On 8/25/22 14:52, Jens Axboe wrote:
On 8/25/22 4:13 AM, Pavel Begunkov wrote:
On 8/25/22 11:11, Pavel Begunkov wrote:
Don't forget to initialise and set addr in io_sendzc(), so if it goes
async we can copy it.
Jens, can you amend it into the last commit?
("io_uring/net: save address for sendzc async execution")
Yes, I'll amend it. But do we have a test case that hits this path?
Because it seems like that would've blown up immediately.
Apparently a test I have only hits io_sendzc_prep_async() callback
and the large buffer test doesn't trigger it. Hard to trigger it
with udp and addresses don't make sense with tcp.
--
Pavel Begunkov
