diff --git a/include/gpu/gl/GrGLFunctions.h b/include/gpu/gl/GrGLFunctions.h
index b0673e13b19a..f50d8ed8ee5a 100644
--- a/include/gpu/gl/GrGLFunctions.h
+++ b/include/gpu/gl/GrGLFunctions.h
@@ -95,6 +95,7 @@ using GrGLGenSamplersFn = GrGLvoid GR_GL_FUNCTION_TYPE(GrGLsizei count, GrGLuint
 using GrGLGenTexturesFn = GrGLvoid GR_GL_FUNCTION_TYPE(GrGLsizei n, GrGLuint* textures);
 using GrGLGenVertexArraysFn = GrGLvoid GR_GL_FUNCTION_TYPE(GrGLsizei n, GrGLuint* arrays);
 using GrGLGetBufferParameterivFn = GrGLvoid GR_GL_FUNCTION_TYPE(GrGLenum target, GrGLenum pname, GrGLint* params);
+using GrGLGetBufferSubDataFn = GrGLvoid GR_GL_FUNCTION_TYPE(GrGLenum target, GrGLintptr offset, GrGLsizeiptr size, GrGLvoid* data);
 using GrGLGetErrorFn = GrGLenum GR_GL_FUNCTION_TYPE();
 using GrGLGetFramebufferAttachmentParameterivFn = GrGLvoid GR_GL_FUNCTION_TYPE(GrGLenum target, GrGLenum attachment, GrGLenum pname, GrGLint* params);
 using GrGLGetFloatvFn = GrGLvoid GR_GL_FUNCTION_TYPE(GrGLenum pname, GrGLfloat* params);
diff --git a/include/gpu/gl/GrGLInterface.h b/include/gpu/gl/GrGLInterface.h
index 862178ec55c0..9e5ab9e219ac 100644
--- a/include/gpu/gl/GrGLInterface.h
+++ b/include/gpu/gl/GrGLInterface.h
@@ -165,6 +165,7 @@ struct SK_API GrGLInterface : public SkRefCnt {
         GrGLFunction<GrGLGenTexturesFn> fGenTextures;
         GrGLFunction<GrGLGenVertexArraysFn> fGenVertexArrays;
         GrGLFunction<GrGLGetBufferParameterivFn> fGetBufferParameteriv;
+        GrGLFunction<GrGLGetBufferSubDataFn> fGetBufferSubData;
         GrGLFunction<GrGLGetErrorFn> fGetError;
         GrGLFunction<GrGLGetFramebufferAttachmentParameterivFn> fGetFramebufferAttachmentParameteriv;
         GrGLFunction<GrGLGetFloatvFn> fGetFloatv;
diff --git a/src/gpu/AsyncReadTypes.h b/src/gpu/AsyncReadTypes.h
index 52d7961b6d4f..14eccd505072 100644
--- a/src/gpu/AsyncReadTypes.h
+++ b/src/gpu/AsyncReadTypes.h
@@ -152,19 +152,36 @@ class TAsyncReadResult : public SkImage::AsyncReadResult {
                            SkISize dimensions,
                            size_t rowBytes,
                            TClientMappedBufferManager<T, IDType>* manager) {
+        size_t size = rowBytes*dimensions.height();
+        auto doConvert = [converter{result.fPixelConverter}, size](const void* src) -> sk_sp<SkData> {
+            if (converter) {
+                sk_sp<SkData> data = SkData::MakeUninitialized(size);
+                converter(data->writable_data(), src);
+                return data;
+            }
+            return nullptr;
+        };
+
         const void* mappedData = result.fTransferBuffer->map();
-        if (!mappedData) {
-            return false;
-        }
-        if (result.fPixelConverter) {
-            size_t size = rowBytes*dimensions.height();
-            sk_sp<SkData> data = SkData::MakeUninitialized(size);
-            result.fPixelConverter(data->writable_data(), mappedData);
-            this->addCpuPlane(std::move(data), rowBytes);
-            result.fTransferBuffer->unmap();
+
+        if (mappedData) {
+            if (auto converted = doConvert(mappedData)) {
+                this->addCpuPlane(std::move(converted), rowBytes);
+                result.fTransferBuffer->unmap();
+            } else {
+                manager->insert(result.fTransferBuffer);
+                this->addMappedPlane(mappedData, rowBytes, std::move(result.fTransferBuffer));
+            }
         } else {
-            manager->insert(result.fTransferBuffer);
-            this->addMappedPlane(mappedData, rowBytes, std::move(result.fTransferBuffer));
+            sk_sp<SkData> tmp = SkData::MakeUninitialized(size);
+            if (!result.fTransferBuffer->getData(tmp->writable_data(), 0, size)) {
+                return false;
+            }
+            if (auto converted = doConvert(tmp->data())) {
+                this->addCpuPlane(std::move(converted), rowBytes);
+            } else {
+                this->addCpuPlane(std::move(tmp), rowBytes);
+            }
         }
         return true;
     }
diff --git a/src/gpu/ganesh/GrGpuBuffer.cpp b/src/gpu/ganesh/GrGpuBuffer.cpp
index 68c0d583bcac..e178192d90ef 100644
--- a/src/gpu/ganesh/GrGpuBuffer.cpp
+++ b/src/gpu/ganesh/GrGpuBuffer.cpp
@@ -83,6 +83,25 @@ bool GrGpuBuffer::updateData(const void* src, size_t offset, size_t size, bool p
     return this->onUpdateData(src, offset, size, preserve);
 }
 
+bool GrGpuBuffer::getData(void* dst, size_t offset, size_t size) {
+    SkASSERT(!this->isMapped());
+    SkASSERT(size > 0 && offset + size <= fSizeInBytes);
+    SkASSERT(dst);
+
+    if (this->wasDestroyed()) {
+        return false;
+    }
+
+    const void* mapped = this->map();
+    if (mapped != nullptr) {
+        memcpy(dst, (const char*)mapped + offset, size);
+        this->unmap();
+        return true;
+    }
+
+    return this->onGetData(dst, offset, size);
+}
+
 void GrGpuBuffer::ComputeScratchKeyForDynamicBuffer(size_t size,
                                                     GrGpuBufferType intendedType,
                                                     skgpu::ScratchKey* key) {
diff --git a/src/gpu/ganesh/GrGpuBuffer.h b/src/gpu/ganesh/GrGpuBuffer.h
index 200d78b5ccbb..867230f07339 100644
--- a/src/gpu/ganesh/GrGpuBuffer.h
+++ b/src/gpu/ganesh/GrGpuBuffer.h
@@ -96,6 +96,22 @@ class GrGpuBuffer : public GrGpuResource, public GrBuffer {
      */
     bool updateData(const void* src, size_t offset, size_t size, bool preserve);
 
+    /**
+     * Get the buffer data.
+     *
+     * WebGL's buffer can't be mapped to client side. Use this function to get the buffer data.
+     *
+     * The data is always copied to client side. Will try copy from mapped if supported.
+     *
+     * The buffer must not be mapped.
+     *
+     * Note that buffer updates do not go through GrContext and therefore are not serialized with
+     * other operations.
+     *
+     * @return returns true if succeeds, false otherwise.
+     */
+    bool getData(void* dst, size_t offset, size_t size);
+
     GrGpuBufferType intendedType() const { return fIntendedType; }
 
 protected:
@@ -129,6 +145,7 @@ class GrGpuBuffer : public GrGpuResource, public GrBuffer {
     virtual void onUnmap(MapType) = 0;
     virtual bool onClearToZero() = 0;
     virtual bool onUpdateData(const void* src, size_t offset, size_t size, bool preserve) = 0;
+    virtual bool onGetData(void* dst, size_t offset, size_t size) { return false; }
 
     size_t onGpuMemorySize() const override { return fSizeInBytes; }
     void onSetLabel() override{}
diff --git a/src/gpu/ganesh/gl/GrGLAssembleGLInterfaceAutogen.cpp b/src/gpu/ganesh/gl/GrGLAssembleGLInterfaceAutogen.cpp
index bbc15630d119..41548e704852 100644
--- a/src/gpu/ganesh/gl/GrGLAssembleGLInterfaceAutogen.cpp
+++ b/src/gpu/ganesh/gl/GrGLAssembleGLInterfaceAutogen.cpp
@@ -151,6 +151,8 @@ sk_sp<const GrGLInterface> GrGLMakeAssembledGLInterface(void *ctx, GrGLGetProc g
     GET_PROC(DrawBuffer);
     GET_PROC(PolygonMode);
 
+    GET_PROC(GetBufferSubData);
+
     if (glVer >= GR_GL_VER(3,0)) {
         GET_PROC(GetStringi);
     }
diff --git a/src/gpu/ganesh/gl/GrGLAssembleWebGLInterfaceAutogen.cpp b/src/gpu/ganesh/gl/GrGLAssembleWebGLInterfaceAutogen.cpp
index 62f8da5f7f54..ed394e8b8e39 100644
--- a/src/gpu/ganesh/gl/GrGLAssembleWebGLInterfaceAutogen.cpp
+++ b/src/gpu/ganesh/gl/GrGLAssembleWebGLInterfaceAutogen.cpp
@@ -137,6 +137,10 @@ sk_sp<const GrGLInterface> GrGLMakeAssembledWebGLInterface(void *ctx, GrGLGetPro
     GET_PROC(VertexAttribPointer);
     GET_PROC(Viewport);
 
+    if (glVer >= GR_GL_VER(2,0)) {
+        GET_PROC(GetBufferSubData);
+    }
+
     if (glVer >= GR_GL_VER(2,0)) {
         GET_PROC(GetStringi);
     }
diff --git a/src/gpu/ganesh/gl/GrGLBuffer.cpp b/src/gpu/ganesh/gl/GrGLBuffer.cpp
index 39080ea9b8dc..7c81bccc8c0c 100644
--- a/src/gpu/ganesh/gl/GrGLBuffer.cpp
+++ b/src/gpu/ganesh/gl/GrGLBuffer.cpp
@@ -275,6 +275,19 @@ bool GrGLBuffer::onUpdateData(const void* src, size_t offset, size_t size, bool
     return true;
 }
 
+bool GrGLBuffer::onGetData(void* dst, size_t offset, size_t size) {
+    SkASSERT(fBufferID);
+
+    if (!this->glCaps().getBufferSubDataSupport()) {
+      return false;
+    }
+
+    // bindbuffer handles dirty context
+    GrGLenum target = this->glGpu()->bindBuffer(fIntendedType, this);
+    GL_CALL(GetBufferSubData(target, offset, size, dst));
+    return true;
+}
+
 void GrGLBuffer::onSetLabel() {
     SkASSERT(fBufferID);
     if (!this->getLabel().empty()) {
diff --git a/src/gpu/ganesh/gl/GrGLBuffer.h b/src/gpu/ganesh/gl/GrGLBuffer.h
index d6ec2cb5df31..bcb61c2a6301 100644
--- a/src/gpu/ganesh/gl/GrGLBuffer.h
+++ b/src/gpu/ganesh/gl/GrGLBuffer.h
@@ -51,6 +51,7 @@ class GrGLBuffer : public GrGpuBuffer {
     void onUnmap(MapType) override;
     bool onClearToZero() override;
     bool onUpdateData(const void* src, size_t offset, size_t size, bool preserve) override;
+    bool onGetData(void* dst, size_t offset, size_t size) override;
 
     void onSetLabel() override;
 
diff --git a/src/gpu/ganesh/gl/GrGLCaps.cpp b/src/gpu/ganesh/gl/GrGLCaps.cpp
index 7bdbe4f72c47..8362ea23df11 100644
--- a/src/gpu/ganesh/gl/GrGLCaps.cpp
+++ b/src/gpu/ganesh/gl/GrGLCaps.cpp
@@ -100,6 +100,7 @@ GrGLCaps::GrGLCaps(const GrContextOptions& contextOptions,
     fFBFetchRequiresEnablePerSample = false;
     fSRGBWriteControl = false;
     fSkipErrorChecks = false;
+    fGetBufferSubDataSupport = false;
 
     fShaderCaps = std::make_unique<GrShaderCaps>();
 
@@ -400,6 +401,12 @@ void GrGLCaps::init(const GrContextOptions& contextOptions,
         return SkToBool(contextFlags & GR_GL_CONTEXT_FLAG_PROTECTED_CONTENT_BIT_EXT);
     }();
 
+    if (GR_IS_GR_GL(standard)) {
+        fGetBufferSubDataSupport = true;
+    } else if (GR_IS_GR_WEBGL(standard)) {
+        fGetBufferSubDataSupport = version >= GR_GL_VER(2, 0);
+    }
+
     /**************************************************************************
     * GrShaderCaps fields
     **************************************************************************/
@@ -559,11 +566,6 @@ void GrGLCaps::init(const GrContextOptions& contextOptions,
         fMapBufferFlags = kNone_MapFlags;
     }
 
-    // Buffers have more restrictions in WebGL than GLES. For example,
-    // https://www.khronos.org/registry/webgl/specs/latest/2.0/#BUFFER_OBJECT_BINDING
-    // We therefore haven't attempted to support mapping or transfers between buffers and surfaces
-    // or between buffers.
-
     if (GR_IS_GR_GL(standard)) {
         if (version >= GR_GL_VER(2, 1) || ctxInfo.hasExtension("GL_ARB_pixel_buffer_object") ||
             ctxInfo.hasExtension("GL_EXT_pixel_buffer_object")) {
@@ -589,6 +591,12 @@ void GrGLCaps::init(const GrContextOptions& contextOptions,
 //            fTransferFromSurfaceToBufferSupport = false;
 //            fTransferBufferType = TransferBufferType::kChromium;
         }
+    } else if (GR_IS_GR_WEBGL(standard)) {
+        if (version >= GR_GL_VER(2, 0)) {
+            fTransferFromBufferToTextureSupport = true;
+            fTransferFromSurfaceToBufferSupport = true;
+            fTransferBufferType = TransferBufferType::kARB_PBO;
+        }
     }
 
     if (GR_IS_GR_GL(standard) &&
@@ -597,6 +605,13 @@ void GrGLCaps::init(const GrContextOptions& contextOptions,
     } else if (GR_IS_GR_GL_ES(standard) &&
                (version >= GR_GL_VER(3, 0) || ctxInfo.hasExtension("GL_NV_copy_buffer"))) {
         fTransferFromBufferToBufferSupport = true;
+    } else if (GR_IS_GR_WEBGL(standard)) {
+        if (version >= GR_GL_VER(2, 0)) {
+            // WebGL has more restriction about buffer binding
+            // https://registry.khronos.org/webgl/specs/latest/2.0/#COPYING_BUFFERS
+            // TODO: make sure index buffer is handled properly to enable this
+            // fTransferFromBufferToBufferSupport = true;
+        }
     }
 
     // On many GPUs, map memory is very expensive, so we effectively disable it here by setting the
diff --git a/src/gpu/ganesh/gl/GrGLCaps.h b/src/gpu/ganesh/gl/GrGLCaps.h
index 16227714260a..23dfb5e14394 100644
--- a/src/gpu/ganesh/gl/GrGLCaps.h
+++ b/src/gpu/ganesh/gl/GrGLCaps.h
@@ -515,6 +515,8 @@ class GrGLCaps : public GrCaps {
 
     bool clientCanDisableMultisample() const { return fClientCanDisableMultisample; }
 
+    bool getBufferSubDataSupport() const { return fGetBufferSubDataSupport; }
+
     GrBackendFormat getBackendFormatFromCompressionType(SkTextureCompressionType) const override;
 
     skgpu::Swizzle getWriteSwizzle(const GrBackendFormat&, GrColorType) const override;
@@ -630,6 +632,7 @@ class GrGLCaps : public GrCaps {
     bool fSRGBWriteControl : 1;
     bool fSkipErrorChecks : 1;
     bool fClientCanDisableMultisample : 1;
+    bool fGetBufferSubDataSupport: 1;
 
     // Driver workarounds
     bool fDoManualMipmapping : 1;
diff --git a/src/gpu/ganesh/gl/GrGLInterfaceAutogen.cpp b/src/gpu/ganesh/gl/GrGLInterfaceAutogen.cpp
index 5d8bd80f6024..4580c1a85855 100644
--- a/src/gpu/ganesh/gl/GrGLInterfaceAutogen.cpp
+++ b/src/gpu/ganesh/gl/GrGLInterfaceAutogen.cpp
@@ -188,6 +188,14 @@ bool GrGLInterface::validate() const {
         }
     }
 
+    if (GR_IS_GR_GL(fStandard) ||
+       (GR_IS_GR_WEBGL(fStandard) && (
+          (glVer >= GR_GL_VER(2,0))))) {
+        if (!fFunctions.fGetBufferSubData) {
+            RETURN_FALSE_INTERFACE;
+        }
+    }
+
     if ((GR_IS_GR_GL(fStandard) && (
           (glVer >= GR_GL_VER(3,0)))) ||
        (GR_IS_GR_GL_ES(fStandard) && (
diff --git a/tests/TransferPixelsTest.cpp b/tests/TransferPixelsTest.cpp
index fead34d340ec..b14763310b1c 100644
--- a/tests/TransferPixelsTest.cpp
+++ b/tests/TransferPixelsTest.cpp
@@ -198,13 +198,25 @@ void basic_transfer_to_test(skiatest::Reporter* reporter,
     if (!buffer) {
         return;
     }
-    void* data = buffer->map();
-    if (!buffer) {
-        ERRORF(reporter, "Could not map buffer");
-        return;
-    }
-    memcpy(data, srcData.get(), size);
-    buffer->unmap();
+    auto updateData = [buffer, caps, gpu, reporter] (const void* src, size_t size, bool preserve) {
+        if (GrCaps::kNone_MapFlags != caps->mapBufferFlags()) {
+            void *map = buffer->map();
+            REPORTER_ASSERT(reporter, map);
+            if (!map) {
+                ERRORF(reporter, "Failed to map transfer buffer.");
+                return;
+            }
+            memcpy(map, src, size);
+            buffer->unmap();
+        } else {
+            if (!buffer->updateData(src, 0, size, preserve)) {
+                ERRORF(reporter, "Could not updateData");
+            }
+            gpu->submitToGpu(GrSyncCpu::kYes);
+        }
+    };
+
+    updateData(srcData.get(), size, false);
 
     //////////////////////////
     // transfer full data
@@ -278,9 +290,7 @@ void basic_transfer_to_test(skiatest::Reporter* reporter,
     // change color of subrectangle
     fill_transfer_data(left, top, width, height, srcRowBytes, allowedSrc.fColorType,
                        srcData.get());
-    data = buffer->map();
-    memcpy(data, srcData.get(), size);
-    buffer->unmap();
+    updateData(srcData.get(), size, true);
 
     result = gpu->transferPixelsTo(tex.get(),
                                    SkIRect::MakeXYWH(left, top, width, height),
@@ -393,7 +403,7 @@ void basic_transfer_from_test(skiatest::Reporter* reporter, const sk_gpu_test::C
 
     sk_sp<GrGpuBuffer> buffer = resourceProvider->createBuffer(bufferSize,
                                                                GrGpuBufferType::kXferGpuToCpu,
-                                                               kDynamic_GrAccessPattern,
+                                                               kStream_GrAccessPattern,
                                                                GrResourceProvider::ZeroInit::kNo);
     REPORTER_ASSERT(reporter, buffer);
     if (!buffer) {
@@ -422,15 +432,12 @@ void basic_transfer_from_test(skiatest::Reporter* reporter, const sk_gpu_test::C
     }
 
     // Copy the transfer buffer contents to a temporary so we can manipulate it.
-    const auto* map = reinterpret_cast<const char*>(buffer->map());
-    REPORTER_ASSERT(reporter, map);
-    if (!map) {
-        ERRORF(reporter, "Failed to map transfer buffer.");
-        return;
+    size_t copy_size = kTexDims.fHeight * fullBufferRowBytes;
+    std::unique_ptr<char[]> transferData(new char[copy_size]);
+
+    if (!buffer->getData(transferData.get(), 0, copy_size)) {
+      ERRORF(reporter, "Could not getData");
     }
-    std::unique_ptr<char[]> transferData(new char[kTexDims.fHeight * fullBufferRowBytes]);
-    memcpy(transferData.get(), map, fullBufferRowBytes * kTexDims.fHeight);
-    buffer->unmap();
 
     GrImageInfo transferInfo(allowedRead.fColorType, kUnpremul_SkAlphaType, nullptr, kTexDims);
 
@@ -468,15 +475,8 @@ void basic_transfer_from_test(skiatest::Reporter* reporter, const sk_gpu_test::C
         gpu->submitToGpu(GrSyncCpu::kYes);
     }
 
-    map = reinterpret_cast<const char*>(buffer->map());
-    REPORTER_ASSERT(reporter, map);
-    if (!map) {
-        ERRORF(reporter, "Failed to map transfer buffer.");
-        return;
-    }
-    const char* bufferStart = reinterpret_cast<const char*>(map) + partialReadOffset;
-    memcpy(transferData.get(), bufferStart, partialBufferRowBytes * kTexDims.fHeight);
-    buffer->unmap();
+    if (!buffer->getData(transferData.get(), partialReadOffset, partialBufferRowBytes *
+                         kTexDims.fHeight)) { ERRORF(reporter, "Could not getData"); }
 
     transferInfo = transferInfo.makeWH(kPartialWidth, kPartialHeight);
     const char* textureDataStart =
diff --git a/tools/gpu/gl/interface/interface.json5 b/tools/gpu/gl/interface/interface.json5
index 44e5163ceb73..04702dd6a009 100644
--- a/tools/gpu/gl/interface/interface.json5
+++ b/tools/gpu/gl/interface/interface.json5
@@ -55,6 +55,15 @@
       "DrawBuffer", "PolygonMode",
     ],
   },
+  {
+    "GL":    [{"ext": "<core>"}],
+    "GLES":  null,
+    "WebGL": [{"min_version": [2, 0], "ext": "<core>"}],
+
+    "functions": [
+      "GetBufferSubData",
+    ],
+  },
   {
     "GL":    [{"min_version": [3, 0], "ext": "<core>"}],
     "GLES":  [{"min_version": [3, 0], "ext": "<core>"}],