On Wed, 2015-07-15 at 11:47 +0200, Johan Hovold wrote:
> On Sun, Jun 28, 2015 at 01:28:18PM -0500, Peter E. Berger wrote:
> > From: "Peter E. Berger" <pberger@xxxxxxxxxxx>
> >
> > The io_ti driver fails to download firmware to Edgeport devices such as
> > the EP/416. One of the problems is that the default 1 second timeout
> > in ti_vsend_sync() is insufficient for download operations. This patch
> > increases the download timeout to 10 seconds.
>
> So only EP/416 appears to affected and this is only needed after fixing
> the firmware version handling as these devices have an e2prom and will
> function anyway?
I found an old EP/421 and discovered that it also demonstrates the
problem, which resolves with the longer timeout. The EP/416 and EP/421
are the only models I've seen with e2prom. I think your assertion that
these will work even with the broken firmware downloads is essentially
correct, but there may be some confusing behavior the first time they
are plugged in.
Some details: If the on-board firmware is already newer than the
version in /lib/firmware/edgeport/down3.bin, then we won't get to this
problematic code at all and the device should work as expected. If the
on-board firmware is older, the download to e2prom will at least start
and probably more: at least in one test here, even though the download
timed out and was noted as a failure so no devices were registered at
the time, it appears to have actually succeeded. This became obvious
when about 60 seconds later the firmware's keepalive timer forced the
devices to reconnect and the driver got another chance to initialize.
>From that point forward, the EP/416 was up-to-date, so subsequent uses
(after a power cycle or USB unplug/plugin cycle) no longer activated the
problematic download code and the devices worked properly.
So, if that behavior holds for all Edgeports with e2prom, this may turn
out to be only a minor inconvenience that just requires a couple of
plug/unplug cycles when the devices are used for the first time, and
only if their onboard firmware is older than what's
in /lib/firmware/edgeport/down3.bin.
>
> I'm trying to determine whether we need this one in stable or not. The
> three last patches are too big and intrusive to meet the stable-kernel
> rules (the sanity one isn't even a fix). We could just say that using
> the latest firmware requires a recent kernel for these devices.
That (telling people to use a recent kernel for these devices) sounds
fine to me.
I'm preparing a v8 patchset and hope to submit it very soon. Thanks.
--Peter
>
> Thanks,
> Johan
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
