Re: Get Different Result for the same core file with different versions of gdb.

On Tue, Feb 2, 2021 at 11:31 PM Simon Marchi <[email protected]>
wrote:

>
>
> On 2021-02-01 9:15 p.m., Andy Fan via Gdb wrote:
> > On Mon, Feb 1, 2021 at 11:57 AM Andy Fan <[email protected]>
> wrote:
> >
> >> Hi:
> >>
> >> You can pay attention to $2 and $3.They check the same addr in the same
> >> core file.
> >> I'm sure I am debugging the same file in the same machine.
> >>
> >> GDB 7.6:
> >>
> >> (gdb) up 3
> >> #3  ResetPlanCache () at plancache.c:1922
> >> 1922 plancache.c: No such file or directory.
> >> (gdb) p ((AllocSet)plansource->query_context)->freelist
> >> $1 = {0x0, *0x5bdd3b8*, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}
> >> (gdb) p *((AllocSet)plansource->query_context)->freelist[1]
> >> *$2 = {size = 10649824, aset = 0xa283c0 <AllocSetFree>}*
> >> (gdb) p plansource->query_context->name
> >> *$3 = 0xbfd929 "unnamed prepared statement"*
> >> (gdb) x /10c plansource->query_context->name
> >> 0xbfd929: 117 'u' 110 'n' 110 'n' 97 'a' 109 'm' 101 'e' 100 'd' 32 ' '
> >> 0xbfd931: 112 'p' 114 'r'
> >> (gdb) quit
> >>
> >>
> >> GDB  9.2:
> >> (gdb) up 3
> >> #3  ResetPlanCache () at plancache.c:1922
> >> 1922 plancache.c: No such file or directory.
> >> (gdb) p ((AllocSet)plansource->query_context)->freelist
> >> $1 = {0x0, *0x5bdd3b8*, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}
> >> (gdb) p *((AllocSet)plansource->query_context)->freelist[1]
> >> *$2 = {size = 16, aset = 0x0}*
> >> (gdb) p plansource->query_context->name
> >> *$3 = 0xbfd929 ""*
> >> (gdb) x /10c plansource->query_context->name
> >> 0xbfd929: 0 '\000' 0 '\000' 0 '\000' 0 '\000' 0 '\000' 0 '\000' 0
> '\000' 0
> >> '\000'
> >> 0xbfd931: 0 '\000' 0 '\000'
> >>
> >>
> >> Any ideas? Thanks!
> >>
> >> --
> >> Best Regards
> >> Andy Fan
> >>
> >
> > Will it have something with compiler?  The compiler I used
> > to build the software (which generated the core file) is LLVM 6.0.0.
>
> At first glance, it sounds like a bug.  Can you provide a reproducer?
>
> Simon
>

Hi Simon:
  Thanks for your reply!  Actually I am debugging postgresql bug, however
I can't reproduce the core at last,  so the simple answer for your question
is
no, I can't provide the reproducer. I'm sorry about that.  But if you want
any
further analysis on the current core file,  I can try my best.

-- 
Best Regards
Andy Fan (https://www.aliyun.com/)