Michael, On Sun, 4 Nov 2018, Michael Shych wrote: Cc: + serial folks and Holger who reported a similar issue. > We are experiencing strange serial IRQ problem on our systems with new kernel. > > We obtain messages like: "do_IRQ: 0.38 No irq handler for vector". > > 1. It's happened with kernel 4.15 and above, tested 4.15 and latest 4.19. > The problem doesn't happened with kernel 4.14 Yes, because we changed interrupt vector management in 4.15 and that unearthed issues on some machines which are not yet fully understood. > 4. The problem exist only in one kind of our platforms with x86 Intel > Celeron 1047UE CPU (Ivy Bridge architecture CPU with QM77 PCH and UART in > Winbond NCT5577D SIO). The problem doesn't happened on our other > platforms with Intel Broadwell-DE D1508 SoC and with Intel Atom-Rangeley > C2558 SoC. > 5. The problem is also happened in appropriate Intel reference board > (Cougar Canyon). It has same Ivy Bridge CPU type but different CPU > model. i7-2715QE instead Celeron 1047UE. > 7. The problem doesn't happened if system is booted only with 1 CPU core > (disabled in BIOS or through boot command maxcpus=1), of course, it's not > a solution. > > Could you advise if this problem was observed before and what can be done? That's a really good question. So one hint is that it's only a few boards and not a really widespread problem. The other is that it goes away when the kernel runs with only one CPU. It looks like an issue when tearing down the serial interrupt, but I need to think more about how to debug that. Thanks, tglx
