| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <CAPHGfUN7k_oLrjvXPJj84Q5NayXFab2QSQ+g4iYjrAsPqO_KnA@mail.gmail.com>
Date: Wed, 9 Jul 2025 20:06:16 +0530
From: Komal Bajaj <komal.bajaj@....qualcomm.com>
To: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc: linux-arm-msm@...r.kernel.org, linux-usb@...r.kernel.org,
linux-kernel@...r.kernel.org,
Souradeep Chowdhury <quic_schowdhu@...cinc.com>,
Konrad Dybcio <konrad.dybcio@....qualcomm.com>
Subject: Re: [PATCH v4] usb: misc: qcom_eud: Access EUD_MODE_MANAGER2 through
secure calls
On Wed, Jul 9, 2025 at 2:14 PM Greg Kroah-Hartman
<gregkh@...uxfoundation.org> wrote:
>
> On Wed, Jul 09, 2025 at 12:25:14PM +0530, Komal Bajaj wrote:
> > EUD_MODE_MANAGER2 register is mapped to a memory region that is marked
> > as read-only for HLOS, enforcing access restrictions that prohibit
> > direct memory-mapped writes via writel().
> >
> > Attempts to write to this region from HLOS can result in silent failures
> > or memory access violations, particularly when toggling EUD (Embedded
> > USB Debugger) state. To ensure secure register access, modify the driver
> > to use qcom_scm_io_writel(), which routes the write operation to Qualcomm
> > Secure Channel Monitor (SCM). SCM has the necessary permissions to access
> > protected memory regions, enabling reliable control over EUD state.
> >
> > SC7280, the only user of EUD is also affected, indicating that this could
> > never have worked on a properly fused device.
> >
> > Fixes: 9a1bf58ccd44 ("usb: misc: eud: Add driver support for Embedded USB Debugger(EUD)")
> > Signed-off-by: Melody Olvera <quic_molvera@...cinc.com>
> > Signed-off-by: Komal Bajaj <komal.bajaj@....qualcomm.com>
> > Reviewed-by: Konrad Dybcio <konrad.dybcio@....qualcomm.com>
> > ---
> > Changes in v4:
> > * Added error logging in disable_eud() for SCM write failures, per Konrad’s suggestion
> > * Link to v3: https://lore.kernel.org/all/20250708085208.19089-1-komal.bajaj@oss.qualcomm.com/
> >
> > Changes in v3:
> > * Moved secure write before normal writes
> > * Added error checking in disable_eud()
> > * Use ENOMEM error code if platform_get_resource() fails
> > * Select QCOM_SCM driver if USB_QCOM_EUD is enabled
> > * Link to v2: https://lore.kernel.org/all/20250627125131.27606-1-komal.bajaj@oss.qualcomm.com/
> >
> > Changes in v2:
> > * Drop separate compatible to be added for secure eud
> > * Use secure call to access EUD mode manager register
> > * Link to v1: https://lore.kernel.org/all/20240807183205.803847-1-quic_molvera@quicinc.com/
> >
> > drivers/usb/misc/Kconfig | 1 +
> > drivers/usb/misc/qcom_eud.c | 27 +++++++++++++++++++++------
> > 2 files changed, 22 insertions(+), 6 deletions(-)
> >
> > diff --git a/drivers/usb/misc/Kconfig b/drivers/usb/misc/Kconfig
> > index 6497c4e81e95..f49d5ac0e957 100644
> > --- a/drivers/usb/misc/Kconfig
> > +++ b/drivers/usb/misc/Kconfig
> > @@ -148,6 +148,7 @@ config USB_QCOM_EUD
> > tristate "QCOM Embedded USB Debugger(EUD) Driver"
> > depends on ARCH_QCOM || COMPILE_TEST
> > select USB_ROLE_SWITCH
> > + select QCOM_SCM
>
> How is this select going to work if COMPILE_TEST is enabled on
> non-ARCH_QCOM configs?
>
> Please don't use select if at all possible, use "depends" properly.
Ack, will switch to depends on QCOM_SCM to avoid build issues with COMPILE_TEST
>
>
> > help
> > This module enables support for Qualcomm Technologies, Inc.
> > Embedded USB Debugger (EUD). The EUD is a control peripheral
> > diff --git a/drivers/usb/misc/qcom_eud.c b/drivers/usb/misc/qcom_eud.c
> > index 83079c414b4f..a58081f53db3 100644
> > --- a/drivers/usb/misc/qcom_eud.c
> > +++ b/drivers/usb/misc/qcom_eud.c
> > @@ -15,6 +15,7 @@
> > #include <linux/slab.h>
> > #include <linux/sysfs.h>
> > #include <linux/usb/role.h>
> > +#include <linux/firmware/qcom/qcom_scm.h>
> >
> > #define EUD_REG_INT1_EN_MASK 0x0024
> > #define EUD_REG_INT_STATUS_1 0x0044
> > @@ -34,7 +35,7 @@ struct eud_chip {
> > struct device *dev;
> > struct usb_role_switch *role_sw;
> > void __iomem *base;
> > - void __iomem *mode_mgr;
> > + phys_addr_t mode_mgr;
> > unsigned int int_status;
> > int irq;
> > bool enabled;
> > @@ -43,18 +44,30 @@ struct eud_chip {
> >
> > static int enable_eud(struct eud_chip *priv)
> > {
> > + int ret;
> > +
> > + ret = qcom_scm_io_writel(priv->mode_mgr + EUD_REG_EUD_EN2, 1);
> > + if (ret)
> > + return ret;
> > +
> > writel(EUD_ENABLE, priv->base + EUD_REG_CSR_EUD_EN);
>
> Nit, why is your local writel() function in backwards order of
> parameters from qcom_scm_io_writel()? That's going to be a major pain
> to maintain over time, don't you think?
The current ordering of parameters in qcom_scm_io_writel is the same as in
qcom_scm_io_readl, possibly to keep them consistent. However, if you believe
it's more appropriate to align the parameter order with the standard writel
convention, I’m open to updating it accordingly.
>
>
> > writel(EUD_INT_VBUS | EUD_INT_SAFE_MODE,
> > priv->base + EUD_REG_INT1_EN_MASK);
> > - writel(1, priv->mode_mgr + EUD_REG_EUD_EN2);
> >
> > return usb_role_switch_set_role(priv->role_sw, USB_ROLE_DEVICE);
> > }
> >
> > static void disable_eud(struct eud_chip *priv)
> > {
> > + int ret;
> > +
> > + ret = qcom_scm_io_writel(priv->mode_mgr + EUD_REG_EUD_EN2, 0);
> > + if (ret) {
> > + dev_err(priv->dev, "failed to disable eud\n");
> > + return;
> > + }
> > +
> > writel(0, priv->base + EUD_REG_CSR_EUD_EN);
> > - writel(0, priv->mode_mgr + EUD_REG_EUD_EN2);
> > }
> >
> > static ssize_t enable_show(struct device *dev,
> > @@ -178,6 +191,7 @@ static void eud_role_switch_release(void *data)
> > static int eud_probe(struct platform_device *pdev)
> > {
> > struct eud_chip *chip;
> > + struct resource *res;
> > int ret;
> >
> > chip = devm_kzalloc(&pdev->dev, sizeof(*chip), GFP_KERNEL);
> > @@ -200,9 +214,10 @@ static int eud_probe(struct platform_device *pdev)
> > if (IS_ERR(chip->base))
> > return PTR_ERR(chip->base);
> >
> > - chip->mode_mgr = devm_platform_ioremap_resource(pdev, 1);
> > - if (IS_ERR(chip->mode_mgr))
> > - return PTR_ERR(chip->mode_mgr);
> > + res = platform_get_resource(pdev, IORESOURCE_MEM, 1);
> > + if (!res)
> > + return -ENODEV;
> > + chip->mode_mgr = res->start;
>
> No ioremap() call needed anymore? Why not?
We’re only using the address for qcom_scm_io_writel(), which operates on
physical addresses and doesn’t require an ioremap’d virtual address.
That’s why devm_ioremap_resource() isn’t needed here.
>
> thanks,
>
> greg k-h
Powered by blists - more mailing lists