Re: [Crash-utility] About displaying virtual memory information of exiting task

----- Original Message ----- > On 12/08/2014 10:28 PM, Dave Anderson wrote: >> >> >> ----- Original Message ----- >>> On 12/06/2014 04:11 AM, Dave Anderson wrote: >>>> Interestingly enough, today I was asked to look at a vmcore in which an >>>> oops >>>> occurred during task exit after tsk->mm had been NULL'd out in exit_mm(): >>> >>> It almost matches what I am facing, when tsk->mm is set to NULL and memory >>> mapping is supposed to be displayed. This is a more simple implementation. >>> I have tried to command like vm [taskp | pid | [-M mm_struct]]. But it have >>> to modify a lot of thing. >>> >>> By the way, I feel the code is becoming more and more complicated, maybe a >>> reconstruction is needed. >> >> Well, the vm_area_dump() function is relatively stable, so let's not go crazy >> here for what's essentially an "experimental" option. > > I see. > >> >>> >>>> >>>> Of course it has its limitations. Since the page tables are being broken down in this case, >>>> "vm -p" fails: >>>> >>>> crash> vm -M ffff880495120dc0 -p >>>> PID: 4563 TASK: ffff88049863f500 CPU: 8 COMMAND: "postgres" >>>> MM PGD RSS TOTAL_VM >>>> 0 0 0k 0k >>>> VMA START END FLAGS FILE >>>> ffff8804a085ce90 400000 f56000 8001875 /usr/local/greenplum-db-4.3.3.1/bin/postgres >>>> VIRTUAL PHYSICAL >>>> vm: invalid kernel virtual address: 50 type: "mm_struct pgd" >>>> crash> >>> >>> After a glance, the pgd comes from the mm of task_struct. We need a lot of work to make it >>> replaced by argument of -M, I don't think it worse it right now. >> >> Actually it doesn't take much work at all. If both tc->mm_struct and tm->mm_struct_addr >> are replaced with the supplied address: >> >> tc->mm_struct = tm->mm_struct_addr = pc->curcmd_private; >> >> then "vm -M ffff880495120dc0 -p" also works OK with my sample vmcore. > > Yes, vm -p will tc->mm_struct to get pgd. But I was afraid permanently changing tc->mm_struct > is not good. > > Take my core into consideration, the case is: > > <cut> > crash> help -t | grep mm_struct > .mm_struct: 0 > mm_struct: 354cc80 > crash> vm -M 0xffff88003ae98ac0 > PID: 4860 TASK: ffff88003ae7eaf0 CPU: 0 COMMAND: "bash" > MM PGD RSS TOTAL_VM > 0 0 0k 0k > VMA START END FLAGS FILE > ffff88003acc3ad8 400000 4d4000 8001875 /bin/bash > ... > crash> help -t | grep mm_struct > .mm_struct: ffff88003ae98ac0 > mm_struct: 354cc80 > crash> > <cut> > > Is it OK to have ".mm_struct" changed here? Probably not... I did take a quick look at the other usages of tc->mm_struct, and I think the only major difference is that the "search -u" command would search the user-space memory of the changed task. But there's also an oddball check for an i386 hypervisor callback from user-space, and it looks like get_task_mem_usage() would use it from that point on, so I agree with you that it should be restored to NULL. Since there's a strong possibility of an error(FATAL, ...) call while executing the command, it's not safe to simply restore it to NULL at the end of the command, but rather the pc->cmd_cleanup() facility should be used with the task address as the pc->cmd_cleanup_arg. >> >>>> >>>> But it does seems like a worthwhile addition. >>>> >>>> The patch doesn't check whether mm->owner or mm->mm_count are legitimate, but I'm not >>>> sure whether it's even worth it? If it fails, it fails, and the help page should just >>>> indicate that the command option is not guaranteed to work. Does the attached patch work >>>> for you? >>> >>> Similar to the core I got. And I modified the patch to add some check. At least I think >>> we need to make sure the address still belongs to a mm_struct object. >> >> I suppose you could, although in all probability it's going to be stay in the mm_struct >> slab cache, and worst case, have been re-used by another task. > > So you like the modification. Sure -- with the pc->cmd_cleanup in place, I don't see how it can hurt. Dave .