set different clock rates separately.
This patch is to allocate a clock rate array to cache every clock
rate.
Signed-off-by: Leo Yan <leo.yan@linaro.org>
Link: https://lore.kernel.org/r/20220416031029.693211-3-leo.yan@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
All nodes within an interconnect bus share interconnect bus clocks, but
every node has its own cached clock rate values, this can lead to
unexpected clock rate setting.
Let's see an example shown in below, in this case, a bus have two nodes
A and B, and its buswidth is 8:
step1: vote bandwidth 1600M for node(A):
aggregated(bw) = 1600M
qcom_icc_node(A)->rate = 1600M / 8 = 200MHz
step2: vote bandwidth 1600M for node(B):
aggregated(bw) = 1600M + 1600M = 3200M
qcom_icc_node(B)->rate = 3200M / 8 = 400MHz
step3: unvote bandwidth 1600M for node(A)
aggregated(bw) = 3200M - 1600M = 1600M
target_clock = 1600M / 8 = 200MHz
The problem is in step 3, the calculated target clock rate is 200MHz,
which equals to the cached clock rate in node(A) (See step 1),
unfortunately, qcom_icc_set() skips to set the new clock rate 200MHz in
this case, so the bus clock rate will continue to stay at 400MHz.
To resolve the issue, one possible solution is to invoke clk_get_rate()
to retrieve the clock rates on the fly, thus we can totally remove the
cached clock rates. But after review the code, many bus clock has set
the flag CLK_GET_RATE_NOCACHE, this results in the retrieving bus clock
rate is time cost for iterating parent clock nodes, and even challenges
bus clock drivers to provide recalc_rate() callbacks.
So this patch moves the cached rates into structure qcom_icc_provider,
we use it as a central place to maintain bus clock handlers and cached
clock rate, therefore, it can smoothly dismiss the mismatching problem.
Signed-off-by: Leo Yan <leo.yan@linaro.org>
Link: https://lore.kernel.org/r/20220416031029.693211-2-leo.yan@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Pointers to struct qcom_icc_node (and similar structures) are not
modified, so they can be made const for safety. The contents of struct
qcom_icc_node must stay non-const.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20220412102623.227607-2-krzysztof.kozlowski@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Add support for attaching to a power domain. This is required
for Aggregate 0 NoC on MSM8996, which is powered by a GDSC.
Signed-off-by: Yassine Oudjana <y.oudjana@protonmail.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Tested-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> #db820c
Link: https://lore.kernel.org/r/20211021132329.234942-3-y.oudjana@protonmail.com
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Drop DEFINE_QNODE macro which has become unused.
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@somainline.org>
Reviewed-by: Marijn Suijten <marijn.suijten@somainline.org>
Link: https://lore.kernel.org/r/20210903232421.1384199-12-dmitry.baryshkov@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
SDM660 driver expects to have QoS registers at the beginning of NoC
address space (sdm660 platform shifts NoC base address). Add support for
using QoS register offset, so that other platforms do not have to change
existing device trees.
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@somainline.org>
Tested-by: Shawn Guo <shawn.guo@linaro.org>
Link: https://lore.kernel.org/r/20210903232421.1384199-6-dmitry.baryshkov@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Other RPM interconnect drivers might also use QoS support. Move AP-owned
nodes support from SDM660 driver to common icc-rpm.c.
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@somainline.org>
Tested-by: Marijn Suijten <marijn.suijten@somainline.org>
Tested-by: Shawn Guo <shawn.guo@linaro.org>
Link: https://lore.kernel.org/r/20210903232421.1384199-5-dmitry.baryshkov@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
All icc-rpm drivers use the same set of bus clocks. Move handling of bus
clocks to qnoc_probe. This both simplifies the code and allows using
qnoc_probe as device's probe function.
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@somainline.org>
Reviewed-by: Marijn Suijten <marijn.suijten@somainline.org>
Tested-by: Shawn Guo <shawn.guo@linaro.org>
Link: https://lore.kernel.org/r/20210903232421.1384199-2-dmitry.baryshkov@linaro.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Add RPM based interconnect driver implements the set and aggregate
functionalities that translates bandwidth requests into RPM messages.
These modules provide a common set of functionalities for all
Qualcomm RPM based interconnect providers and should help reduce code
duplication when adding new providers.
Signed-off-by: Jun Nie <jun.nie@linaro.org>
Link: https://lore.kernel.org/r/20201204075345.5161-2-jun.nie@linaro.org
Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
