Re: [PATCH v2] fbdev: bfin-lq035q1-fb: new Blackfin Landscape LCD EZ-Extender driver

[Mike Frysinger <vapier.adi@gmail.com>]

On Fri, Sep 25, 2009 at 16:24, Mike Frysinger wrote:
> On Thu, Sep 24, 2009 at 19:32, Andrew Morton wrote:
>> On Thu, 17 Sep 2009 17:37:06 -0400 Mike Frysinger wrote:
>>> +static int lq035q1_control(unsigned char reg, unsigned short value)
>>> +{
>>> +     int ret;
>>> +     u8 regs[3] = {LQ035_INDEX, 0, 0};
>>> +     u8 dat[3] = {LQ035_DATA, 0, 0};
>>> +
>>> +     if (spi_control.spidev) {
>>> +             regs[2] = reg;
>>> +             dat[1] = value >> 8;
>>> +             dat[2] = value & 0xFF;
>>> +
>>> +             ret = spi_write(spi_control.spidev, regs, ARRAY_SIZE(regs));
>>> +             ret |= spi_write(spi_control.spidev, dat, ARRAY_SIZE(dat));
>>> +     } else
>>> +             return -ENODEV;
>>> +
>>> +     return ret;
>>> +}
>>
>> I am suspecting that this function (and the similar ones below) rely
>> upon state within the hardware and will hence misbehave if two
>> instances are run concurrently.
>>
>> Is that correct>  If so, is there locking to prevent this from occurring?
>
> if by "instances" you mean "users" as in "multiple programs
> reading/writing the framebuffer concurrently", then probably.  rather
> than handle the locking ourselves, it can be pushed to the SPI bus by
> having the regs/dat be transfers in a single message.

hmm there shouldnt be any locking problems here actually.  the spi
access is only to initialize/shutdown, and there is a dedicated CS for
each device.  so i dont think anything here really needs changing
(other than to allocate the spi_control global variable dynamically in
the probe).
-mike