CS-350代写、C++编程语言代做
CS-350 - Fundamentals of Computing Systems
Homework Assignment #8 - BUILD
Due on November 14, 2024 — Late deadline: November 16, 2024 EoD at 11:59 pm
Prof. Renato Mancuso
Renato Mancuso
1CS-350 - Fundamentals of Computing Systems::Homework Assignment #8 - BUILD Problem 1
BUILD Problem 1
We now have a server capable of performing non-trivial operations on images! All we have to do now is to
make the server multi-threaded. And yes, we already have the infrastructure to spawn multiple threads. So
what’s missing exactly?
Output File: server_mimg.c
Overview. As mentioned above, the main idea is to allow multple workers to perform operations on images
in parallel. Everything you have developed in BUILD 6 will be reused, but we must make sure that when
multiple worker threads are spawned, the correctness of the operations on the images is still guaranteed.
But what could be jeopardizing the correctness of these operations? Let us consider a concrete case. Imaging
the following sequence of requests queued at the server: (1) Retrieve image A, (2) blur image A, (3) detect
the vertical edges of image A, and (4) send back the result of the operations performed on image A.
With only one worker, the operations are carried out in this sequence and the result sent back to the client is
an image with the cumulative operations (2) and (3) correctly applied to the source image. With 2 workers
(unless we ffx our implementation) we could have worker #1 and #2 working in parallel on operations (1)
and (2) with the result of the operations being some weird mix of the two operations.
In this assignment, the goal is to allow safe multi-threading where the semantics of sequential operations on
the images is preserved even if multiple threads are spawned and operate in parallel. For this task, we will
use semaphores to perform inter-thread synchronization.
Design. One of the main problem that we have to solve is un-arbitrated access to shared data structures.
To verify that there is a problem unless we insert synchronization primitives accordingly, start with your (or
my) solution for HW6, rename it appropriately, and enable spawning more than 1 worker threads.
Then, run the following simple experiment. First, run the client to generate the sequence of operations listed
above with 1 worker thread and look carefully at the output report generated by the client:
./server_mimg -q 100 -w 1 2222 & ./client -I images/ -L 1:R:1:0,0:b:1:0,0:v:1:0,0:T:1:0 2222
You will notice that the ffrst hash reported by the client (9f3363f0249c15163d52e60fd9544c31) is simply
the hash of the original test1.bmp image. The second (and last) hash reported by the client is the hash
(00e4fc4b9c7c71ee2ca3946053f78793) of the blur+vertical edge detection operations applied in sequence
to the image. However, if we increase the number of worker to 2, the ffnal hash will be different! For instance,
when running:
./server_mimg -q 100 -w 2 2222 & ./client -I images/ -L 1:R:1:0,0:b:1:0,0:v:1:0,0:T:1:0 2222
The last hash obtained on the reference machine changes to b5932c2bcb0a64121def911286c706e2, but
might be something else entirely on a different machine. Also in some cases, the server crashes entirely.
To solve the problem, the cleanest way is to introduce semaphore-based synchronization between threads.
In order to deffne a semaphore, you should use the type sem_t deffned in semaphore.h. Before a semaphore
can be used, it must be initialized.
This can be done with the sem_init(sem_t * semaphore, int pshared, unsigned int init_value).
Here, semaphore is pointer to the semaphore to be initialized, pshared can be set to 0, and init_value is
a non-negative initialization value of the semaphore, following the semantics we have covered in class.
Once your semaphore has been correctly initialized (make sure to check for the error value of the sem_init(...)
call!), the wait and signal operations can be performed over it, following the semantics we have discussed
Problem 1 continued on next page. . . 2CS-350 - Fundamentals of Computing Systems::Homework Assignment #8 - BUILD Problem 1 (continued)
in class. To wait on a semaphore, you must use the sem_wait(sem_t * semaphore) call; to signal on a
semaphore, you must use the sem_post(sem_t * semaphore).
Shared Data Structures. Of course, the main question is what data structures must be protected?. Here
is a list of things that can be problematic, but your own implementation could be different, so try to map
the statement below to your own code.
(1) Image Objects: One obvious place to start is to protect the image objects that are registered with
the server and upon which operations are requested by the client. We want to prevent different worker
to simultaneously change the content of an image, so a good idea is to introduce one semaphore per each
registered image! These must be created and/or initialized dynamically at image registration time.
(2) Image Registration Array: Another shared data structure is the global array of registered images.
Concurrent operations over that array is not a good idea, so all the threads will need to synchronize when
trying to access that shared data structure.
(3) Connection Socket: What? The connection socket has always been shared, so why is that a problem
now? The truth is that it has always been a problem, but we did not care because the responses from
the workers to the client were always a one-shot send(...) operation. But now, there are cases where the
server follows a two-step approach in the protocol it follows with the client. For instance, when handling an
IMG_RETRIEVE operation, a worker ffrst provides a positive acknowledgment of completed request and then
the payload of the image being retrieved. What if another worker starts sending other data while a retrieve
operation is in progress? Careful: the same goes for the parent when handling IMG_REGISTER operations.
(4) Request Queue and STDOUT Console: We already know that the shared request queue and the
shared STDOUT console require the use of semaphores to ensure correctness. Perhaps take inspiration from
the use of semaphores in those cases to handle the other shared data structures listed above.
Desired Output. The expected server output is pretty much what you already constructed in HW6. Here
is it summarized again for reference. You should print queue status dumps, rejection and completion notices.
Queue status dumps and rejection notice are identical in format to HW5 and HW6. Once again, the queue
dump status is printed when any of the worker threads completes processing of any of the requests.
Just like HW6, when a request successfully completes service, the thread ID of the worker thread that has
completed the request will need to be added at the beginning of the line following the format below. You can
assign thread ID = (number of workers + 1) to the parent thread. If multiple worker threads are available to
process a pending request, any one of them (but only at most one!) can begin processing the next request.
T R:,,,,,,
,
Here, is a string representing the requested operation over an image. For instance, if the operation
was IMG_REGISTER, then the server should output the string “IMG REGISTER” (no quotes) for this ffeld.
should just be 0 or 1, depending on what the client requested. should be
the image ID for which the client has requested an operation. If the server is ignoring any of these values
in the response, set these ffelds to 0. Finally, should report the image ID on which the
server has performed the operation requested by the client. Recall that this might be different from what
sent by the client if overwrite = 0 in the client’s request, but it must be the same if overwrite = 1.
Additional Help. You might have noticed, from the commands recommended above, that the client (v4.2)
now allows you to deffne a script of image operation requests. This is useful to test the correctness of your
server under a controlled workload.
To use this feature, you should still provide the path to the folder containing the test images using the
-I parameter. Next, you should also provide the -L
parameter, where the is a comma-separated list of image operations with the
following format:
Problem 1 continued on next page. . . 3CS-350 - Fundamentals of Computing Systems::Homework Assignment #8 - BUILD Problem 1 (continued)

热门主题

课程名

mktg2509 csci 2600 38170 lng302 csse3010 phas3226 77938 arch1162 engn4536/engn6536 acx5903 comp151101 phl245 cse12 comp9312 stat3016/6016 phas0038 comp2140 6qqmb312 xjco3011 rest0005 ematm0051 5qqmn219 lubs5062m eee8155 cege0100 eap033 artd1109 mat246 etc3430 ecmm462 mis102 inft6800 ddes9903 comp6521 comp9517 comp3331/9331 comp4337 comp6008 comp9414 bu.231.790.81 man00150m csb352h math1041 eengm4100 isys1002 08 6057cem mktg3504 mthm036 mtrx1701 mth3241 eeee3086 cmp-7038b cmp-7000a ints4010 econ2151 infs5710 fins5516 fin3309 fins5510 gsoe9340 math2007 math2036 soee5010 mark3088 infs3605 elec9714 comp2271 ma214 comp2211 infs3604 600426 sit254 acct3091 bbt405 msin0116 com107/com113 mark5826 sit120 comp9021 eco2101 eeen40700 cs253 ece3114 ecmm447 chns3000 math377 itd102 comp9444 comp(2041|9044) econ0060 econ7230 mgt001371 ecs-323 cs6250 mgdi60012 mdia2012 comm221001 comm5000 ma1008 engl642 econ241 com333 math367 mis201 nbs-7041x meek16104 econ2003 comm1190 mbas902 comp-1027 dpst1091 comp7315 eppd1033 m06 ee3025 msci231 bb113/bbs1063 fc709 comp3425 comp9417 econ42915 cb9101 math1102e chme0017 fc307 mkt60104 5522usst litr1-uc6201.200 ee1102 cosc2803 math39512 omp9727 int2067/int5051 bsb151 mgt253 fc021 babs2202 mis2002s phya21 18-213 cege0012 mdia1002 math38032 mech5125 07 cisc102 mgx3110 cs240 11175 fin3020s eco3420 ictten622 comp9727 cpt111 de114102d mgm320h5s bafi1019 math21112 efim20036 mn-3503 fins5568 110.807 bcpm000028 info6030 bma0092 bcpm0054 math20212 ce335 cs365 cenv6141 ftec5580 math2010 ec3450 comm1170 ecmt1010 csci-ua.0480-003 econ12-200 ib3960 ectb60h3f cs247—assignment tk3163 ics3u ib3j80 comp20008 comp9334 eppd1063 acct2343 cct109 isys1055/3412 math350-real math2014 eec180 stat141b econ2101 msinm014/msing014/msing014b fit2004 comp643 bu1002 cm2030
联系我们
EMail: 99515681@qq.com
QQ: 99515681
留学生作业帮-留学生的知心伴侣！
工作时间：08:00-21:00
python代写
微信客服：codinghelp
站长地图