On Mon, Jul 01, 2024 at 09:20:12AM -0700, Rob Clark wrote:
> From: Rob Clark <robdclark@xxxxxxxxxxxx>
>
> Parse out the bitfields for easier-to-read fault messages.
>
> Signed-off-by: Rob Clark <robdclark@xxxxxxxxxxxx>
> ---
> .../iommu/arm/arm-smmu/arm-smmu-qcom-debug.c | 52 +++++---------
> drivers/iommu/arm/arm-smmu/arm-smmu.c | 70 +++++++++++++++----
> drivers/iommu/arm/arm-smmu/arm-smmu.h | 21 ++++++
> 3 files changed, 92 insertions(+), 51 deletions(-)
>
[...]
> diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu.c b/drivers/iommu/arm/arm-smmu/arm-smmu.c
> index 23cf91ac409b..79ec911ae151 100644
> --- a/drivers/iommu/arm/arm-smmu/arm-smmu.c
> +++ b/drivers/iommu/arm/arm-smmu/arm-smmu.c
> @@ -405,32 +405,72 @@ static const struct iommu_flush_ops arm_smmu_s2_tlb_ops_v1 = {
> .tlb_add_page = arm_smmu_tlb_add_page_s2_v1,
> };
>
> +
> +void arm_smmu_read_context_fault_info(struct arm_smmu_device *smmu, int idx,
> + struct arm_smmu_context_fault_info *cfi)
> +{
> + cfi->iova = arm_smmu_cb_readq(smmu, idx, ARM_SMMU_CB_FAR);
> + cfi->fsr = arm_smmu_cb_read(smmu, idx, ARM_SMMU_CB_FSR);
> + cfi->fsynr = arm_smmu_cb_read(smmu, idx, ARM_SMMU_CB_FSYNR0);
> + cfi->cbfrsynra = arm_smmu_gr1_read(smmu, ARM_SMMU_GR1_CBFRSYNRA(idx));
> +}
> +
> +void arm_smmu_print_context_fault_info(struct arm_smmu_device *smmu, int idx,
> + const struct arm_smmu_context_fault_info *cfi)
> +{
> + dev_dbg(smmu->dev,
> + "Unhandled context fault: fsr=0x%x, iova=0x%08lx, fsynr=0x%x, cbfrsynra=0x%x, cb=%d\n",
> + cfi->fsr, cfi->iova, cfi->fsynr, cfi->cbfrsynra, idx);
> +
I just happen to hit an IOMMU fault in 6.11-rc4 and I did not see the 'Unhandled
context fault' log, but just the register dump in dmesg. Tracking that lead me
to this patch.
May I know the reasoning behind making the actual error message _dbg()? It is
intentional first place?
- Mani
> + dev_err(smmu->dev, "FSR = %08x [%s%sFormat=%u%s%s%s%s%s%s%s%s], SID=0x%x\n",
> + cfi->fsr,
> + (cfi->fsr & ARM_SMMU_CB_FSR_MULTI) ? "MULTI " : "",
> + (cfi->fsr & ARM_SMMU_CB_FSR_SS) ? "SS " : "",
> + (u32)FIELD_GET(ARM_SMMU_CB_FSR_FORMAT, cfi->fsr),
> + (cfi->fsr & ARM_SMMU_CB_FSR_UUT) ? " UUT" : "",
> + (cfi->fsr & ARM_SMMU_CB_FSR_ASF) ? " ASF" : "",
> + (cfi->fsr & ARM_SMMU_CB_FSR_TLBLKF) ? " TLBLKF" : "",
> + (cfi->fsr & ARM_SMMU_CB_FSR_TLBMCF) ? " TLBMCF" : "",
> + (cfi->fsr & ARM_SMMU_CB_FSR_EF) ? " EF" : "",
> + (cfi->fsr & ARM_SMMU_CB_FSR_PF) ? " PF" : "",
> + (cfi->fsr & ARM_SMMU_CB_FSR_AFF) ? " AFF" : "",
> + (cfi->fsr & ARM_SMMU_CB_FSR_TF) ? " TF" : "",
> + cfi->cbfrsynra);
> +
> + dev_err(smmu->dev, "FSYNR0 = %08x [S1CBNDX=%u%s%s%s%s%s%s PLVL=%u]\n",
> + cfi->fsynr,
> + (u32)FIELD_GET(ARM_SMMU_CB_FSYNR0_S1CBNDX, cfi->fsynr),
> + (cfi->fsynr & ARM_SMMU_CB_FSYNR0_AFR) ? " AFR" : "",
> + (cfi->fsynr & ARM_SMMU_CB_FSYNR0_PTWF) ? " PTWF" : "",
> + (cfi->fsynr & ARM_SMMU_CB_FSYNR0_NSATTR) ? " NSATTR" : "",
> + (cfi->fsynr & ARM_SMMU_CB_FSYNR0_IND) ? " IND" : "",
> + (cfi->fsynr & ARM_SMMU_CB_FSYNR0_PNU) ? " PNU" : "",
> + (cfi->fsynr & ARM_SMMU_CB_FSYNR0_WNR) ? " WNR" : "",
> + (u32)FIELD_GET(ARM_SMMU_CB_FSYNR0_PLVL, cfi->fsynr));
> +}
> +
> static irqreturn_t arm_smmu_context_fault(int irq, void *dev)
> {
> - u32 fsr, fsynr, cbfrsynra;
> - unsigned long iova;
> + struct arm_smmu_context_fault_info cfi;
> struct arm_smmu_domain *smmu_domain = dev;
> struct arm_smmu_device *smmu = smmu_domain->smmu;
> + static DEFINE_RATELIMIT_STATE(rs, DEFAULT_RATELIMIT_INTERVAL,
> + DEFAULT_RATELIMIT_BURST);
> int idx = smmu_domain->cfg.cbndx;
> int ret;
>
> - fsr = arm_smmu_cb_read(smmu, idx, ARM_SMMU_CB_FSR);
> - if (!(fsr & ARM_SMMU_CB_FSR_FAULT))
> - return IRQ_NONE;
> + arm_smmu_read_context_fault_info(smmu, idx, &cfi);
>
> - fsynr = arm_smmu_cb_read(smmu, idx, ARM_SMMU_CB_FSYNR0);
> - iova = arm_smmu_cb_readq(smmu, idx, ARM_SMMU_CB_FAR);
> - cbfrsynra = arm_smmu_gr1_read(smmu, ARM_SMMU_GR1_CBFRSYNRA(idx));
> + if (!(cfi.fsr & ARM_SMMU_CB_FSR_FAULT))
> + return IRQ_NONE;
>
> - ret = report_iommu_fault(&smmu_domain->domain, NULL, iova,
> - fsynr & ARM_SMMU_CB_FSYNR0_WNR ? IOMMU_FAULT_WRITE : IOMMU_FAULT_READ);
> + ret = report_iommu_fault(&smmu_domain->domain, NULL, cfi.iova,
> + cfi.fsynr & ARM_SMMU_CB_FSYNR0_WNR ? IOMMU_FAULT_WRITE : IOMMU_FAULT_READ);
>
> - if (ret == -ENOSYS)
> - dev_err_ratelimited(smmu->dev,
> - "Unhandled context fault: fsr=0x%x, iova=0x%08lx, fsynr=0x%x, cbfrsynra=0x%x, cb=%d\n",
> - fsr, iova, fsynr, cbfrsynra, idx);
> + if (ret == -ENOSYS && __ratelimit(&rs))
> + arm_smmu_print_context_fault_info(smmu, idx, &cfi);
>
> - arm_smmu_cb_write(smmu, idx, ARM_SMMU_CB_FSR, fsr);
> + arm_smmu_cb_write(smmu, idx, ARM_SMMU_CB_FSR, cfi.fsr);
> return IRQ_HANDLED;
> }
>
> diff --git a/drivers/iommu/arm/arm-smmu/arm-smmu.h b/drivers/iommu/arm/arm-smmu/arm-smmu.h
> index b04a00126a12..e2aeb511ae90 100644
> --- a/drivers/iommu/arm/arm-smmu/arm-smmu.h
> +++ b/drivers/iommu/arm/arm-smmu/arm-smmu.h
> @@ -198,6 +198,7 @@ enum arm_smmu_cbar_type {
> #define ARM_SMMU_CB_FSR 0x58
> #define ARM_SMMU_CB_FSR_MULTI BIT(31)
> #define ARM_SMMU_CB_FSR_SS BIT(30)
> +#define ARM_SMMU_CB_FSR_FORMAT GENMASK(10, 9)
> #define ARM_SMMU_CB_FSR_UUT BIT(8)
> #define ARM_SMMU_CB_FSR_ASF BIT(7)
> #define ARM_SMMU_CB_FSR_TLBLKF BIT(6)
> @@ -223,7 +224,14 @@ enum arm_smmu_cbar_type {
> #define ARM_SMMU_CB_FAR 0x60
>
> #define ARM_SMMU_CB_FSYNR0 0x68
> +#define ARM_SMMU_CB_FSYNR0_PLVL GENMASK(1, 0)
> #define ARM_SMMU_CB_FSYNR0_WNR BIT(4)
> +#define ARM_SMMU_CB_FSYNR0_PNU BIT(5)
> +#define ARM_SMMU_CB_FSYNR0_IND BIT(6)
> +#define ARM_SMMU_CB_FSYNR0_NSATTR BIT(8)
> +#define ARM_SMMU_CB_FSYNR0_PTWF BIT(10)
> +#define ARM_SMMU_CB_FSYNR0_AFR BIT(11)
> +#define ARM_SMMU_CB_FSYNR0_S1CBNDX GENMASK(23, 16)
>
> #define ARM_SMMU_CB_FSYNR1 0x6c
>
> @@ -533,4 +541,17 @@ struct arm_smmu_device *qcom_smmu_impl_init(struct arm_smmu_device *smmu);
> void arm_smmu_write_context_bank(struct arm_smmu_device *smmu, int idx);
> int arm_mmu500_reset(struct arm_smmu_device *smmu);
>
> +struct arm_smmu_context_fault_info {
> + unsigned long iova;
> + u32 fsr;
> + u32 fsynr;
> + u32 cbfrsynra;
> +};
> +
> +void arm_smmu_read_context_fault_info(struct arm_smmu_device *smmu, int idx,
> + struct arm_smmu_context_fault_info *cfi);
> +
> +void arm_smmu_print_context_fault_info(struct arm_smmu_device *smmu, int idx,
> + const struct arm_smmu_context_fault_info *cfi);
> +
> #endif /* _ARM_SMMU_H */
> --
> 2.45.2
>
>
--
மணிவண்ணன் சதாசிவம்
