From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-alma10-1.taild15c8.ts.net [100.103.45.18]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id B04604EDCAD; Fri, 2 Oct 2026 16:43:47 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=100.103.45.18 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1790959428; cv=none; b=ewBsG7NgMIbeG22CfTIarKYqddEP8GZQwVT7AVN9KmJ0BXiKKspN3Zekk6WBPbWcGkd+OaOwg0Bo/OZEPrMExaECay30hJvqvMkscTI+LSpUaE8FBVkj9J35Y9hfeh0XAVK7AaQSYUcsZlrEQksyV7V4k0ztt6vnZLZln8SgFC0= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1790959428; c=relaxed/simple; bh=ANnzqIE88PpAbXvMgNi1vRap06Y/mTqIF6czgADO0g4=; h=Date:From:To:Cc:Subject:Message-ID:References:MIME-Version: Content-Type:Content-Disposition:In-Reply-To; b=Vrla6UOYBCJ8ahG2b4qjxjj8bIy8o5IH9YlreIID6ixNrdNumRNEX9djaQgQ/OZXzOzKabFmrVOYbRnD7Afm5m4PaHRvrPtZjXa8kCivZCmfV1Q9XsICKhzIeRhpA/logKNX/hUNR3oqt0+8v2T4d5qWOJ4reOpmIaPsclzRQ78= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b=QVkvXH4M; arc=none smtp.client-ip=100.103.45.18 Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="QVkvXH4M" Received: by smtp.kernel.org (Postfix) with ESMTPSA id EE9D71F000FF; Fri, 2 Oct 2026 16:43:44 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=kernel.org; s=k20260515; t=1790959427; bh=y9MRhlz4pzaoGqz6XBjvYnXMN79/O0P66ZDWtWt6Lv4=; h=Date:From:To:Cc:Subject:References:In-Reply-To; b=QVkvXH4MIkXiJq0i0bW6Gqqdu15m800MonrF8VWPSFxRDFxCfnKP9lVLCvJHIlwDF DuUH5d68DiioC64U3/jjVHOkJyKYeMLCRpsowq7friAXvljxxrtfWWhjSHP84YnWoB LD4qMtBBoJ+UWRUhpZssXzxjN2AhJ5QxKPUxF0AhDaplBSpuy/fYSIyGrW7BXGRyaj 5erZhJi3pLpFiENetIzPd/QJOvb5fNVhWkMBsRJLtrcEKxIhvG2Cyl+4ercWd1CieS uV/oEuGTs+oBCAgQMruT/CBXs+6Bn7Rj5cJwF4dJ+VWLIsbfue0LQ0+c+Qe7Bfp990 eRlT/l+AABUeA== Date: Fri, 2 Oct 2026 17:43:42 +0100 From: Will Deacon To: Prakash Gupta Cc: Robin Murphy , Joerg Roedel , Rob Clark , Connor Abbott , linux-arm-msm@vger.kernel.org, linux-arm-kernel@lists.infradead.org, iommu@lists.linux.dev, linux-kernel@vger.kernel.org, Akhil P Oommen , Pranjal Shrivastava , Pratyush Brahma Subject: Re: [PATCH v4] iommu/arm-smmu: Use pm_runtime in fault handlers Message-ID: References: <20260806-smmu-rpm-v4-1-8183d007331c@oss.qualcomm.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20260806-smmu-rpm-v4-1-8183d007331c@oss.qualcomm.com> On Thu, Aug 06, 2026 at 03:52:41PM +0530, Prakash Gupta wrote: > Commit d4a44f0750bb ("iommu/arm-smmu: Invoke pm_runtime across the driver") > enabled pm_runtime for the arm-smmu device. On systems where the SMMU > sits in a power domain, all register accesses must be done while the > device is runtime active to avoid unclocked register reads and > potential NoC errors. > > So far, this has not been an issue for most SMMU clients because > stall-on-fault is enabled by default. While a translation fault is > being handled, the SMMU stalls further translations for that context > bank, so the fault handler would not race with a powered-down SMMU. > > Adreno SMMU now disables stall-on-fault in the presence of fault > storms to avoid saturating SMMU resources and hanging the GMU. With > stall-on-fault disabled, the SMMU can generate faults while its power > domain may no longer be enabled, which makes unclocked accesses to > fault-status registers in the SMMU fault handlers possible. > > Guard the context and global fault handlers with > arm_smmu_rpm_get_if_active() and arm_smmu_rpm_put() so that all SMMU > fault register accesses are done with the SMMU powered. If the SMMU is > not runtime active, the fault can be safely ignored as > arm_smmu_device_reset() clears fault registers on resume. > > Additionally, disable fault reporting in arm_smmu_runtime_suspend() > before powering down. pm_runtime_get_if_active() returns 0 during > RPM_SUSPENDING, so without this, level-triggered fault interrupts would > cause an interrupt storm while the device is being suspended. > arm_smmu_device_reset() re-enables fault reporting on resume. Separate patch? > Also call synchronize_irq() for each context IRQ after masking fault > reporting but before clk_bulk_disable(). This closes a race where a > fault handler that passed arm_smmu_rpm_get_if_active() before the IRQ > was masked could still be executing MMIO accesses after the clocks are > cut. Separate patch? > Fixes: b13044092c1e ("drm/msm: Temporarily disable stall-on-fault after a page fault") > Co-developed-by: Pratyush Brahma > Signed-off-by: Pratyush Brahma > Signed-off-by: Prakash Gupta > Reviewed-by: Pranjal Shrivastava > --- > Changes in v4: > - Add synchronize_irq() loop in arm_smmu_runtime_suspend() after masking > fault reporting but before clk_bulk_disable(), closing a race where an > in-flight fault handler could access MMIO registers after clocks are cut > - Link to v3: https://patch.msgid.link/20260630-smmu-rpm-v3-1-f69874a580fa@oss.qualcomm.com > > Changes in v3: > - Add arm_smmu_rpm_get_if_active() wrapper that returns 1 when pm_runtime > is disabled, ensuring fault handlers work on non-pm_runtime systems > - Disable fault reporting in arm_smmu_runtime_suspend() before powering > down to prevent interrupt storms during RPM_SUSPENDING state > - Use pm_runtime_put_autosuspend() in arm_smmu_rpm_put() instead of > private __pm_runtime_put_autosuspend() > - Link to v2: https://patch.msgid.link/20260313-smmu-rpm-v2-1-8c2236b402b0@oss.qualcomm.com > > Changes in v2: > - Switched from arm_smmu_rpm_get()/arm_smmu_rpm_put() wrappers to > pm_runtime_get_if_active()/pm_runtime_put_autosuspend() APIs > - Added support for smmu->impl->global_fault callback in global fault handler > - Remove threaded irq context fault restriction to allow modifying stall > mode for adreno smmu > - Link to v1: https://patch.msgid.link/20260127-smmu-rpm-v1-1-2ef2f4c85305@oss.qualcomm.com > --- > drivers/iommu/arm/arm-smmu/arm-smmu.c | 101 +++++++++++++++++++++++++--------- > 1 file changed, 76 insertions(+), 25 deletions(-) > > diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu.c b/drivers/iommu/arm/arm-smmu/arm-smmu.c > index 0bd21d206eb3..2b692a5293a2 100644 > --- a/drivers/iommu/arm/arm-smmu/arm-smmu.c > +++ b/drivers/iommu/arm/arm-smmu/arm-smmu.c > @@ -79,11 +79,16 @@ static inline int arm_smmu_rpm_get(struct arm_smmu_device *smmu) > > static inline void arm_smmu_rpm_put(struct arm_smmu_device *smmu) > { > - if (pm_runtime_enabled(smmu->dev)) { > - pm_runtime_mark_last_busy(smmu->dev); > - __pm_runtime_put_autosuspend(smmu->dev); > + if (pm_runtime_enabled(smmu->dev)) > + pm_runtime_put_autosuspend(smmu->dev); > +} > > - } > +static inline int arm_smmu_rpm_get_if_active(struct arm_smmu_device *smmu) > +{ > + if (!pm_runtime_enabled(smmu->dev)) > + return 1; > + > + return pm_runtime_get_if_active(smmu->dev); > } Can this race with pm_runtime_disable(), perhaps via arm_smmu_device_shutdown()? If so, then we could return -EINVAL, which doesn't look like it's handled correctly by the caller here: > static void arm_smmu_rpm_use_autosuspend(struct arm_smmu_device *smmu) > @@ -462,10 +467,20 @@ static irqreturn_t arm_smmu_context_fault(int irq, void *dev) > int idx = smmu_domain->cfg.cbndx; > int ret; > > + if (!arm_smmu_rpm_get_if_active(smmu)) > + return IRQ_NONE; ^^^ Is this assuming that the irq is edge-triggered? Is that guaranteed? Will