Skip to contentSkip to conversation

cryptoapi compression modules & JFFSx

DISCUSSION Kluba, Patrik 3 participants 3 messages Jun 23, 2005 – Jun 24, 2005

lore
Share & tools
lore
#0 lore raw

Hi everybody,

I'm going to port JFFS2's compression modules to CryptoApi except {in|de}flate, which Artem is working(?) on.
I've noticed that the pcompress thing (slen <-> *slen and partial compression which about a discussion was on the list) is in Herbert's repository. Does it mean that it will get into the kernel once? I just would like to be sure whether should I implement pcompress or not.

The second thing is that we would like to use CryptoApi from user space. This way it won't be necessary to reimplement compression algorithms in user space filesystem image creation programs (mkfs.jffsx), and it would make using & distributing closed-source proprietary compression methods easier. There's a patch at http://www.logix.cz/michal/devel/cryptodev/, written by Michal Ludvig, which adds a /dev/crypto device for this purpose, as on *BSD. Is there a chance that this can get into the kernel?

Regards,

Patrik Kluba

#1 lore raw
Quoted 7 lines · Patrik

I'm going to port JFFS2's compression modules to CryptoApi except {in|de}flate, which Artem is working(?) on.
I've noticed that the pcompress thing (slen <-> *slen and partial compression which about a discussion was on the list) is in Herbert's repository. Does it mean that it will get into the kernel once? I just would like to be sure whether should I implement pcompress or not.

Well I've removed it actually because Artem said he didn't need it anymore.

However, if you can provide an implementation of pcompress for deflate that's generic enough then I'm happy to put it back. By genericity I mean not making assumptions such as the input buffer size being less than 4K.

Quoted 8 lines

The second thing is that we would like to use CryptoApi from user space. This way it won't be necessary to reimplement compression algorithms in user space filesystem image creation programs (mkfs.jffsx), and it would make using & distributing closed-source proprietary compression methods easier.
There's a patch at http://www.logix.cz/michal/devel/cryptodev/, written by Michal Ludvig, which adds a /dev/crypto device for this purpose, as on *BSD. Is there a chance that this can get into the kernel?

Something liks this will be included once we have async crypto. However, that could be a while yet so I suggest that you start by including the deflate implementation in user-space.

Cheers,

Signature · 4 lines
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
#2 lore raw
In reply to Kluba, Patrik
Quoted 4 lines · Patrik

Hi everybody,

I'm going to port JFFS2's compression modules to CryptoApi except {in|de}flate, which Artem is working(?) on.

No, I don't work on that anymore, so you probably want to port deflate as well. My investigation showed that it can't be easily done without hacking zlib internals, unless you greatly lose the compression ratio by flushing the zstream and do other ugly things.

Nevertheless, you're the compression specialist(s), so you should do this better :-)

Signature · 3 lines
Best Regards,
Artem B. Bityuckiy,
St.-Petersburg, Russia.