# Diskussion:Lösungsvorschlag Operating Systems And Computer Networks FS16

Wechseln zu: Navigation, Suche

## 2 Scheduling

### c)

The question implies that RMS finds a satisfiable schedule, but ${\displaystyle \sum _{i=1}^{4}{\frac {c_{i}}{p_{i}}}={\frac {1}{3}}+{\frac {1}{15}}+{\frac {2}{6}}+{\frac {2}{11}}\approx 0.95>4\cdot (2^{\frac {1}{4}}-1)\approx 0.76}$. Am I missing something? --Thedot (Diskussion) 19:13, 26. Jun. 2017 (CEST)

Note: As far as I remember: If the condition of this mathematical calculation finds a solution it is guaranteed that there is a feasible schedule. But if the condition of the calculation is not satisfied, that does not imply that there is not a feasible schedule. So the Calculation is a sufficient and not necessary condition. If you try to find a schedule by hand according to the rules, you will see that it is possible. (Jérémy Scheurer)

## 3 Memory Management

### d)

Maybe it is an error that VPO != PFO for the first translation.

## 4 Filesystems/IO

### b)

One Element (data block) is 1024 B. Four Elements are 4096 B. (byte_offset / 1024) should be the index of the Element. Why is the solution dividing by four for the direct block? I think it should be

int i = byte_offset / 1024; // index of data block

if (i < 4) {
return in->direct_block[i];
}


One Data block has a size of 4 KiB, not 1 KiB, as it says in the description. (Antolos)

### c)

Isn't it the other way around? Meaning, column-wise ACL is a list for each object and the entries of those lists are principals and their rights over that object. Row-wise would be a list for each principal with entries being the objects they have any rights over and what rights those are. We define the access control matrix to have a row for each principal and a column for each object.

## 5 Knowledge Bank

5.5 According to the solution of exercise 4 of 2018, (i) is not correct. no!: According to the solution of this exercise (i), (iii) and (iv) are correct! See: [1]; Not my fault 🤷🏼‍♂️ net-eth.slack.com
5.7. a is correct because ping definitely requires the use of icmp. See: [2], [3],[4]
5.7 c is questionable because traceroute itself doesn't require the user to use the icmp protocol. In fact the standard implementation of traceroute on linux or mac uses UDP by default. However traceroute relies on icmp responses from the hosts where the ttl runs out. See: [5], [6]

## 8 Layer 3

### 2.

Why did you change 1.2.1.0/26 to /27? Why did you change 1.2.5.0/24 to /29? I get that this is more specific, but is that neccessary? --Minker (Diskussion) 10:30, 9. Aug. 2017 (CEST)
I would say, it is not necessary, because the routers probably use shortest-matching-prefix routing anyway. So overlapping prefixes are not a problem (except if they have the same length?). -- Antolos , 29. Aug. 2017

## 9 Layer 4

### b)

2. Shouldn't it be [16, 22], and [26, 32] instead of [17, 22], and [27, 32]? -- 2639, 25. Jul. 2018