Judging by the segfault, you're using an earlier version of M5 that
didn't handle unconnected ports very gracefully. In the versions in the
stable and unstable repositories a port will be connected to a
default/dummy peer. If you ever try sending something to that peer, it
will complain at you instead of segfaulting.

A possible reason you're getting a bad addr would be that Linux is
looking for a device were none exists. It could also be that address
translation is going wrong for some reason and it's accessing junk
addresses instead of physical memory.

If possible, I would recommend updating to the stable repository. This
problem may already be fixed, and if you haven't introduced any new bugs
it may just start working.

Gabe

ef wrote:
> Hello,
>
> I am trying to run m5, and I am getting a segfault during linux boot
> up. I am hoping anyone can give me any advice and insight into fixing
> this error,as there are some issues with  the changes in my M5 version.
>
> Debug information is listed below including Trace Flags and GDB
> information:
>
>
> M5 output:
> 397217555000: system.physmem: Read of size 8 on address 0x78c088 data
> 0x4180000010c
> .....
> 397217555000: system.membus: recvFunctional: packet src 4 dest -1 addr
> 0x4180000010c cmd ReadReq
> 397217555000: system.membus: Unable to find destination for addr:
> 0x4180000010c, will use default port
> Segmentation Fault
>
> GDB output:
> #0  0x00000000004c9bdc in Port::sendFunctional (this=0x0,
> pkt=0x7fffc259cd30) at build/ALPHA_FS/mem/port.hh:215
>
>
>   I am pretty inexperienced when it comes to M5, does anyone know
> anything else I can do to try to isolate the error and get more
> information in hopes of fixing it. The real issue is not the segfault,
> but the bad addr being received from physical memory.
>
> Also shouldn't M5 stop running as soon as the destination is not
> found, as the membus does not have a default port, which is causing
> the seg fault.
>
>
> Thanks
> EF
> ------------------------------------------------------------------------
>
> _______________________________________________
> m5-users mailing list
> [email protected]
> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users

_______________________________________________
m5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users

Reply via email to