On Sun, Jun 30, 2024 at 11:25 AM Jiaxun Yang <jiaxun.yang@xxxxxxxxxxx> wrote: > > > > 在2024年6月30日六月 上午4:01，Jiaxun Yang写道： > > 在2024年6月30日六月 上午3:22，Jiaxun Yang写道： > >> 在2024年6月28日六月 下午6:57，Yu Zhao写道： > >>> Hi, > >>> > >>> OpenWrt folks ran into MCEs caused by multiple matching TLB entries > >>> [1], after they updated their kernel from v6.1 to v6.6. > >>> > >>> I reported similar crashes previously [2], on v6.4. So they asked me > >>> whether I'm aware of a fix from the mainline, which I am not. > >>> on > >>> I took a quick look from the MM's POV and found nothing obviously > >>> wrong. I'm hoping they have better luck with the MIPS experts. > >> > >> Hi Yu, > >> > >> I never hit such problem on my (non-bcm) 74Kc systems. > >> > >> However a quick glance suggested it may be related to Wired TLB entries > >> on your platform. > >> > >> Both duplicated TLB entries, Index 2 and 3, are all below "Wired" setting, > >> which means they are not managed by mm, but platform code. > > > > I just tried to dig into bcm47xx platform code and I think we should blame > > bcm47xx_prom_highmem_init, which created wired entry for high mem and may > > conflict with kernel's mapping. > > > > Nowadays, MIPS mm code can handle highmem on it's own, so there is no need > > to create such entry IMO. > > Sorry, I think I made a wrong diagnoses, it's actually a problem in our cache > alias code. > > Will try to fix. Thanks for looking into this!
