Fix clk issue - #1045
Open
Jie Gan (jiegan0107) wants to merge 3 commits into
Open
Conversation
…ailure Each CoreSight platform_driver's probe() enables its clocks via coresight_get_enable_clocks() -> devm_clk_get_optional_enabled(), which registers a devm cleanup to run clk_disable_unprepare() on driver detach. The probe wrapper then unconditionally calls pm_runtime_put() regardless of whether the inner probe succeeded, so on failure this also fires runtime_suspend() and disables the same clocks a first time. The driver core then unwinds the failed probe and runs the devm cleanup, disabling them a second time and underflowing the refcount: coresight-etm4x etm0: probe with driver coresight-etm4x failed with error -22 ------------[ cut here ]------------ qdss_clk already disabled WARNING: CPU: 1 PID: 432 at drivers/clk/clk.c:1188 clk_core_disable+0x1d0/0x218 ... ------------[ cut here ]------------ Unpreparing enabled qdss_clk WARNING: CPU: 0 PID: 432 at drivers/clk/clk.c:1061 clk_core_unprepare+0x248/0x268 ... qdss_clk is shared by every CoreSight node, so the extra disable drives its refcount to 0 while sibling devices still expect it enabled. The next funnel to probe then touches unclocked hardware and panics: SError Interrupt on CPU1, code 0x00000000be000000 -- SError Kernel panic - not syncing: Asynchronous SError Interrupt ... coresight_clear_self_claim_tag+0x7c/0x1e0 [coresight] (P) funnel_probe+0x114/0x2e0 [coresight_funnel] dynamic_funnel_probe+0x24/0x70 [coresight_funnel] Use pm_runtime_put_noidle() instead of pm_runtime_put() on the failure path so it drops the usage count without invoking runtime_suspend(), leaving the devm cleanup as the sole disabler. Affects catu, ctcu, etm4x, funnel, replicator, stm, tmc, tpiu and tnoc, all of which share this probe skeleton. Link: https://lore.kernel.org/all/20260907-fix-clk-issue-v1-1-efe81fa2b697@oss.qualcomm.com/ Fixes: 1abc1b2 ("coresight: Appropriately disable programming clocks") Signed-off-by: Jie Gan <jie.gan@oss.qualcomm.com>
coresight_get_enable_clocks() enables the programming clock and the optional AT clock through devm_clk_get_optional_enabled(), which also registers a devm action to call clk_disable_unprepare() when the driver detaches. After probe, pm_runtime_put() allows the device to suspend and the runtime suspend callback disables the same clocks. During remove the device is left runtime suspended, so pm_runtime_disable() freezes it with the clocks already disabled. The devm cleanup that runs afterwards calls clk_disable_unprepare() a second time, underflowing the clock enable refcount. Resume the device with pm_runtime_get_sync() before tearing it down so the clocks are enabled again and balance the devm-managed disable. Then pm_runtime_set_suspended() and pm_runtime_put_noidle() leave the device in a coherent runtime PM state (suspended, usage count balanced) once the devm action has disabled the clocks. This affects all CoreSight platform drivers that obtain their clocks through coresight_get_enable_clocks(): catu, cpu-debug, ctcu, etm4x, funnel, replicator, stm, tmc and tpiu. Link: https://lore.kernel.org/all/20260710-fix-clock-refcount-unbalance-v3-1-a37a1fb17981@oss.qualcomm.com/ Fixes: 1abc1b2 ("coresight: Appropriately disable programming clocks") Reviewed-by: Yeoreum Yun <yeoreum.yun@arm.com> Reviewed-by: Leo Yan <leo.yan@arm.com> Signed-off-by: Jie Gan <jie.gan@oss.qualcomm.com>
…emove coresight_get_enable_clocks() enables the programming clock through devm_clk_get_optional_enabled(), which also registers a devm action to call clk_disable_unprepare() when the driver detaches. After probe, pm_runtime_put() allows the device to suspend and the runtime suspend callback disables the clock. During remove the device is left runtime suspended, so pm_runtime_disable() freezes it with the clock already disabled. The devm cleanup that runs afterwards calls clk_disable_unprepare() a second time, underflowing the clock enable refcount. Resume the device with pm_runtime_get_sync() before tearing it down so the clock is enabled again and balances the devm-managed disable. Then pm_runtime_set_suspended() and pm_runtime_put_noidle() leave the device in a coherent runtime PM state (suspended, usage count balanced) once the devm action has disabled the clock. Link: https://lore.kernel.org/all/20260710-fix-clock-refcount-unbalance-v3-2-a37a1fb17981@oss.qualcomm.com/ Fixes: 1abc1b2 ("coresight: Appropriately disable programming clocks") Reviewed-by: Leo Yan <leo.yan@arm.com> Signed-off-by: Jie Gan <jie.gan@oss.qualcomm.com>
Jie Gan (jiegan0107)
requested review from
a team,
Amit Kucheria (idlethread),
knaveen-qc and
Mukesh Ojha (mukeshojha-linux)
September 7, 2026 06:40
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
CRs-Fixed: 4668530