public inbox for devel@edk2.groups.io
 help / color / mirror / Atom feed
From: "Philippe Mathieu-Daudé" <philmd@redhat.com>
To: devel@edk2.groups.io, ard.biesheuvel@linaro.org
Cc: leif.lindholm@linaro.org
Subject: Re: [edk2-devel] [PATCH v2 1/2] EmbeddedPkg/NonCoherentDmaLib: implement support for DMA range limits
Date: Tue, 26 Nov 2019 12:59:12 +0100	[thread overview]
Message-ID: <08876497-a2ce-4b55-6f50-f332ffd45ddd@redhat.com> (raw)
In-Reply-To: <ee6f5062-eeb1-fe89-ec8a-5ab2e6367b7a@redhat.com>

On 11/26/19 11:44 AM, Philippe Mathieu-Daudé wrote:
> On 11/26/19 12:12 AM, Ard Biesheuvel via Groups.Io wrote:
>> Implement support for driving peripherals with limited DMA ranges to
>> NonCoherentDmaLib, by adding a device address limit, and taking it,
>> along with the device offset, into account when allocating or mapping
>> DMA buffers.
>>
>> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
>> ---
>>   EmbeddedPkg/EmbeddedPkg.dec                                 |   6 +
>>   EmbeddedPkg/Library/NonCoherentDmaLib/NonCoherentDmaLib.c   | 165 
>> ++++++++++++++++++--
>>   EmbeddedPkg/Library/NonCoherentDmaLib/NonCoherentDmaLib.inf |   1 +
>>   3 files changed, 160 insertions(+), 12 deletions(-)
>>
>> diff --git a/EmbeddedPkg/EmbeddedPkg.dec b/EmbeddedPkg/EmbeddedPkg.dec
>> index 8812a6db7c30..69922802f473 100644
>> --- a/EmbeddedPkg/EmbeddedPkg.dec
>> +++ b/EmbeddedPkg/EmbeddedPkg.dec
>> @@ -186,6 +186,12 @@ [PcdsFixedAtBuild.common, PcdsDynamic.common]
>>     #
>>     gEmbeddedTokenSpaceGuid.PcdDmaDeviceOffset|0x0|UINT64|0x0000058
>> +  #
>> +  # Highest address value supported by the device for DMA addressing. 
>> Note
>> +  # that this value should be strictly greater than PcdDmaDeviceOffset.
>> +  #
>> +  
>> gEmbeddedTokenSpaceGuid.PcdDmaDeviceLimit|0xFFFFFFFFFFFFFFFF|UINT64|0x000005A 
>>
>> +
>>     #
>>     # Selection between DT and ACPI as a default
>>     #
>> diff --git a/EmbeddedPkg/Library/NonCoherentDmaLib/NonCoherentDmaLib.c 
>> b/EmbeddedPkg/Library/NonCoherentDmaLib/NonCoherentDmaLib.c
>> index 78220f6358aa..115345765435 100644
>> --- a/EmbeddedPkg/Library/NonCoherentDmaLib/NonCoherentDmaLib.c
>> +++ b/EmbeddedPkg/Library/NonCoherentDmaLib/NonCoherentDmaLib.c
>> @@ -40,6 +40,8 @@ typedef struct {
>>   STATIC EFI_CPU_ARCH_PROTOCOL      *mCpu;
>>   STATIC LIST_ENTRY                 UncachedAllocationList;
>> +STATIC PHYSICAL_ADDRESS           mDmaHostAddressLimit;
>> +
>>   STATIC
>>   PHYSICAL_ADDRESS
>>   HostToDeviceAddress (
>> @@ -49,6 +51,102 @@ HostToDeviceAddress (
>>     return (PHYSICAL_ADDRESS)(UINTN)Address + PcdGet64 
>> (PcdDmaDeviceOffset);
>>   }
>> +/**
>> +  Allocates one or more 4KB pages of a certain memory type at a 
>> specified
>> +  alignment.
>> +
>> +  Allocates the number of 4KB pages specified by Pages of a certain 
>> memory type
>> +  with an alignment specified by Alignment. The allocated buffer is 
>> returned.
>> +  If Pages is 0, then NULL is returned. If there is not enough memory 
>> at the
>> +  specified alignment remaining to satisfy the request, then NULL is 
>> returned.
>> +  If Alignment is not a power of two and Alignment is not zero, then 
>> ASSERT().
>> +  If Pages plus EFI_SIZE_TO_PAGES (Alignment) overflows, then ASSERT().
>> +
>> +  @param  MemoryType            The type of memory to allocate.
>> +  @param  Pages                 The number of 4 KB pages to allocate.
>> +  @param  Alignment             The requested alignment of the 
>> allocation.
>> +                                Must be a power of two.
>> +                                If Alignment is zero, then byte 
>> alignment is
>> +                                used.
>> +
>> +  @return A pointer to the allocated buffer or NULL if allocation fails.
>> +
>> +**/
>> +STATIC
>> +VOID *
>> +InternalAllocateAlignedPages (
>> +  IN EFI_MEMORY_TYPE  MemoryType,
>> +  IN UINTN            Pages,
>> +  IN UINTN            Alignment
>> +  )
>> +{
>> +  EFI_STATUS            Status;
>> +  EFI_PHYSICAL_ADDRESS  Memory;
>> +  UINTN                 AlignedMemory;
>> +  UINTN                 AlignmentMask;
>> +  UINTN                 UnalignedPages;
>> +  UINTN                 RealPages;
>> +
>> +  //
>> +  // Alignment must be a power of two or zero.
>> +  //
>> +  ASSERT ((Alignment & (Alignment - 1)) == 0);
>> +
>> +  if (Pages == 0) {
>> +    return NULL;
>> +  }
>> +  if (Alignment > EFI_PAGE_SIZE) {
>> +    //
>> +    // Calculate the total number of pages since alignment is larger 
>> than page
>> +    // size.
>> +    //
>> +    AlignmentMask  = Alignment - 1;
>> +    RealPages      = Pages + EFI_SIZE_TO_PAGES (Alignment);
>> +    //
>> +    // Make sure that Pages plus EFI_SIZE_TO_PAGES (Alignment) does not
>> +    // overflow.
>> +    //
>> +    ASSERT (RealPages > Pages);
>> +
>> +    Memory = mDmaHostAddressLimit;
>> +    Status = gBS->AllocatePages (AllocateMaxAddress, MemoryType, 
>> RealPages,
>> +                    &Memory);
>> +    if (EFI_ERROR (Status)) {
>> +      return NULL;
>> +    }
>> +    AlignedMemory  = ((UINTN)Memory + AlignmentMask) & ~AlignmentMask;
>> +    UnalignedPages = EFI_SIZE_TO_PAGES (AlignedMemory - (UINTN)Memory);

The previous line made me think twice (due to the cast), but I can't 
find a simpler way to write the same code.

>> +    if (UnalignedPages > 0) {
>> +      //
>> +      // Free first unaligned page(s).
>> +      //
>> +      Status = gBS->FreePages (Memory, UnalignedPages);
>> +      ASSERT_EFI_ERROR (Status);
>> +    }
>> +    Memory         = AlignedMemory + EFI_PAGES_TO_SIZE (Pages);
>> +    UnalignedPages = RealPages - Pages - UnalignedPages;
>> +    if (UnalignedPages > 0) {
>> +      //
>> +      // Free last unaligned page(s).
>> +      //
>> +      Status = gBS->FreePages (Memory, UnalignedPages);
>> +      ASSERT_EFI_ERROR (Status);
>> +    }
>> +  } else {
>> +    //
>> +    // Do not over-allocate pages in this case.
>> +    //
>> +    Memory = mDmaHostAddressLimit;
>> +    Status = gBS->AllocatePages (AllocateMaxAddress, MemoryType, Pages,
>> +                    &Memory);
>> +    if (EFI_ERROR (Status)) {
>> +      return NULL;
>> +    }
>> +    AlignedMemory = (UINTN)Memory;
>> +  }
>> +  return (VOID *)AlignedMemory;
>> +}
>> +
>>   /**
>>     Provides the DMA controller-specific addresses needed to access 
>> system memory.
>> @@ -111,7 +209,30 @@ DmaMap (
>>       return  EFI_OUT_OF_RESOURCES;
>>     }
>> -  if (Operation != MapOperationBusMasterRead &&
>> +  if (((UINTN)HostAddress + *NumberOfBytes) > mDmaHostAddressLimit) {
>> +
>> +    if (Operation == MapOperationBusMasterCommonBuffer) {
>> +      goto CommonBufferError;
>> +    }
>> +
>> +    AllocSize = ALIGN_VALUE (*NumberOfBytes, mCpu->DmaBufferAlignment);
>> +    Map->BufferAddress = InternalAllocateAlignedPages 
>> (EfiBootServicesData,
>> +                           EFI_SIZE_TO_PAGES (AllocSize),
>> +                           mCpu->DmaBufferAlignment);
>> +    if (Map->BufferAddress == NULL) {
>> +      Status = EFI_OUT_OF_RESOURCES;
>> +      goto FreeMapInfo;
>> +    }
>> +
>> +    if (Map->Operation == MapOperationBusMasterRead) {
>> +      CopyMem (Map->BufferAddress, (VOID *)(UINTN)Map->HostAddress,
>> +        *NumberOfBytes);
>> +    }
>> +    mCpu->FlushDataCache (mCpu, (UINTN)Map->BufferAddress, AllocSize,
>> +            EfiCpuFlushTypeWriteBack);
>> +
>> +    *DeviceAddress = HostToDeviceAddress (Map->BufferAddress);
>> +  } else if (Operation != MapOperationBusMasterRead &&
>>         ((((UINTN)HostAddress & (mCpu->DmaBufferAlignment - 1)) != 0) ||
>>          ((*NumberOfBytes & (mCpu->DmaBufferAlignment - 1)) != 0))) {
>> @@ -128,12 +249,7 @@ DmaMap (
>>         // on uncached buffers.
>>         //
>>         if (Operation == MapOperationBusMasterCommonBuffer) {
>> -        DEBUG ((DEBUG_ERROR,
>> -          "%a: Operation type 'MapOperationBusMasterCommonBuffer' is 
>> only "
>> -          "supported\non memory regions that were allocated using "
>> -          "DmaAllocateBuffer ()\n", __FUNCTION__));
>> -        Status = EFI_UNSUPPORTED;
>> -        goto FreeMapInfo;
>> +        goto CommonBufferError;
>>         }
>>         //
>> @@ -199,6 +315,12 @@ DmaMap (
>>     return EFI_SUCCESS;
>> +CommonBufferError:
>> +  DEBUG ((DEBUG_ERROR,
>> +    "%a: Operation type 'MapOperationBusMasterCommonBuffer' is only "
>> +    "supported\non memory regions that were allocated using "
>> +    "DmaAllocateBuffer ()\n", __FUNCTION__));
>> +  Status = EFI_UNSUPPORTED;
>>   FreeMapInfo:
>>     FreePool (Map);
>> @@ -229,6 +351,7 @@ DmaUnmap (
>>     MAP_INFO_INSTANCE *Map;
>>     EFI_STATUS        Status;
>>     VOID              *Buffer;
>> +  UINTN             AllocSize;
>>     if (Mapping == NULL) {
>>       ASSERT (FALSE);
>> @@ -238,7 +361,17 @@ DmaUnmap (
>>     Map = (MAP_INFO_INSTANCE *)Mapping;
>>     Status = EFI_SUCCESS;
>> -  if (Map->DoubleBuffer) {
>> +  if (((UINTN)Map->HostAddress + Map->NumberOfBytes) > 
>> mDmaHostAddressLimit) {
>> +    AllocSize = ALIGN_VALUE (Map->NumberOfBytes, 
>> mCpu->DmaBufferAlignment);
>> +    if (Map->Operation == MapOperationBusMasterWrite) {
>> +      mCpu->FlushDataCache (mCpu, (UINTN)Map->BufferAddress, AllocSize,
>> +              EfiCpuFlushTypeInvalidate);
>> +      CopyMem ((VOID *)(UINTN)Map->HostAddress, Map->BufferAddress,
>> +        Map->NumberOfBytes);
>> +    }
>> +    FreePages (Map->BufferAddress, EFI_SIZE_TO_PAGES (AllocSize));
>> +  } else if (Map->DoubleBuffer) {
>> +
>>       ASSERT (Map->Operation == MapOperationBusMasterWrite);
>>       if (Map->Operation != MapOperationBusMasterWrite) {
>> @@ -335,10 +468,9 @@ DmaAllocateAlignedBuffer (
>>       return EFI_INVALID_PARAMETER;
>>     }
>> -  if (MemoryType == EfiBootServicesData) {
>> -    Allocation = AllocateAlignedPages (Pages, Alignment);
>> -  } else if (MemoryType == EfiRuntimeServicesData) {
>> -    Allocation = AllocateAlignedRuntimePages (Pages, Alignment);
>> +  if (MemoryType == EfiBootServicesData ||
>> +      MemoryType == EfiRuntimeServicesData) {
>> +    Allocation = InternalAllocateAlignedPages (MemoryType, Pages, 
>> Alignment);
>>     } else {
>>       return EFI_INVALID_PARAMETER;
>>     }
>> @@ -479,6 +611,15 @@ NonCoherentDmaLibConstructor (
>>   {
>>     InitializeListHead (&UncachedAllocationList);
>> +  //
>> +  // Ensure that the combination of DMA addressing offset and limit 
>> produces
>> +  // a sane value.
>> +  //
>> +  ASSERT (PcdGet64 (PcdDmaDeviceLimit) > PcdGet64 (PcdDmaDeviceOffset));
>> +
>> +  mDmaHostAddressLimit = PcdGet64 (PcdDmaDeviceLimit) -
>> +                         PcdGet64 (PcdDmaDeviceOffset);
>> +
>>     // Get the Cpu protocol for later use
>>     return gBS->LocateProtocol (&gEfiCpuArchProtocolGuid, NULL, (VOID 
>> **)&mCpu);
>>   }
>> diff --git 
>> a/EmbeddedPkg/Library/NonCoherentDmaLib/NonCoherentDmaLib.inf 
>> b/EmbeddedPkg/Library/NonCoherentDmaLib/NonCoherentDmaLib.inf
>> index 2db751afee91..1a21cfe4ff23 100644
>> --- a/EmbeddedPkg/Library/NonCoherentDmaLib/NonCoherentDmaLib.inf
>> +++ b/EmbeddedPkg/Library/NonCoherentDmaLib/NonCoherentDmaLib.inf
>> @@ -38,6 +38,7 @@ [Protocols]
>>   [Pcd]
>>     gEmbeddedTokenSpaceGuid.PcdDmaDeviceOffset
>> +  gEmbeddedTokenSpaceGuid.PcdDmaDeviceLimit
>>   [Depex]
>>     gEfiCpuArchProtocolGuid
>>
> 
> For the 82% of the patch I understood:
> Reviewed-by: Philippe Mathieu-Daude <philmd@redhat.com>

Leif explained me that "I've reviewed it, but don't fully understand it, 
but I think it's valuable" translate into a Acked-by on EDK2, so please 
replace the previous Reviewed-by by:
Acked-by: Philippe Mathieu-Daude <philmd@redhat.com>

Thanks,

Phil.


  reply	other threads:[~2019-11-26 11:59 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-11-25 23:12 [PATCH v2 0/2] EmbeddedPkg: support for RPi4 PCI and platform DMA Ard Biesheuvel
2019-11-25 23:12 ` [PATCH v2 1/2] EmbeddedPkg/NonCoherentDmaLib: implement support for DMA range limits Ard Biesheuvel
2019-11-26 10:28   ` [edk2-devel] " Pete Batard
2019-11-26 10:44   ` Philippe Mathieu-Daudé
2019-11-26 11:59     ` Philippe Mathieu-Daudé [this message]
2019-11-26 12:02       ` Ard Biesheuvel
2019-11-26 12:04   ` Leif Lindholm
2019-11-26 12:11     ` Ard Biesheuvel
2019-11-26 12:22       ` Leif Lindholm
2019-11-25 23:12 ` [PATCH v2 2/2] EmbeddedPkg: implement EDK2 IoMmu protocol wrapping DmaLib Ard Biesheuvel
2019-11-26  8:19   ` [edk2-devel] " Philippe Mathieu-Daudé
2019-11-26 10:28     ` Pete Batard

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=08876497-a2ce-4b55-6f50-f332ffd45ddd@redhat.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