Re: GDB bug or something else?

[Sad Clouds <cryintothebluesky%gmail.com@localhost>]

On Fri, 23 Oct 2026 16:08:27 +0300
Valery Ushakov <uwe%stderr.spb.ru@localhost> wrote:

> On Thu, Oct 22, 2026 at 09:29:04 +0100, Sad Clouds wrote:
> 
> > The test program is attached. Can anyone think of reasons why single
> > stepping through the program with GDB results in incorrect value read
> > from the stack, instead of 0x12345? Using GDB x/xg command to examine
> > the value at the specified address also gives the same incorrect value.
> 
> Register window flushed?  Print $i6 after writing to the stack.  You
> are probably getting _taht_ value back when you read and print it.
> 
> -uwe

Yes you may be right. Looks like GDB is causing side effects which
affect the memory image of the debugged program. I'm cautious now when
trying to examine the register window save area of a running program via
GDB on sparc. I think this should not affect debug sessions on core
dumps, as the process is dead and not running on CPU.