Re: [edk2-devel] [PATCH v3 2/3] UefiCpuPkg: RISC-V: MMU: Support Svpbmt extension

["Laszlo Ersek" <lersek@redhat.com>]

On 3/2/24 00:20, Tuan Phan wrote:
> Thanks for the detailed review. Please see my comments below.
> 
> On Fri, Mar 1, 2024 at 4:14 AM Laszlo Ersek <lersek@redhat.com
> <mailto:lersek@redhat.com>> wrote:
> 
>     On 3/1/24 02:29, Tuan Phan wrote:
>     > The GCD EFI_MEMORY_UC and EFI_MEMORY_WC memory attributes will be
>     > supported when Svpbmt extension available.
>     >
>     > Cc: Gerd Hoffmann <kraxel@redhat.com <mailto:kraxel@redhat.com>>
>     > Cc: Laszlo Ersek <lersek@redhat.com <mailto:lersek@redhat.com>>
>     > Cc: Rahul Kumar <rahul1.kumar@intel.com
>     <mailto:rahul1.kumar@intel.com>>
>     > Cc: Ray Ni <ray.ni@intel.com <mailto:ray.ni@intel.com>>
>     > Signed-off-by: Tuan Phan <tphan@ventanamicro.com
>     <mailto:tphan@ventanamicro.com>>
>     > ---
>     >  .../Library/BaseRiscVMmuLib/BaseRiscVMmuLib.c | 101
>     +++++++++++++++---
>     >  .../BaseRiscVMmuLib/BaseRiscVMmuLib.inf       |   1 +
>     >  2 files changed, 88 insertions(+), 14 deletions(-)
>     >
>     > diff --git a/UefiCpuPkg/Library/BaseRiscVMmuLib/BaseRiscVMmuLib.c
>     b/UefiCpuPkg/Library/BaseRiscVMmuLib/BaseRiscVMmuLib.c
>     > index 826a1d32a1d4..f4419bb8f380 100644
>     > --- a/UefiCpuPkg/Library/BaseRiscVMmuLib/BaseRiscVMmuLib.c
>     > +++ b/UefiCpuPkg/Library/BaseRiscVMmuLib/BaseRiscVMmuLib.c
>     > @@ -36,6 +36,11 @@
>     >  #define PTE_PPN_SHIFT         10
>     >  #define RISCV_MMU_PAGE_SHIFT  12
>     > 
>     > +#define RISCV_CPU_FEATURE_PBMT_BITMASK  BIT2
>     > +#define PTE_PBMT_NC                     BIT61
>     > +#define PTE_PBMT_IO                     BIT62
>     > +#define PTE_PBMT_MASK                   (PTE_PBMT_NC | PTE_PBMT_IO)
>     > +
>     >  STATIC UINTN  mModeSupport[] = { SATP_MODE_SV57, SATP_MODE_SV48,
>     SATP_MODE_SV39, SATP_MODE_OFF };
>     >  STATIC UINTN  mMaxRootTableLevel;
>     >  STATIC UINTN  mBitPerLevel;
>     > @@ -489,32 +494,89 @@ UpdateRegionMapping (
>     >  /**
>     >    Convert GCD attribute to RISC-V page attribute.
>     > 
>     > -  @param  GcdAttributes The GCD attribute.
>     > +  @param  GcdAttributes   The GCD attribute.
>     > +  @param  RiscVAttribtues The pointer of RISC-V page attribute.
>     > 
>     > -  @return               The RISC-V page attribute.
>     > +  @retval EFI_INVALID_PARAMETER   The RiscVAttribtues is NULL or
>     cache type mask not valid.
>     > +  @retval EFI_SUCCESS             The operation succesfully.
>     > 
>     >  **/
>     >  STATIC
>     > -UINTN
>     > +EFI_STATUS
>     >  GcdAttributeToPageAttribute (
>     > -  IN UINTN  GcdAttributes
>     > +  IN UINTN   GcdAttributes,
> 
>     Just noticing: why is GcdAttributes *not* UINT64 in the first place?
> 
>     All the bit macros we test against it, such as EFI_MEMORY_RO
>     (0x0000000000020000ULL) are of type unsigned long long (UINT64).
> 
> Good catch. Will fix it. 
> 
> 
>     > +  OUT UINTN  *RiscVAttributes
>     >    )
>     >  {
>     > -  UINTN  RiscVAttributes;
>     > +  UINT64   CacheTypeMask;
>     > +  BOOLEAN  PmbtExtEnabled = (PcdGet64 (PcdRiscVFeatureOverride) &
>     RISCV_CPU_FEATURE_PBMT_BITMASK) ? TRUE : FALSE;
> 
>     - Per the edk2 coding style, locals should not be initialized (separate
>     assignment is needed).
> 
>     - Bitmask checks always need an explicit comparison, such as
> 
>       ((a & b) != 0)
> 
>     or similar. Implicitly interpreting (a & b) as a truth value is not
>     appropriate.
> 
>     - "(whatever) ? TRUE : FALSE" is both bad style and unnecessary.
> 
>       BOOLEAN  PmbtExtEnabled;
> 
>       PmbtExtEnabled = (PcdGet64 (PcdRiscVFeatureOverride) &
>                         RISCV_CPU_FEATURE_PBMT_BITMASK) != 0;
> 
> Will fix it. 
> 
>     > 
>     > -  RiscVAttributes = RISCV_PG_R | RISCV_PG_W | RISCV_PG_X;
>     > +  if (!RiscVAttributes) {
> 
>     - The coding style requires an explicit nullity check:
> 
>       if (RiscVAttributes == NULL) {
> 
> Will fix it.  
> 
> 
>     > +    return EFI_INVALID_PARAMETER;
>     > +  }
>     > +
>     > +  *RiscVAttributes = RISCV_PG_R | RISCV_PG_W | RISCV_PG_X;
>     > 
>     >    // Determine protection attributes
>     >    if ((GcdAttributes & EFI_MEMORY_RO) != 0) {
>     > -    RiscVAttributes &= ~(RISCV_PG_W);
>     > +    *RiscVAttributes &= ~(RISCV_PG_W);
>     >    }
>     > 
>     >    // Process eXecute Never attribute
>     >    if ((GcdAttributes & EFI_MEMORY_XP) != 0) {
>     > -    RiscVAttributes &= ~RISCV_PG_X;
>     > +    *RiscVAttributes &= ~RISCV_PG_X;
>     > +  }
>     > +
> 
>     My next comment is unrelated to the patch, it's just something that
>     catches my eye, and I think is worth fixing:
> 
>     RISCV_PG_W is BIT2 (0x00000004), and RISCV_PG_X is BIT3 (0x00000008).
>     Meaning, they are of type *signed int* (INT32). Applying the bit-neg
>     operator on them produces a negative value (because it flips the sign
>     bit), which is very ugly.
> 
>     I suggest a separate patch for changing these into
> 
>       ~(UINTN)RISCV_PG_W
>       ~(UINTN)RISCV_PG_X
> 
>     Alternatively, you could do
> 
> Will fix it in a separate patch along with the above change.
> 
> 
>       *RiscVAttributes = RISCV_PG_R;
>       if ((GcdAttributes & EFI_MEMORY_RO) == 0) {
>         *RiscVAttributes |= RISCV_PG_W;
>       }
>       if ((GcdAttributes & EFI_MEMORY_XP) == 0) {
>         *RiscVAttributes |= RISCV_PG_X;
>       }
> 
>     Either way: separate patch.
> 
>     > +  CacheTypeMask = GcdAttributes & EFI_CACHE_ATTRIBUTE_MASK;
>     > +  if ((CacheTypeMask != 0) &&
>     > +      (((CacheTypeMask - 1) & CacheTypeMask) != 0))
> 
>     This is not what I recommended in my previous review
>     <https://edk2.groups.io/g/devel/message/115243
>     <https://edk2.groups.io/g/devel/message/115243>>.
> 
>     Compare:
> 
>       (CacheTypeMask != 0) && ...
> 
>     versus
> 
>       (CacheTypeMask == 0) || ...
> 
>     Both of these ensure that the power-of-two check in the second
>     subcondition (i.e., the subtraction of 1) is avoided when CacheTypeMask
>     is zero. In the first variant, you get (FALSE && ...), in the second
>     variant, you get (TRUE || ...); therefore, the power-of-two check is
>     short-circuited for a zero input in both variants.
> 
>     However, considering the larger CacheTypeMask validation, your variant
>     is incorrect, because a zero CacheTypeMask will ultimately evaluate the
>     condition to FALSE -- (FALSE && ...) is FALSE --, and so the "return
>     EFI_INVALID_PARAMETER" statement will not be reached. Whereas (TRUE ||
>     ...) is TRUE, and so we return EFI_INVALID_PARAMETER for
>     CacheTypeMask==0.
> 
> Actually the EDK2 passes (CacheTypeMask == 0) to this API during my
> debug session.
> Given that situation, this function doesn't do anything when
> CacheTypeMask  == 0 so I think
> it should not give the warning message.

I would be curious how that can happen; to me a CacheTypeMask==0 input
looks somewhat invalid.

Either way, if such an input *is* valid, then there is a different
problem with the patch: in the debug message we say that the cache type
mask should contain *exactly one* bit set. That's not correct then: it
should say *at most one* bit set. (Because the value 0 has 0 bits set,
and apparently that is valid input.)


> 
> 
>     > +  {
>     > +    DEBUG (
>     > +      (
>     > +       DEBUG_ERROR,
>     > +       "%a: The cache type mask (0x%llX) should contain exactly
>     one bit set\n",
> 
>     - Edk2's PrintLib does not use "ll" length modifiers. %u, %x and %X are
>     for UINT32, and %lu, %lx and %lX are for UINT64. Furthermore, you may
>     replace "l" with "L" freely.
> 
> Will fix it. 
> 
> 
>     - We generally group together the double parens for DEBUG invocations:
> 
>       DEBUG ((
>         DEBUG_ERROR,
>         "%a: The cache type mask (0x%lX) ...\n",
>         __func__,
>         CacheTypeMask
>         ));
> 
>     > +       __func__,
>     > +       CacheTypeMask
>     > +      )
>     > +      );
> 
>     The indentation of the closing parens is not correct either; please put
>     your patches through uncrustify first. (CI will reject these issues
>     anyway, in github pull requests.)
> 
> Actually this code is the result of uncrustify modification. Let me
> check if anything
> wrong with uncrustify. 

It's very strange. Do you know what your original code (the input to
uncrustify) looked like? I wonder if uncrustify produces strange output
if it sees unexpected input. Normally I wouldn't expect uncrustify to
change the "((" format that I'm proposing. If it still does, then my
request is invalid of course (uncrustify has priority, whatever it does).

Thanks!
Laszlo

> 
> 
>     For running uncrustify locally:
> 
>     - clone
>     <https://projectmu@dev.azure.com/projectmu/Uncrustify/_git/Uncrustify <https://projectmu@dev.azure.com/projectmu/Uncrustify/_git/Uncrustify>>
> 
>     - check it out at tag 73.0.8 (the tag that edk2 CI uses on github is in
>     ".pytool/Plugin/UncrustifyCheck/uncrustify_ext_dep.yaml")
> 
>     - build it (IIRC it uses cmake)
> 
>     - with nothing dirty in the working tree (i.e., everything committed, or
>     at least stashed to the index), run
> 
>       uncrustify \
>         -c .pytool/Plugin/UncrustifyCheck/uncrustify.cfg \
>         --replace \
>         --no-backup \
>         --if-changed \
>         -F file-list.txt
> 
>     > +    return EFI_INVALID_PARAMETER;
>     >    }
>     > 
>     > -  return RiscVAttributes;
>     > +  switch (CacheTypeMask) {
>     > +    case EFI_MEMORY_UC:
>     > +      if (PmbtExtEnabled) {
>     > +        *RiscVAttributes |= PTE_PBMT_IO;
>     > +      } else {
>     > +        DEBUG (
>     > +          (
>     > +           DEBUG_VERBOSE,
>     > +           "%a: EFI_MEMORY_UC set but Pmbt extension not
>     available\n",
>     > +           __func__
>     > +          )
>     > +          );
>     > +      }
>     > +
>     > +      break;
>     > +    case EFI_MEMORY_WC:
>     > +      if (PmbtExtEnabled) {
>     > +        *RiscVAttributes |= PTE_PBMT_NC;
>     > +      } else {
>     > +        DEBUG (
>     > +          (
>     > +           DEBUG_VERBOSE,
>     > +           "%a: EFI_MEMORY_WC set but Pmbt extension not
>     available\n",
>     > +           __func__
>     > +          )
>     > +          );
>     > +      }
>     > +
>     > +      break;
>     > +    default:
>     > +      // Default PMA mode
>     > +      break;
>     > +  }
>     > +
>     > +  return EFI_SUCCESS;
>     >  }
>     > 
>     >  /**
>     > @@ -537,21 +599,32 @@ RiscVSetMemoryAttributes (
>     >    IN UINTN                 Attributes
>     >    )
>     >  {
>     > -  UINTN  PageAttributesSet;
>     > +  UINTN       PageAttributesSet;
>     > +  UINTN       PageAttributesClear;
>     > +  EFI_STATUS  Status;
>     > 
>     > -  PageAttributesSet = GcdAttributeToPageAttribute (Attributes);
>     > +  Status = GcdAttributeToPageAttribute (Attributes,
>     &PageAttributesSet);
>     > +  if (EFI_ERROR (Status)) {
>     > +    return Status;
>     > +  }
>     > 
>     >    if (!RiscVMmuEnabled ()) {
>     >      return EFI_SUCCESS;
>     >    }
>     > 
>     > +  PageAttributesClear = PTE_ATTRIBUTES_MASK;
>     > +  if ((PcdGet64 (PcdRiscVFeatureOverride) &
>     RISCV_CPU_FEATURE_PBMT_BITMASK) != 0) {
>     > +    PageAttributesClear |= PTE_PBMT_MASK;
>     > +  }
>     > +
>     >    DEBUG (
>     >      (
>     >       DEBUG_VERBOSE,
>     > -     "%a: Set %llX page attribute 0x%X\n",
>     > +     "%a: %llX: set attributes 0x%X, clear attributes 0x%X\n",
>     >       __func__,
>     >       BaseAddress,
>     > -     PageAttributesSet
>     > +     PageAttributesSet,
>     > +     PageAttributesClear
>     >      )
>     >      );
> 
>     - UINT64 should be formatted with %[Ll][uxX].
> 
>     - UINT32 should be formatted with %[uxX].
> 
>     - UINTN is trickier, there is no dedicated conversion specifier. The
>     portable solution (between 32-bit and 64-bit platforms in edk2) is to
>     (a) cast the UINTN value to UINT64, (b) format the latter with
>     %[Ll][uxX].
> 
>     So you need something like
> 
>       DEBUG ((
>         DEBUG_VERBOSE,
>         "%a: %LX: set attributes 0x%LX, clear attributes 0x%LX\n",
>         __func__,
>         BaseAddress,                // this is UINT64
>         (UINT64)PageAttributesSet,  // originally UINTN
>         (UINT64)PageAttributesClear // originally UINTN
>         ));
> 
> Thanks for the suggestion. Will fix it. 
> 
> 
>     > 
>     > @@ -559,7 +632,7 @@ RiscVSetMemoryAttributes (
>     >             BaseAddress,
>     >             Length,
>     >             PageAttributesSet,
>     > -           PTE_ATTRIBUTES_MASK,
>     > +           PageAttributesClear,
>     >             (UINTN *)RiscVGetRootTranslateTable (),
>     >             TRUE
>     >             );
>     > diff --git
>     a/UefiCpuPkg/Library/BaseRiscVMmuLib/BaseRiscVMmuLib.inf
>     b/UefiCpuPkg/Library/BaseRiscVMmuLib/BaseRiscVMmuLib.inf
>     > index 51ebe1750e97..1dbaa81f3608 100644
>     > --- a/UefiCpuPkg/Library/BaseRiscVMmuLib/BaseRiscVMmuLib.inf
>     > +++ b/UefiCpuPkg/Library/BaseRiscVMmuLib/BaseRiscVMmuLib.inf
>     > @@ -28,3 +28,4 @@
>     > 
>     >  [Pcd]
>     >    gUefiCpuPkgTokenSpaceGuid.PcdCpuRiscVMmuMaxSatpMode  ## CONSUMES
>     > +  gEfiMdePkgTokenSpaceGuid.PcdRiscVFeatureOverride     ## CONSUMES
> 
>     Laszlo
> 
> 



-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#116337): https://edk2.groups.io/g/devel/message/116337
Mute This Topic: https://groups.io/mt/104656466/7686176
Group Owner: devel+owner@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [rebecca@openfw.io]
-=-=-=-=-=-=-=-=-=-=-=-