[Crash-utility] [PATCH] Also search module init section for symbols

Dave Anderson anderson at redhat.com
Tue Oct 5 18:46:45 UTC 2010


----- "Hu Tao" <hutao at cn.fujitsu.com> wrote:

> Hi Dave,
> 
>    When I was investigating the backtrace problem(crash doesn't show
>    some functions from backtrace output), I found the reason was that
>    there was a dead-looping(or anything that would block module init
>    function) in module init function, in which case kernel had no chance
>    to update mod->symtab from mod->core_symtab, and mod->symtab was
>    still referring into the module init section which was not freed
>    until the end of module init function.
> 
>    Since crash never searches module init function for symbols, in the
>    case we can't see any symbol from module from the backtrace output.
> 
>    Following patch makes crash search the module init section for
>    symbols too if the section is not null.
> 
> -- 
> Thanks,
> Hu Tao

In addition to the initial suggestions I had re: your patch, I made 
several additions so that more functions than value_search() and
symbol_dump() would find the init symbols.  With your patch, commands
like these still did not find the init symbols:

 sym <symbol-name>
 sym -q <string>
 sym -n <symbol>
 sym -p <symbol>

I'm running a sanity test now, and if all goes well, will issue a new
release shortly.

Thanks,
  Dave




More information about the Crash-utility mailing list