Migrate reference external HAL implementation to AIDL

Android T migrated the HAL interface to AIDL, but the reference HAL
was never updated to use AIDL. This CL migrates the reference
HAL implementation for external cameras to use AIDL as well. The
external HAL uses V4L2 standard to expose USB cameras to the
cameraserver.

The reference HAL implementation for internal camera was dropped
because it is not possible to write a generic HAL that works with some
large percentage of internal cameras.

Bug: 219974678
Test: Existing CTS tests pass with external camera connected.
Change-Id: I35f3dc32c16670eca7735a4ac00fed3daf36aa65
diff --git a/camera/device/default/ExternalCameraUtils.cpp b/camera/device/default/ExternalCameraUtils.cpp
new file mode 100644
index 0000000..cfb95f2
--- /dev/null
+++ b/camera/device/default/ExternalCameraUtils.cpp
@@ -0,0 +1,860 @@
+/*
+ * Copyright (C) 2022 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#define LOG_TAG "ExtCamUtils"
+// #define LOG_NDEBUG 0
+
+#include "ExternalCameraUtils.h"
+
+#include <aidlcommonsupport/NativeHandle.h>
+#include <jpeglib.h>
+#include <linux/videodev2.h>
+#include <log/log.h>
+#include <algorithm>
+#include <cinttypes>
+#include <cmath>
+
+#define HAVE_JPEG  // required for libyuv.h to export MJPEG decode APIs
+#include <libyuv.h>
+
+namespace android {
+namespace hardware {
+namespace camera {
+
+namespace external {
+namespace common {
+
+namespace {
+const int kDefaultCameraIdOffset = 100;
+const int kDefaultJpegBufSize = 5 << 20;  // 5MB
+const int kDefaultNumVideoBuffer = 4;
+const int kDefaultNumStillBuffer = 2;
+const int kDefaultOrientation = 0;  // suitable for natural landscape displays like tablet/TV
+                                    // For phone devices 270 is better
+}  // anonymous namespace
+
+const char* ExternalCameraConfig::kDefaultCfgPath = "/vendor/etc/external_camera_config.xml";
+
+ExternalCameraConfig ExternalCameraConfig::loadFromCfg(const char* cfgPath) {
+    using namespace tinyxml2;
+    ExternalCameraConfig ret;
+
+    XMLDocument configXml;
+    XMLError err = configXml.LoadFile(cfgPath);
+    if (err != XML_SUCCESS) {
+        ALOGE("%s: Unable to load external camera config file '%s'. Error: %s", __FUNCTION__,
+              cfgPath, XMLDocument::ErrorIDToName(err));
+        return ret;
+    } else {
+        ALOGI("%s: load external camera config succeeded!", __FUNCTION__);
+    }
+
+    XMLElement* extCam = configXml.FirstChildElement("ExternalCamera");
+    if (extCam == nullptr) {
+        ALOGI("%s: no external camera config specified", __FUNCTION__);
+        return ret;
+    }
+
+    XMLElement* providerCfg = extCam->FirstChildElement("Provider");
+    if (providerCfg == nullptr) {
+        ALOGI("%s: no external camera provider config specified", __FUNCTION__);
+        return ret;
+    }
+
+    XMLElement* cameraIdOffset = providerCfg->FirstChildElement("CameraIdOffset");
+    if (cameraIdOffset != nullptr) {
+        ret.cameraIdOffset = std::atoi(cameraIdOffset->GetText());
+    }
+
+    XMLElement* ignore = providerCfg->FirstChildElement("ignore");
+    if (ignore == nullptr) {
+        ALOGI("%s: no internal ignored device specified", __FUNCTION__);
+        return ret;
+    }
+
+    XMLElement* id = ignore->FirstChildElement("id");
+    while (id != nullptr) {
+        const char* text = id->GetText();
+        if (text != nullptr) {
+            ret.mInternalDevices.insert(text);
+            ALOGI("%s: device %s will be ignored by external camera provider", __FUNCTION__, text);
+        }
+        id = id->NextSiblingElement("id");
+    }
+
+    XMLElement* deviceCfg = extCam->FirstChildElement("Device");
+    if (deviceCfg == nullptr) {
+        ALOGI("%s: no external camera device config specified", __FUNCTION__);
+        return ret;
+    }
+
+    XMLElement* jpegBufSz = deviceCfg->FirstChildElement("MaxJpegBufferSize");
+    if (jpegBufSz == nullptr) {
+        ALOGI("%s: no max jpeg buffer size specified", __FUNCTION__);
+    } else {
+        ret.maxJpegBufSize = jpegBufSz->UnsignedAttribute("bytes", /*Default*/ kDefaultJpegBufSize);
+    }
+
+    XMLElement* numVideoBuf = deviceCfg->FirstChildElement("NumVideoBuffers");
+    if (numVideoBuf == nullptr) {
+        ALOGI("%s: no num video buffers specified", __FUNCTION__);
+    } else {
+        ret.numVideoBuffers =
+                numVideoBuf->UnsignedAttribute("count", /*Default*/ kDefaultNumVideoBuffer);
+    }
+
+    XMLElement* numStillBuf = deviceCfg->FirstChildElement("NumStillBuffers");
+    if (numStillBuf == nullptr) {
+        ALOGI("%s: no num still buffers specified", __FUNCTION__);
+    } else {
+        ret.numStillBuffers =
+                numStillBuf->UnsignedAttribute("count", /*Default*/ kDefaultNumStillBuffer);
+    }
+
+    XMLElement* fpsList = deviceCfg->FirstChildElement("FpsList");
+    if (fpsList == nullptr) {
+        ALOGI("%s: no fps list specified", __FUNCTION__);
+    } else {
+        if (!updateFpsList(fpsList, ret.fpsLimits)) {
+            return ret;
+        }
+    }
+
+    XMLElement* depth = deviceCfg->FirstChildElement("Depth16Supported");
+    if (depth == nullptr) {
+        ret.depthEnabled = false;
+        ALOGI("%s: depth output is not enabled", __FUNCTION__);
+    } else {
+        ret.depthEnabled = depth->BoolAttribute("enabled", false);
+    }
+
+    if (ret.depthEnabled) {
+        XMLElement* depthFpsList = deviceCfg->FirstChildElement("DepthFpsList");
+        if (depthFpsList == nullptr) {
+            ALOGW("%s: no depth fps list specified", __FUNCTION__);
+        } else {
+            if (!updateFpsList(depthFpsList, ret.depthFpsLimits)) {
+                return ret;
+            }
+        }
+    }
+
+    XMLElement* minStreamSize = deviceCfg->FirstChildElement("MinimumStreamSize");
+    if (minStreamSize == nullptr) {
+        ALOGI("%s: no minimum stream size specified", __FUNCTION__);
+    } else {
+        ret.minStreamSize = {
+                static_cast<int32_t>(minStreamSize->UnsignedAttribute("width", /*Default*/ 0)),
+                static_cast<int32_t>(minStreamSize->UnsignedAttribute("height", /*Default*/ 0))};
+    }
+
+    XMLElement* orientation = deviceCfg->FirstChildElement("Orientation");
+    if (orientation == nullptr) {
+        ALOGI("%s: no sensor orientation specified", __FUNCTION__);
+    } else {
+        ret.orientation = orientation->IntAttribute("degree", /*Default*/ kDefaultOrientation);
+    }
+
+    ALOGI("%s: external camera cfg loaded: maxJpgBufSize %d,"
+          " num video buffers %d, num still buffers %d, orientation %d",
+          __FUNCTION__, ret.maxJpegBufSize, ret.numVideoBuffers, ret.numStillBuffers,
+          ret.orientation);
+    for (const auto& limit : ret.fpsLimits) {
+        ALOGI("%s: fpsLimitList: %dx%d@%f", __FUNCTION__, limit.size.width, limit.size.height,
+              limit.fpsUpperBound);
+    }
+    for (const auto& limit : ret.depthFpsLimits) {
+        ALOGI("%s: depthFpsLimitList: %dx%d@%f", __FUNCTION__, limit.size.width, limit.size.height,
+              limit.fpsUpperBound);
+    }
+    ALOGI("%s: minStreamSize: %dx%d", __FUNCTION__, ret.minStreamSize.width,
+          ret.minStreamSize.height);
+    return ret;
+}
+
+bool ExternalCameraConfig::updateFpsList(tinyxml2::XMLElement* fpsList,
+                                         std::vector<FpsLimitation>& fpsLimits) {
+    using namespace tinyxml2;
+    std::vector<FpsLimitation> limits;
+    XMLElement* row = fpsList->FirstChildElement("Limit");
+    while (row != nullptr) {
+        FpsLimitation prevLimit{{0, 0}, 1000.0};
+        FpsLimitation limit = {
+                {/* width */ static_cast<int32_t>(row->UnsignedAttribute("width", /*Default*/ 0)),
+                 /* height */ static_cast<int32_t>(
+                         row->UnsignedAttribute("height", /*Default*/ 0))},
+                /* fpsUpperBound */ row->DoubleAttribute("fpsBound", /*Default*/ 1000.0)};
+        if (limit.size.width <= prevLimit.size.width ||
+            limit.size.height <= prevLimit.size.height ||
+            limit.fpsUpperBound >= prevLimit.fpsUpperBound) {
+            ALOGE("%s: FPS limit list must have increasing size and decreasing fps!"
+                  " Prev %dx%d@%f, Current %dx%d@%f",
+                  __FUNCTION__, prevLimit.size.width, prevLimit.size.height,
+                  prevLimit.fpsUpperBound, limit.size.width, limit.size.height,
+                  limit.fpsUpperBound);
+            return false;
+        }
+        limits.push_back(limit);
+        row = row->NextSiblingElement("Limit");
+    }
+    fpsLimits = limits;
+    return true;
+}
+
+ExternalCameraConfig::ExternalCameraConfig()
+    : cameraIdOffset(kDefaultCameraIdOffset),
+      maxJpegBufSize(kDefaultJpegBufSize),
+      numVideoBuffers(kDefaultNumVideoBuffer),
+      numStillBuffers(kDefaultNumStillBuffer),
+      depthEnabled(false),
+      orientation(kDefaultOrientation) {
+    fpsLimits.push_back({/* size */ {/* width */ 640, /* height */ 480}, /* fpsUpperBound */ 30.0});
+    fpsLimits.push_back({/* size */ {/* width */ 1280, /* height */ 720}, /* fpsUpperBound */ 7.5});
+    fpsLimits.push_back(
+            {/* size */ {/* width */ 1920, /* height */ 1080}, /* fpsUpperBound */ 5.0});
+    minStreamSize = {0, 0};
+}
+
+}  // namespace common
+}  // namespace external
+
+namespace device {
+namespace implementation {
+
+double SupportedV4L2Format::FrameRate::getFramesPerSecond() const {
+    return static_cast<double>(durationDenominator) / durationNumerator;
+}
+
+Frame::Frame(uint32_t width, uint32_t height, uint32_t fourcc)
+    : mWidth(width), mHeight(height), mFourcc(fourcc) {}
+Frame::~Frame() {}
+
+V4L2Frame::V4L2Frame(uint32_t w, uint32_t h, uint32_t fourcc, int bufIdx, int fd, uint32_t dataSize,
+                     uint64_t offset)
+    : Frame(w, h, fourcc), mBufferIndex(bufIdx), mFd(fd), mDataSize(dataSize), mOffset(offset) {}
+
+V4L2Frame::~V4L2Frame() {
+    unmap();
+}
+
+int V4L2Frame::getData(uint8_t** outData, size_t* dataSize) {
+    return map(outData, dataSize);
+}
+
+int V4L2Frame::map(uint8_t** data, size_t* dataSize) {
+    if (data == nullptr || dataSize == nullptr) {
+        ALOGI("%s: V4L2 buffer map bad argument: data %p, dataSize %p", __FUNCTION__, data,
+              dataSize);
+        return -EINVAL;
+    }
+
+    std::lock_guard<std::mutex> lk(mLock);
+    if (!mMapped) {
+        void* addr = mmap(nullptr, mDataSize, PROT_READ, MAP_SHARED, mFd, mOffset);
+        if (addr == MAP_FAILED) {
+            ALOGE("%s: V4L2 buffer map failed: %s", __FUNCTION__, strerror(errno));
+            return -EINVAL;
+        }
+        mData = static_cast<uint8_t*>(addr);
+        mMapped = true;
+    }
+    *data = mData;
+    *dataSize = mDataSize;
+    ALOGV("%s: V4L map FD %d, data %p size %zu", __FUNCTION__, mFd, mData, mDataSize);
+    return 0;
+}
+
+int V4L2Frame::unmap() {
+    std::lock_guard<std::mutex> lk(mLock);
+    if (mMapped) {
+        ALOGV("%s: V4L unmap data %p size %zu", __FUNCTION__, mData, mDataSize);
+        if (munmap(mData, mDataSize) != 0) {
+            ALOGE("%s: V4L2 buffer unmap failed: %s", __FUNCTION__, strerror(errno));
+            return -EINVAL;
+        }
+        mMapped = false;
+    }
+    return 0;
+}
+
+AllocatedFrame::AllocatedFrame(uint32_t w, uint32_t h) : Frame(w, h, V4L2_PIX_FMT_YUV420) {}
+AllocatedFrame::~AllocatedFrame() {}
+
+int AllocatedFrame::getData(uint8_t** outData, size_t* dataSize) {
+    YCbCrLayout layout;
+    int ret = allocate(&layout);
+    if (ret != 0) {
+        return ret;
+    }
+    *outData = mData.data();
+    *dataSize = mBufferSize;
+    return 0;
+}
+
+int AllocatedFrame::allocate(YCbCrLayout* out) {
+    std::lock_guard<std::mutex> lk(mLock);
+    if ((mWidth % 2) || (mHeight % 2)) {
+        ALOGE("%s: bad dimension %dx%d (not multiple of 2)", __FUNCTION__, mWidth, mHeight);
+        return -EINVAL;
+    }
+
+    // This frame might be sent to jpeglib to be encoded. Since AllocatedFrame only contains YUV420,
+    // jpeglib expects height and width of Y component to be an integral multiple of 2*DCTSIZE,
+    // and heights and widths of Cb and Cr components to be an integral multiple of DCTSIZE. If the
+    // image size does not meet this requirement, libjpeg expects its input to be padded to meet the
+    // constraints. This padding is removed from the final encoded image so the content in the
+    // padding doesn't matter. What matters is that the memory is accessible to jpeglib at the time
+    // of encoding.
+    // For example, if the image size is 1500x844 and DCTSIZE is 8, jpeglib expects a YUV 420
+    // frame with components of following sizes:
+    //   Y:      1504x848 because 1504 and 848 are the next smallest multiples of 2*8
+    //   Cb/Cr:  752x424 which are the next smallest multiples of 8
+
+    // jpeglib takes an array of row pointers which makes vertical padding trivial when setting up
+    // the pointers. Padding horizontally is a bit more complicated. AllocatedFrame holds the data
+    // in a flattened buffer, which means memory accesses past a row will flow into the next logical
+    // row. For any row of a component, we can consider the first few bytes of the next row as
+    // padding for the current one. This is true for Y and Cb components and all but last row of the
+    // Cr component. Reading past the last row of Cr component will lead to undefined behavior as
+    // libjpeg attempts to read memory past the allocated buffer. To prevent undefined behavior,
+    // the buffer allocated here is padded such that libjpeg never accesses unallocated memory when
+    // reading the last row. Effectively, we only need to ensure that the last row of Cr component
+    // has width that is an integral multiple of DCTSIZE.
+
+    size_t dataSize = mWidth * mHeight * 3 / 2;  // YUV420
+
+    size_t cbWidth = mWidth / 2;
+    size_t requiredCbWidth = DCTSIZE * ((cbWidth + DCTSIZE - 1) / DCTSIZE);
+    size_t padding = requiredCbWidth - cbWidth;
+    size_t finalSize = dataSize + padding;
+
+    if (mData.size() != finalSize) {
+        mData.resize(finalSize);
+        mBufferSize = dataSize;
+    }
+
+    if (out != nullptr) {
+        out->y = mData.data();
+        out->yStride = mWidth;
+        uint8_t* cbStart = mData.data() + mWidth * mHeight;
+        uint8_t* crStart = cbStart + mWidth * mHeight / 4;
+        out->cb = cbStart;
+        out->cr = crStart;
+        out->cStride = mWidth / 2;
+        out->chromaStep = 1;
+    }
+    return 0;
+}
+
+int AllocatedFrame::getLayout(YCbCrLayout* out) {
+    IMapper::Rect noCrop = {0, 0, static_cast<int32_t>(mWidth), static_cast<int32_t>(mHeight)};
+    return getCroppedLayout(noCrop, out);
+}
+
+int AllocatedFrame::getCroppedLayout(const IMapper::Rect& rect, YCbCrLayout* out) {
+    if (out == nullptr) {
+        ALOGE("%s: null out", __FUNCTION__);
+        return -1;
+    }
+
+    std::lock_guard<std::mutex> lk(mLock);
+    if ((rect.left + rect.width) > static_cast<int>(mWidth) ||
+        (rect.top + rect.height) > static_cast<int>(mHeight) || (rect.left % 2) || (rect.top % 2) ||
+        (rect.width % 2) || (rect.height % 2)) {
+        ALOGE("%s: bad rect left %d top %d w %d h %d", __FUNCTION__, rect.left, rect.top,
+              rect.width, rect.height);
+        return -1;
+    }
+
+    out->y = mData.data() + mWidth * rect.top + rect.left;
+    out->yStride = mWidth;
+    uint8_t* cbStart = mData.data() + mWidth * mHeight;
+    uint8_t* crStart = cbStart + mWidth * mHeight / 4;
+    out->cb = cbStart + mWidth * rect.top / 4 + rect.left / 2;
+    out->cr = crStart + mWidth * rect.top / 4 + rect.left / 2;
+    out->cStride = mWidth / 2;
+    out->chromaStep = 1;
+    return 0;
+}
+
+bool isAspectRatioClose(float ar1, float ar2) {
+    constexpr float kAspectRatioMatchThres = 0.025f;  // This threshold is good enough to
+                                                      // distinguish 4:3/16:9/20:9 1.33/1.78/2
+    return std::abs(ar1 - ar2) < kAspectRatioMatchThres;
+}
+
+aidl::android::hardware::camera::common::Status importBufferImpl(
+        /*inout*/ std::map<int, CirculatingBuffers>& circulatingBuffers,
+        /*inout*/ HandleImporter& handleImporter, int32_t streamId, uint64_t bufId,
+        buffer_handle_t buf,
+        /*out*/ buffer_handle_t** outBufPtr) {
+    using ::aidl::android::hardware::camera::common::Status;
+    if (buf == nullptr && bufId == BUFFER_ID_NO_BUFFER) {
+        ALOGE("%s: bufferId %" PRIu64 " has null buffer handle!", __FUNCTION__, bufId);
+        return Status::ILLEGAL_ARGUMENT;
+    }
+
+    CirculatingBuffers& cbs = circulatingBuffers[streamId];
+    if (cbs.count(bufId) == 0) {
+        if (buf == nullptr) {
+            ALOGE("%s: bufferId %" PRIu64 " has null buffer handle!", __FUNCTION__, bufId);
+            return Status::ILLEGAL_ARGUMENT;
+        }
+        // Register a newly seen buffer
+        buffer_handle_t importedBuf = buf;
+        handleImporter.importBuffer(importedBuf);
+        if (importedBuf == nullptr) {
+            ALOGE("%s: output buffer for stream %d is invalid!", __FUNCTION__, streamId);
+            return Status::INTERNAL_ERROR;
+        } else {
+            cbs[bufId] = importedBuf;
+        }
+    }
+    *outBufPtr = &cbs[bufId];
+    return Status::OK;
+}
+
+uint32_t getFourCcFromLayout(const YCbCrLayout& layout) {
+    intptr_t cb = reinterpret_cast<intptr_t>(layout.cb);
+    intptr_t cr = reinterpret_cast<intptr_t>(layout.cr);
+    if (std::abs(cb - cr) == 1 && layout.chromaStep == 2) {
+        // Interleaved format
+        if (layout.cb > layout.cr) {
+            return V4L2_PIX_FMT_NV21;
+        } else {
+            return V4L2_PIX_FMT_NV12;
+        }
+    } else if (layout.chromaStep == 1) {
+        // Planar format
+        if (layout.cb > layout.cr) {
+            return V4L2_PIX_FMT_YVU420;  // YV12
+        } else {
+            return V4L2_PIX_FMT_YUV420;  // YU12
+        }
+    } else {
+        return FLEX_YUV_GENERIC;
+    }
+}
+
+int getCropRect(CroppingType ct, const Size& inSize, const Size& outSize, IMapper::Rect* out) {
+    if (out == nullptr) {
+        ALOGE("%s: out is null", __FUNCTION__);
+        return -1;
+    }
+
+    uint32_t inW = inSize.width;
+    uint32_t inH = inSize.height;
+    uint32_t outW = outSize.width;
+    uint32_t outH = outSize.height;
+
+    // Handle special case where aspect ratio is close to input but scaled
+    // dimension is slightly larger than input
+    float arIn = ASPECT_RATIO(inSize);
+    float arOut = ASPECT_RATIO(outSize);
+    if (isAspectRatioClose(arIn, arOut)) {
+        out->left = 0;
+        out->top = 0;
+        out->width = static_cast<int32_t>(inW);
+        out->height = static_cast<int32_t>(inH);
+        return 0;
+    }
+
+    if (ct == VERTICAL) {
+        uint64_t scaledOutH = static_cast<uint64_t>(outH) * inW / outW;
+        if (scaledOutH > inH) {
+            ALOGE("%s: Output size %dx%d cannot be vertically cropped from input size %dx%d",
+                  __FUNCTION__, outW, outH, inW, inH);
+            return -1;
+        }
+        scaledOutH = scaledOutH & ~0x1;  // make it multiple of 2
+
+        out->left = 0;
+        out->top = static_cast<int32_t>((inH - scaledOutH) / 2) & ~0x1;
+        out->width = static_cast<int32_t>(inW);
+        out->height = static_cast<int32_t>(scaledOutH);
+        ALOGV("%s: crop %dx%d to %dx%d: top %d, scaledH %d", __FUNCTION__, inW, inH, outW, outH,
+              out->top, static_cast<int32_t>(scaledOutH));
+    } else {
+        uint64_t scaledOutW = static_cast<uint64_t>(outW) * inH / outH;
+        if (scaledOutW > inW) {
+            ALOGE("%s: Output size %dx%d cannot be horizontally cropped from input size %dx%d",
+                  __FUNCTION__, outW, outH, inW, inH);
+            return -1;
+        }
+        scaledOutW = scaledOutW & ~0x1;  // make it multiple of 2
+
+        out->left = static_cast<int32_t>((inW - scaledOutW) / 2) & ~0x1;
+        out->top = 0;
+        out->width = static_cast<int32_t>(scaledOutW);
+        out->height = static_cast<int32_t>(inH);
+        ALOGV("%s: crop %dx%d to %dx%d: top %d, scaledW %d", __FUNCTION__, inW, inH, outW, outH,
+              out->top, static_cast<int32_t>(scaledOutW));
+    }
+
+    return 0;
+}
+
+int formatConvert(const YCbCrLayout& in, const YCbCrLayout& out, Size sz, uint32_t format) {
+    int ret = 0;
+    switch (format) {
+        case V4L2_PIX_FMT_NV21:
+            ret = libyuv::I420ToNV21(
+                    static_cast<uint8_t*>(in.y), static_cast<int32_t>(in.yStride),
+                    static_cast<uint8_t*>(in.cb), static_cast<int32_t>(in.cStride),
+                    static_cast<uint8_t*>(in.cr), static_cast<int32_t>(in.cStride),
+                    static_cast<uint8_t*>(out.y), static_cast<int32_t>(out.yStride),
+                    static_cast<uint8_t*>(out.cr), static_cast<int32_t>(out.cStride),
+                    static_cast<int32_t>(sz.width), static_cast<int32_t>(sz.height));
+            if (ret != 0) {
+                ALOGE("%s: convert to NV21 buffer failed! ret %d", __FUNCTION__, ret);
+                return ret;
+            }
+            break;
+        case V4L2_PIX_FMT_NV12:
+            ret = libyuv::I420ToNV12(
+                    static_cast<uint8_t*>(in.y), static_cast<int32_t>(in.yStride),
+                    static_cast<uint8_t*>(in.cb), static_cast<int32_t>(in.cStride),
+                    static_cast<uint8_t*>(in.cr), static_cast<int32_t>(in.cStride),
+                    static_cast<uint8_t*>(out.y), static_cast<int32_t>(out.yStride),
+                    static_cast<uint8_t*>(out.cb), static_cast<int32_t>(out.cStride),
+                    static_cast<int32_t>(sz.width), static_cast<int32_t>(sz.height));
+            if (ret != 0) {
+                ALOGE("%s: convert to NV12 buffer failed! ret %d", __FUNCTION__, ret);
+                return ret;
+            }
+            break;
+        case V4L2_PIX_FMT_YVU420:  // YV12
+        case V4L2_PIX_FMT_YUV420:  // YU12
+            // TODO: maybe we can speed up here by somehow save this copy?
+            ret = libyuv::I420Copy(static_cast<uint8_t*>(in.y), static_cast<int32_t>(in.yStride),
+                                   static_cast<uint8_t*>(in.cb), static_cast<int32_t>(in.cStride),
+                                   static_cast<uint8_t*>(in.cr), static_cast<int32_t>(in.cStride),
+                                   static_cast<uint8_t*>(out.y), static_cast<int32_t>(out.yStride),
+                                   static_cast<uint8_t*>(out.cb), static_cast<int32_t>(out.cStride),
+                                   static_cast<uint8_t*>(out.cr), static_cast<int32_t>(out.cStride),
+                                   static_cast<int32_t>(sz.width), static_cast<int32_t>(sz.height));
+            if (ret != 0) {
+                ALOGE("%s: copy to YV12 or YU12 buffer failed! ret %d", __FUNCTION__, ret);
+                return ret;
+            }
+            break;
+        case FLEX_YUV_GENERIC:
+            // TODO: b/72261744 write to arbitrary flexible YUV layout. Slow.
+            ALOGE("%s: unsupported flexible yuv layout"
+                  " y %p cb %p cr %p y_str %d c_str %d c_step %d",
+                  __FUNCTION__, out.y, out.cb, out.cr, out.yStride, out.cStride, out.chromaStep);
+            return -1;
+        default:
+            ALOGE("%s: unknown YUV format 0x%x!", __FUNCTION__, format);
+            return -1;
+    }
+    return 0;
+}
+
+int encodeJpegYU12(const Size& inSz, const YCbCrLayout& inLayout, int jpegQuality,
+                   const void* app1Buffer, size_t app1Size, void* out, size_t maxOutSize,
+                   size_t& actualCodeSize) {
+    /* libjpeg is a C library so we use C-style "inheritance" by
+     * putting libjpeg's jpeg_destination_mgr first in our custom
+     * struct. This allows us to cast jpeg_destination_mgr* to
+     * CustomJpegDestMgr* when we get it passed to us in a callback */
+    struct CustomJpegDestMgr {
+        struct jpeg_destination_mgr mgr;
+        JOCTET* mBuffer;
+        size_t mBufferSize;
+        size_t mEncodedSize;
+        bool mSuccess;
+    } dmgr;
+
+    jpeg_compress_struct cinfo = {};
+    jpeg_error_mgr jerr;
+
+    /* Initialize error handling with standard callbacks, but
+     * then override output_message (to print to ALOG) and
+     * error_exit to set a flag and print a message instead
+     * of killing the whole process */
+    cinfo.err = jpeg_std_error(&jerr);
+
+    cinfo.err->output_message = [](j_common_ptr cinfo) {
+        char buffer[JMSG_LENGTH_MAX];
+
+        /* Create the message */
+        (*cinfo->err->format_message)(cinfo, buffer);
+        ALOGE("libjpeg error: %s", buffer);
+    };
+    cinfo.err->error_exit = [](j_common_ptr cinfo) {
+        (*cinfo->err->output_message)(cinfo);
+        if (cinfo->client_data) {
+            auto& dmgr = *reinterpret_cast<CustomJpegDestMgr*>(cinfo->client_data);
+            dmgr.mSuccess = false;
+        }
+    };
+
+    /* Now that we initialized some callbacks, let's create our compressor */
+    jpeg_create_compress(&cinfo);
+
+    /* Initialize our destination manager */
+    dmgr.mBuffer = static_cast<JOCTET*>(out);
+    dmgr.mBufferSize = maxOutSize;
+    dmgr.mEncodedSize = 0;
+    dmgr.mSuccess = true;
+    cinfo.client_data = static_cast<void*>(&dmgr);
+
+    /* These lambdas become C-style function pointers and as per C++11 spec
+     * may not capture anything */
+    dmgr.mgr.init_destination = [](j_compress_ptr cinfo) {
+        auto& dmgr = reinterpret_cast<CustomJpegDestMgr&>(*cinfo->dest);
+        dmgr.mgr.next_output_byte = dmgr.mBuffer;
+        dmgr.mgr.free_in_buffer = dmgr.mBufferSize;
+        ALOGV("%s:%d jpeg start: %p [%zu]", __FUNCTION__, __LINE__, dmgr.mBuffer, dmgr.mBufferSize);
+    };
+
+    dmgr.mgr.empty_output_buffer = [](j_compress_ptr cinfo __unused) {
+        ALOGV("%s:%d Out of buffer", __FUNCTION__, __LINE__);
+        return 0;
+    };
+
+    dmgr.mgr.term_destination = [](j_compress_ptr cinfo) {
+        auto& dmgr = reinterpret_cast<CustomJpegDestMgr&>(*cinfo->dest);
+        dmgr.mEncodedSize = dmgr.mBufferSize - dmgr.mgr.free_in_buffer;
+        ALOGV("%s:%d Done with jpeg: %zu", __FUNCTION__, __LINE__, dmgr.mEncodedSize);
+    };
+    cinfo.dest = reinterpret_cast<struct jpeg_destination_mgr*>(&dmgr);
+
+    /* We are going to be using JPEG in raw data mode, so we are passing
+     * straight subsampled planar YCbCr and it will not touch our pixel
+     * data or do any scaling or anything */
+    cinfo.image_width = inSz.width;
+    cinfo.image_height = inSz.height;
+    cinfo.input_components = 3;
+    cinfo.in_color_space = JCS_YCbCr;
+
+    /* Initialize defaults and then override what we want */
+    jpeg_set_defaults(&cinfo);
+
+    jpeg_set_quality(&cinfo, jpegQuality, 1);
+    jpeg_set_colorspace(&cinfo, JCS_YCbCr);
+    cinfo.raw_data_in = 1;
+    cinfo.dct_method = JDCT_IFAST;
+
+    /* Configure sampling factors. The sampling factor is JPEG subsampling 420
+     * because the source format is YUV420. Note that libjpeg sampling factors
+     * are... a little weird. Sampling of Y=2,U=1,V=1 means there is 1 U and
+     * 1 V value for each 2 Y values */
+    cinfo.comp_info[0].h_samp_factor = 2;
+    cinfo.comp_info[0].v_samp_factor = 2;
+    cinfo.comp_info[1].h_samp_factor = 1;
+    cinfo.comp_info[1].v_samp_factor = 1;
+    cinfo.comp_info[2].h_samp_factor = 1;
+    cinfo.comp_info[2].v_samp_factor = 1;
+
+    /* Start the compressor */
+    jpeg_start_compress(&cinfo, TRUE);
+
+    /* Let's not hardcode YUV420 in 6 places... 5 was enough */
+    int maxVSampFactor = cinfo.max_v_samp_factor;
+    int cVSubSampling = cinfo.comp_info[0].v_samp_factor / cinfo.comp_info[1].v_samp_factor;
+
+    /* Compute our macroblock height, so we can pad our input to be vertically
+     * macroblock aligned. No need to for horizontal alignment since AllocatedFrame already
+     * pads horizontally */
+
+    size_t mcuV = DCTSIZE * maxVSampFactor;
+    size_t paddedHeight = mcuV * ((inSz.height + mcuV - 1) / mcuV);
+
+    /* libjpeg uses arrays of row pointers, which makes it really easy to pad
+     * data vertically (unfortunately doesn't help horizontally) */
+    std::vector<JSAMPROW> yLines(paddedHeight);
+    std::vector<JSAMPROW> cbLines(paddedHeight / cVSubSampling);
+    std::vector<JSAMPROW> crLines(paddedHeight / cVSubSampling);
+
+    uint8_t* py = static_cast<uint8_t*>(inLayout.y);
+    uint8_t* pcb = static_cast<uint8_t*>(inLayout.cb);
+    uint8_t* pcr = static_cast<uint8_t*>(inLayout.cr);
+
+    for (int32_t i = 0; i < paddedHeight; i++) {
+        /* Once we are in the padding territory we still point to the last line
+         * effectively replicating it several times ~ CLAMP_TO_EDGE */
+        int li = std::min(i, inSz.height - 1);
+        yLines[i] = static_cast<JSAMPROW>(py + li * inLayout.yStride);
+        if (i < paddedHeight / cVSubSampling) {
+            li = std::min(i, (inSz.height - 1) / cVSubSampling);
+            cbLines[i] = static_cast<JSAMPROW>(pcb + li * inLayout.cStride);
+            crLines[i] = static_cast<JSAMPROW>(pcr + li * inLayout.cStride);
+        }
+    }
+
+    /* If APP1 data was passed in, use it */
+    if (app1Buffer && app1Size) {
+        jpeg_write_marker(&cinfo, JPEG_APP0 + 1, static_cast<const JOCTET*>(app1Buffer), app1Size);
+    }
+
+    /* While we still have padded height left to go, keep giving it one
+     * macroblock at a time. */
+    while (cinfo.next_scanline < cinfo.image_height) {
+        const uint32_t batchSize = DCTSIZE * maxVSampFactor;
+        const uint32_t nl = cinfo.next_scanline;
+        JSAMPARRAY planes[3]{&yLines[nl], &cbLines[nl / cVSubSampling],
+                             &crLines[nl / cVSubSampling]};
+
+        uint32_t done = jpeg_write_raw_data(&cinfo, planes, batchSize);
+
+        if (done != batchSize) {
+            ALOGE("%s: compressed %u lines, expected %u (total %u/%u)", __FUNCTION__, done,
+                  batchSize, cinfo.next_scanline, cinfo.image_height);
+            return -1;
+        }
+    }
+
+    /* This will flush everything */
+    jpeg_finish_compress(&cinfo);
+
+    /* Grab the actual code size and set it */
+    actualCodeSize = dmgr.mEncodedSize;
+
+    return 0;
+}
+
+Size getMaxThumbnailResolution(const common::V1_0::helper::CameraMetadata& chars) {
+    Size thumbSize{0, 0};
+    camera_metadata_ro_entry entry = chars.find(ANDROID_JPEG_AVAILABLE_THUMBNAIL_SIZES);
+    for (uint32_t i = 0; i < entry.count; i += 2) {
+        Size sz{.width = entry.data.i32[i], .height = entry.data.i32[i + 1]};
+        if (sz.width * sz.height > thumbSize.width * thumbSize.height) {
+            thumbSize = sz;
+        }
+    }
+
+    if (thumbSize.width * thumbSize.height == 0) {
+        ALOGW("%s: non-zero thumbnail size not available", __FUNCTION__);
+    }
+
+    return thumbSize;
+}
+
+void freeReleaseFences(std::vector<CaptureResult>& results) {
+    for (auto& result : results) {
+        native_handle_t* inputReleaseFence =
+                ::android::makeFromAidl(result.inputBuffer.releaseFence);
+        if (inputReleaseFence != nullptr) {
+            native_handle_close(inputReleaseFence);
+            native_handle_delete(inputReleaseFence);
+        }
+        for (auto& buf : result.outputBuffers) {
+            native_handle_t* outReleaseFence = ::android::makeFromAidl(buf.releaseFence);
+            if (outReleaseFence != nullptr) {
+                native_handle_close(outReleaseFence);
+                native_handle_delete(outReleaseFence);
+            }
+        }
+    }
+}
+
+#define ARRAY_SIZE(a) (sizeof(a) / sizeof((a)[0]))
+#define UPDATE(md, tag, data, size)               \
+    do {                                          \
+        if ((md).update((tag), (data), (size))) { \
+            ALOGE("Update " #tag " failed!");     \
+            return BAD_VALUE;                     \
+        }                                         \
+    } while (0)
+
+status_t fillCaptureResultCommon(CameraMetadata& md, nsecs_t timestamp,
+                                 camera_metadata_ro_entry& activeArraySize) {
+    if (activeArraySize.count < 4) {
+        ALOGE("%s: cannot find active array size!", __FUNCTION__);
+        return -EINVAL;
+    }
+    // android.control
+    // For USB camera, we don't know the AE state. Set the state to converged to
+    // indicate the frame should be good to use. Then apps don't have to wait the
+    // AE state.
+    const uint8_t aeState = ANDROID_CONTROL_AE_STATE_CONVERGED;
+    UPDATE(md, ANDROID_CONTROL_AE_STATE, &aeState, 1);
+
+    const uint8_t ae_lock = ANDROID_CONTROL_AE_LOCK_OFF;
+    UPDATE(md, ANDROID_CONTROL_AE_LOCK, &ae_lock, 1);
+
+    // Set AWB state to converged to indicate the frame should be good to use.
+    const uint8_t awbState = ANDROID_CONTROL_AWB_STATE_CONVERGED;
+    UPDATE(md, ANDROID_CONTROL_AWB_STATE, &awbState, 1);
+
+    const uint8_t awbLock = ANDROID_CONTROL_AWB_LOCK_OFF;
+    UPDATE(md, ANDROID_CONTROL_AWB_LOCK, &awbLock, 1);
+
+    const uint8_t flashState = ANDROID_FLASH_STATE_UNAVAILABLE;
+    UPDATE(md, ANDROID_FLASH_STATE, &flashState, 1);
+
+    // This means pipeline latency of X frame intervals. The maximum number is 4.
+    const uint8_t requestPipelineMaxDepth = 4;
+    UPDATE(md, ANDROID_REQUEST_PIPELINE_DEPTH, &requestPipelineMaxDepth, 1);
+
+    // android.scaler
+    const int32_t crop_region[] = {
+            activeArraySize.data.i32[0],
+            activeArraySize.data.i32[1],
+            activeArraySize.data.i32[2],
+            activeArraySize.data.i32[3],
+    };
+    UPDATE(md, ANDROID_SCALER_CROP_REGION, crop_region, ARRAY_SIZE(crop_region));
+
+    // android.sensor
+    UPDATE(md, ANDROID_SENSOR_TIMESTAMP, &timestamp, 1);
+
+    // android.statistics
+    const uint8_t lensShadingMapMode = ANDROID_STATISTICS_LENS_SHADING_MAP_MODE_OFF;
+    UPDATE(md, ANDROID_STATISTICS_LENS_SHADING_MAP_MODE, &lensShadingMapMode, 1);
+
+    const uint8_t sceneFlicker = ANDROID_STATISTICS_SCENE_FLICKER_NONE;
+    UPDATE(md, ANDROID_STATISTICS_SCENE_FLICKER, &sceneFlicker, 1);
+
+    return OK;
+}
+
+#undef ARRAY_SIZE
+#undef UPDATE
+
+AllocatedV4L2Frame::AllocatedV4L2Frame(std::shared_ptr<V4L2Frame> frameIn)
+    : Frame(frameIn->mWidth, frameIn->mHeight, frameIn->mFourcc) {
+    uint8_t* dataIn;
+    size_t dataSize;
+    if (frameIn->getData(&dataIn, &dataSize) != 0) {
+        ALOGE("%s: map input V4L2 frame failed!", __FUNCTION__);
+        return;
+    }
+
+    mData.resize(dataSize);
+    std::memcpy(mData.data(), dataIn, dataSize);
+}
+
+AllocatedV4L2Frame::~AllocatedV4L2Frame() {}
+
+int AllocatedV4L2Frame::getData(uint8_t** outData, size_t* dataSize) {
+    if (outData == nullptr || dataSize == nullptr) {
+        ALOGE("%s: outData(%p)/dataSize(%p) must not be null", __FUNCTION__, outData, dataSize);
+        return -1;
+    }
+
+    *outData = mData.data();
+    *dataSize = mData.size();
+    return 0;
+}
+
+}  // namespace implementation
+}  // namespace device
+}  // namespace camera
+}  // namespace hardware
+}  // namespace android
\ No newline at end of file