linux-kernel - Re: [PATCH v1 06/11] mmc: mmci: Qcomm: Add 3 clock cycle delay after register write

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

Date:	Tue, 13 May 2014 09:29:04 +0200
From:	Linus Walleij <linus.walleij@...aro.org>
To:	Srinivas Kandagatla <srinivas.kandagatla@...aro.org>
Cc:	Russell King <linux@....linux.org.uk>,
	"linux-mmc@...r.kernel.org" <linux-mmc@...r.kernel.org>,
	Chris Ball <chris@...ntf.net>,
	Ulf Hansson <ulf.hansson@...aro.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	agross@...cinc.com,
	"linux-arm-msm@...r.kernel.org" <linux-arm-msm@...r.kernel.org>
Subject: Re: [PATCH v1 06/11] mmc: mmci: Qcomm: Add 3 clock cycle delay after
 register write

On Tue, Apr 29, 2014 at 10:20 AM,  <srinivas.kandagatla@...aro.org> wrote:

> From: Srinivas Kandagatla <srinivas.kandagatla@...aro.org>
>
> Most of the Qcomm SD card controller registers must be updated to the MCLK
> domain so subsequent writes to registers will be ignored until 3 clock cycles
> have passed.
>
> This patch adds a 3 clock cycle delay required after writing to controller
> registers on Qualcomm SOCs. Without this delay all the register writes are not
> successfull, resulting in not detecting cards.
>
> Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@...aro.org>

Sounds like someone decided to clock the internal state machine
in the MMCI using MCLK instead of PCLK :-(

A bit nasty if this ends up in the fastpath (irq) though. Which it
invariably does, right?

> +       /*
> +        * On QCom SD card controller, registers must be updated to the
> +        * MCLK domain so subsequent writes to this register will be ignored
> +        * for 3 clk cycles.
> +        */
> +       if (host->hw_designer == AMBA_VENDOR_QCOM)
> +               udelay(1 + ((3 * USEC_PER_SEC)/host->mclk));

Add a new field in vendor data instead, and use DIV_ROUND_UP():

static struct variant_data variant_qcom = {
         .mclk_delayed_writes  = true,
(...)

if (host->vendor->mclk_delayed_writes)
    udelay(DIV_ROUND_UP((3 * USEC_PER_SEC), host->mclk));

You get the idea.

Yours,
Linus Walleij
--
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/