Re: [PATCH v4 05/43] arm64: RME: Add SMC definitions for calling the RMM

[Gavin Shan]

On 9/6/24 10:11 AM, Gavin Shan wrote:
> On 8/22/24 1:38 AM, Steven Price wrote:
> > The RMM (Realm Management Monitor) provides functionality that can be
> > accessed by SMC calls from the host.
> >
> > The SMC definitions are based on DEN0137[1] version 1.0-rel0-rc1
> >
> > [1] https://developer.arm.com/-/cdn-downloads/permalink/PDF/Architectures/DEN0137_1.0-rel0-rc1_rmm-arch_external.pdf
> >
> > Signed-off-by: Steven Price <steven.price@xxxxxxx>
> > ---
> > Changes since v3:
> >   * Update to match RMM spec v1.0-rel0-rc1.
> > Changes since v2:
> >   * Fix specification link.
> >   * Rename rec_entry->rec_enter to match spec.
> >   * Fix size of pmu_ovf_status to match spec.
> > ---
> >   arch/arm64/include/asm/rmi_smc.h | 253 +++++++++++++++++++++++++++++++
> >   1 file changed, 253 insertions(+)
> >   create mode 100644 arch/arm64/include/asm/rmi_smc.h
>
> [...]
>
> > +
> > +#define RMI_FEATURE_REGISTER_0_S2SZ        GENMASK(7, 0)
> > +#define RMI_FEATURE_REGISTER_0_LPA2        BIT(8)
> > +#define RMI_FEATURE_REGISTER_0_SVE_EN        BIT(9)
> > +#define RMI_FEATURE_REGISTER_0_SVE_VL        GENMASK(13, 10)
> > +#define RMI_FEATURE_REGISTER_0_NUM_BPS        GENMASK(19, 14)
> > +#define RMI_FEATURE_REGISTER_0_NUM_WPS        GENMASK(25, 20)
> > +#define RMI_FEATURE_REGISTER_0_PMU_EN        BIT(26)
> > +#define RMI_FEATURE_REGISTER_0_PMU_NUM_CTRS    GENMASK(31, 27)
> > +#define RMI_FEATURE_REGISTER_0_HASH_SHA_256    BIT(32)
> > +#define RMI_FEATURE_REGISTER_0_HASH_SHA_512    BIT(33)
> > +#define RMI_FEATURE_REGISTER_0_GICV3_NUM_LRS    GENMASK(37, 34)
> > +#define RMI_FEATURE_REGISTER_0_MAX_RECS_ORDER    GENMASK(41, 38)
> > +
>
> Those definitions aren't consistent to tf-rmm at least. For example, the latest tf-rmm
> has bit-28 and bit-29 for RMI_FEATURE_REGISTER_0_HASH_SHA_{256, 512}. I didn't check the
> specification yet, but they need to be corrected in Linux host or tf-rmm.
>
>    git@xxxxxxxxxx:TF-RMM/tf-rmm.git
>    head: 258b7952640b Merge "fix(tools/clang-tidy): ignore header include check" into integration
>
>    [gshan@gshan tf-rmm]$ git grep RMI_FEATURE_REGISTER_0_HASH_SHA.*_SHIFT
>    lib/smc/include/smc-rmi.h:#define RMI_FEATURE_REGISTER_0_HASH_SHA_256_SHIFT     UL(28)
>    lib/smc/include/smc-rmi.h:#define RMI_FEATURE_REGISTER_0_HASH_SHA_512_SHIFT     UL(29)
>
> Due to the inconsistent definitions, I'm unable to start a guest with the following
> combination: linux-host/cca-host/v4, linux-guest/cca-guest/v5, kvmtool/cca/v2.
>
>    # ./start_guest.sh
>    Info: # lkvm run -k Image -m 256 -c 2 --name guest-152
>    [  145.894085] config_realm_hash_algo: unsupported ALGO_SHA256 by rmm_feat_reg0=0x0000000034488e30
>    KVM_CAP_RME(KVM_CAP_ARM_RME_CONFIG_REALM) hash_algo: Invalid argument

Please ignore above comments. As Steven pointed out in another thread, the TF-RMM needs to
be something other than the latest upstream one. With the TF-RMM, I'm able to boot the guest
with cca/host-v4 and cca/guest-v5.

  git fetch https://git.trustedfirmware.org/TF-RMM/tf-rmm.git \
  refs/changes/85/30485/11

Thanks,
Gavin