On Tue, Sep 12, 2023 at 05:37:23PM +0200, Sabrina Dubroca wrote:
We'd have to do pretty much what Jakub suggested [1] (handle ENOSPC by
waiting for all current requests) and then resubmit the failed
request. So I think keeping the MAY_BACKLOG flag and handling EBUSY
this way is simpler. With this, we send one request to the backlog,
then we wait until the queue drains.
[1] https://lore.kernel.org/netdev/20230908142602.2ced0631@kernel.org/ (local)
You need to handle something like ENOSPC anyhow because the underlying
driver may experience a real failure (e.g., someone unplugged the
hardware) which would have pretty much the same effect of ENOSPC
(i.e., an error with no retries). So if the tls code can't cope with
that then it has to be fixed.
Cheers,
--
Email: Herbert Xu [off-list ref]
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt