On Thu, 11 Mar 2021 17:05:53 +0000 "Luck, Tony" <tony.luck@xxxxxxxxx> wrote: > > I guess that p->mce_vaddr stores the virtual address of the error here. > > If so, sending SIGBUS with the address looks enough as we do now, so why > > do you walk page table to find the error virtual address? > > p->mce_vaddr only has the virtual address for the COPYIN case. In that code > path we decode the kernel instruction that hit the fault in order to find the virtual > address. That's easy because: > > 1) The kernel RIP is known to be good (can't page fault etc. on kernel address). > 2) There are only a half dozen instructions used by the kernel for get_user() or > copy_from_user(). > > When the machine check happens during user execution accessing poison data > we only have the physical address (from MCi_ADDR). > > -Tony Sorry to interrupt as I am really confused here: If it's a copyin case, has the page been mapped for the current process? will memory_failure() find it and unmap it? if succeed, then the current will be signaled with correct vaddr and shift? Maybe the mce_vaddr is set correctly, but we may lost the correct page shift? And for copyin case, we don't need to call set_mce_nospec()? -- Thanks! Aili Yao
