CS代考 Operating Systems – CSCI 402 – cscodehelp代写

Operating Systems – CSCI 402
4.1 A Simple System (Monolithic Kernel)
A Framework for Devices
Low-level Kernel (will come back to talk about this after Ch 7) Processes & Threads
Storage Management (will come back to talk about this after Ch 5)
1
321 0
Copyright ý . T100
Operating Systems – CSCI 402
Computer Terminal
2
321 0
Copyright ý . Systems – CSCI 402
A “tty”
3
321 0
Copyright ý . Systems – CSCI 402
Devices
Challenges in supporting devices
device independence device discovery
Device naming
two choices
independent name space (i.e., named independently from other things in the system)
devices are named as files
4
321 0
Copyright ý . Cheng

a special file does not contain data
it refers to devices by their major and minor device numbers
if you do “ls -l”, you can see the device numbers
Data structure in the early Unix systems
statically allocated array in the kernel called cdevsw
(character device switch)
321 0
Operating Systems – CSCI 402
A Framework for Devices
Device driver:
every device is identified by a device “number”, which is actually a pair of numbers
a major device number – identifies the device driver a minor device number – device index for all devices managed by the same device driver
Special entries were created in the file system to refer to devices
usually in the /dev directory
e.g., /dev/disk1, /dev/disk2 each marked as a special file
5
Copyright ý . Devices
Operating Systems – CSCI 402
unix
etc
home
pro
dev
disk1
disk2

1 major = 6 2
device number: minor = 1
0
3 4
entries in cdevsw contains addresses of the device driver entry points
a device driver maintains
its own data structure
cdevsw
read entry point write entry point mmap

321 0
6
Copyright ý . Systems – CSCI 402
Device Drivers in Early Unix Systems
The kernel was statically configured to contain device-specific information such as:
interrupt-vector locations
locations of device-control registeres on whatever bus the device was attached to
Static approach was simple, but cannot be easily extended
a kernel must be custom configured for each installation
7
321 0
Copyright ý . Systems – CSCI 402
Device Probing
First step to improve the old way
allow the devices to to be found and automatically configured when the system booted
(still require that a kernel contain all necessary device drivers)
Each device driver includes a probe routine
invoked at boot time
probe the relevant buses for devices and configure them
including identifying and recording interrupt-vector and device-control-register locations
This allowed one kernel image to be built that could be useful for a number of similar but not identical installations
boot time is kind of long
impractical as the number of supported devices gets big
8
321 0
Copyright ý . Systems – CSCI 402
Device Probing
What¡¯s the right thing to do?
Step 1: discover the device without the benefit of having the
relevant device driver in the kernel
Step 2: find the needed device drivers and dynamically link
them into the kernel
but how do you achieve this?
Solution: use meta-drivers
a meta-drive handles a particular kind of bus e.g., USB (Universal Serial Bus)
a USB meta-driver is installed into the kernel
any device that goes onto a USB (Universal Serial Bus)
must know how to interact with the USB meta-driver via the USB protocol
once a connected device is identified, system software would select the appropriate device driver and load into the kernel what about applications? how can they reference
dynamically discovered devices?
9
321 0
Copyright ý . Cheng

some current Linux systems use udev
user-level application assigns names based on rules provided by an administrator
321 0
Operating Systems – CSCI 402
Discovering Devices
So, you plug in a new device to your computer on a particular bus
OS would notice find a device driver
what kind of device is it?
where is the driver?
assign a name, but how is it chosen?
multiple similar devices, but how does application choose?
In some Linux systems, entries are added into /dev as the kernel discovers them
lookup the names from a database of names known as devfs downside of this approach is that device naming conventions not universally accepted
what¡¯s an application to do?
10
Copyright ý . Systems – CSCI 402
Discovering Devices
What about the case where different devices acted similarly?
e.g., touchpad on a laptop and USB mouse
how should the choice be presented to applications?
Windows has the notion of interface classes
a device can register itself as members of one or more such classes
an application can enumerate all currently connected members of such a class and choose among them (or use them all)
11
321 0
Copyright ý . Systems – CSCI 402
4.1 A Simple System (Monolithic Kernel)
A Framework for Devices
Low-level Kernel (will come back to talk about this after Ch 7) Processes & Threads
Storage Management (will come back to talk about this after Ch 5)
12
321 0
Copyright ý . process is:
Processes and Threads
a holder for an address space
a collection of other information shared by a set of threads a collection of references to open files and other “execution context”
As discussed in Ch 1, processes related APIs include
fork(), exec(), wait(), exit()
Operating Systems – CSCI 402
13
321 0
Copyright ý . and Threads
Operating Systems – CSCI 402
address space description
open file descriptors
list of threads
current state
stack pointer stack pointer
stack pointer
other registers other registers
other registers
state state
state
thread control blocks
process control block
Note: all these are relevant to your Kernel Assignment 1
although we are only doing one thread per process Copyright ý . Cheng
stacks
14
321 0

Pretty simple
Process Life Cycle
a process starts in the run state
Run Zombie
Operating Systems – CSCI 402
15
321 0
Copyright ý . Relationships (1)
Process hierarchy
run “pstree” on Linux
Init
Operating Systems – CSCI 402
cmd
Login 1 Login 2
cmd cmd cmd
Login 3
cmd
Sub proc. Sub proc.
16
321 0
Copyright ý . Relationships (2)
If a process dies, you must reparent all its child processes Init
Operating Systems – CSCI 402
cmd
Login 1 Login 2
cmd cmd cmd
Login 3
cmd
Sub proc. Sub proc.
17
321 0
Copyright ý . Relationships (3)
If a process dies, you must reparent all its child processes new parent is the INIT process
Operating Systems – CSCI 402
cmd
Init
Login 1 Login 2
cmd cmd cmd
Login 3
cmd
Sub proc. Sub proc.
18
321 0
Copyright ý . and Threads
fork
Or fork
Solaris uses the 2nd approach
expensive to fork a process
Problem with 1st approach
thread 1 called fork() and thread 2 has a mutex locked who will unlock the mutex?
POSIX solution is to provide a way to unlock all mutex
before fork() Copyright ý . Cheng
321 0
19
Operating Systems – CSCI 402
12

Running
Operating Systems – CSCI 402
Thread Life Cycle
scheduler
thread itself
thread itself
Runnable
Terminated
other
Waiting
a thread starts in the runnable state
sleeps in the run queue (or “ready queue”)
threads sleep in the run queue to wait to use the CPU
20
321 0
Copyright ý .
Operating Systems – CSCI 402
Thread Life Cycle
scheduler
thread itself
thread itself
Runnable
Terminated
other
Waiting
the scheduler switches a thread¡¯s state from runnable to running
21
321 0
Copyright ý . Life Cycle
Operating Systems – CSCI 402
Running
scheduler
thread itself
thread itself
Runnable
Terminated
other
the scheduler is not involved here
Waiting
a thread goes from running to waiting when a blocking call is made by the thread itself
22
321 0
Copyright ý . Life Cycle
Operating Systems – CSCI 402
Running
scheduler
thread itself
thread itself
Runnable
Terminated
other
Waiting
the scheduler switches a thread¡¯s state from running to runnable when the thread used up its execution quantum
a thread can also “yield” the CPU (see examples in faber_thread_test() in kernel 1)
23
321 0
Copyright ý . Life Cycle
Operating Systems – CSCI 402
Running
scheduler
thread itself
thread itself
Runnable
Terminated
other
the scheduler is not involved here
Waiting
a thread gets unblocked by the action of another thread or by an interrupt handler
24
321 0
Copyright ý .
Operating Systems – CSCI 402
Thread Life Cycle
scheduler
thread itself
thread itself
Runnable
Terminated
other
Waiting
in order for a thread to enter the terminated state, it has to be in the running state just before that
what if something like pthread_cancel() is invoked when the thread is not in the running state?
321 0
25
Copyright ý . Systems – CSCI 402
Thread Life Cycle
Does pthread_exit() delete the thread (completely) that calls it? no, the thread goes into a zombie state (i.e., “terminated”)
What¡¯s left in the thread after it calls pthread_exit()? its thread control block
needs to keep thread ID and return code around
its stack
how can a thread delete its own stack? no way! which stack are we talking about anyway?
26
321 0
Copyright ý . Systems – CSCI 402
Thread Life Cycle
Who is deleting the thread control block and freeing up the thread¡¯s stack space?
If a thread is not detached
it can be taken care of in the pthread_join() code
the thread that calls pthread_join() does the clean up
If a thread is detached (our simple OS does not support this)
can do this is one of two ways
1) use a special reaper thread
basically doing pthread_join()
2) queue these threads on a list and have other threads
free them when it¡¯s convenient (e.g., when the scheduler schedule a thread to run)
27
321 0
Copyright ý . Systems – CSCI 402
Kernel 1 Process & Thread Life Cycles
Part of the kernel 1 assignment is to implement the life cycles of processes and threads
process/thread creation/termination
since we are only doing one thread per process (MTP=0),
when a thread dies, the process must die as well
process/thread cancellation
process waiting (and no thread joining since MTP=0) etc.
Unlike warmup2, in kernel assignments, first procedures of almost all kernel threads have been written for you already!
the thread code there make function calls and some of these functions are not-yet-implemented
your job is to implement those functions so that these kernel threads can run perfectly
321 0
28
Copyright ý . Systems – CSCI 402
Kernel 1 Process & Thread Life Cycles
Hint on how to do this is by reading kernel code
read the code in “kernel/proc/faber_test.c”
if it calls a function that you are suppose to implement, it¡¯s telling you what it¡¯s expecting from that function!
feel free to discuss things like that in the class Google
Group
you need to understand what every line of code is doing there you need to pass every test there (see grading guidelines)
you must not change anything there
make sure the printout is correct (you may want to discuss it in the class Google Group)
if you need to do something similar in another module, just copy the code from it
you can copy code that¡¯s given to you as course material and you don¡¯t have to cite your source
321 0
29
Copyright ý . Cheng

Leave a Reply

Your email address will not be published. Required fields are marked *