Re: [PATCH net-next v6 2/2] net: wwan: mhi_wwan_ctrl: drive DTR/RTS via the IP_CTRL channel

[Peter Hunt]

Thanks for the review. I'll send a v7 addressing these.

[High] No in-tree controller declares IP_CTRL: the IP_CTRL channel entry
for the Sierra EM919x/EM929x is commit 83c29a55b89e ("bus: mhi: host:
pci_generic: Add IP_CTRL channel for Sierra EM919x/EM929x"), applied to
mhi-next from v5 of this series, so it isn't in net-next yet. The cover
letter mentions it, but the commit message didn't, so v7 notes the
dependency in patch 2.

[Medium] Ignored mhi_queue_buf() failure in mhi_wwan_dtr_dl_xfer_cb():
agreed. v7 retries a failed requeue from a delayed work item on -ENOMEM
and logs other failures. -EIO and -ENODEV mean the channel is going away,
and the next probe queues a fresh buffer.

[Medium] DTR dropped on last close regardless of HUPCL (patch 1/2):
agreed. v7 only drops DTR/RTS on last close if HUPCL is set, and defaults
HUPCL on for AT ports, since port->at_data.termios is otherwise zeroed.
Close behaviour is unchanged by default, and userspace can clear HUPCL
with TCSETS to keep DTR asserted, as on a TTY.

[Low] wwan_remove_port() passing 0 (patch 1/2): agreed, v7 passes the
masked mdmbits through a helper shared with the close path.

pw-bot: cr