Re: [PATCH V2] gpu: host1x: handle the correct # of syncpt regs

[Stephen Warren <swarren@wwwdotorg.org>]

On 04/07/2014 02:18 AM, Thierry Reding wrote:
> On Fri, Apr 04, 2014 at 04:31:05PM -0600, Stephen Warren wrote:
>> From: Stephen Warren <swarren@nvidia.com>
>>
>> BIT_WORD() truncates rather than rounds, so the loops in
>> syncpt_thresh_isr() and _host1x_intr_disable_all_syncpt_intrs() use <=
>> rather than < in an attempt to process the correct number of registers
>> when rounding of the conversion of count of bits to count of words is
>> necessary. However, when rounding isn't necessary because the value is
>> already a multiple of the divisor (as is the case for all values of
>> nb_pts the code actually sees), this causes one too many registers to
>> be processed.
>>
>> Solve this by using and explicit DIV_ROUND_UP() call, rather than
>> BIT_WORD(), and comparing with < rather than <=.
>>
>> Signed-off-by: Stephen Warren <swarren@nvidia.com>
>> ---
>> v2: Use DIV_ROUND_UP rather than BITS_TO_LONGS to avoid problems on 64-bit.
>> ---
>>  drivers/gpu/host1x/hw/intr_hw.c | 4 ++--
>>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> If I understand correctly there's no immediate need for this to go to
> stable kernels, nor for it to be queued for 3.15, right? That is the
> potential extra write isn't causing any harm on actual hardware, is it?
> 
> In that case I'll queue this up for 3.16.

We should definitely apply this, and as far back as the code exists,
since the SW is touching non-existent registers, and that is presumably
undefined behaviour, which could potentially cause hard-to-diagnose bugs.

Besides, I want the mainline kernel to run on our simulator without
having to maintain patches for fixed issues.