The 'icc_bw_lock' mutex is introduced in commit af42269c35
("interconnect: Fix locking for runpm vs reclaim") in order to decouple
serialization of bw aggregation from codepaths that require memory
allocation.
However commit d30f83d278 ("interconnect: core: Add dynamic id
allocation support") added a devm_kasprintf() call into a path protected
by the 'icc_bw_lock' which causes the following lockdep warning on
machines like the Lenovo ThinkPad X13s:
======================================================
WARNING: possible circular locking dependency detected
6.16.0-rc3 #15 Not tainted
------------------------------------------------------
(udev-worker)/342 is trying to acquire lock:
ffffb973f7ec4638 (fs_reclaim){+.+.}-{0:0}, at: __kmalloc_node_track_caller_noprof+0xa0/0x3e0
but task is already holding lock:
ffffb973f7f7f0e8 (icc_bw_lock){+.+.}-{4:4}, at: icc_node_add+0x44/0x154
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (icc_bw_lock){+.+.}-{4:4}:
icc_init+0x48/0x108
do_one_initcall+0x64/0x30c
kernel_init_freeable+0x27c/0x500
kernel_init+0x20/0x1d8
ret_from_fork+0x10/0x20
-> #0 (fs_reclaim){+.+.}-{0:0}:
__lock_acquire+0x136c/0x2114
lock_acquire+0x1c8/0x354
fs_reclaim_acquire+0x74/0xa8
__kmalloc_node_track_caller_noprof+0xa0/0x3e0
devm_kmalloc+0x54/0x124
devm_kvasprintf+0x74/0xd4
devm_kasprintf+0x58/0x80
icc_node_add+0xb4/0x154
qcom_osm_l3_probe+0x20c/0x314 [icc_osm_l3]
platform_probe+0x68/0xd8
really_probe+0xc0/0x38c
__driver_probe_device+0x7c/0x160
driver_probe_device+0x40/0x110
__driver_attach+0xfc/0x208
bus_for_each_dev+0x74/0xd0
driver_attach+0x24/0x30
bus_add_driver+0x110/0x234
driver_register+0x60/0x128
__platform_driver_register+0x24/0x30
osm_l3_driver_init+0x20/0x1000 [icc_osm_l3]
do_one_initcall+0x64/0x30c
do_init_module+0x58/0x23c
load_module+0x1df8/0x1f70
init_module_from_file+0x88/0xc4
idempotent_init_module+0x188/0x280
__arm64_sys_finit_module+0x6c/0xd8
invoke_syscall+0x48/0x110
el0_svc_common.constprop.0+0xc0/0xe0
do_el0_svc+0x1c/0x28
el0_svc+0x4c/0x158
el0t_64_sync_handler+0xc8/0xcc
el0t_64_sync+0x198/0x19c
other info that might help us debug this:
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(icc_bw_lock);
lock(fs_reclaim);
lock(icc_bw_lock);
lock(fs_reclaim);
*** DEADLOCK ***
The icc_node_add() functions is not designed to fail, and as such it
should not do any memory allocation. In order to avoid this, add a new
helper function for the name generation to be called by drivers which
are using the new dynamic id feature.
Fixes: d30f83d278 ("interconnect: core: Add dynamic id allocation support")
Signed-off-by: Gabor Juhos <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Co-developed-by: Johan Hovold <[email protected]>
Signed-off-by: Johan Hovold <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
When memory allocation fails during creating the name of the nodes in
icc_clk_register(), the code continues on the error path and it calls
icc_nodes_remove() to destroy the already created nodes. However that
function only destroys the nodes which were already added to the provider
and the newly created nodes are never destroyed in case of error.
In order to avoid a memory leaks, change the code to destroy the newly
created nodes explicitly in case of memory allocation failures.
Fixes: 44c5aa73cc ("interconnect: icc-clk: check return values of devm_kasprintf()")
Signed-off-by: Gabor Juhos <[email protected]>
Reviewed-by: Bartosz Golaszewski <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
Since commit d30f83d278 ("interconnect: core: Add dynamic id allocation
support"), interconnect node ids greater than or equal to ICC_DYN_ID_START
are reserved for dynamic id allocation. Yet the icc_node_create_nolock()
function allows to directly use such ids for creating nodes. This can
cause problems by executing dynamic id related codepaths even for nodes
intended to use static ids.
For example, the 'nsscc-ipq9574' driver creates interconnect nodes with
static ids starting from 19148. Because these ids belongs to the dynamic
id range, the icc_node_add() function replaces the node names
unexpectedly.
The node names looked like this before the change:
# grep nss_cc /sys/kernel/debug/interconnect/interconnect_summary
nss_cc_nssnoc_ppe_clk_master 0 0
nss_cc_nssnoc_ppe_clk_slave 0 0
nss_cc_nssnoc_ppe_cfg_clk_master 0 0
...
And those have an unexpected suffix now:
# grep nss_cc /sys/kernel/debug/interconnect/interconnect_summary
[email protected] 0 0
[email protected] 0 0
[email protected] 0 0
...
Increase the value of ICC_DYN_ID_START to avoid this.
Also, add sanity check to the icc_node_create_nolock() function to prevent
directly creating nodes with ids reserved for dynamic allocation in order
to detect these kind of problems.
Fixes: d30f83d278 ("interconnect: core: Add dynamic id allocation support")
Signed-off-by: Gabor Juhos <[email protected]>
Reviewed-by: Dmitry Baryshkov <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
Add Epoch Subsystem (EPSS) L3 provider support on SA8775P SoCs.
Current interconnect framework is based on static IDs for creating node
and registering with framework. This becomes a limitation for topologies
where there are multiple instances of same interconnect provider.
Modified interconnect framework APIs to create and link icc node with
dynamic IDs, this will help to overcome the dependency on static IDs.
* icc-sa8775p
dt-bindings: interconnect: Add EPSS L3 compatible for SA8775P
interconnect: core: Add dynamic id allocation support
interconnect: qcom: Add multidev EPSS L3 support
interconnect: qcom: icc-rpmh: Add dynamic icc node id support
interconnect: qcom: sa8775p: Add dynamic icc node id support
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
To facilitate dynamic id allocation, discard the static IDs
from node data and set alloc_dyn_id in descriptor structure
to indicate dynamic ID allocation. Update the topology to use
node pointers for links instead of static IDs and forward declare
the node pointer to avoid undefined references.
Signed-off-by: Raviteja Laggyshetty <[email protected]>
Reviewed-by: Dmitry Baryshkov <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
Interconnect framework relies on static IDs for creating,
linking and maintaning the topology. This dependency on static
IDs prevents creating two instances of same provider. To overcome
the dependency on static IDs, dynamic ID support is being added.
To facilitate dynamic node ID support, the driver now uses
node pointers for links instead of static node IDs and icc_node
pointer is added as member in qcom_icc_node structure to track
the node creation.
Signed-off-by: Raviteja Laggyshetty <[email protected]>
Reviewed-by: Dmitry Baryshkov <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
EPSS on SA8775P has two instances, necessitating the creation of two
device nodes with different compatibles due to the unique ICC node ID
and name limitations in the interconnect framework. Add multidevice
support for the OSM-L3 provider to dynamically obtain unique node IDs
and register with the framework.
EPSS topology includes a single master-slave pair within the same
provider, the node linking logic is simplified by directly connecting
the master node to the slave node.
Signed-off-by: Raviteja Laggyshetty <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
The current interconnect framework relies on static IDs for node
creation and registration, which limits topologies with multiple
instances of the same interconnect provider. To address this,
introduce icc_node_create_dyn() and icc_link_nodes() APIs to
dynamically allocate IDs for interconnect nodes during creation
and link. This change removes the dependency on static IDs,
allowing multiple instances of the same hardware, such as EPSS L3.
Signed-off-by: Raviteja Laggyshetty <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
Pull interconnect changes from Georgi:
interconnect changes for 6.14
This pull request contains the interconnect changes for the 6.14-rc1 merge
window. It contains one new driver and DT documentation updates for L3
and bandwidth monitors.
Driver changes:
- New driver for the SM8750 platform
- Add DT compatibles for QCS615 BWMON and SM8650 OSM
Signed-off-by: Georgi Djakov <[email protected]>
* tag 'icc-6.14-rc1' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/djakov/icc:
dt-bindings: interconnect: qcom,msm8998-bwmon: Add SM8750 CPU BWMONs
dt-bindings: interconnect: OSM L3: Document sm8650 OSM L3 compatible
dt-bindings: interconnect: qcom-bwmon: Document QCS615 bwmon compatibles
interconnect: sm8750: Add missing const to static qcom_icc_desc
interconnect: qcom: Add interconnect provider driver for SM8750
dt-bindings: interconnect: add interconnect bindings for SM8750
Add driver for the network of connects present on the SAR2130P platform.
* icc-sar2130p
dt-bindings: interconnect: qcom: document SAR2130P NoC
interconnect: qcom: add support for SAR2130P
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
Add driver for the Qualcomm interconnect buses found in QCS8300
based platforms. The topology consists of several NoCs that are
controlled by a remote processor that collects the aggregated
bandwidth for each master-slave pairs.
Signed-off-by: Raviteja Laggyshetty <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
After commit 0edb555a65 ("platform: Make platform_driver::remove()
return void") .remove() is (again) the right callback to implement for
platform drivers.
Convert all platform drivers below drivers/interconnect to use
.remove(), with the eventual goal to drop struct
platform_driver::remove_new(). As .remove() and .remove_new() have the
same prototypes, conversion is done by just changing the structure
member name in the driver initializer.
Signed-off-by: Uwe Kleine-König <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
Add driver for the Qualcomm interconnect buses found in QCS615
based platforms. The topology consists of several NoCs that are
controlled by a remote processor that collects the aggregated
bandwidth for each master-slave pairs.
Signed-off-by: Raviteja Laggyshetty <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
rename qos_clks_required flag to qos_requires_clocks in qcom_icc_desc
structure. This flag indicates that clocks are required for accessing
and programming the QoS registers.
Suggested-by: Bjorn Andersson <[email protected]>
Signed-off-by: Raviteja Laggyshetty <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
Return -EPROBE_DEFER from interconnect provider incase probe defer is
received from devm_clk_bulk_get_all(). This would help in reattempting
the inteconnect driver probe, once the required QoS clocks are
available.
Suggested-by: Bjorn Andersson <[email protected]>
Signed-off-by: Raviteja Laggyshetty <[email protected]>
Reviewed-by: Konrad Dybcio <[email protected]>
Fixes: 0a7be6b35d ("interconnect: qcom: icc-rpmh: Add QoS configuration support")
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
A set of fixes that target stability of the SM8350 platform.
* icc-sm8350
interconnect: qcom: sm8350: drop DISP nodes
dt-bindings: interconnect: qcom,sm8350: drop DISP nodes
interconnect: qcom: sm8250: Enable sync_state
Link: https://lore.kernel.org/r/[email protected]/
Signed-off-by: Georgi Djakov <[email protected]>
This series introduce new ICC drivers for some legacy socs
while at it also updates a bit of qcs404 driver which seems
to not receive much attention lately.
* icc-misc
dt-bindings: interconnect: qcom: Add Qualcomm MSM8976 NoC
interconnect: qcom: Add MSM8976 interconnect provider driver
dt-bindings: interconnect: qcom: Add Qualcomm MSM8937 NoC
interconnect: qcom: Add MSM8937 interconnect provider driver
interconnect: qcom: qcs404: Mark AP-owned nodes as such
interconnect: qcom: qcs404: Add regmaps and more bus descriptions
dt-bindings: interconnect: qcom: msm8939: Fix example
interconnect: qcom: msm8953: Add ab_coeff
dt-bindings: interconnect: qcom: msm8953: Fix 'See also' in description
Link: https://lore.kernel.org/all/[email protected]/
Signed-off-by: Georgi Djakov <[email protected]>
Enable the generic icc sync_state callback to ensure interconnect votes
are actually taken into account, instead of being forced to the maximum
value.
Fixes: b95b668eaa ("interconnect: qcom: icc-rpmh: Add BCMs to commit list in pre_aggregate")
Signed-off-by: Dmitry Baryshkov <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
The msm-5.x kernels have additional display RSC and separate display BCM
voter. Since upstream kernel doesn't yet provide display RSC, we end up
duplicating several nodes, which can result in incorrect votes being
cast. Drop *_DISP nodes.
Fixes: d26a566744 ("interconnect: qcom: Add SM8350 interconnect provider driver")
Signed-off-by: Dmitry Baryshkov <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
With the new __counted_by annotation, the "num_nodes" struct member must
be set before accessing the "nodes" array. This initialization was done
in other places where a new struct icc_onecell_data is allocated, but this
case in icc_clk_register() was missed. Set "num_nodes" after allocation.
Fixes: dd4904f3b9 ("interconnect: qcom: Annotate struct icc_onecell_data with __counted_by")
Signed-off-by: Kees Cook <[email protected]>
Reviewed-by: Gustavo A. R. Silva <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
Pull char / misc and other driver updates from Greg KH:
"Here is the "big" set of char/misc and other driver subsystem changes
for 6.11-rc1. Nothing major in here, just loads of new drivers and
updates. Included in here are:
- IIO api updates and new drivers added
- wait_interruptable_timeout() api cleanups for some drivers
- MODULE_DESCRIPTION() additions for loads of drivers
- parport out-of-bounds fix
- interconnect driver updates and additions
- mhi driver updates and additions
- w1 driver fixes
- binder speedups and fixes
- eeprom driver updates
- coresight driver updates
- counter driver update
- new misc driver additions
- other minor api updates
All of these, EXCEPT for the final Kconfig build fix for 32bit
systems, have been in linux-next for a while with no reported issues.
The Kconfig fixup went in 29 hours ago, so might have missed the
latest linux-next, but was acked by everyone involved"
* tag 'char-misc-6.11-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc: (330 commits)
misc: Kconfig: exclude mrvl-cn10k-dpi compilation for 32-bit systems
misc: delete Makefile.rej
binder: fix hang of unregistered readers
misc: Kconfig: add a new dependency for MARVELL_CN10K_DPI
virtio: add missing MODULE_DESCRIPTION() macro
agp: uninorth: add missing MODULE_DESCRIPTION() macro
spmi: add missing MODULE_DESCRIPTION() macros
dev/parport: fix the array out-of-bounds risk
samples: configfs: add missing MODULE_DESCRIPTION() macro
misc: mrvl-cn10k-dpi: add Octeon CN10K DPI administrative driver
misc: keba: Fix missing AUXILIARY_BUS dependency
slimbus: Fix struct and documentation alignment in stream.c
MAINTAINERS: CC dri-devel list on Qualcomm FastRPC patches
misc: fastrpc: use coherent pool for untranslated Compute Banks
misc: fastrpc: support complete DMA pool access to the DSP
misc: fastrpc: add missing MODULE_DESCRIPTION() macro
misc: fastrpc: Add missing dev_err newlines
misc: fastrpc: Use memdup_user()
nvmem: core: Implement force_ro sysfs attribute
nvmem: Use sysfs_emit() for type attribute
...
Presently, icc-clk driver autogenerates the master and slave ids.
However, devices with multiple nodes on the interconnect could
have other constraints and may not match with the auto generated
node ids.
Hence, modify the driver to use the master/slave ids provided by
the caller instead of auto generating.
Also, update clk-cbf-8996 accordingly.
Acked-by: Georgi Djakov <[email protected]>
Signed-off-by: Varadarajan Narayanan <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Bjorn Andersson <[email protected]>
This series adds QoS support for QNOC type device which can be found on
SC7280 platform. It adds support for programming priority,
priority forward disable and urgency forwarding. This helps in
priortizing the traffic originating from different interconnect masters
at NOC (Network On Chip).
* icc-rpmh-qos
dt-bindings: interconnect: add clock property to enable QOS on SC7280
interconnect: qcom: icc-rpmh: Add QoS configuration support
interconnect: qcom: sc7280: enable QoS configuration
interconnect: qcom: Fix DT backwards compatibility for QoS
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
Add driver for interconnect busses found in MSM8953 based platforms.
The topology consists of four NoCs that are partially controlled by a
RPM processor.
Note that one of NoCs (System NoC) has a counterpart (System NoC MM)
that is modelled as child device to avoid resource conflicts, since it
uses same MMIO space for configuration.
Signed-off-by: Vladimir Lypak <[email protected]>
Reviewed-by: Krzysztof Kozlowski <[email protected]>
Signed-off-by: Barnabás Czémán <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Georgi Djakov <[email protected]>
Gabor Juhos