Re: [PATCH] mtd: cfi_cmdset_0002: Support Persistent Protection Bits (PPB) locking

[Artem Bityutskiy <dedekind1@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1197 bytes --]

On Fri, 2012-12-07 at 08:22 +0100, Stefan Roese wrote:
> +       /*
> +        * Wait for some time as unlocking of all sectors takes quite long
> +        */
> +       timeo = jiffies + (2 * HZ);     /* 2s max (un)locking */

Please, use msecs_to_jiffies() instead.

> +       for (;;) {
> +               if (chip_ready(map, adr))
> +                       break;
> +
> +               if (time_after(jiffies, timeo)) {
> +                       printk(KERN_ERR "Waiting for chip to be ready timed out.\n");
> +                       ret = -EIO;
> +                       break;
> +               }
> +               mutex_unlock(&chip->mutex);
> +               cfi_udelay(1);
> +               mutex_lock(&chip->mutex);
> +       } 

Would you please educate me a bit and explain what is protected by
'chip->mutex' and by 'get_chip()'.

Why you need to drop the mutex here?

Why is it not an ABBA deadlock to do this:

Task 1: In the loop above, has chip locked, doing
        mutex_lock(&chip->mutex);

Task 2: done mutex_lock(&chip->mutex), now doing
        ret = get_chip(map, chip, adr + chip->start, FL_LOCKING);

-- 
Best Regards,
Artem Bityutskiy

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]