Re: vmstat -m and netstat -m dumping core when run on vmcores

On Mon, Oct 01, 2018 at 09:15:46AM +0300, Yuri Pankov wrote:
> Hi,
> 
> I've noticed the following rebooting after a panic:
> 
> pid 41246 (vmstat), uid 0: exited on signal 11 (core dumped)
> pid 47091 (netstat), uid 0: exited on signal 11
> 
> And indeed, trying to manually run those on the resulting vmcore makes 
> them crash:
> 
> # vmstat -m -M /var/crash/vmcore.0
> Segmentation fault (core dumped)
> # netstat -m -M /var/crash/vmcore.0
> Segmentation fault
> 
> Backtrace is below:
> 
> * thread #1, name = 'vmstat', stop reason = signal SIGSEGV
>    * frame #0: 0x000000080026d19f 
> libmemstat.so.3`memstat_kvm_malloc(list=0x00000008007c5000, 
> kvm_handle=0x000000080070e000) at memstat_malloc.c:351
>      frame #1: 0x0000000000204b15 vmstat`main [inlined] domemstat_malloc 
> at vmstat.c:1406
>      frame #2: 0x0000000000204af4 vmstat`main(argc=<unavailable>, 
> argv=0x00007fffffffeb28) at vmstat.c:386
>      frame #3: 0x0000000000204095 vmstat`_start(ap=<unavailable>, 
> cleanup=<unavailable>) at crt1.c:74
> 
> Wonder if it's just me or something is broken here?

I think this is due to r338899.  libmemstat needs an adjustments to handle
that.