Re: [PATCH v4] gdb: ADI support

[Wei-min Pan <weimin dot pan at oracle dot com>]

On 7/28/2017 3:30 AM, Yao Qi wrote:
> On Thu, Jul 27, 2017 at 5:34 PM, Wei-min Pan <weimin.pan@oracle.com> wrote:
>
> > The first "adi x shmaddr" command is mainly used to test that the ADI
> > information is available
> > at the address. Its value which can be anything is not tested. The second
> > "examine" command,
> > on the other hand, makes sure the new ADI version tag is being set correctly
> > at the address.
> Can you show me the output of these adi commands?
> (the output in gdb.log for example)Then, I can give some
> concrete suggestions on the patterns to match output
> in these tests.

Here are the outputs of these commands. Thanks.

....
Breakpoint 2, main () at 
/home/wepan/work/binutils-gdb/gdb/testsuite/gdb.arch/sparc64-adi.c:86^M
86        if (memory_fill (shmaddr, SHMSIZE, PAT) != 0) /* line 
breakpoint here */^M
(gdb) PASS: gdb.arch/sparc64-adi.exp: continue to breakpoint: continue 
to line breakpoint in main
adi x shmaddr^M
0xfff800010002c000:     0 ^M
(gdb) PASS: gdb.arch/sparc64-adi.exp: examine ADI
adi a/100 shmaddr=7^M
(gdb) PASS: gdb.arch/sparc64-adi.exp: assign ADI
adi x/100 shmaddr^M
0xfff800010002c000:     7 7 ^M
(gdb) PASS: gdb.arch/sparc64-adi.exp: examine new ADI
adi a/100 shmaddr=0x0^M
(gdb) PASS: gdb.arch/sparc64-adi.exp: reset ADI
continue^M
Continuing.^M
^M
Program received signal SIGSEGV, Segmentation fault^M
ADI precise mismatch while accessing address 0xfff8000100048000.^M
0xfff8000100129f38 in adi_set_version () from /usr/lib64/libadi.so^M
(gdb) PASS: gdb.arch/sparc64-adi.exp: continue to sigsegv