Re: [PATCHv2 2/2] PCAP regulator driver (for 2.6.32)

From: Mark Brown
Date: Fri Jun 26 2009 - 21:04:56 EST


On Fri, Jun 26, 2009 at 08:08:41PM -0300, Daniel Ribeiro wrote:
> Changed: Removed workaround for regulator use_count issues.

The usual place for these comments is after the --- with the diffstat
(the tools can then deal automatically with them). Incidentally, the
code you had would only help pxamci with this particular regulator
driver.

> Add (partial) support for the voltage regulators on the PCAP2 PMIC.
>
> Signed-off-by: Daniel Ribeiro <drwyrm@xxxxxxxxx>

Sorry, I've noticed a race condition below that I didn't spot first time
round below:

> +static int pcap_regulator_set_voltage(struct regulator_dev *rdev,
> + int min_uV, int max_uV)
> +{
> + struct pcap_regulator *vreg = &vreg_table[rdev_get_id(rdev)];
> + void *pcap = rdev_get_drvdata(rdev);
> + int uV;
> + u32 tmp;
> + u8 i;
> +
> + if (vreg->n_voltages == 1)
> + return -EINVAL;

It'd be a little more friendly to check if the supported voltage is in
the requested range. However, that'd only be an issue if constraints
allowed voltage changes which is obviously broken so a comment to that
effect would be enough. I wouldn't have mentioned it but...

> +{
> + struct pcap_regulator *vreg = &vreg_table[rdev_get_id(rdev)];
> + void *pcap = rdev_get_drvdata(rdev);
> + u32 tmp;
> +
> + if (vreg->en == NA)
> + return -EINVAL;
> +
> + ezx_pcap_read(pcap, vreg->reg, &tmp);
> + tmp |= (1 << vreg->en);
> + ezx_pcap_write(pcap, vreg->reg, tmp);

This read/modify/write cycle is racy; the individual regulator is locked
by the core but this register is shared between all regulators on the
chip so if two are being updated at once things will go wrong. Most of
the MFDs have a set_bits() function that does an atomic read/modify/write
cycle for cases like this.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/