 hwwang
|
46b4325dba
|
添加图测试文件
|
9 months ago |
|
hwwang
|
5711145b3b
|
修改hash表和sandbox_request的互斥锁为读写锁,但现在觉得map不加锁可能会有隐患
|
9 months ago |
|
hwwang
|
1854297530
|
将沙盒输出挂载在一个链表中,修改锁的范围,但是感觉结果不是很好
|
10 months ago |
|
hwwang
|
bae018314f
|
修改数据拷贝方式,关于50%的响应获取问题还有待商榷
|
10 months ago |
|
hwwang
|
0f9b75d784
|
asd
|
10 months ago |
|
hwwang
|
8ae955fbfc
|
修改内存释放错误的bug
|
10 months ago |
|
hwwang
|
702f5de60d
|
为hash表添加锁
|
10 months ago |
|
hwwang
|
0b1fb6c5d7
|
修一部分logbug
|
10 months ago |
|
hwwang
|
7abc07445d
|
修改log参数,未修改完
|
10 months ago |
|
hwwang
|
3aeed6f94b
|
已添加DAG,需要修改各部份时间测试代码
|
10 months ago |
|
hwwang
|
70fdbb348c
|
need modify thirdparty hashmap
|
11 months ago |
|
hwwang
|
73da853ad4
|
main
|
12 months ago |
|
hwwang
|
a6e464db44
|
add test
|
12 months ago |
|
hwwang
|
75bf451df0
|
modify module to tree
|
12 months ago |
xiaosuGW
|
9559c92c68
|
submit changes for workshop paper
|
3 years ago |
lyuxiaosu
|
9b97152632
|
apply traiged sending sig alarm to other worker thread for SRSF
|
3 years ago |
|
lyuxiaosu
|
f2853d2783
|
record each thread worker's queuelength when an alarm signal come in
|
3 years ago |
|
lyuxiaosu
|
d831af5615
|
expose cpu speed as an environment value
|
3 years ago |
|
xiaosuGW
|
1ac9e5a881
|
set local runqueue's maximum queuelength is 10240
|
4 years ago |
|
xiaosuGW
|
3bf42c4b6c
|
set maximum global queuelength is 40960
|
4 years ago |
|
xiaosuGW
|
be2e0c28b0
|
comment log
|
4 years ago |
|
xiaosuGW
|
9c8e7f0b88
|
make running_time more accurate
|
4 years ago |
|
xiaosuGW
|
1eceb58278
|
remove assert(current_sandbox->state == SANDBOX_RUNNING from software_interrupt.c and substract scheduling time from running time
|
4 years ago |
|
xiaosuGW
|
d1af64845b
|
if SIGINT is sent by kill, it's also valid
|
4 years ago |
|
xiaosuGW
|
9740e08525
|
remove debug log
|
4 years ago |
|
xiaosuGW
|
7005dd1467
|
A simple solution to fix bug: sledge reply 200 even though the function code return a negative value
|
4 years ago |
|
xiaosuGW
|
9ff138f0f8
|
fix bug: execution time unit is ms and deadline is cpu cycles
|
4 years ago |
|
xiaosuGW
|
66b20d20c0
|
add workload counter for each worker thread
|
4 years ago |
|
xiaosuGW
|
42e3628a10
|
fix bug: miss one module when calcuate the total execution time of the chain
|
4 years ago |
|
xiaosuGW
|
bf306fc0c9
|
fix assert failed problem: assert failed at current_sandbox is not in RUNNING_STATE
|
4 years ago |
|
xiaosuGW
|
e2104ecf97
|
recover assert(\!listener_thread_is_running())
|
4 years ago |
|
xiaosuGW
|
4184fe9384
|
comment mem_log to fix segment fault
|
4 years ago |
|
xiaosuGW
|
e70c74cfbd
|
remove + to fix compiling error
|
4 years ago |
|
xiaosuGW
|
7d7e28853c
|
add more debug log
|
4 years ago |
|
xiaosuGW
|
f53255bbf3
|
fix bug: negative remaining slack turns to be a large number when assign it to a uint64_t type of value
|
4 years ago |
|
xiaosuGW
|
c64a45b214
|
fix one bug of SRSF
|
4 years ago |
|
xiaosuGW
|
5af9fffd55
|
modify priority function for srsf: if remaining slack is negative, return 0 priority
|
4 years ago |
|
xiaosuGW
|
9ab4001f8d
|
handle some unfeasible assert() and negative remaining slack vs the return type of uint64_t of priority function
|
4 years ago |
|
xiaosuGW
|
c57cfe0b20
|
modifiy main.c to support SRSF
|
4 years ago |
|
xiaosuGW
|
da7f5506d6
|
implement srsf
|
4 years ago |
|
xiaosuGW
|
5a03420071
|
replace "tail_module" with "tail-module" in json file parse
|
4 years ago |
|
xiaosuGW
|
0862bd248a
|
hardcode runtime_processor_speed_MHz with the maximum value of CPU speed, this is the temporary solution
|
4 years ago |
|
xiaosuGW
|
416b1ae80b
|
1. move operations for perf_window out of the admission control macro. 2. memory printf the execution times when window is full
|
4 years ago |
|
xiaosuGW
|
44e4026df2
|
implement multiple linear chain
|
4 years ago |
|
xiaosuGW
|
b648b62bfe
|
replace "assert" with "return" if FILE pointer is NULL in memlogging.c
|
4 years ago |
|
lyuxiaosu
|
6f1c4b2901
|
remove assert(runtime_preemption_enabled) since the signal might be SIGINT
|
4 years ago |
|
lyuxiaosu
|
9a1ae46a1a
|
recover creating 100M memory as init for logging
|
4 years ago |
|
lyuxiaosu
|
ecec9fbdcf
|
1. recover to w mode to fopen the perf log file. 2. enrich comments
|
4 years ago |
|
xiaosuGW
|
b4a5fc7af2
|
1.Implement memory-based logging for each worker thread. 2. Dump all memory-based log to a file when receiving SIGINT signal
|
4 years ago |
|
xiaosuGW
|
417427ab39
|
1. optimization: avoid global queue if the next sandbox has the same deadline with the current sandbox. 2. correct format and enrich comments
|
4 years ago |
