[PATCH v3 2/3] memory: mtk-smi: Add enable IOMMU SMC command for MM master
Chengci.Xu
chengci.xu at mediatek.com
Fri Jul 29 19:12:42 PDT 2022
On Thu, 2022-07-28 at 13:11 +0200, AngeloGioacchino Del Regno wrote:
> Il 27/07/22 12:45, Chengci.Xu ha scritto:
> > For concerns about security, the register to enable/disable IOMMU
> > of
> > SMI LARB should only be configured in secure world. Thus, we add
> > some
> > SMC command for multimedia master to enable/disable MM IOMMU in ATF
> > by
> > setting the register of SMI LARB. This function is prepared for
> > MT8188.
> >
> > Signed-off-by: Chengci.Xu <chengci.xu at mediatek.com>
> > ---
> > drivers/memory/mtk-smi.c | 11 +++++++++++
> > include/linux/soc/mediatek/mtk_sip_svc.h | 3 +++
> > include/soc/mediatek/smi.h | 7 +++++++
> > 3 files changed, 21 insertions(+)
> >
> > diff --git a/drivers/memory/mtk-smi.c b/drivers/memory/mtk-smi.c
> > index d7cb7ead2ac7..41ce66c39123 100644
> > --- a/drivers/memory/mtk-smi.c
> > +++ b/drivers/memory/mtk-smi.c
> > @@ -3,6 +3,7 @@
> > * Copyright (c) 2015-2016 MediaTek Inc.
> > * Author: Yong Wu <yong.wu at mediatek.com>
> > */
> > +#include <linux/arm-smccc.h>
> > #include <linux/clk.h>
> > #include <linux/component.h>
> > #include <linux/device.h>
> > @@ -14,6 +15,7 @@
> > #include <linux/of_platform.h>
> > #include <linux/platform_device.h>
> > #include <linux/pm_runtime.h>
> > +#include <linux/soc/mediatek/mtk_sip_svc.h>
> > #include <soc/mediatek/smi.h>
> > #include <dt-bindings/memory/mt2701-larb-port.h>
> > #include <dt-bindings/memory/mtk-memory-port.h>
> > @@ -89,6 +91,7 @@
> > #define MTK_SMI_FLAG_THRT_UPDATE BIT(0)
> > #define MTK_SMI_FLAG_SW_FLAG BIT(1)
> > #define MTK_SMI_FLAG_SLEEP_CTL BIT(2)
> > +#define MTK_SMI_FLAG_SEC_REG BIT(3)
> > #define MTK_SMI_CAPS(flags, _x) (!!((flags) & (_x)))
> >
> > struct mtk_smi_reg_pair {
> > @@ -235,6 +238,7 @@ static void
> > mtk_smi_larb_config_port_gen2_general(struct device *dev)
> > struct mtk_smi_larb *larb = dev_get_drvdata(dev);
> > u32 reg, flags_general = larb->larb_gen->flags_general;
> > const u8 *larbostd = larb->larb_gen->ostd ? larb->larb_gen-
> > >ostd[larb->larbid] : NULL;
> > + struct arm_smccc_res res;
> > int i;
> >
> > if (BIT(larb->larbid) & larb->larb_gen-
> > >larb_direct_to_common_mask)
> > @@ -259,6 +263,13 @@ static void
> > mtk_smi_larb_config_port_gen2_general(struct device *dev)
> > reg |= BANK_SEL(larb->bank[i]);
> > writel(reg, larb->base + SMI_LARB_NONSEC_CON(i));
> > }
> > +
> > + if (MTK_SMI_CAPS(flags_general, MTK_SMI_FLAG_SEC_REG)) {
> > + arm_smccc_smc(MTK_SIP_KERNEL_IOMMU_CONTROL,
> > IOMMU_ATF_CMD_CONFIG_SMI_LARB,
> > + larb->larbid, (u32)(*larb->mmu), 0, 0, 0,
> > 0, &res);
> > + if (res.a0 != 0)
> > + dev_err(dev, "Enable iommu fail, ret %ld\n",
> > res.a0);
>
> This means that the system will eventually crash or anyway be
> unstable: in this
> case, we should not allow further interaction with the IOMMUs and/or
> SMI.
>
> So, if you place this here, you will have to change this function to
> return
> something for the caller to take action.
Thanks, a new patch will add a return value for this function.
And, we will return -EINVAL when enable iommu failed.
>
> > + }
> > }
> >
> > static const u8 mtk_smi_larb_mt8195_ostd[][SMI_LARB_PORT_NR_MAX]
> > = {
> > diff --git a/include/linux/soc/mediatek/mtk_sip_svc.h
> > b/include/linux/soc/mediatek/mtk_sip_svc.h
> > index 082398e0cfb1..0761128b4354 100644
> > --- a/include/linux/soc/mediatek/mtk_sip_svc.h
> > +++ b/include/linux/soc/mediatek/mtk_sip_svc.h
> > @@ -22,4 +22,7 @@
> > ARM_SMCCC_CALL_VAL(ARM_SMCCC_FAST_CALL, MTK_SIP_SMC_CONVENTION,
> > \
> > ARM_SMCCC_OWNER_SIP, fn_id)
> >
> > +/* IOMMU related SMC call */
> > +#define MTK_SIP_KERNEL_IOMMU_CONTROL MTK_SIP_SMC_CMD(0x514)
> > +
> > #endif
> > diff --git a/include/soc/mediatek/smi.h
> > b/include/soc/mediatek/smi.h
> > index 11f7d6b59642..8c781b7bd88d 100644
> > --- a/include/soc/mediatek/smi.h
> > +++ b/include/soc/mediatek/smi.h
> > @@ -9,6 +9,13 @@
> > #include <linux/bitops.h>
> > #include <linux/device.h>
> >
> > +/* IOMMU & SMI ATF CMD */
> > +
> > +enum IOMMU_ATF_CMD {
>
> Why do you have an enumeration when you're defining just one command?
> Is it expected to have more?
Yes, we will have another command for INFRA IOMMU enable/disable.
>
> Besides, the enum name should be lower case, and...
Thanks, got it.
>
> > + IOMMU_ATF_CMD_CONFIG_SMI_LARB, /* For mm master to
> > en/disable iommu */
> > + IOMMU_ATF_CMD_COUNT,
>
> if IOMMU_ATF_CMD_COUNT means "end of this enumeration", please call
> it
> IOMMU_ATF_CMD_MAX instead.
Thanks, got it.
>
> > +};
> > +
> > #if IS_ENABLED(CONFIG_MTK_SMI)
> >
> > #define MTK_SMI_MMU_EN(port) BIT(port)
>
>
More information about the Linux-mediatek
mailing list