lostecho/ostep-projects
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

clarity for return codes

Browse Source
This commit is contained in:
Remzi Arpaci-Dusseau
2018-04-08 14:56:36 -05:00
parent 62bd769180
commit c05a2e3482
1 changed files with 12 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
concurrency-xv6-threads/README.md
View File

@@ -36,23 +36,25 @@ share the address space with this process. It should also free the address
space if this is last reference to it. Also, `exit()` should work as before space if this is last reference to it. Also, `exit()` should work as before
but for both processes and threads; little change is required here. but for both processes and threads; little change is required here.
Your thread library will be built on top of this, and just have a simple Your thread library will be built on top of this, and just have a simple `int
`thread_create(void (*start_routine)(void *, void *), void *arg1, void *arg2)` thread_create(void (*start_routine)(void *, void *), void *arg1, void *arg2)`
routine. This routine should call `malloc()` to create a new user stack, use routine. This routine should call `malloc()` to create a new user stack, use
`clone()` to create the child thread and get it running. A `thread_join()` `clone()` to create the child thread and get it running. It returns the newly
call should also be created, which calls the underlying `join()` system call, created PID to the parent and 0 to the child (if successful), -1 otherwise.
frees the user stack, and then returns. An `int thread_join()` call should also be created, which calls the underlying
`join()` system call, frees the user stack, and then returns. It returns the
waited-for PID (when successful), -1 otherwise.
Your thread library should also have a simple *ticket lock* (read [this book Your thread library should also have a simple *ticket lock* (read [this book
chapter](http://pages.cs.wisc.edu/~remzi/OSTEP/threads-locks.pdf) for more chapter](http://pages.cs.wisc.edu/~remzi/OSTEP/threads-locks.pdf) for more
information on this). There should be a type `lock_t` that one uses to declare information on this). There should be a type `lock_t` that one uses to declare
a lock, and two routines `lock_acquire(lock_t *)` and `lock_release(lock_t a lock, and two routines `void lock_acquire(lock_t *)` and `void
*)`, which acquire and release the lock. The spin lock should use x86 atomic lock_release(lock_t *)`, which acquire and release the lock. The spin lock
add to build the lock -- see [this wikipedia should use x86 atomic add to build the lock -- see [this wikipedia
page](https://en.wikipedia.org/wiki/Fetch-and-add) for a way to create an page](https://en.wikipedia.org/wiki/Fetch-and-add) for a way to create an
atomic fetch-and-add routine using the x86 `xaddl` instruction. One last atomic fetch-and-add routine using the x86 `xaddl` instruction. One last
routine, `lock_init(lock_t *)`, is used to initialize the lock as need be (it routine, `void lock_init(lock_t *)`, is used to initialize the lock as need be
should only be called by one thread). (it should only be called by one thread).
The thread library should be available as part of every program that runs in The thread library should be available as part of every program that runs in
xv6. Thus, you should add prototypes to `user/user.h` and the actual code to xv6. Thus, you should add prototypes to `user/user.h` and the actual code to