This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH 2/3] record-btrace, disas: omit pc prefix
On Thu, 28 Feb 2013 08:03:04 +0100, Metzger, Markus T wrote:
> Within a loop, you will see repeating => very frequently.
>
> (gdb) rec inst -
> 0x00000000004006ec <_ZL4loopPv+8>: mov 0x200955(%rip),%rax # 0x601048 <_ZL6shared>
> => 0x00000000004006f3 <_ZL4loopPv+15>: add $0x1,%rax
> 0x00000000004006f7 <_ZL4loopPv+19>: mov %rax,0x20094a(%rip) # 0x601048 <_ZL6shared>
> 0x00000000004006fe <_ZL4loopPv+26>: jmp 0x4006ec <_ZL4loopPv+8>
> 0x00000000004006ec <_ZL4loopPv+8>: mov 0x200955(%rip),%rax # 0x601048 <_ZL6shared>
> => 0x00000000004006f3 <_ZL4loopPv+15>: add $0x1,%rax
> 0x00000000004006f7 <_ZL4loopPv+19>: mov %rax,0x20094a(%rip) # 0x601048 <_ZL6shared>
> 0x00000000004006fe <_ZL4loopPv+26>: jmp 0x4006ec <_ZL4loopPv+8>
> 0x00000000004006ec <_ZL4loopPv+8>: mov 0x200955(%rip),%rax # 0x601048 <_ZL6shared>
> => 0x00000000004006f3 <_ZL4loopPv+15>: add $0x1,%rax
> (gdb)
I did not think about this case but I still find it correct, => should display
the current PC and it is the current PC.
> > (gdb) record instruction-history
> > warning: Recorded trace may be corrupted.
> > warning: Recorded trace may be corrupted.
> > warning: Recorded trace may be corrupted.
>
> That's unexpected. If you can reproduce this, would you please send
> me the reproducer so I can have a look at it?
I have patched intel_supports_btrace on my model 0x1a, couldn't it be the
reason?
Jan