> List: linux-serial > Subject: RE: "No irq handler for vector" problem > From: Michael Shych <michaelsh () mellanox ! com> > Date: 2018-11-06 13:35:40 > Message-ID: DB5PR05MB1320C5210856FDF10425A9BED9CB0 () > DB5PR05MB1320 ! eurprd05 ! prod ! outlook ! com > > We are experienced problem only on Ivy Bridge CPU with Winbond SIO. > There are no problem on Atom Rangeley and Broadwell-DE SoCs. > Interrupt for serial port, IRQ3, is delivered via Serial IRQ on all these > platforms not only on problem IvyBridge. Hello to all, Based on the info above we found one experimental Com Express with Broadwell U (exact model: i3-5010U). We have the exact what we had on HSW 2980U, and the problem repeats itself/persists, as well as on HSW i5-4300U. Exact the same C-Fast used with Ubuntu kernel 4.15.0. We see these messages using Winbond SIO, with the same baseboard with already mentioned Com Express modules with named CPUs. In regards what I know about PCHs used on these Com Expresses, I know that HSW U/Y uses Series 8 Lynx Point PCH. 2980U is SoC, but not HSW i5-4300 (PCH is a separate silicon on the same package). Also, for the BDW-U i3-5010U, the same story. My best bet (not sure) is that it uses the same Series 8 Lynx Point PCH (on the same package, but BDW-U is smaller - 14 nm HSW shrink). BDW-DE is true SoC as I know it, and probably uses different Series 9 Wildcat Point PCH. So, this is the possible reason why we see different behaviour on Broadwell i3-5010U (different PCHs between i3-5010U and BDW-DE Soc). Best Regards, Zoran _______
