On 23/02/2025 10:52, Leon Romanovsky wrote:
On Thu, Feb 20, 2025 at 09:53:52AM -0400, Jason Gunthorpe wrote:
On Thu, Feb 20, 2025 at 11:06:52AM +0200, Leon Romanovsky wrote:
Mainly -EBUSY from FW command interface, so users can safely call again
to modify QP.
Forget about this comment, I was distracted, and it is -EBUSY from
uverbs_try_lock_object() and not from FW command interface.
Userspace is doing something really wrong if it is triggering that..
And right now, userspace isn't aware of it. Users will be aware of it, after
we will return real error code and not mask everything under same -EINVAL.
Yes.
Currently, most libibverbs calls pass the kernel-masked return value
directly to the application.
Thanks.
Thanks
Jason
