Re: io_uring_prep_timeout_update on linked timeouts

[Pavel Begunkov <asml.silence@gmail.com>]

On 8/31/21 12:36 PM, Victor Stewart wrote:
>>> _Not tested_, but something like below should do. will get it
>>> done properly later, but even better if we already have a test
>>> case. Victor?
> 
> sorry been traveling! i can extract out a simple test today...
> 
>> FWIW, I wrote a simple test case for it, and it seemed to work fine.
>> Nothing fancy, just a piped read that would never finish with a linked
>> timeout (1s), submit, then submit a ltimeout update that changes it to
>> 2s instead. Test runs and update completes first with res == 0 as
>> expected, and 2s later the ltimeout completes with -EALREADY (because
>> the piped read went async) and the piped read gets canceled.
> 
> ...unless?

Extra tests are always appreciated, there is never too many of them :)
FWIW, don't see any new related added by Jens (yet?).

-- 
Pavel Begunkov