[Bug 788] Starting or stopping gsmd completely locks up the Neo

bugzilla-daemon at bugzilla.openmoko.org bugzilla-daemon at bugzilla.openmoko.org
Mon Sep 17 22:20:50 CEST 2007


------- Additional Comments From mwester at dls.net  2007-09-17 22:20 -------
Proposed new patch is attached.

This proposed patch changes the behavior of serial port 0 on the GTA01 device
such that hardware flow control is not permitted when in console mode, and
permitted when switched to the GSM modem.  If an application (such as gsmd)
turns on hardware flow control, the operation will succeed as expected. 
However, the serial driver will only actually enable hardware flow control when
the serial port is switched to the GSM modem.

The behavior is exactly as if the hardware RTS line on the console port was in
the on state (as opposed to the current GTA01 hardware, where the RTS line is off.

This patch resolves the issue originally reported by this bug.  It also resolves
potential issues that might arise from mis-use or operator error, such as
starting gsmd while the GSM modem is not turned on and enabled (currently, this
action will lock up the device if the user does not use the /etc/init.d/gsmd
script to start/stop the gsmd service.)

[Re: Harald's comment #18: Changing the console to register/unregister may
resolve the issue by properly resetting the flow control state of the port when
the GSM modem changes state, and if it works, it is a fairly simple solution
that will address most cases.  However, this proposed patch also protects in the
case where flow control is enabled but the GSM modem has not been switched, and
also permits the serial port to be switched between the modem and the console
and back again without loss of any port settings (consider the ability to switch
on the fly to print a kernel "oops" output and switch back again, without gsmd
noticing anything more than a few dropped characters).] 

------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

More information about the buglog mailing list