On Thursday 10 May 2012, 20:34:50 Hui Kang wrote: > On 5/10/12 4:21 PM, "Stephan Diestelhorst" > > <[email protected]> wrote: > >Am Montag 07 Mai 2012, 19:57:05 schrieb Hui Kang: > >> More information Š. > >> The code snippet is from ooo-pipe.cpp is > >> > >> if unlikely (isclass(uop.opcode, OPCLASS_BRANCH)) { > >> > >> assert(uop.eom); > >> > >> If the uop is a branch class opcode and it should be set an eom. This > >> > >>looks an exception for this case. > >> > >> Any comment? > > > >Hi Hui, > > > > maybe you should also post this on the Marss86 mailing list. It > > > >would be interesting which instruction actually triggered the fault. > >You can either do an objdump of the right binary (library / kernel it > >seems?) or (at least that is the way it works in PTLsim) use the > >-ringbuf to log the ROB's content and then browse backwards through > >the log and see which uops were decoded. [...] > >This certainly looks like a decoder problem, did you tweak anything > >there? The unknown opcode is just an artefact of the assertion. That > >is the way simulation is brought to a rather immediate halt. > > I did not tweak anything in the marssx86 code. The assert stops the > simulation immediately.
The assertion however indicates that some instruction you are simulating decodes badly. That one would be good to find out. What benchmark are you running? Stephan -- Stephan Diestelhorst, AMD Operating System Research Center [email protected] Tel. +49 (0)351 448 356 719 Advanced Micro Devices GmbH Einsteinring 24 85609 Aschheim Germany Geschaeftsfuehrer: Alberto Bozzo u. Andrew Bowd Sitz: Dornach, Gemeinde Aschheim, Landkreis Muenchen Registergericht Muenchen, HRB Nr. 43632, WEEE-Reg-Nr: DE 12919551 _______________________________________________ http://www.marss86.org Marss86-Devel mailing list [email protected] https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel
