ilo-pona/xv6-65oo2
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Activity

x

Browse source
This commit is contained in:
Frans Kaashoek 2019-07-30 13:07:17 -04:00
parent 848d1906e8
commit 9c4f62e8e3
1 changed files with 16 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

18
labs/lock.html
View file

@ -98,7 +98,7 @@ workloads.
and run test0 from bcachetest and you will see "!"s. and run test0 from bcachetest and you will see "!"s.
<p>Modify <tt>bget</tt> so that a lookup for a buffer that is in the <p>Modify <tt>bget</tt> so that a lookup for a buffer that is in the
bcache doesn't need to acquire <tt>bcache.lock</tt>. This more bcache doesn't need to acquire <tt>bcache.lock</tt>. This is more
tricky than the kalloc assignment, because bcache buffers are truly tricky than the kalloc assignment, because bcache buffers are truly
shared among processes. You must maintain the invariant that a shared among processes. You must maintain the invariant that a
buffer is only once in memory. buffer is only once in memory.
@ -121,8 +121,13 @@ against, including:
<p>Here are some hints: <p>Here are some hints:
<ul> <ul>
<li>Read the description of buffer cache in the xv6 book (Section 7.2).
<li>Use a simple design: i.e., don't design a lock-free implementation. <li>Use a simple design: i.e., don't design a lock-free implementation.
<li>Use a simple hash table with locks per bucket <li>Use a simple hash table with locks per bucket.
<li>Searching in hash table for a buffer and allocating an entry
for that buffer when the buffer is not found must be atomic.
<li>It is fine to acquire <tt>bcache.lock</tt> in <tt>brelse</tt>
to update the LRU/MRU list.
</ul> </ul>
<p>Check that your implementation has less contention <p>Check that your implementation has less contention
@ -130,5 +135,14 @@ against, including:
<p>Make sure your implementation passes bcachetest and usertests. <p>Make sure your implementation passes bcachetest and usertests.
<p>Optional:
<ul>
<li>make the buffer cache more scalable (e.g., avoid taking
out <tt>bcache.lock</tt> on <tt>brelse</tt>).
<li>make lookup lock-free (Hint: use gcc's <tt>__sync_*</tt>
functions.) How do you convince yourself that your implementation is correct?
</ul>
</body> </body>
</html> </html>