/*
* Copyright (c) Meta Platforms, Inc. and affiliates.
* All rights reserved.
*
* This source code is licensed under both the BSD-style license (found in the
* LICENSE file in the root directory of this source tree) and the GPLv2 (found
* in the COPYING file in the root directory of this source tree).
* You may select, at your option, one of the above-listed licenses.
*/

#include "platform.h"
#include <stdio.h>      /* fprintf, open, fdopen, fread, _fileno, stdin, stdout */
#include <stdlib.h>     /* malloc, free */
#include <assert.h>
#include <errno.h>      /* errno */

#if defined (_MSC_VER)
#  include <sys/stat.h>
#  include <io.h>
#endif

#include "fileio_asyncio.h"
#include "fileio_common.h"

/* **********************************************************************
*  Sparse write
************************************************************************/

/** AIO_fwriteSparse() :
*  @return : storedSkips,
*            argument for next call to AIO_fwriteSparse() or AIO_fwriteSparseEnd() */
static unsigned
AIO_fwriteSparse(FILE* file,
                const void* buffer, size_t bufferSize,
                const FIO_prefs_t* const prefs,
                unsigned storedSkips)
{
   const size_t* const bufferT = (const size_t*)buffer;   /* Buffer is supposed malloc'ed, hence aligned on size_t */
   size_t bufferSizeT = bufferSize / sizeof(size_t);
   const size_t* const bufferTEnd = bufferT + bufferSizeT;
   const size_t* ptrT = bufferT;
   static const size_t segmentSizeT = (32 KB) / sizeof(size_t);   /* check every 32 KB */

   if (prefs->testMode) return 0;  /* do not output anything in test mode */

   if (!prefs->sparseFileSupport) {  /* normal write */
       size_t const sizeCheck = fwrite(buffer, 1, bufferSize, file);
       if (sizeCheck != bufferSize)
           EXM_THROW(70, "Write error : cannot write block : %s",
                     strerror(errno));
       return 0;
   }

   /* avoid int overflow */
   if (storedSkips > 1 GB) {
       if (LONG_SEEK(file, 1 GB, SEEK_CUR) != 0)
       EXM_THROW(91, "1 GB skip error (sparse file support)");
       storedSkips -= 1 GB;
   }

   while (ptrT < bufferTEnd) {
       size_t nb0T;

       /* adjust last segment if < 32 KB */
       size_t seg0SizeT = segmentSizeT;
       if (seg0SizeT > bufferSizeT) seg0SizeT = bufferSizeT;
       bufferSizeT -= seg0SizeT;

       /* count leading zeroes */
       for (nb0T=0; (nb0T < seg0SizeT) && (ptrT[nb0T] == 0); nb0T++) ;
       storedSkips += (unsigned)(nb0T * sizeof(size_t));

       if (nb0T != seg0SizeT) {   /* not all 0s */
           size_t const nbNon0ST = seg0SizeT - nb0T;
           /* skip leading zeros */
           if (LONG_SEEK(file, storedSkips, SEEK_CUR) != 0)
               EXM_THROW(92, "Sparse skip error ; try --no-sparse");
           storedSkips = 0;
           /* write the rest */
           if (fwrite(ptrT + nb0T, sizeof(size_t), nbNon0ST, file) != nbNon0ST)
               EXM_THROW(93, "Write error : cannot write block : %s",
                         strerror(errno));
       }
       ptrT += seg0SizeT;
   }

   {   static size_t const maskT = sizeof(size_t)-1;
       if (bufferSize & maskT) {
           /* size not multiple of sizeof(size_t) : implies end of block */
           const char* const restStart = (const char*)bufferTEnd;
           const char* restPtr = restStart;
           const char* const restEnd = (const char*)buffer + bufferSize;
           assert(restEnd > restStart && restEnd < restStart + sizeof(size_t));
           for ( ; (restPtr < restEnd) && (*restPtr == 0); restPtr++) ;
           storedSkips += (unsigned) (restPtr - restStart);
           if (restPtr != restEnd) {
               /* not all remaining bytes are 0 */
               size_t const restSize = (size_t)(restEnd - restPtr);
               if (LONG_SEEK(file, storedSkips, SEEK_CUR) != 0)
                   EXM_THROW(92, "Sparse skip error ; try --no-sparse");
               if (fwrite(restPtr, 1, restSize, file) != restSize)
                   EXM_THROW(95, "Write error : cannot write end of decoded block : %s",
                             strerror(errno));
               storedSkips = 0;
           }   }   }

   return storedSkips;
}

static void
AIO_fwriteSparseEnd(const FIO_prefs_t* const prefs, FILE* file, unsigned storedSkips)
{
   if (prefs->testMode) assert(storedSkips == 0);
   if (storedSkips>0) {
       assert(prefs->sparseFileSupport > 0);  /* storedSkips>0 implies sparse support is enabled */
       (void)prefs;   /* assert can be disabled, in which case prefs becomes unused */
       if (LONG_SEEK(file, storedSkips-1, SEEK_CUR) != 0)
           EXM_THROW(69, "Final skip error (sparse file support)");
       /* last zero must be explicitly written,
        * so that skipped ones get implicitly translated as zero by FS */
       {   const char lastZeroByte[1] = { 0 };
           if (fwrite(lastZeroByte, 1, 1, file) != 1)
               EXM_THROW(69, "Write error : cannot write last zero : %s", strerror(errno));
       }   }
}


/* **********************************************************************
*  AsyncIO functionality
************************************************************************/

/* AIO_supported:
* Returns 1 if AsyncIO is supported on the system, 0 otherwise. */
int AIO_supported(void) {
#ifdef ZSTD_MULTITHREAD
   return 1;
#else
   return 0;
#endif
}

/* ***********************************
*  Generic IoPool implementation
*************************************/

static IOJob_t *AIO_IOPool_createIoJob(IOPoolCtx_t *ctx, size_t bufferSize) {
   IOJob_t* const job  = (IOJob_t*) malloc(sizeof(IOJob_t));
   void* const buffer = malloc(bufferSize);
   if(!job || !buffer)
       EXM_THROW(101, "Allocation error : not enough memory");
   job->buffer = buffer;
   job->bufferSize = bufferSize;
   job->usedBufferSize = 0;
   job->file = NULL;
   job->ctx = ctx;
   job->offset = 0;
   return job;
}


/* AIO_IOPool_createThreadPool:
* Creates a thread pool and a mutex for threaded IO pool.
* Displays warning if asyncio is requested but MT isn't available. */
static void AIO_IOPool_createThreadPool(IOPoolCtx_t* ctx, const FIO_prefs_t* prefs) {
   ctx->threadPool = NULL;
   ctx->threadPoolActive = 0;
   if(prefs->asyncIO) {
       if (ZSTD_pthread_mutex_init(&ctx->ioJobsMutex, NULL))
           EXM_THROW(102,"Failed creating ioJobsMutex mutex");
       /* We want MAX_IO_JOBS-2 queue items because we need to always have 1 free buffer to
        * decompress into and 1 buffer that's actively written to disk and owned by the writing thread. */
       assert(MAX_IO_JOBS >= 2);
       ctx->threadPool = POOL_create(1, MAX_IO_JOBS - 2);
       ctx->threadPoolActive = 1;
       if (!ctx->threadPool)
           EXM_THROW(104, "Failed creating I/O thread pool");
   }
}

/* AIO_IOPool_init:
* Allocates and sets and a new I/O thread pool including its included availableJobs. */
static void AIO_IOPool_init(IOPoolCtx_t* ctx, const FIO_prefs_t* prefs, POOL_function poolFunction, size_t bufferSize) {
   int i;
   AIO_IOPool_createThreadPool(ctx, prefs);
   ctx->prefs = prefs;
   ctx->poolFunction = poolFunction;
   ctx->totalIoJobs = ctx->threadPool ? MAX_IO_JOBS : 2;
   ctx->availableJobsCount = ctx->totalIoJobs;
   for(i=0; i < ctx->availableJobsCount; i++) {
       ctx->availableJobs[i] = AIO_IOPool_createIoJob(ctx, bufferSize);
   }
   ctx->jobBufferSize = bufferSize;
   ctx->file = NULL;
}


/* AIO_IOPool_threadPoolActive:
* Check if current operation uses thread pool.
* Note that in some cases we have a thread pool initialized but choose not to use it. */
static int AIO_IOPool_threadPoolActive(IOPoolCtx_t* ctx) {
   return ctx->threadPool && ctx->threadPoolActive;
}


/* AIO_IOPool_lockJobsMutex:
* Locks the IO jobs mutex if threading is active */
static void AIO_IOPool_lockJobsMutex(IOPoolCtx_t* ctx) {
   if(AIO_IOPool_threadPoolActive(ctx))
       ZSTD_pthread_mutex_lock(&ctx->ioJobsMutex);
}

/* AIO_IOPool_unlockJobsMutex:
* Unlocks the IO jobs mutex if threading is active */
static void AIO_IOPool_unlockJobsMutex(IOPoolCtx_t* ctx) {
   if(AIO_IOPool_threadPoolActive(ctx))
       ZSTD_pthread_mutex_unlock(&ctx->ioJobsMutex);
}

/* AIO_IOPool_releaseIoJob:
* Releases an acquired job back to the pool. Doesn't execute the job. */
static void AIO_IOPool_releaseIoJob(IOJob_t* job) {
   IOPoolCtx_t* const ctx = (IOPoolCtx_t *) job->ctx;
   AIO_IOPool_lockJobsMutex(ctx);
   assert(ctx->availableJobsCount < ctx->totalIoJobs);
   ctx->availableJobs[ctx->availableJobsCount++] = job;
   AIO_IOPool_unlockJobsMutex(ctx);
}

/* AIO_IOPool_join:
* Waits for all tasks in the pool to finish executing. */
static void AIO_IOPool_join(IOPoolCtx_t* ctx) {
   if(AIO_IOPool_threadPoolActive(ctx))
       POOL_joinJobs(ctx->threadPool);
}

/* AIO_IOPool_setThreaded:
* Allows (de)activating threaded mode, to be used when the expected overhead
* of threading costs more than the expected gains. */
static void AIO_IOPool_setThreaded(IOPoolCtx_t* ctx, int threaded) {
   assert(threaded == 0 || threaded == 1);
   assert(ctx != NULL);
   if(ctx->threadPoolActive != threaded) {
       AIO_IOPool_join(ctx);
       ctx->threadPoolActive = threaded;
   }
}

/* AIO_IOPool_free:
* Release a previously allocated IO thread pool. Makes sure all tasks are done and released. */
static void AIO_IOPool_destroy(IOPoolCtx_t* ctx) {
   int i;
   if(ctx->threadPool) {
       /* Make sure we finish all tasks and then free the resources */
       AIO_IOPool_join(ctx);
       /* Make sure we are not leaking availableJobs */
       assert(ctx->availableJobsCount == ctx->totalIoJobs);
       POOL_free(ctx->threadPool);
       ZSTD_pthread_mutex_destroy(&ctx->ioJobsMutex);
   }
   assert(ctx->file == NULL);
   for(i=0; i<ctx->availableJobsCount; i++) {
       IOJob_t* job = (IOJob_t*) ctx->availableJobs[i];
       free(job->buffer);
       free(job);
   }
}

/* AIO_IOPool_acquireJob:
* Returns an available io job to be used for a future io. */
static IOJob_t* AIO_IOPool_acquireJob(IOPoolCtx_t* ctx) {
   IOJob_t *job;
   assert(ctx->file != NULL || ctx->prefs->testMode);
   AIO_IOPool_lockJobsMutex(ctx);
   assert(ctx->availableJobsCount > 0);
   job = (IOJob_t*) ctx->availableJobs[--ctx->availableJobsCount];
   AIO_IOPool_unlockJobsMutex(ctx);
   job->usedBufferSize = 0;
   job->file = ctx->file;
   job->offset = 0;
   return job;
}


/* AIO_IOPool_setFile:
* Sets the destination file for future files in the pool.
* Requires completion of all queued jobs and release of all otherwise acquired jobs. */
static void AIO_IOPool_setFile(IOPoolCtx_t* ctx, FILE* file) {
   assert(ctx!=NULL);
   AIO_IOPool_join(ctx);
   assert(ctx->availableJobsCount == ctx->totalIoJobs);
   ctx->file = file;
}

static FILE* AIO_IOPool_getFile(const IOPoolCtx_t* ctx) {
   return ctx->file;
}

/* AIO_IOPool_enqueueJob:
* Enqueues an io job for execution.
* The queued job shouldn't be used directly after queueing it. */
static void AIO_IOPool_enqueueJob(IOJob_t* job) {
   IOPoolCtx_t* const ctx = (IOPoolCtx_t *)job->ctx;
   if(AIO_IOPool_threadPoolActive(ctx))
       POOL_add(ctx->threadPool, ctx->poolFunction, job);
   else
       ctx->poolFunction(job);
}

/* ***********************************
*  WritePool implementation
*************************************/

/* AIO_WritePool_acquireJob:
* Returns an available write job to be used for a future write. */
IOJob_t* AIO_WritePool_acquireJob(WritePoolCtx_t* ctx) {
   return AIO_IOPool_acquireJob(&ctx->base);
}

/* AIO_WritePool_enqueueAndReacquireWriteJob:
* Queues a write job for execution and acquires a new one.
* After execution `job`'s pointed value would change to the newly acquired job.
* Make sure to set `usedBufferSize` to the wanted length before call.
* The queued job shouldn't be used directly after queueing it. */
void AIO_WritePool_enqueueAndReacquireWriteJob(IOJob_t **job) {
   AIO_IOPool_enqueueJob(*job);
   *job = AIO_IOPool_acquireJob((IOPoolCtx_t *)(*job)->ctx);
}

/* AIO_WritePool_sparseWriteEnd:
* Ends sparse writes to the current file.
* Blocks on completion of all current write jobs before executing. */
void AIO_WritePool_sparseWriteEnd(WritePoolCtx_t* ctx) {
   assert(ctx != NULL);
   AIO_IOPool_join(&ctx->base);
   AIO_fwriteSparseEnd(ctx->base.prefs, ctx->base.file, ctx->storedSkips);
   ctx->storedSkips = 0;
}

/* AIO_WritePool_setFile:
* Sets the destination file for future writes in the pool.
* Requires completion of all queues write jobs and release of all otherwise acquired jobs.
* Also requires ending of sparse write if a previous file was used in sparse mode. */
void AIO_WritePool_setFile(WritePoolCtx_t* ctx, FILE* file) {
   AIO_IOPool_setFile(&ctx->base, file);
   assert(ctx->storedSkips == 0);
}

/* AIO_WritePool_getFile:
* Returns the file the writePool is currently set to write to. */
FILE* AIO_WritePool_getFile(const WritePoolCtx_t* ctx) {
   return AIO_IOPool_getFile(&ctx->base);
}

/* AIO_WritePool_releaseIoJob:
* Releases an acquired job back to the pool. Doesn't execute the job. */
void AIO_WritePool_releaseIoJob(IOJob_t* job) {
   AIO_IOPool_releaseIoJob(job);
}

/* AIO_WritePool_closeFile:
* Ends sparse write and closes the writePool's current file and sets the file to NULL.
* Requires completion of all queues write jobs and release of all otherwise acquired jobs.  */
int AIO_WritePool_closeFile(WritePoolCtx_t* ctx) {
   FILE* const dstFile = ctx->base.file;
   assert(dstFile!=NULL || ctx->base.prefs->testMode!=0);
   AIO_WritePool_sparseWriteEnd(ctx);
   AIO_IOPool_setFile(&ctx->base, NULL);
   return fclose(dstFile);
}

/* AIO_WritePool_executeWriteJob:
* Executes a write job synchronously. Can be used as a function for a thread pool. */
static void AIO_WritePool_executeWriteJob(void* opaque){
   IOJob_t* const job = (IOJob_t*) opaque;
   WritePoolCtx_t* const ctx = (WritePoolCtx_t*) job->ctx;
   ctx->storedSkips = AIO_fwriteSparse(job->file, job->buffer, job->usedBufferSize, ctx->base.prefs, ctx->storedSkips);
   AIO_IOPool_releaseIoJob(job);
}

/* AIO_WritePool_create:
* Allocates and sets and a new write pool including its included jobs. */
WritePoolCtx_t* AIO_WritePool_create(const FIO_prefs_t* prefs, size_t bufferSize) {
   WritePoolCtx_t* const ctx = (WritePoolCtx_t*) malloc(sizeof(WritePoolCtx_t));
   if(!ctx) EXM_THROW(100, "Allocation error : not enough memory");
   AIO_IOPool_init(&ctx->base, prefs, AIO_WritePool_executeWriteJob, bufferSize);
   ctx->storedSkips = 0;
   return ctx;
}

/* AIO_WritePool_free:
* Frees and releases a writePool and its resources. Closes destination file if needs to. */
void AIO_WritePool_free(WritePoolCtx_t* ctx) {
   /* Make sure we finish all tasks and then free the resources */
   if(AIO_WritePool_getFile(ctx))
       AIO_WritePool_closeFile(ctx);
   AIO_IOPool_destroy(&ctx->base);
   assert(ctx->storedSkips==0);
   free(ctx);
}

/* AIO_WritePool_setAsync:
* Allows (de)activating async mode, to be used when the expected overhead
* of asyncio costs more than the expected gains. */
void AIO_WritePool_setAsync(WritePoolCtx_t* ctx, int async) {
   AIO_IOPool_setThreaded(&ctx->base, async);
}


/* ***********************************
*  ReadPool implementation
*************************************/
static void AIO_ReadPool_releaseAllCompletedJobs(ReadPoolCtx_t* ctx) {
   int i;
   for(i=0; i<ctx->completedJobsCount; i++) {
       IOJob_t* job = (IOJob_t*) ctx->completedJobs[i];
       AIO_IOPool_releaseIoJob(job);
   }
   ctx->completedJobsCount = 0;
}

static void AIO_ReadPool_addJobToCompleted(IOJob_t* job) {
   ReadPoolCtx_t* const ctx = (ReadPoolCtx_t *)job->ctx;
   AIO_IOPool_lockJobsMutex(&ctx->base);
   assert(ctx->completedJobsCount < MAX_IO_JOBS);
   ctx->completedJobs[ctx->completedJobsCount++] = job;
   if(AIO_IOPool_threadPoolActive(&ctx->base)) {
       ZSTD_pthread_cond_signal(&ctx->jobCompletedCond);
   }
   AIO_IOPool_unlockJobsMutex(&ctx->base);
}

/* AIO_ReadPool_findNextWaitingOffsetCompletedJob_locked:
* Looks through the completed jobs for a job matching the waitingOnOffset and returns it,
* if job wasn't found returns NULL.
* IMPORTANT: assumes ioJobsMutex is locked. */
static IOJob_t* AIO_ReadPool_findNextWaitingOffsetCompletedJob_locked(ReadPoolCtx_t* ctx) {
   IOJob_t *job = NULL;
   int i;
   /* This implementation goes through all completed jobs and looks for the one matching the next offset.
    * While not strictly needed for a single threaded reader implementation (as in such a case we could expect
    * reads to be completed in order) this implementation was chosen as it better fits other asyncio
    * interfaces (such as io_uring) that do not provide promises regarding order of completion. */
   for (i=0; i<ctx->completedJobsCount; i++) {
       job = (IOJob_t *) ctx->completedJobs[i];
       if (job->offset == ctx->waitingOnOffset) {
           ctx->completedJobs[i] = ctx->completedJobs[--ctx->completedJobsCount];
           return job;
       }
   }
   return NULL;
}

/* AIO_ReadPool_numReadsInFlight:
* Returns the number of IO read jobs currently in flight. */
static size_t AIO_ReadPool_numReadsInFlight(ReadPoolCtx_t* ctx) {
   const int jobsHeld = (ctx->currentJobHeld==NULL ? 0 : 1);
   return (size_t)(ctx->base.totalIoJobs - (ctx->base.availableJobsCount + ctx->completedJobsCount + jobsHeld));
}

/* AIO_ReadPool_getNextCompletedJob:
* Returns a completed IOJob_t for the next read in line based on waitingOnOffset and advances waitingOnOffset.
* Would block. */
static IOJob_t* AIO_ReadPool_getNextCompletedJob(ReadPoolCtx_t* ctx) {
   IOJob_t *job = NULL;
   AIO_IOPool_lockJobsMutex(&ctx->base);

   job = AIO_ReadPool_findNextWaitingOffsetCompletedJob_locked(ctx);

   /* As long as we didn't find the job matching the next read, and we have some reads in flight continue waiting */
   while (!job && (AIO_ReadPool_numReadsInFlight(ctx) > 0)) {
       assert(ctx->base.threadPool != NULL); /* we shouldn't be here if we work in sync mode */
       ZSTD_pthread_cond_wait(&ctx->jobCompletedCond, &ctx->base.ioJobsMutex);
       job = AIO_ReadPool_findNextWaitingOffsetCompletedJob_locked(ctx);
   }

   if(job) {
       assert(job->offset == ctx->waitingOnOffset);
       ctx->waitingOnOffset += job->usedBufferSize;
   }

   AIO_IOPool_unlockJobsMutex(&ctx->base);
   return job;
}


/* AIO_ReadPool_executeReadJob:
* Executes a read job synchronously. Can be used as a function for a thread pool. */
static void AIO_ReadPool_executeReadJob(void* opaque){
   IOJob_t* const job = (IOJob_t*) opaque;
   ReadPoolCtx_t* const ctx = (ReadPoolCtx_t *)job->ctx;
   if(ctx->reachedEof) {
       job->usedBufferSize = 0;
       AIO_ReadPool_addJobToCompleted(job);
       return;
   }
   job->usedBufferSize = fread(job->buffer, 1, job->bufferSize, job->file);
   if(job->usedBufferSize < job->bufferSize) {
       if(ferror(job->file)) {
           EXM_THROW(37, "Read error");
       } else if(feof(job->file)) {
           ctx->reachedEof = 1;
       } else {
           EXM_THROW(37, "Unexpected short read");
       }
   }
   AIO_ReadPool_addJobToCompleted(job);
}

static void AIO_ReadPool_enqueueRead(ReadPoolCtx_t* ctx) {
   IOJob_t* const job = AIO_IOPool_acquireJob(&ctx->base);
   job->offset = ctx->nextReadOffset;
   ctx->nextReadOffset += job->bufferSize;
   AIO_IOPool_enqueueJob(job);
}

static void AIO_ReadPool_startReading(ReadPoolCtx_t* ctx) {
   while(ctx->base.availableJobsCount) {
       AIO_ReadPool_enqueueRead(ctx);
   }
}

/* AIO_ReadPool_setFile:
* Sets the source file for future read in the pool. Initiates reading immediately if file is not NULL.
* Waits for all current enqueued tasks to complete if a previous file was set. */
void AIO_ReadPool_setFile(ReadPoolCtx_t* ctx, FILE* file) {
   assert(ctx!=NULL);
   AIO_IOPool_join(&ctx->base);
   AIO_ReadPool_releaseAllCompletedJobs(ctx);
   if (ctx->currentJobHeld) {
       AIO_IOPool_releaseIoJob((IOJob_t *)ctx->currentJobHeld);
       ctx->currentJobHeld = NULL;
   }
   AIO_IOPool_setFile(&ctx->base, file);
   ctx->nextReadOffset = 0;
   ctx->waitingOnOffset = 0;
   ctx->srcBuffer = ctx->coalesceBuffer;
   ctx->srcBufferLoaded = 0;
   ctx->reachedEof = 0;
   if(file != NULL)
       AIO_ReadPool_startReading(ctx);
}

/* AIO_ReadPool_create:
* Allocates and sets and a new readPool including its included jobs.
* bufferSize should be set to the maximal buffer we want to read at a time, will also be used
* as our basic read size. */
ReadPoolCtx_t* AIO_ReadPool_create(const FIO_prefs_t* prefs, size_t bufferSize) {
   ReadPoolCtx_t* const ctx = (ReadPoolCtx_t*) malloc(sizeof(ReadPoolCtx_t));
   if(!ctx) EXM_THROW(100, "Allocation error : not enough memory");
   AIO_IOPool_init(&ctx->base, prefs, AIO_ReadPool_executeReadJob, bufferSize);

   ctx->coalesceBuffer = (U8*) malloc(bufferSize * 2);
   if(!ctx->coalesceBuffer) EXM_THROW(100, "Allocation error : not enough memory");
   ctx->srcBuffer = ctx->coalesceBuffer;
   ctx->srcBufferLoaded = 0;
   ctx->completedJobsCount = 0;
   ctx->currentJobHeld = NULL;

   if(ctx->base.threadPool)
       if (ZSTD_pthread_cond_init(&ctx->jobCompletedCond, NULL))
           EXM_THROW(103,"Failed creating jobCompletedCond cond");

   return ctx;
}

/* AIO_ReadPool_free:
* Frees and releases a readPool and its resources. Closes source file. */
void AIO_ReadPool_free(ReadPoolCtx_t* ctx) {
   if(AIO_ReadPool_getFile(ctx))
       AIO_ReadPool_closeFile(ctx);
   if(ctx->base.threadPool)
       ZSTD_pthread_cond_destroy(&ctx->jobCompletedCond);
   AIO_IOPool_destroy(&ctx->base);
   free(ctx->coalesceBuffer);
   free(ctx);
}

/* AIO_ReadPool_consumeBytes:
* Consumes byes from srcBuffer's beginning and updates srcBufferLoaded accordingly. */
void AIO_ReadPool_consumeBytes(ReadPoolCtx_t* ctx, size_t n) {
   assert(n <= ctx->srcBufferLoaded);
   ctx->srcBufferLoaded -= n;
   ctx->srcBuffer += n;
}

/* AIO_ReadPool_releaseCurrentlyHeldAndGetNext:
* Release the current held job and get the next one, returns NULL if no next job available. */
static IOJob_t* AIO_ReadPool_releaseCurrentHeldAndGetNext(ReadPoolCtx_t* ctx) {
   if (ctx->currentJobHeld) {
       AIO_IOPool_releaseIoJob((IOJob_t *)ctx->currentJobHeld);
       ctx->currentJobHeld = NULL;
       AIO_ReadPool_enqueueRead(ctx);
   }
   ctx->currentJobHeld = AIO_ReadPool_getNextCompletedJob(ctx);
   return (IOJob_t*) ctx->currentJobHeld;
}

/* AIO_ReadPool_fillBuffer:
* Tries to fill the buffer with at least n or jobBufferSize bytes (whichever is smaller).
* Returns if srcBuffer has at least the expected number of bytes loaded or if we've reached the end of the file.
* Return value is the number of bytes added to the buffer.
* Note that srcBuffer might have up to 2 times jobBufferSize bytes. */
size_t AIO_ReadPool_fillBuffer(ReadPoolCtx_t* ctx, size_t n) {
   IOJob_t *job;
   int useCoalesce = 0;
   if(n > ctx->base.jobBufferSize)
       n = ctx->base.jobBufferSize;

   /* We are good, don't read anything */
   if (ctx->srcBufferLoaded >= n)
       return 0;

   /* We still have bytes loaded, but not enough to satisfy caller. We need to get the next job
    * and coalesce the remaining bytes with the next job's buffer */
   if (ctx->srcBufferLoaded > 0) {
       useCoalesce = 1;
       memcpy(ctx->coalesceBuffer, ctx->srcBuffer, ctx->srcBufferLoaded);
       ctx->srcBuffer = ctx->coalesceBuffer;
   }

   /* Read the next chunk */
   job = AIO_ReadPool_releaseCurrentHeldAndGetNext(ctx);
   if(!job)
       return 0;
   if(useCoalesce) {
       assert(ctx->srcBufferLoaded + job->usedBufferSize <= 2*ctx->base.jobBufferSize);
       memcpy(ctx->coalesceBuffer + ctx->srcBufferLoaded, job->buffer, job->usedBufferSize);
       ctx->srcBufferLoaded += job->usedBufferSize;
   }
   else {
       ctx->srcBuffer = (U8 *) job->buffer;
       ctx->srcBufferLoaded = job->usedBufferSize;
   }
   return job->usedBufferSize;
}

/* AIO_ReadPool_consumeAndRefill:
* Consumes the current buffer and refills it with bufferSize bytes. */
size_t AIO_ReadPool_consumeAndRefill(ReadPoolCtx_t* ctx) {
   AIO_ReadPool_consumeBytes(ctx, ctx->srcBufferLoaded);
   return AIO_ReadPool_fillBuffer(ctx, ctx->base.jobBufferSize);
}

/* AIO_ReadPool_getFile:
* Returns the current file set for the read pool. */
FILE* AIO_ReadPool_getFile(const ReadPoolCtx_t* ctx) {
   return AIO_IOPool_getFile(&ctx->base);
}

/* AIO_ReadPool_closeFile:
* Closes the current set file. Waits for all current enqueued tasks to complete and resets state. */
int AIO_ReadPool_closeFile(ReadPoolCtx_t* ctx) {
   FILE* const file = AIO_ReadPool_getFile(ctx);
   AIO_ReadPool_setFile(ctx, NULL);
   return fclose(file);
}

/* AIO_ReadPool_setAsync:
* Allows (de)activating async mode, to be used when the expected overhead
* of asyncio costs more than the expected gains. */
void AIO_ReadPool_setAsync(ReadPoolCtx_t* ctx, int async) {
   AIO_IOPool_setThreaded(&ctx->base, async);
}