On Tue, Jun 2, 2015 at 10:00 AM, Ian Campbell <[email protected]> wrote:
> On Tue, 2015-06-02 at 08:40 +0100, Ian Campbell wrote: > > On Sun, 2015-05-31 at 15:55 +0200, JM wrote: > > > Yes, I managed to catch it in the act: > > > > > > http://pastebin.com/PYqqrcLX > > > > > > > > > Basically, the device stalls during the flashing process (which takes > > > maybe three minutes), and that seems to cause sata to time out, and > > > eventually everything gets remounted read-only. Now if I recall > > > correctly, I'd get sata link timeouts with 3.16 as well, but it > > > behaved more gracefully (reset a couple of times and was back in > > > business after the flashing was done). > > > > I wonder if that somehow relates to this coherency issue: > > > http://lists.infradead.org/pipermail/linux-arm-kernel/2015-May/347665.html > > which turned out to be fixed by: > > http://article.gmane.org/gmane.linux.kernel.stable/137371 > > FWIW I couldn't boot 4.0+ on my TS-419P without that patch (SATA drives > not detected) and with it I don't see any stall/hang with flash-kernel. > > > If not then I think this would be something worth raising with the > > upstream developers. > > Ian. > > I tested the patch you suggested, and somewhat surprisingly it fixed the xhci/USB3.0 initialization problem, but the stalling and sata resets during kernel/initramfs flashing persist. http://pastebin.com/gERQKnDg Best regards, Jan
