1) there's no loop.c-2.6.original available -- that's my major complaint
-- so I can't see where it came from (with all the kernel changes).
There were some api changes I saw, but there seemed to be some "other
changes" -- for example:
@@ -709,6 +734,9 @@
sector_t IV;
struct page *pg;
+ if(!bio->bi_size)
+ return 0;
+
pos = ((loff_t) bio->bi_sector << 9) + lo->lo_offset;
buf = page_address(LDE_lo_bio_free0->bi_io_vec[0].bv_page);
IV = bio->bi_sector;
I'm not sure if that's a reason to upgrade or not...
2) Some stuff had to be cleaned up -- this was just a feeler to see if
anyone had a use for this...
I'm not sure about the future of loopAES with all the other methods
of encryption in the kernel (I started working on this several months
ago -- and since we used loopAES in 2.4 machines, I was asked to get it
working on 2.6). Note I'm new to this...
Another advantage is I had a framework to load multiple loopAES drivers
at the same time (which does different things -- for example, to compare
the
C code and the assembler code on x86).
Just replacing should compile it as device loop/major 7 -- but this can
be
changed at compile and or runtime...
Here's the patch with the extraneous stuff -- against 2.3a...feel free
to
use/comment/flush (I didn't see any patches on sourceforge in the
patchmanager,
so I wasn't sure if that was a place to put patches...)
marty
-----Original Message-----
From: Jari Ruusu [mailto:jariruusu@xxxxxxxxxxxxxxxxxxxxx]
Sent: Wednesday, November 14, 2007 9:46 AM
To: Leisner, Martin
Subject: Re: linux 2.6 driver
"Leisner, Martin" wrote:
> There's no loop.c-2.6.original in the 3.2a/3.2b distributions.
>
> I'm using 3.2a and made some mods to the driver.
>
> Looking at linus's git tree, there were substantial changes made to
> loop.c since May.
>
> FYI:
> diff -u loop-AES-v3.2a/loop.c-2.6.patched
> loop-AES-v3.2b/loop.c-2.6.patched | diffstat
> loop.c-2.6.patched | 89
> +++++++++++++++++++++++++++++++++++++++++++++--------
> 1 files changed, 77 insertions(+), 12 deletions(-)
>
> I've modified the loop driver so it can run alongside the standard
loop
> driver in the kernel (this was when systems had them build in instead
of
> modulized).
>
> I had the ability to:
> default to another name besides /dev/loop* (I used loopAES)
> runtime assignment of major numbers (so there's no conflict with
> the
> standard loop)
>
> This was easier than to recompile/modify all the kernels I was using
> (and tell everyone I was supporting they "had to run a new kernel".
[ This is private email, but feel free to CC a public mailing list
on your possible reply. ]
I'm not sure how to interpret your message. It could mean:
1) You are unhappy with amount of changes to loop.c-2.6.patched file?
- Most of those changes are needed because of block layer interface
changes on linux-2.6.24-rc and later kernels. It would be cool if
interfaces didn't change, but they did.
2) You wanted me and/or others to look at your changes?
- Your email didn't include a patch.
3) You are making public statement that you can hack code?
- Fine. Nothing wrong with that.
--
Jari Ruusu 1024R/3A220F51 5B 4B F9 BB D3 3F 52 E9 DB 1D EB E3 24 0E A9
DD
Attachment:
loop.c.diff
Description: loop.c.diff