MC7455 in X220 with ModemManager 1.6.4

[Bjørn Mork]

Dan Williams <dcbw at redhat.com> writes:

> On Fri, 2017-06-09 at 01:25 +0200, Dominik Strnad wrote:
>> Hello Bjorn,
>> 
>> thank you. I meanwhile succeeded to get car into online mode by
>> isolating
>> pin20 on minPCIe card. But I at least confirm, that after using
>> ENTERCND
>> the PCOFFEN do not report error anymore (but the card was already not
>> in
>> disable state because of pin isolation. But I have a question - what
>> is
>> better? To use your command (probably as part of some udev rule?) or
>> keep
>> that pin isolated? Thank you.
>
> Ideally figure out why BIOS isn't letting the card be enabled :)

I still think this is a BIOS flag which needs to be set or cleared, but
I don't know the exact magic spell.  Looking at the thinkpad_acpi driver
now, I get a distinct deja vu feeling...  If I only had taken some notes
:-(

I wonder if I simply might have temporarily disabled the "HWPRESENT"
test in the driver, and let the default wan_shutdown() just update the
state?  In any case, I do believe the solution is hidden somewhere here:



static void wan_shutdown(void)
{
	/* Order firmware to save current state to NVRAM */
	if (!acpi_evalf(NULL, NULL, "\\WGSV", "vd",
			TP_ACPI_WGSV_SAVE_STATE))
		pr_notice("failed to save WWAN state to NVRAM\n");
	else
		vdbg_printk(TPACPI_DBG_RFKILL,
			"WWAN state saved to NVRAM\n");
}

static void wan_exit(void)
{
	sysfs_remove_group(&tpacpi_pdev->dev.kobj,
		&wan_attr_group);

	tpacpi_destroy_rfkill(TPACPI_RFK_WWAN_SW_ID);

	wan_shutdown();
}

static int __init wan_init(struct ibm_init_struct *iibm)
{
	int res;
	int status = 0;

	vdbg_printk(TPACPI_DBG_INIT | TPACPI_DBG_RFKILL,
			"initializing wan subdriver\n");

	TPACPI_ACPIHANDLE_INIT(hkey);

	tp_features.wan = hkey_handle &&
	    acpi_evalf(hkey_handle, &status, "GWAN", "qd");

	vdbg_printk(TPACPI_DBG_INIT | TPACPI_DBG_RFKILL,
		"wan is %s, status 0x%02x\n",
		str_supported(tp_features.wan),
		status);

#ifdef CONFIG_THINKPAD_ACPI_DEBUGFACILITIES
	if (dbg_wwanemul) {
		tp_features.wan = 1;
		pr_info("wwan switch emulation enabled\n");
	} else
#endif
	if (tp_features.wan &&
	    !(status & TP_ACPI_WANCARD_HWPRESENT)) {
		/* no wan hardware present in system */
		tp_features.wan = 0;
		dbg_printk(TPACPI_DBG_INIT | TPACPI_DBG_RFKILL,
			   "wan hardware not installed\n");
	}

	if (!tp_features.wan)
		return 1;

..


> But failing that, either PCOFFEN or the taping of pin20 are probably
> fine.  PCOFFEN is likely the better solution, though that may be reset
> on firmware updates.

I don't think it will be reset. NVRAM settings are usually kept, unless
the firmware upgrade includes a specific setting. And they normally
don't touch PCOFFEN.


Bjørn