[Bug gdb/30246] New: gdb record and replay , sometime variable operate wrong. but rr works, based on ryzen 4800h cpu(zen architecture)

["czgf2v at 163 dot com" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=30246

            Bug ID: 30246
           Summary: gdb record and replay , sometime variable operate
                    wrong. but rr works, based on ryzen 4800h cpu(zen
                    architecture)
           Product: gdb
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: gdb
          Assignee: unassigned at sourceware dot org
          Reporter: czgf2v at 163 dot com
  Target Milestone: 14.1

run 
---------------------
$make -f Makefile;gdb ./main -ex "source ./debug.gdb"
---------------------

after a while,(this bug happens after doing many operations with random numbers
and ‘rc’ command)

output on my machine (the output is wrong)
-------------
...
Active record target: record-full
Replay mode:
Lowest recorded instruction number is 1.
Current instruction number is 257733.
Highest recorded instruction number is 257773.
Log contains 257773 instructions.
Max logged instructions is 2000000000.
$1 = 10100110110111001001101
$2 = 101001101101110010011010
$3 = 0
-------------

but with rr
-------------
$ rr ./main
$ rr replay /home/czg/.local/share/rr/main-3
...
source ./debug_rr.gdb
-------------
(because in rr use 'next' to run and no ‘record’ command needed [rr has
recorded all before], so gdb command file slightly differ)


output on my machine (the output is right)
-------------
$1 = 111111010111001001000010
$2 = 111111010111001001000010
$3 = 1
-------------

-- 
You are receiving this mail because:
You are on the CC list for the bug.