jp9000/obs-studio
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

libobs: Implement and use better scaling function for 64-bit integers

Browse Source 
As os_gettime_ns() gets large the current scaling methods, mostly by casting
to uint64_t, may lead to numerical overflows. Sweep the code and use
util_mul_div64() where applicable.

Signed-off-by: Hans Petter Selasky <hps@selasky.org>
This commit is contained in:
Hans Petter Selasky
2020-03-21 10:55:12 +01:00
parent 47058d9b69
commit 0e4ea14ba1
22 changed files with 97 additions and 57 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
UI/window-basic-auto-config-test.cpp
View File

@@ -4,6 +4,7 @@
#include <obs.hpp>
#include <util/platform.h>
#include <util/util_uint64.h>
#include <graphics/vec4.h>
#include <graphics/graphics.h>
#include <graphics/math-extra.h>
@@ -413,12 +414,14 @@ void AutoConfigTestPage::TestBandwidthThread()
cv.wait(ul);
uint64_t total_time = os_gettime_ns() - t_start;
if (total_time == 0)
total_time = 1;
int total_bytes =
(int)obs_output_get_total_bytes(output) - start_bytes;
uint64_t bitrate = (uint64_t)total_bytes * 8 * 1000000000 /
total_time / 1000;
uint64_t bitrate = util_mul_div64(
total_bytes, 8ULL * 1000000000ULL / 1000ULL,
total_time);
if (obs_output_get_frames_dropped(output) ||
(int)bitrate < (wiz->startingBitrate * 75 / 100)) {
server.bitrate = (int)bitrate * 70 / 100;