Hi Tony, > Am 12.04.2018 um 17:03 schrieb Tony Lindgren <t...@atomide.com>: > > * H. Nikolaus Schaller <h...@goldelico.com> [180411 07:05]: >>> Am 10.04.2018 um 22:56 schrieb Ladislav Michl <la...@linux-mips.org>: >>> Well, that's a shame... However, this code is in production for several >>> months now, >> >> Well, we could have tested earlier release-candidates... > > I've found it's way easier to test Linux next on regular basis > and report the regressions compared to chasing regressions after > the merge window or tagged kernels. By that time you may have > already multiple regressions to deal with which is always fun. > > If you have products running with mainline kernel, maybe set up > automated boot testing for core devices to catch regressions > early?
Well, it is rare that we face upstream regressions of this severe kind. Every 2-3 years... We much more often have to face missing features in upstream so that we add our own patch sets on top of each -rc. In this case here we could have found this right after the merge window when v4.16-rc1 came out. But by bad chance (and holidays and other more important tasks) we did only test on GTA04 boards with standard NAND. And just now did boot a board with OneNAND installed and were surprised by this boot problem. BR, Nikolaus