linux-kernel - RFC: [PATCH 0/9 Integration of SmartMedia/xD into mtd subsystem

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <1262814216.14552.22.camel@maxim-laptop>
Date:	Wed, 06 Jan 2010 23:43:36 +0200
From:	Maxim Levitsky <maximlevitsky@...il.com>
To:	linux-kernel <linux-kernel@...r.kernel.org>
Cc:	linux-mtd <linux-mtd@...ts.infradead.org>,
	Alex Dubov <oakad@...oo.com>, joern <joern@...fs.org>
Subject: RFC: [PATCH 0/9 Integration of SmartMedia/xD into mtd subsystem

Here is the result of my work of supporting an xD card reader that I
have on my notebook.

I had to fix few problems in mtd translation layer, add few workarounds.
This is by no way a final version of the patches, there is still lot of
cleanup to do.
The patches weren't tested with checkpatch.pl for example.

I can now read and write an xD card, and I have no crashes.
Still following issues remain:

1 - Write speed IS very SLOW, just about 200 Kbytes/s
In fact a reader in my printer gives me about 330 Kbytes/s, thus
partially card is to blame.
Also, hardware doesn't support interrupts to test when card is ready,
thus writes consume 100% of one cpu.

2 - I rely on mtd driver to have empty oob layout so I can read both oob
and data using ->read_oob  with MTD_OOB_PLACE.
I thinking to add new mode to allow to read whole oob + data and check
ecc, or I will have to do pointless copying of data from 'censored' oob
to normal structure. (This applies to FTL driver I also include in this
patchset)

3 - Using the mtd device directly with anything but supplied FTL or
SSFDC, on 'modern' xD cards just doesn't. These cards implement a fake
nand command set, thus don't have a real oob.
It was disappointing for me too.
I suspect that all 'Type M' cards are of this fake type, but older cards
are OK.

4 - Suspend/resume support works, but relies on all mtd users not to
suspend in middle of card access, bacause in this case its very
difficult/impossible to know card state.
Driver will refuse suspend in that case.
Access via all block devices is safe.

5 - And of course there are bugs, thus I warn you now that this driver
has high probability to erase data from your card.
Thus DON'T USE THIS DRIVER IF ANYTHING IMPORTANT IS STORED ON THE CARD.

6 - driver only detects PCI ID of my notebook, so it might not load on
compatable chips with different ID.

7 - patches developed against stable 2.6.32, and won't compile against
latest git due to changes in kfifo api.

Feedback is welcome, flames too :-)

Best regards,
Maxim Levitsky

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/