[ntuple] factor out loading of zero pages

Author: jblomer

tree/ntuple/inc/ROOT/RPageStorage.hxx

@@ -567,6 +567,16 @@ The page source also gives access to the ntuple's metadata.
 */
 // clang-format on
 class RPageSource : public RPageStorage {
+protected:
+   /// Summarizes meta-data necessary to load a certain page. Used by LoadPageImpl().
+   struct RPageSummary {
+      ROOT::DescriptorId_t fClusterId = 0;
+      /// The first element number of the page's column in the given cluster
+      std::uint64_t fColumnOffset = 0;
+      /// Location of the page on disk
+      ROOT::RClusterDescriptor::RPageInfoExtended fPageInfo;
+   };
+
 public:
    /// Used in SetEntryRange / GetEntryRange
    struct REntryRange {
@@ -640,6 +650,9 @@ private:
    /// Must not be called when the descriptor guard is taken.
    void UpdateLastUsedCluster(ROOT::DescriptorId_t clusterId);
 
+   // Common treatment of zero pages that would otherwise need to be handled in LoadPageImpl()
+   ROOT::Internal::RPageRef LoadZeroPage(ColumnHandle_t columnHandle, const RPageSummary &pageSummary);
+
 protected:
    /// Default I/O performance counters that get registered in `fMetrics`
    struct RCounters {
@@ -693,15 +706,6 @@ protected:
       }
    };
 
-   /// Summarizes meta-data is necessary to load a certain page. Used by LoadPageImpl().
-   struct RPageSummary {
-      ROOT::DescriptorId_t fClusterId = 0;
-      /// The first element number of the page's column in the given cluster
-      std::uint64_t fColumnOffset = 0;
-      /// Location of the page on disk
-      ROOT::RClusterDescriptor::RPageInfoExtended fPageInfo;
-   };
-
    std::unique_ptr<RCounters> fCounters;
 
    ROOT::RNTupleReadOptions fOptions;
@@ -725,7 +729,7 @@ protected:
    virtual std::unique_ptr<RPageSource> CloneImpl() const = 0;
    // Only called if a task scheduler is set. No-op be default.
    virtual void UnzipClusterImpl(ROOT::Internal::RCluster *cluster);
-   // Returns a page from storage if not found in the page pool. Should be able to handle zero page locators.
+   // Returns a page from storage if not found in the page pool. Will never receive requests for zero pages.
    virtual ROOT::Internal::RPageRef LoadPageImpl(ColumnHandle_t columnHandle, const RPageSummary &pageSummary) = 0;
 
    /// Prepare a page range read for the column set in `clusterKey`.  Specifically, pages referencing the

tree/ntuple/src/RPageStorage.cxx

@@ -392,6 +392,24 @@ void ROOT::Internal::RPageSource::UpdateLastUsedCluster(ROOT::DescriptorId_t clu
    fLastUsedCluster = clusterId;
 }
 
+ROOT::Internal::RPageRef
+ROOT::Internal::RPageSource::LoadZeroPage(ColumnHandle_t columnHandle, const RPageSummary &pageSummary)
+{
+   const auto &pageInfo = pageSummary.fPageInfo;
+   assert(pageInfo.GetLocator().GetType() == RNTupleLocator::kTypePageZero);
+
+   const auto element = columnHandle.fColumn->GetElement();
+   const auto elementSize = element->GetSize();
+   const auto elementInMemoryType = element->GetIdentifier().fInMemoryType;
+
+   auto pageZero = fPageAllocator->NewPage(elementSize, pageInfo.GetNElements());
+   pageZero.GrowUnchecked(pageInfo.GetNElements());
+   std::memset(pageZero.GetBuffer(), 0, pageZero.GetNBytes());
+   pageZero.SetWindow(pageSummary.fColumnOffset + pageInfo.GetFirstElementIndex(),
+                      RPage::RClusterInfo(pageSummary.fClusterId, pageSummary.fColumnOffset));
+   return fPagePool.RegisterPage(std::move(pageZero), RPagePool::RKey{columnHandle.fPhysicalId, elementInMemoryType});
+}
+
 ROOT::Internal::RPageRef
 ROOT::Internal::RPageSource::LoadPage(ColumnHandle_t columnHandle, ROOT::NTupleSize_t globalIndex)
 {
@@ -422,8 +440,11 @@ ROOT::Internal::RPageSource::LoadPage(ColumnHandle_t columnHandle, ROOT::NTupleS
       pageSummary.fPageInfo = clusterDescriptor.GetPageRange(columnId).Find(globalIndex - pageSummary.fColumnOffset);
    }
 
-   if (pageSummary.fPageInfo.GetLocator().GetType() == RNTupleLocator::kTypeUnknown)
+   if (pageSummary.fPageInfo.GetLocator().GetType() == RNTupleLocator::kTypeUnknown) {
       throw RException(R__FAIL("tried to read a page with an unknown locator"));
+   } else if (pageSummary.fPageInfo.GetLocator().GetType() == RNTupleLocator::kTypePageZero) {
+      return LoadZeroPage(columnHandle, pageSummary);
+   }
 
    UpdateLastUsedCluster(pageSummary.fClusterId);
    return LoadPageImpl(columnHandle, pageSummary);
@@ -458,8 +479,11 @@ ROOT::Internal::RPageSource::LoadPage(ColumnHandle_t columnHandle, RNTupleLocalI
       pageSummary.fPageInfo = clusterDescriptor.GetPageRange(columnId).Find(localIndex.GetIndexInCluster());
    }
 
-   if (pageSummary.fPageInfo.GetLocator().GetType() == RNTupleLocator::kTypeUnknown)
+   if (pageSummary.fPageInfo.GetLocator().GetType() == RNTupleLocator::kTypeUnknown) {
       throw RException(R__FAIL("tried to read a page with an unknown locator"));
+   } else if (pageSummary.fPageInfo.GetLocator().GetType() == RNTupleLocator::kTypePageZero) {
+      return LoadZeroPage(columnHandle, pageSummary);
+   }
 
    UpdateLastUsedCluster(clusterId);
    return LoadPageImpl(columnHandle, pageSummary);

tree/ntuple/src/RPageStorageDaos.cxx

@@ -561,16 +561,6 @@ ROOT::Internal::RPageRef ROOT::Experimental::Internal::RPageSourceDaos::LoadPage
    const auto elementSize = element->GetSize();
    const auto elementInMemoryType = element->GetIdentifier().fInMemoryType;
 
-   if (pageInfo.GetLocator().GetType() == RNTupleLocator::kTypePageZero) {
-      auto pageZero = fPageAllocator->NewPage(elementSize, pageInfo.GetNElements());
-      pageZero.GrowUnchecked(pageInfo.GetNElements());
-      memset(pageZero.GetBuffer(), 0, pageZero.GetNBytes());
-      pageZero.SetWindow(pageSummary.fColumnOffset + pageInfo.GetFirstElementIndex(),
-                         ROOT::Internal::RPage::RClusterInfo(clusterId, pageSummary.fColumnOffset));
-      return fPagePool.RegisterPage(std::move(pageZero),
-                                    ROOT::Internal::RPagePool::RKey{columnId, elementInMemoryType});
-   }
-
    RSealedPage sealedPage;
    sealedPage.SetNElements(pageInfo.GetNElements());
    sealedPage.SetHasChecksum(pageInfo.HasChecksum());

tree/ntuple/src/RPageStorageFile.cxx

@@ -558,15 +558,6 @@ ROOT::Internal::RPageSourceFile::LoadPageImpl(ColumnHandle_t columnHandle, const
    const auto elementSize = element->GetSize();
    const auto elementInMemoryType = element->GetIdentifier().fInMemoryType;
 
-   if (pageInfo.GetLocator().GetType() == RNTupleLocator::kTypePageZero) {
-      auto pageZero = fPageAllocator->NewPage(elementSize, pageInfo.GetNElements());
-      pageZero.GrowUnchecked(pageInfo.GetNElements());
-      memset(pageZero.GetBuffer(), 0, pageZero.GetNBytes());
-      pageZero.SetWindow(pageSummary.fColumnOffset + pageInfo.GetFirstElementIndex(),
-                         ROOT::Internal::RPage::RClusterInfo(clusterId, pageSummary.fColumnOffset));
-      return fPagePool.RegisterPage(std::move(pageZero), RPagePool::RKey{columnId, elementInMemoryType});
-   }
-
    RSealedPage sealedPage;
    sealedPage.SetNElements(pageInfo.GetNElements());
    sealedPage.SetHasChecksum(pageInfo.HasChecksum());