[Crash-utility] crash start in CentOS 8

[HAGIO KAZUHITO(萩尾 一仁)]

Hi Bhupesh,

-----Original Message-----
> > Just FYI, with respect to the System.map file, the following Dave's post
> > would be helpful:
> > https://www.redhat.com/archives/crash-utility/2018-June/msg00002.html
> >
> > If you have the vmlinux corresponding to a running kernel, you don't need
> > the System.map file.
> 
> Indeed, I remember discussing this with Dave A. some months back.
> Infact with newer KASLR enabled kernels it makes little sense to use
> System.map file as it contains the non-relocated symbol values that
> were compiled into the vmlinux file - that's why I suggested just to
> use to the vmlinux and vmcore file as an arguments while invoking the
> crash tool to Patrick.
> 
> However, going back to the discussion I had with Dave A., I think the
> main point of confusion was the 'crash_whitepaper' which is still
> located at [1], and which provides the following example to invoke the
> crash utility with the System.map file, vmlinux and vmcore
> combination:
> 
>  Examples when running on a dumpfile:
> 
>       $ crash /boot/System.map vmlinux.dbg vmcore
> 
> I think while this was ok for older kernels which did not support
> KASLR, for newer kernels this needs an update.
> I am trying to see if I can modify the 'crash_whitepaper' and push the
> same to github, so that users are not confused with the System.map
> option.
> 
> What's your views on the same? Please let me know.
> 
> [1]. https://crash-utility.github.io/crash_whitepaper.html

There are several command lines with System.map in "4. Invocation" chapter and
I think that the ones in "RHEL2.1 Kernels (or kernels built without -g flag)"
section are no problem because the reason why it's needed is explained well.

The others in the "crash -h" output in "Invocation output" section might be
a little confusing as you say if they see only the -S option's explanation..

I think it would be fine to omit the part of the output below the [mapfile]
argument and itself because the output looks to be intended almost only to
introduce the "crash -h" option.  Then users see the current crash -h output
which doesn't have command lines with System.map.  So, for instance:

diff --git a/crash_whitepaper.html b/crash_whitepaper.html
index 9b2e88c8acb4..7cb2608e6062 100644
--- a/crash_whitepaper.html
+++ b/crash_whitepaper.html
@@ -1166,7 +1166,8 @@ Usage:
     file.  If the [dumpfile] argument is not entered, the session will be
     invoked on the live system using /dev/mem, which usually requires root
     privileges.
- 
+...
+<!-- Omit this part so that users don't use System.map needlessly.
   [mapfile]
     If the live system kernel, or the kernel from which the [dumpfile] 
     was derived, was not compiled with the -g switch, then the additional
@@ -1215,7 +1216,7 @@ Usage:
   [-d num]
     Set crash debug level [num].  The higher the number, the more debug data
     will be printed during crash runtime.
-
+-->
 </pre>
 </td>
 </tr>

How about this?

Thanks,
Kazu