There's no need to use u64 over there. In a matter of fact, the div is not even needed, as it is multiplying by 1000 and dividing by 1000.
So, simplify the logic. While here, constrain the buffer size to a certain range (between the current value and 10 times it) Signed-off-by: Mauro Carvalho Chehab <[email protected]> --- .../media/test-drivers/vidtv/vidtv_bridge.c | 20 ++++++++++--------- 1 file changed, 11 insertions(+), 9 deletions(-) diff --git a/drivers/media/test-drivers/vidtv/vidtv_bridge.c b/drivers/media/test-drivers/vidtv/vidtv_bridge.c index fe4e496acc34..74b054947bbe 100644 --- a/drivers/media/test-drivers/vidtv/vidtv_bridge.c +++ b/drivers/media/test-drivers/vidtv/vidtv_bridge.c @@ -106,26 +106,28 @@ static unsigned int mux_buf_sz_pkts; module_param(mux_buf_sz_pkts, uint, 0); MODULE_PARM_DESC(mux_buf_sz_pkts, "Size for the internal mux buffer in multiples of 188 bytes"); +#define MUX_BUF_MIN_SZ 90164 +#define MUX_BUF_MAX_SZ (MUX_BUF_MIN_SZ * 10) + static u32 vidtv_bridge_mux_buf_sz_for_mux_rate(void) { - u64 max_elapsed_time_msecs = VIDTV_MAX_SLEEP_USECS / 1000; + u32 max_elapsed_time_msecs = VIDTV_MAX_SLEEP_USECS / USEC_PER_MSEC; u32 nbytes_expected; u32 mux_buf_sz = mux_buf_sz_pkts * TS_PACKET_LEN; - u32 slack; - nbytes_expected = div64_u64(mux_rate_kbytes_sec * 1000, MSEC_PER_SEC); + nbytes_expected = mux_rate_kbytes_sec; nbytes_expected *= max_elapsed_time_msecs; mux_buf_sz = roundup(nbytes_expected, TS_PACKET_LEN); - slack = mux_buf_sz / 10; + mux_buf_sz += mux_buf_sz / 10; - //if (mux_buf_sz < MUX_BUF_MIN_SZ) - // mux_buf_sz = MUX_BUF_MIN_SZ; + if (mux_buf_sz < MUX_BUF_MIN_SZ) + mux_buf_sz = MUX_BUF_MIN_SZ; - //if (mux_buf_sz > MUX_BUF_MAX_SZ) - // mux_buf_sz = MUX_BUF_MAX_SZ; + if (mux_buf_sz > MUX_BUF_MAX_SZ) + mux_buf_sz = MUX_BUF_MAX_SZ; - return mux_buf_sz + slack; + return mux_buf_sz; } static bool vidtv_bridge_check_demod_lock(struct vidtv_dvb *dvb, u32 n) -- 2.26.2
