linux-kernel - Re: Questions on mmc

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-prev] [day] [month] [year] [list]

Date:	Sat, 25 Apr 2009 22:24:15 +0200
From:	Pierre Ossman <pierre@...man.eu>
To:	Martin Fuzzey <mfuzzey@...il.com>
Cc:	Sascha Hauer <s.hauer@...gutronix.de>, linux-kernel@...r.kernel.org
Subject: Re: Questions on mmc_test suite

On Wed, 15 Apr 2009 11:13:44 +0200
Martin Fuzzey <mfuzzey@...il.com> wrote:

> Hi,
> I'm currently updating the mxcmmc driver for the i.MX21 platform
> [patches have been posted to the arm list] and have a couple of
> questions about the test suite.
> 
> 1) I'm getting an error on testcase 10 (weird reads) which does reads
> of 3 to 512 bytes by steps of 7.
> Modifying the test so it continues in case of error shows that it
> _only_ fails on the 395 byte read.
> 
> It fails with a software timeout (-110)
> The card sends the same response to CMD17 (READ_SINGLE_BLOCK) as the
> reads that work
> but hardware never indicates transfer complete.
> 
> Changing the test to start at 395 makes it fail immediately (so not
> dependent on previous reads).
> Has anyone seen this before?
> 

Nothing I recognize, no. Have you tested more than one card in case
it's a card bug?

> 2) I'm getting "Warning: Host did not wait for busy state to end" on
> the multiblock write tests (Tests 5 and 13) but they still pass.
> Seems to be timing related since this doesn't occur with MMC debugging enabled.
> My understanding of this is that the driver should wait for the
> hardware to stop signalling busy on the data line (response R1b) after
> sending
> command 12 (stop transmission) before replying - is this correct?
> However the MX2 SDHC doesn't have a register for this.
> 
> Should I:
> a) ignore it? [apart from the testcase warning everything seems to work]
> b) try reading the data line as GPIO  - seems a bit of a hack
> c) issue CMD13 (SEND_STATUS) from the driver to poll the card  - seems
> to be the wrong layer.
> 
> If b) is the way to go how long can the delay be - is polling
> acceptable or must it be interrupt driven?
> 

I'd say a). If the hardware doesn't support it then I don't think we
can do that much better than the polling that mmc_block does.

Rgds
-- 
     -- Pierre Ossman

  WARNING: This correspondence is being monitored by the
  Swedish government. Make sure your server uses encryption
  for SMTP traffic and consider using PGP for end-to-end
  encryption.

Download attachment "signature.asc" of type "application/pgp-signature" (199 bytes)