Hi, Ruoyao, The problem still exists in 5.16-rc8, can you try to change cpu_relax() definition to smp_mb()? It seems can fix the problem. Huacai On Tue, Dec 28, 2021 at 4:34 PM Xi Ruoyao <xry111@xxxxxxxxxxxxxxxx> wrote: > > On Tue, 2021-12-21 at 15:53 +0800, Huacai Chen wrote: > > > On Mon, Dec 20, 2021 at 5:04 PM Xi Ruoyao <xry111@xxxxxxxxxxxxxxxx> > > wrote: > > > > > > The snapshot panics on my system under high pressure (building and > > > testing GCC). The panic message is pasted, but it's kind of broken up > > > likely because multiple CPU cores were outputing to the serial console > > > simutaniously. > > > > > > The config is attached. I'm not sure the reason of the panic (bug in > > > the patches or bug in mainline kernel?) Do you have some pointers to > > > diagnostic and fix the issue? > > > > > > [ 5391.004745] CPU 1 Unable to handle kernel paging request at virtual address 000000000040007e, era == 90000000003aa07c, ra == 90000000003aa84c > > /* snip */ > > > We also found that the latest github kernel has some stable issues, > > and we are investigating. > > I rebased the patches onto 438645193e59e91761ccb3fa55f6ce70b615ff93 and > the problem *seems* gone. Not sure if it's something fixed in the > mainline or some more strange thing. > -- > Xi Ruoyao <xry111@xxxxxxxxxxxxxxxx> > School of Aerospace Science and Technology, Xidian University
