From: "Oliver Smith-Denny" <osde@linux.microsoft.com>
To: devel@edk2.groups.io, nishant.sharma@arm.com
Cc: Ard Biesheuvel <ardb+tianocore@kernel.org>,
Sami Mujawar <sami.mujawar@arm.com>,
Thomas Abraham <thomas.abraham@arm.com>,
Sayanta Pattanayak <sayanta.pattanayak@arm.com>,
Achin Gupta <achin.gupta@arm.com>
Subject: Re: [edk2-devel] [edk2-platforms][PATCH V1 06/20] ArmPkg: Add support for FFA_MEM_PERM_GET/SET ABIs
Date: Wed, 12 Jul 2023 10:43:34 -0700 [thread overview]
Message-ID: <c7114034-a823-9e24-8754-96edeefb4d5a@linux.microsoft.com> (raw)
In-Reply-To: <20230711143658.781597-7-nishant.sharma@arm.com>
On 7/11/2023 7:36 AM, Nishant Sharma wrote:
> From: Achin Gupta <achin.gupta@arm.com>
>
> This patch uses the FFA_MEM_PERM_GET/SET ABIs to tweak the permissions of a
> set of pages if FF-A v1.1 and above is supported by the SPMC. For FF-A v1.0
> the previous method through FFA_MSG_SEND_DIRECT_REQ/RESP is used.
>
> Signed-off-by: Achin Gupta <achin.gupta@arm.com>
> Signed-off-by: Nishant Sharma <nishant.sharma@arm.com>
> ---
> ArmPkg/Include/IndustryStandard/ArmFfaSvc.h | 2 +
> ArmPkg/Library/StandaloneMmMmuLib/ArmMmuStandaloneMmLib.c | 132 +++++++++++++++++---
> 2 files changed, 120 insertions(+), 14 deletions(-)
>
> diff --git a/ArmPkg/Include/IndustryStandard/ArmFfaSvc.h b/ArmPkg/Include/IndustryStandard/ArmFfaSvc.h
> index c80d783fad3f..7987678c996e 100644
> --- a/ArmPkg/Include/IndustryStandard/ArmFfaSvc.h
> +++ b/ArmPkg/Include/IndustryStandard/ArmFfaSvc.h
> @@ -23,6 +23,8 @@
> #define ARM_SVC_ID_FFA_MSG_SEND_DIRECT_RESP_AARCH64 0xC4000070
> #define ARM_SVC_ID_FFA_SUCCESS_AARCH32 0x84000061
> #define ARM_SVC_ID_FFA_SUCCESS_AARCH64 0xC4000061
> +#define ARM_SVC_ID_FFA_MEM_PERM_SET_AARCH32 0x84000089
> +#define ARM_SVC_ID_FFA_MEM_PERM_GET_AARCH32 0x84000088
> #define ARM_SVC_ID_FFA_ERROR_AARCH32 0x84000060
> #define ARM_SVC_ID_FFA_ERROR_AARCH64 0xC4000060
> #define ARM_SVC_ID_FFA_MSG_WAIT_AARCH32 0x8400006B
> diff --git a/ArmPkg/Library/StandaloneMmMmuLib/ArmMmuStandaloneMmLib.c b/ArmPkg/Library/StandaloneMmMmuLib/ArmMmuStandaloneMmLib.c
> index 1a41a289ef17..76ef214bcb85 100644
> --- a/ArmPkg/Library/StandaloneMmMmuLib/ArmMmuStandaloneMmLib.c
> +++ b/ArmPkg/Library/StandaloneMmMmuLib/ArmMmuStandaloneMmLib.c
> @@ -25,6 +25,66 @@
> #include <Library/DebugLib.h>
> #include <Library/PcdLib.h>
>
> +
> +/**
> + Utility function to determine whether ABIs in FF-A v1.1 to set and get
> + memory permissions can be used. Ideally, this should be invoked once in the
> + library constructor and set a flag that can be used at runtime. However, the
> + StMM Core invokes this library before constructors are called and before the
> + StMM image itself is relocated.
> +
> + @retval EFI_SUCCESS The availability of ABIs was correctly determined.
> + @retval Other value Software stack is misconfigured.
> +
> +**/
> +STATIC
> +BOOLEAN
> +UseFfaMemPermAbis (
> + VOID
> + )
> +{
> + ARM_SVC_ARGS SvcArgs;
> + UINT32 SpmcFfaVersion;
> + STATIC UINT16 SpmcMajorVer = 0;
> + STATIC UINT16 SpmcMinorVer = 0;
> +
> + // Use prefetched version info. if either is not 0, then the version is
> + // already fetched.
> + if ((SpmcMajorVer | SpmcMinorVer) != 0) {
> + return (SpmcMajorVer == SPM_MAJOR_VERSION_FFA) && (SpmcMinorVer >= SPM_MINOR_VERSION_FFA);
> + }
> +
> + // Nothing to do if FF-A has not be enabled
nit: "been enabled"
> + if (FixedPcdGet32 (PcdFfaEnable) == 0) {
> + return FALSE;
> + }
> +
> + // Prepare the message parameters.
> + ZeroMem (&SvcArgs, sizeof (ARM_SVC_ARGS));
> + SvcArgs.Arg0 = ARM_SVC_ID_FFA_VERSION_AARCH32;
> + SvcArgs.Arg1 = FFA_VERSION_COMPILED;
> +
> + // Invoke the ABI
> + ArmCallSvc (&SvcArgs);
> +
> + // Check if FF-A is supported and what version
> + SpmcFfaVersion = SvcArgs.Arg0;
> +
> + // Damn! FF-A is not supported at all even though we specified v1.0 as our
> + // version. However, the feature flag has been turned on. This is a > + // misconfigured software stack. So, return an error and assert in
a debug build.
> + if (SpmcFfaVersion == ARM_FFA_SPM_RET_NOT_SUPPORTED) {
> + ASSERT (0);
It would be nice to either have the assert be self documenting
(ASSERT (SpmcFfaVersion == ARM_FFA_SPM_RET_NOT_SUPPORTED)) or
to add a print here.
> + return FALSE;
> + }
> +
> + SpmcMajorVer = (SpmcFfaVersion >> FFA_VERSION_MAJOR_SHIFT) & FFA_VERSION_MAJOR_MASK;
> + SpmcMinorVer = (SpmcFfaVersion >> FFA_VERSION_MINOR_SHIFT) & FFA_VERSION_MINOR_MASK;
> +
> + return (SpmcMajorVer == SPM_MAJOR_VERSION_FFA) && (SpmcMinorVer >= SPM_MINOR_VERSION_FFA);
> +}
> +
> +
> /** Send memory permission request to target.
>
> @param [in, out] SvcArgs Pointer to SVC arguments to send. On
> @@ -55,6 +115,36 @@ SendMemoryPermissionRequest (
>
> ArmCallSvc (SvcArgs);
> if (FixedPcdGet32 (PcdFfaEnable) != 0) {
> +
> + // Check if FF-A memory permission ABIs were used.
> + if (UseFfaMemPermAbis()) {
> + switch (SvcArgs->Arg0) {
> +
> + case ARM_SVC_ID_FFA_ERROR_AARCH32:
> + case ARM_SVC_ID_FFA_ERROR_AARCH64:
> + switch (SvcArgs->Arg2) {
> + case ARM_FFA_SPM_RET_INVALID_PARAMETERS:
> + return EFI_INVALID_PARAMETER;
> + case ARM_FFA_SPM_RET_NOT_SUPPORTED:
> + return EFI_UNSUPPORTED;
> + default:
> + // Undefined error code received.
> + ASSERT (0);
For these two default cases, can we print out what
error code we received?
Thanks,
Oliver
> + return EFI_INVALID_PARAMETER;
> + }
> +
> + case ARM_SVC_ID_FFA_SUCCESS_AARCH32:
> + case ARM_SVC_ID_FFA_SUCCESS_AARCH64:
> + *RetVal = SvcArgs->Arg2;
> + return EFI_SUCCESS;
> +
> + default:
> + // Undefined error code received.
> + ASSERT (0);
> + return EFI_INVALID_PARAMETER;
> + }
> + }
> +
> // Get/Set memory attributes is an atomic call, with
> // StandaloneMm at S-EL0 being the caller and the SPM
> // core being the callee. Thus there won't be a
> @@ -164,12 +254,18 @@ GetMemoryPermissions (
> // See [1], Section 13.5.5.1 MM_SP_MEMORY_ATTRIBUTES_GET_AARCH64.
> ZeroMem (&SvcArgs, sizeof (ARM_SVC_ARGS));
> if (FixedPcdGet32 (PcdFfaEnable) != 0) {
> - // See [2], Section 10.2 FFA_MSG_SEND_DIRECT_REQ.
> - SvcArgs.Arg0 = ARM_SVC_ID_FFA_MSG_SEND_DIRECT_REQ;
> - SvcArgs.Arg1 = ARM_FFA_DESTINATION_ENDPOINT_ID;
> - SvcArgs.Arg2 = 0;
> - SvcArgs.Arg3 = ARM_SVC_ID_SP_GET_MEM_ATTRIBUTES;
> - SvcArgs.Arg4 = BaseAddress;
> + // Check if FF-A memory permission ABIs can be used.
> + if (UseFfaMemPermAbis()) {
> + SvcArgs.Arg0 = ARM_SVC_ID_FFA_MEM_PERM_GET_AARCH32;
> + SvcArgs.Arg1 = BaseAddress;
> + } else {
> + // See [2], Section 10.2 FFA_MSG_SEND_DIRECT_REQ.
> + SvcArgs.Arg0 = ARM_SVC_ID_FFA_MSG_SEND_DIRECT_REQ;
> + SvcArgs.Arg1 = ARM_FFA_DESTINATION_ENDPOINT_ID;
> + SvcArgs.Arg2 = 0;
> + SvcArgs.Arg3 = ARM_SVC_ID_SP_GET_MEM_ATTRIBUTES;
> + SvcArgs.Arg4 = BaseAddress;
> + }
> } else {
> SvcArgs.Arg0 = ARM_SVC_ID_SP_GET_MEM_ATTRIBUTES;
> SvcArgs.Arg1 = BaseAddress;
> @@ -219,14 +315,22 @@ RequestMemoryPermissionChange (
> // See [1], Section 13.5.5.2 MM_SP_MEMORY_ATTRIBUTES_SET_AARCH64.
> ZeroMem (&SvcArgs, sizeof (ARM_SVC_ARGS));
> if (FixedPcdGet32 (PcdFfaEnable) != 0) {
> - // See [2], Section 10.2 FFA_MSG_SEND_DIRECT_REQ.
> - SvcArgs.Arg0 = ARM_SVC_ID_FFA_MSG_SEND_DIRECT_REQ;
> - SvcArgs.Arg1 = ARM_FFA_DESTINATION_ENDPOINT_ID;
> - SvcArgs.Arg2 = 0;
> - SvcArgs.Arg3 = ARM_SVC_ID_SP_SET_MEM_ATTRIBUTES;
> - SvcArgs.Arg4 = BaseAddress;
> - SvcArgs.Arg5 = EFI_SIZE_TO_PAGES (Length);
> - SvcArgs.Arg6 = Permissions;
> + // Check if FF-A memory permission ABIs can be used.
> + if (UseFfaMemPermAbis()) {
> + SvcArgs.Arg0 = ARM_SVC_ID_FFA_MEM_PERM_SET_AARCH32;
> + SvcArgs.Arg1 = BaseAddress;
> + SvcArgs.Arg2 = EFI_SIZE_TO_PAGES (Length);
> + SvcArgs.Arg3 = Permissions;
> + } else {
> + // See [2], Section 10.2 FFA_MSG_SEND_DIRECT_REQ.
> + SvcArgs.Arg0 = ARM_SVC_ID_FFA_MSG_SEND_DIRECT_REQ;
> + SvcArgs.Arg1 = ARM_FFA_DESTINATION_ENDPOINT_ID;
> + SvcArgs.Arg2 = 0;
> + SvcArgs.Arg3 = ARM_SVC_ID_SP_SET_MEM_ATTRIBUTES;
> + SvcArgs.Arg4 = BaseAddress;
> + SvcArgs.Arg5 = EFI_SIZE_TO_PAGES (Length);
> + SvcArgs.Arg6 = Permissions;
> + }
> } else {
> SvcArgs.Arg0 = ARM_SVC_ID_SP_SET_MEM_ATTRIBUTES;
> SvcArgs.Arg1 = BaseAddress;
next prev parent reply other threads:[~2023-07-12 17:43 UTC|newest]
Thread overview: 41+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-07-11 14:36 [edk2-platforms][PATCH V1 00/20] Add the support for ARM Firmware First Framework Nishant Sharma
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 01/20] ArmPkg: Change PcdFfaEnable flag datatype Nishant Sharma
2023-07-12 17:21 ` [edk2-devel] " Oliver Smith-Denny
2023-07-12 17:23 ` Chris Fernald
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 02/20] StandaloneMmPkg: Allocate and initialise SP stack from internal memory Nishant Sharma
2023-07-12 17:47 ` [edk2-devel] " Chris Fernald
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 03/20] StandaloneMmPkg: Include libfdt in the StMM Nishant Sharma
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 04/20] ArmPkg: Add data structures to receive FF-A boot information Nishant Sharma
2023-07-12 17:27 ` [edk2-devel] " Oliver Smith-Denny
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 05/20] ArmPkg/ArmFfaSvc: Add helper macros and fids Nishant Sharma
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 06/20] ArmPkg: Add support for FFA_MEM_PERM_GET/SET ABIs Nishant Sharma
2023-07-12 17:43 ` Oliver Smith-Denny [this message]
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 07/20] StandaloneMmPkg: define new data structure to stage FF-A boot information Nishant Sharma
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 08/20] StandaloneMmPkg: Add backwards compatible support to detect FF-A v1.1 Nishant Sharma
2023-07-12 20:31 ` [edk2-devel] " Oliver Smith-Denny
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 09/20] StandaloneMmPkg: parse SP manifest and populate new boot information Nishant Sharma
2023-07-13 15:24 ` [edk2-devel] " Girish Mahadevan
2023-07-13 16:48 ` Chris Fernald
2023-07-13 20:49 ` Achin Gupta
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 10/20] StandaloneMmPkg: Populate Hoblist for SP init from StMM " Nishant Sharma
2023-07-12 20:52 ` [edk2-devel] " Oliver Smith-Denny
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 11/20] StandaloneMmPkg: Skip zero sized sections while tweaking page permissions Nishant Sharma
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 12/20] StandaloneMmPkg: Add global check for FF-A abis Nishant Sharma
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 13/20] ArmPkg: Bump the StMM SP FF-A minor version to 1 Nishant Sharma
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 14/20] ArmPkg/MmCommunicationDxe: Introduce FF-A version check Nishant Sharma
2023-07-13 16:56 ` [edk2-devel] " Chris Fernald
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 15/20] ArmPkg/MmCommunicationDxe: Add support for obtaining FF-A partition ID Nishant Sharma
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 16/20] ArmPkg/MmCommunicationDxe: Register FF-A RX/TX buffers Nishant Sharma
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 17/20] ArmPkg/MmCommunicationDxe: Unmap FF-A RX/TX buffers during ExitBootServices Nishant Sharma
2023-07-12 20:59 ` [edk2-devel] " Oliver Smith-Denny
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 18/20] ArmPkg/MmCommunicationDxe: Discover the StMM SP Nishant Sharma
2023-07-12 21:10 ` [edk2-devel] " Oliver Smith-Denny
2023-07-12 21:48 ` Girish Mahadevan
2023-07-13 17:16 ` Chris Fernald
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 19/20] ArmPkg/MmCommunicationDxe: Use the FF-A transport for MM requests Nishant Sharma
2023-07-11 19:22 ` [edk2-devel] " Kun Qin
2023-07-12 14:21 ` achin.gupta
2023-07-12 17:13 ` Kun Qin
2023-07-12 21:49 ` Girish Mahadevan
2023-07-11 14:36 ` [edk2-platforms][PATCH V1 20/20] StandaloneMmPkg: Add support for MM requests as FF-A direct messages Nishant Sharma
2023-07-12 21:13 ` [edk2-devel] [edk2-platforms][PATCH V1 00/20] Add the support for ARM Firmware First Framework Oliver Smith-Denny
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-list from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=c7114034-a823-9e24-8754-96edeefb4d5a@linux.microsoft.com \
--to=devel@edk2.groups.io \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox