Re: [PATCH] 8250_dw: do not int overflow when rate can not be aplied

[Ed Blake]

On 11/01/18 18:03, Ed Blake wrote:
> On 11/01/18 17:55, Nuno GonÃalves wrote:
>> So, for me clk_round_rate() always returns 24000000, and only the loop
>> variable i changes, so the search is monotonic, from the highest baud
>> to the lowest (increasing divider).
>>
>> I am using a Allwiner H2+, with the serial port configuration from
>> sunxi-h3-h5.dtsi.
>>
>> Are you sure that clk_round_rate can return differet values? Is that
>> because some boards might have several clock options beside the
>> adjustable divider?
> Yes I'm sure. Some platforms allow the clock rate to be varied, hence
> the existence of clk_round_rate() and clk_set_rate().
>
>> I really need to understand what is the problem, to be able to suggest
>> a solution to the integer overflow that is being allowed to happen.
> Some sort of overflow check on i * max_rate could work?

Actually I have another suggestion. I'll submit a separate patch.

-- 
Ed