mirrors/libstb-hal
1
0
Fork 0
mirror of https://github.com/tuxbox-neutrino/libstb-hal.git synced 2025-08-26 23:13:16 +02:00
Code Issues Packages Projects Releases Wiki Activity

spark/record: move buffer allocation outside record thread

Browse Source 
move the allocation of the recording buffer outside the
recording thread, so that allocation failures can be
communicated as "failure to start recording" instantly
This commit is contained in:
Stefan Seyfried
2015-03-07 18:59:04 +01:00
parent ef2df97bde
commit 49a9834ac5
1 changed files with 18 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

32
libspark/record.cpp
View File

@@ -32,6 +32,8 @@
#define lt_debug(args...) _lt_debug(TRIPLE_DEBUG_RECORD, this, args)
#define lt_info(args...) _lt_info(TRIPLE_DEBUG_RECORD, this, args)
#define BUFSIZE (2 << 20) /* 2MB */
static const int bufsize = BUFSIZE;
typedef enum {
RECORD_RUNNING,
@@ -47,6 +49,7 @@ class RecData
public:
RecData(int num) {
dmx = NULL;
buf = NULL;
record_thread_running = false;
file_fd = -1;
exit_flag = RECORD_STOPPED;
@@ -56,6 +59,7 @@ public:
int file_fd;
int dmx_num;
cDemux *dmx;
uint8_t *buf;
pthread_t record_thread;
bool record_thread_running;
record_state_t exit_flag;
@@ -121,7 +125,13 @@ bool cRecord::Start(int fd, unsigned short vpid, unsigned short *apids, int nump
if (posix_fadvise(pd->file_fd, 0, 0, POSIX_FADV_DONTNEED))
perror("posix_fadvise");
i = pthread_create(&pd->record_thread, 0, execute_record_thread, pd);
pd->buf = (uint8_t *)malloc(bufsize);
if (!pd->buf) {
i = errno;
lt_info("%s: unable to allocate read buffer of %d bytes (%m)\n", __func__, bufsize);
}
else
i = pthread_create(&pd->record_thread, 0, execute_record_thread, pd);
if (i != 0)
{
pd->exit_flag = RECORD_FAILED_READ;
@@ -129,6 +139,8 @@ bool cRecord::Start(int fd, unsigned short vpid, unsigned short *apids, int nump
lt_info("%s: error creating thread! (%m)\n", __func__);
delete pd->dmx;
pd->dmx = NULL;
free(pd->buf);
pd->buf = NULL;
return false;
}
pd->record_thread_running = true;
@@ -147,6 +159,10 @@ bool cRecord::Stop(void)
pthread_join(pd->record_thread, NULL);
pd->record_thread_running = false;
if (pd->buf)
free(pd->buf);
pd->buf = NULL;
/* We should probably do that from the destructor... */
if (!pd->dmx)
lt_info("%s: dmx == NULL?\n", __func__);
@@ -223,22 +239,11 @@ void RecData::RecordThread()
{
lt_info("%s: begin\n", __func__);
hal_set_threadname("hal:record");
#define BUFSIZE (2 << 20) /* 2MB */
#define READSIZE (BUFSIZE / 16)
const int bufsize = BUFSIZE;
const int readsize = READSIZE;
const int readsize = bufsize / 16;
int buf_pos = 0;
int queued = 0;
uint8_t *buf;
struct aiocb a;
buf = (uint8_t *)malloc(bufsize);
if (!buf)
{
exit_flag = RECORD_FAILED_MEMORY;
lt_info("%s: unable to allocate buffer! (out of memory)\n", __func__);
}
int val = fcntl(file_fd, F_GETFL);
if (fcntl(file_fd, F_SETFL, val|O_APPEND))
lt_info("%s: O_APPEND? (%m)\n", __func__);
@@ -350,7 +355,6 @@ void RecData::RecordThread()
a.aio_nbytes = queued;
r = aio_write(&a);
}
free(buf);
#if 0
// TODO: do we need to notify neutrino about failing recording?