Re: [m5-users] Version 1.1 Full System Benchmark Segmentation Faultwith Multiple CPUs

Ali Saidi Wed, 30 Aug 2006 14:03:11 -0700

Almost certainly your problem is that you hacked the ALPHA_ACCESS_VERSION number. Alpha access 1305 and 1303 are very different versions of the alpha access device, pal code, and console code. Some opcodes were introduced to allow booting of MPs faster, etc. The simulator still shouldn't segfault, so please let us know what segfaulted (m5 or the application) and a provide backtrace if possible.

As for your options with with M5 1.1 you probably should use the console and palcode from the m5 system 1.1 tarball.

Ali

On Aug 30, 2006, at 1:57 PM, Lisa Hsu wrote:

So, what's segfaulting, M5 or the OS? I had assumed M5 was segfaulting since I didn't see a segfault message on the console. There are two gdb's to use when debugging M5 - gdb for the program running on M5, and gdb for M5 itself. Attaching gdb is for debugging the software that M5 is running.

lisa

On 8/30/06, Stewart, Chris (Solutions Alliances) <[EMAIL PROTECTED]> wrote:

The program seg-faults too quickly to get gdb attached. Basically, the main output indicates that gdb is listening and then immediately the Segmentation Fault occurs.

Also, I have as yet been unable to get statistics (m5stats.txt) from the version 2 beta, so I am working on both simultaneously.
Chris Stewart

From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] On Behalf Of Lisa Hsu
Sent: Wednesday, August 30, 2006 12:50 PM
To: M5 users mailing list
Subject: Re: [m5-users] Version 1.1 Full System Benchmark Segmentation Faultwith Multiple CPUs

hi chris,

have you tried running with gdb and doing a backtrace at the fault? that would be more helpful for diagnosis.

also - if you have 2.0 installed - any reason why you're using 1.1?

lisa

On 8/30/06, Stewart, Chris (Solutions Alliances) <[EMAIL PROTECTED]> wrote:

I've successfully run the surge-specweb benchmark with version 1.1 with the following command line:
./build/ALPHA_FS/m5.opt -d myout -ETEST=SURGE_SPECWEB -EFREQUENCY=4GHz configs/fullsys/run.py
./build/ALPHA_FS/m5.opt -d myout -ETEST=SURGE_SPECWEB configs/fullsys/run.py
However, I'm getting a Segmentation Fault when I attempt the following:
./build/ALPHA_FS/m5.opt -d myout -ETEST=SURGE_SPECWEB -ENUMCPUS=2 configs/fullsys/run.py
If it's helpful, here are the console outputs:
Server:
M5 console: m5AlphaAccess @ 0xFFFFFD0200000000 Got Configuration 127 memsize 20000000 pages 10000 First free page after ROM 0xFFFFFC0000018000 HWRPB 0xFFFFFC0000018000 l1pt 0xFFFFFC0000040000 l2pt 0xFFFFFC0000042000 l3pt_rpb 0xFFFFFC0000044000 l3pt_kernel 0xFFFFFC0000048000 l2reserv 0xFFFFFC0000046000 kstart = 0xFFFFFC0000310000, kend = 0xFFFFFC00008064E8, kentry = 0xFFFFFC0000310000, numCPUs = 0x2 CPU Clock at 2000 MHz IntrClockFrequency=1024 Booting with 2 processor(s)
KSP: 0x20043FE8 PTBR 0x20
KSP: 0x20043FE8 PTBR 0x20
Console Callback at 0x0, fixup at 0x0, crb offset: 0x790 Memory cluster 0 [0 - 392] Memory cluster 1 [392 - 65144] Initalizing mdt_bitmap addr 0xFFFFFC0000038000 mem_pages 10000 ConsoleDispatch at virt 100008D8 phys 188D8 val FFFFFC00000100A8 Bootstraping CPU 1 with sp=0xFFFFFC0000076000
Client:
M5 console: m5AlphaAccess @ 0xFFFFFD0200000000 Got Configuration 127 memsize 10000000 pages 8000 First free page after ROM 0xFFFFFC0000018000 HWRPB 0xFFFFFC0000018000 l1pt 0xFFFFFC0000040000 l2pt 0xFFFFFC0000042000 l3pt_rpb 0xFFFFFC0000044000 l3pt_kernel 0xFFFFFC0000048000 l2reserv 0xFFFFFC0000046000 kstart = 0xFFFFFC0000310000, kend = 0xFFFFFC00008064E8, kentry = 0xFFFFFC0000310000, numCPUs = 0x1 CPU Clock at 2000 MHz IntrClockFrequency=1024 Booting with 1 processor(s)
KSP: 0x20043FE8 PTBR 0x20
Console Callback at 0x0, fixup at 0x0, crb offset: 0x510 Memory cluster 0 [0 - 392] Memory cluster 1 [392 - 32376] Initalizing mdt_bitmap addr 0xFFFFFC0000038000 mem_pages 8000 ConsoleDispatch at virt 10000658 phys 18658 val FFFFFC00000100A8 unix_boot_mem ends at FFFFFC0000076000 k_argc = 0 jumping to kernel at 0xFFFFFC0000310000, (PCBB 0xFFFFFC0000018180 pfn 1028) CallbackFixup 0 18000, t7=FFFFFC0000700000
I also had to hack dev/alpha_access.h to change the following:
#define ALPHA_ACCESS_VERSION (1303)
...into...
#define ALPHA_ACCESS_VERSION (1305)
...because I was getting console version mismatch errors before doing so.
A final note: I've got the patched version 2.0 beta installed on the same system as version 1.1, but the system paths are different for each (and so also are the binaries, disk images, etc.)

Chris Stewart

_______________________________________________
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

_______________________________________________
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

Re: [m5-users] Version 1.1 Full System Benchmark Segmentation Faultwith Multiple CPUs

Reply via email to