Hello

Thank you for your reply.

Is there no way to get a valid checkpoint in a simulation with caches enabled? 
Is there nothing that I can do (hack)?

Alexandru



________________________________
From: Ali Saidi <[email protected]>
To: Iordan Alexandru <[email protected]>; gem5 users mailing list 
<[email protected]>
Sent: Tuesday, August 9, 2011 9:31 PM
Subject: Re: [gem5-users] About checkpointing


You can't checkpoint the simulation with caches enabled. Any dirty state in the 
caches isn't saved in the checkpoint.
 
Ali
 
 
On Tue, 9 Aug 2011 03:29:47 -0700 (PDT), Iordan Alexandru <[email protected]> 
wrote:
Hello
>I have been having problems using the checkpoint feature in GEM5. I tried to 
>run a very simple scenario:
>1. I launched the simulator with the command:
>./build/ALPHA_FS/gem5.opt --outdir=test/ configs/example/fs.py -n 2 --caches 
>--l2cache
>2. using m5term I created a checkpoint (witch was created with no errors) and 
>then I exited the simulation
>3. I tried to restore using the checkpoint:
>./build/ALPHA_FS/gem5.opt --outdir=test/ configs/example/fs.py -n 2 --caches 
>--l2cache -r 1
>The message that I got in m5term is the following:
>==== m5 slave terminal: Terminal 0 ====
>Unable to handle kernel paging request at virtual address  000000011fd2dd40
>Fixing recursive fault but reboot is needed!
>Unable to handle kernel paging request at virtual address 0000000000000004
>CPU 0 mount(761): Oops 0
>pc = []  ra = []  ps = 0000    Not tainted
>pc is at do_exit+0xf4/0x950
>ra is at do_exit+0x80/0x950
>v0 = 0000000000000007  t0 = 0000000000000004  t1 = fffffc001f4a96d0
>t2 = fffffc000085eca0  t3 = 0000000000000000  t4 = fffffc00008191c0
>t5 = 0000000000000ad7  t6 = 0000000000000031  t7 = fffffc001f354000
>s0 = 0000000000000009  s1 = fffffc001f4a9500  s2 = 0000000000000000
>s3 = 0000000000000000  s4 = 0000000000000000  s5 = 000000011fd2dd40
>s6 = 0000000000000000
>a0 = fffffc001f4a9500  a1 = fffffc000083c108  a2 = ffffffffffffffff
>a3 = 0000000000000000  a4 = 0000000000000008  a5 = 0000000000000009
>t8 =  0000000000000000  t9 = fffffc00006c9534  t10= 0000000000007fff
>t11= 00000000000003ff  pv = fffffc000034f030  at = 0000000000000000
>gp = fffffc000085bf40  sp = fffffc001f357da0
>Trace:
>[] do_page_fault+0x3c0/0x4a0
>[] do_page_fault+0x3ac/0x4a0
>[] entMM+0x9c/0xc0
>[] do_group_exit+0x54/0x100
>
>Code: 402203a1  e4200001  b44400b0  a42a0358  20210004  60004000  40603122 
>Fixing recursive fault but reboot is needed!
>Unable to handle kernel paging request at virtual address 0000000000000004
>CPU 0 mount(761): Oops 0
>pc = []  ra = []  ps = 0000    Tainted: G      D  
>pc is at do_exit+0xf4/0x950
>ra is at do_exit+0x80/0x950
>v0 = 0000000000000007  t0 =  0000000000000004  t1 = fffffc001f4a96d0
>t2 = fffffc000085eca0  t3 = 0000000000000000  t4 = fffffc00008191c0
>t5 = 0000000000000fed  t6 = 0000000000000031  t7 = fffffc001f354000
>s0 = 000000000000000b  s1 = fffffc001f4a9500  s2 = 0000000000000000
>s3 = fffffc001f357c38  s4 = 0000000000000000  s5 = 0000000000000004
>s6 = 0000000000000000
>a0 = fffffc001f4a9500  a1 = fffffc000083c108  a2 = ffffffffffffffff
>a3 = 0000000000000000  a4 = 0000000000000008  a5 = 0000000000000008
>t8 = 0000000000000000  t9 = fffffc00006c9534  t10= 0000000000007fff
>t11= 00000000000003ff  pv = fffffc000034f030  at = 0000000000000000
>gp = fffffc000085bf40  sp = fffffc001f357b10
>Trace:
>[] die_if_kernel+0x12c/0x130
>[] die_if_kernel+0xd4/0x130
>[]  do_page_fault+0x3ac/0x4a0
>[] entMM+0x9c/0xc0
>[] release_console_sem+0x20c/0x310
>[] release_console_sem+0x2b4/0x310
>[] vprintk+0x224/0x530
>[] vprintk+0x224/0x530
>[] do_exit+0x80/0x950
>[] exit_signals+0x0/0x190
>[] do_exit+0xf4/0x950
>[] do_page_fault+0x3c0/0x4a0
>[] do_page_fault+0x3ac/0x4a0
>[] entMM+0x9c/0xc0
>[] do_group_exit+0x54/0x100
>
>Code: 402203a1  e4200001  b44400b0  a42a0358  20210004  60004000  40603122 
>Did I did something wrong here? Do I need to add something in the 
>configuration file before I can use the checkpoint  feature?
>Thank you in advance!
>Alexandru
>
>die_if_kernel recursion detected.
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

Reply via email to