[GSOC] HAMMER2 compression feature week3 report

[Daniel Flores <daniel5555@xxxxxxxxx>]

--089e0149397acb424604e0d9b14a
Content-Type: text/plain; charset=windows-1252
Content-Transfer-Encoding: quoted-printable

Hello everyone,

Here is my report for third week. This week I started to actually implement
LZ4 compression in HAMMER2 file system. To do so, I need to significantly
change writing and reading internal functions for cases when the
compression function is set. The reason is that the current write path and
read path are not prepared for compression of data inside them, so I'll
have to implement them differently. This will also help for the future
encryption function, because, I think, the same paths will serve for that
too.

Currently I=92m working on write path. It doesn't work at this point, becau=
se
I'm doing something really wrong and the system crashes, but hopefully I'll
will find what the problem is and fix it soon. For now it =93assumes=94 tha=
t
the compression always fails and writes data as it is. Then I'll have to
implement a read path which, similarly, will =93assume=94 that compression
failed and will simply read the data as it is.

When both paths will be implemented correctly, it should be very simple to
insert there the compression/decompression functions. At least, it should
be straightforward.

Hopefully, during next week at some point I will have the write path
working.

All the code related to this work is now in my repository, branch
=93hammer2_LZ4=94 [1].

I'll appreciate any suggestions and criticism.


Daniel

[1] git://leaf.dragonflybsd.org/~iostream/dragonfly.git

--089e0149397acb424604e0d9b14a
Content-Type: text/html; charset=windows-1252
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">

<p>Hello everyone,</p><p>Here is my report for third week. This week I star=
ted to actually implement LZ4 compression in HAMMER2 file system. To do so,=
 I need to significantly change writing and reading internal functions for =
cases when the compression function is set. The reason is that the current =
write path and read path are not prepared for compression of data inside th=
em, so I&#39;ll have to implement them differently. This will also help for=
 the future encryption function, because, I think, the same paths will serv=
e for that too.</p>
<p>Currently I=92m working on write path. It doesn&#39;t work at this point=
, because I&#39;m doing something really wrong and the system crashes, but =
hopefully I&#39;ll will find what the problem is and fix it soon. For now i=
t =93assumes=94 that the compression always fails and writes data as it is.=
 Then I&#39;ll have to implement a read path which, similarly, will =93assu=
me=94 that compression failed and will simply read the data as it is.</p>
<p>When both paths will be implemented correctly, it should be very simple =
to insert there the compression/decompression functions. At least, it shoul=
d be straightforward.</p><p>Hopefully, during next week at some point I wil=
l have the write path working.</p>
<p>All the code related to this work is now in my repository, branch =93ham=
mer2_LZ4=94 [1].</p><p>I&#39;ll appreciate any suggestions and criticism.</=
p><p><br></p><p>Daniel</p><p>[1] git://<a href=3D"http://leaf.dragonflybsd.=
org/~iostream/dragonfly.git">leaf.dragonflybsd.org/~iostream/dragonfly.git<=
/a></p>


</div>

--089e0149397acb424604e0d9b14a--