On Fri, Apr 4, 2025 at 11:37 AM Stefano Garzarella <sgarz...@redhat.com> wrote:
>
> On Fri, 4 Apr 2025 at 19:32, Dionna Amalie Glaze <dionnagl...@google.com>
> wrote:
> >
> > On Thu, Apr 3, 2025 at 3:10 AM Stefano Garzarella <sgarz...@redhat.com>
> > wrote:
> > >
> > > From: Stefano Garzarella <sgarz...@redhat.com>
> > >
> > > Add driver for the vTPM defined by the AMD SVSM spec [1].
> > >
> > > The specification defines a protocol that a SEV-SNP guest OS can use to
> > > discover and talk to a vTPM emulated by the Secure VM Service Module
> > > (SVSM)
> > > in the guest context, but at a more privileged level (VMPL0).
> > >
> > > The new tpm-svsm platform driver uses two functions exposed by x86/sev
> > > to verify that the device is actually emulated by the platform and to
> > > send commands and receive responses.
> > >
> > > The device cannot be hot-plugged/unplugged as it is emulated by the
> > > platform, so we can use module_platform_driver_probe(). The probe
> > > function will only check whether in the current runtime configuration,
> > > SVSM is present and provides a vTPM.
> > >
> > > This device does not support interrupts and sends responses to commands
> > > synchronously. In order to have .recv() called just after .send() in
> > > tpm_try_transmit(), the .status() callback returns 0, and both
> > > .req_complete_mask and .req_complete_val are set to 0.
> > >
> > > [1] "Secure VM Service Module for SEV-SNP Guests"
> > > Publication # 58019 Revision: 1.00
> > >
> > > Signed-off-by: Stefano Garzarella <sgarz...@redhat.com>
> > > ---
> > > v6:
> > > - removed the `locality` field (set to 0) and the FIXME comment [Jarkko]
> > > v5:
> > > - removed cancel/status/req_* ops after rebase on master that cotains
> > > commit 980a573621ea ("tpm: Make chip->{status,cancel,req_canceled} opt")
> > > v4:
> > > - moved "asm" includes after the "linux" includes [Tom]
> > > - allocated buffer separately [Tom/Jarkko/Jason]
> > > v3:
> > > - removed send_recv() ops and followed the ftpm driver implementing
> > > .status,
> > > .req_complete_mask, .req_complete_val, etc. [Jarkko]
> > > - removed link to the spec because those URLs are unstable [Borislav]
> > > ---
> > > drivers/char/tpm/tpm_svsm.c | 128 ++++++++++++++++++++++++++++++++++++
> > > drivers/char/tpm/Kconfig | 10 +++
> > > drivers/char/tpm/Makefile | 1 +
> > > 3 files changed, 139 insertions(+)
> > > create mode 100644 drivers/char/tpm/tpm_svsm.c
> > >
> > > diff --git a/drivers/char/tpm/tpm_svsm.c b/drivers/char/tpm/tpm_svsm.c
> > > new file mode 100644
> > > index 000000000000..b9242c9eab87
> > > --- /dev/null
> > > +++ b/drivers/char/tpm/tpm_svsm.c
> > > @@ -0,0 +1,128 @@
> > > +// SPDX-License-Identifier: GPL-2.0-only
> > > +/*
> > > + * Copyright (C) 2025 Red Hat, Inc. All Rights Reserved.
> > > + *
> > > + * Driver for the vTPM defined by the AMD SVSM spec [1].
> > > + *
> > > + * The specification defines a protocol that a SEV-SNP guest OS can use
> > > to
> > > + * discover and talk to a vTPM emulated by the Secure VM Service Module
> > > (SVSM)
> > > + * in the guest context, but at a more privileged level (usually VMPL0).
> > > + *
> > > + * [1] "Secure VM Service Module for SEV-SNP Guests"
> > > + * Publication # 58019 Revision: 1.00
> > > + */
> > > +
> > > +#include <linux/module.h>
> > > +#include <linux/kernel.h>
> > > +#include <linux/platform_device.h>
> > > +#include <linux/tpm_svsm.h>
> > > +
> > > +#include <asm/sev.h>
> > > +
> > > +#include "tpm.h"
> > > +
> > > +struct tpm_svsm_priv {
> > > + void *buffer;
> > > +};
> > > +
> > > +static int tpm_svsm_send(struct tpm_chip *chip, u8 *buf, size_t len)
> > > +{
> > > + struct tpm_svsm_priv *priv = dev_get_drvdata(&chip->dev);
> > > + int ret;
> > > +
> > > + ret = svsm_vtpm_cmd_request_fill(priv->buffer, 0, buf, len);
> > > + if (ret)
> > > + return ret;
> > > +
> > > + /*
> > > + * The SVSM call uses the same buffer for the command and for the
> > > + * response, so after this call, the buffer will contain the
> > > response
> > > + * that can be used by .recv() op.
> > > + */
> > > + return snp_svsm_vtpm_send_command(priv->buffer);
> > > +}
> > > +
> > > +static int tpm_svsm_recv(struct tpm_chip *chip, u8 *buf, size_t len)
> > > +{
> > > + struct tpm_svsm_priv *priv = dev_get_drvdata(&chip->dev);
> > > +
> > > + /*
> > > + * The internal buffer contains the response after we send the
> > > command
> > > + * to SVSM.
> > > + */
> > > + return svsm_vtpm_cmd_response_parse(priv->buffer, buf, len);
> > > +}
> > > +
> > > +static struct tpm_class_ops tpm_chip_ops = {
> > > + .flags = TPM_OPS_AUTO_STARTUP,
> > > + .recv = tpm_svsm_recv,
> > > + .send = tpm_svsm_send,
> > > +};
> > > +
> > > +static int __init tpm_svsm_probe(struct platform_device *pdev)
> > > +{
> > > + struct device *dev = &pdev->dev;
> > > + struct tpm_svsm_priv *priv;
> > > + struct tpm_chip *chip;
> > > + int err;
> > > +
> > > + if (!snp_svsm_vtpm_probe())
> > > + return -ENODEV;
> > > +
> > > + priv = devm_kmalloc(dev, sizeof(*priv), GFP_KERNEL);
> > > + if (!priv)
> > > + return -ENOMEM;
> > > +
> > > + /*
> > > + * The maximum buffer supported is one page (see
> > > SVSM_VTPM_MAX_BUFFER
> > > + * in tpm_svsm.h).
> > > + */
> > > + priv->buffer = (void *)devm_get_free_pages(dev, GFP_KERNEL, 0);
> > > + if (!priv->buffer)
> > > + return -ENOMEM;
> > > +
> > > + chip = tpmm_chip_alloc(dev, &tpm_chip_ops);
> > > + if (IS_ERR(chip))
> > > + return PTR_ERR(chip);
> > > +
> > > + dev_set_drvdata(&chip->dev, priv);
> > > +
> > > + err = tpm2_probe(chip);
> >
> > Our testing is showing that tpm2_probe is hitting a null pointer deref
> > in tpm_transmit.
>
> Next time, please share a backtrace.
Right, my bad.
>
> BTW I suspect you're not using Linus' tree, so be sure to backport
> also commit 980a573621ea ("tpm: Make
> chip->{status,cancel,req_canceled} opt").
>
> Without that, you will have a null ptr dereference since .status() is
> NULL from v5 of this series (as specified in the changelog).
Thanks, I missed that detail. Will report back.
>
> Stefano
>
--
-Dionna Glaze, PhD, CISSP, CCSP (she/her)
