Re: [PATCH] stm: the number of masters should be (sw_end - sw_start + 1)

[Alexander Shishkin <alexander.shishkin@linux.intel.com>]

Chunyan Zhang <zhang.chunyan@linaro.org> writes:

> On Fri, Dec 11, 2015 at 4:51 PM, Alexander Shishkin
> <alexander.shishkin@linux.intel.com> wrote:
>> Chunyan Zhang <zhang.chunyan@linaro.org> writes:
>>
>>> On Fri, Dec 11, 2015 at 3:31 PM, Alexander Shishkin
>>> <alexander.shishkin@linux.intel.com> wrote:
>>>> Chunyan Zhang <zhang.chunyan@linaro.org> writes:
>>>>
>>>>> sw_end represents the last software master, sw_start is index of the
>>>>> first master, so the number of software masters should be
>>>>> sw_end - sw_start + 1.
>>>>
>>>> Looks about right, but it needs to be in two separate patches.
>>>>
>>>>> Signed-off-by: Chunyan Zhang <zhang.chunyan@linaro.org>
>>>>> ---
>>>>>  drivers/hwtracing/intel_th/sth.c | 2 +-
>>>>>  drivers/hwtracing/stm/core.c     | 2 +-
>>>>>  2 files changed, 2 insertions(+), 2 deletions(-)
>>>>>
>>>>> diff --git a/drivers/hwtracing/intel_th/sth.c b/drivers/hwtracing/intel_th/sth.c
>>>>> index 56101c3..28917d7 100644
>>>>> --- a/drivers/hwtracing/intel_th/sth.c
>>>>> +++ b/drivers/hwtracing/intel_th/sth.c
>>>>> @@ -173,7 +173,7 @@ static int intel_th_sw_init(struct sth_device *sth)
>>>>>       sth->stm.sw_start = reg & 0xffff;
>>>>>       sth->stm.sw_end = reg >> 16;
>>>>>
>>>>> -     sth->sw_nmasters = sth->stm.sw_end - sth->stm.sw_start;
>>>>> +     sth->sw_nmasters = sth->stm.sw_end - sth->stm.sw_start + 1;
>>>>>       dev_dbg(sth->dev, "sw_start: %x sw_end: %x masters: %x nchannels: %x\n",
>>>>>               sth->stm.sw_start, sth->stm.sw_end, sth->sw_nmasters,
>>>>>               sth->stm.sw_nchannels);
>>>>> diff --git a/drivers/hwtracing/stm/core.c b/drivers/hwtracing/stm/core.c
>>>>> index 7f7bdb3..cb676f2 100644
>>>>> --- a/drivers/hwtracing/stm/core.c
>>>>> +++ b/drivers/hwtracing/stm/core.c
>>>>> @@ -632,7 +632,7 @@ int stm_register_device(struct device *parent, struct stm_data *stm_data,
>>>>>       if (!stm_data->packet || !stm_data->sw_nchannels)
>>>>>               return -EINVAL;
>>>>>
>>>>> -     nmasters = stm_data->sw_end - stm_data->sw_start;
>>>>> +     nmasters = stm_data->sw_end - stm_data->sw_start + 1;
>>>>>       stm = kzalloc(sizeof(*stm) + nmasters * sizeof(void *), GFP_KERNEL);
>>>>
>>>> Or even offsetof(struct stm_device, masters[stm_data->sw_end]).
>>>>
>>>
>>> This should use 'offsetofend()'.
>>
>> No, actually, just scratch my previous comment as it was completely
>> wrong, just fix the off-by-one. If we were to use offsetof(), it should
>> rather be of masters[nmasters], but all we need is to fix the off-by-one
>> right now.
>>
>
> Sorry, you may lose me here, what's 'off-by-one' ?

It's an error when the result of your calculation is off by 1 (one too
few or one too many).

https://www.google.com/search?q=off-by-one

Regards,
--
Alex