[Gems-users] workload checkpoints problem


Date: Tue, 18 Nov 2008 18:22:10 +0530
From: "Krishnaiah Gummidipudi" <krishna@xxxxxxxxxxxxxx>
Subject: [Gems-users] workload checkpoints problem
Hi,

I am trying to setup gems-2.1 and simics-2.2.19 with the help of wiki documentation and the scripts provided in gems website.

I have downloaded solaris-10 images and created a checkpoint using the script "naked-check-create.sh" and then used a solaris machine to compile the apache workload as directed in the script "workload-check-create" for the same machine (sarek 16p 65536mb). But I am facing a problem while creating the cold and warm checkpoints for the same workload.

Below is the output that I observe on the xterm of the simulated (target) machine. I am not sure why the "mount" command is not working in the simulated environment.

Could it be because of the corrupt image that was created for solaris-10? Because when I tried to create the image I observed some weird messages like "CRAFF warning: File was not closed cleanly; may be corrupted" and " [qlc24B_2 info] Note: Abort Command Received - Something probably went wrong" on the simics console. But finally the installation was successful with disk images created.

Can you please help how to run benchmarks on simulated machines?

Thanks,
Krishna

///////////////////////Xterm//////////////////////////////

 /mnt/cdrom
# mount -F hsfs -o ro /dev/dsk/c1t6d0s0 /mnt/cdrom
mount: No such device
mount: cannot mount /dev/dsk/c1t6d0s0
# cp /mnt/cdrom/mount_simicsfs /usr/lib/fs/simicsfs/mount
cp: cannot access /mnt/cdrom/mount_simicsfs
# cp /mnt/cdrom/simicsfs-sol10 /usr/kernel/fs/sparcv9/simicsfs
cp: cannot access /mnt/cdrom/simicsfs-sol10
# umount /mnt/cdrom
umount: warning: /mnt/cdrom not in mnttab
umount: /mnt/cdrom not mounted
# echo "simicsfs - /host simicsfs - no -" >> /etc/vfstab
# mkdir /host
# /usr/sbin/ndd -set /dev/tcp tcp_smallest_anon_port 2000
# /usr/sbin/ndd -set /dev/tcp tcp_conn_req_max_q 10240
/usr/sbin/ndd -set /dev/tcp tcp_conn_req_max_q0 10240
# mount /host
mkdir -p /home/phd/# krishna/gems/workloads/build
cp -r /host/extra/gems-2.1/workmountl: Operation not applicable to FSType simicsfs


Below is the output on simics console. Hope there is nothing unexpected here.
/////////////////////////Simics console/////////////////////////
:
[qlc24B_2 info] Disk registering: Loop id: 0x8
[qlc24B_2 info] Disk registering: Loop id: 0x9
[qlc24B_2 info] Disk registering: Loop id: 0xa
[qlc24B_2 info] Disk registering: Loop id: 0xb
[cpu9] v:0x0000000001098010 p:0x0000000802c98010  cmp %g5, 0
Setting new inspection cpu: cpu9
[cpu5] v:0x00000000010416d0 p:0x0000000802c416d0  jmpl [%o7 + 8], %g0
Setting new inspection cpu: cpu5
[cpu2] v:0x00000000010329f4 p:0x0000000802c329f4  sub %i5, 176, %i2
Setting new inspection cpu: cpu2
[cpu1] v:0x0000000001001880 p:0x0000000802c01880  srl %sp, 0, %sp
Setting new inspection cpu: cpu1
[cpu0] v:0x0000000001000480 p:0x0000000802c00480  rdpr %cleanwin, %l0
Setting new inspection cpu: cpu0
[cpu1] v:0x0000000001000c80 p:0x0000000802c00c80  mov 48, %g6
:

[← Prev in Thread] Current Thread [Next in Thread→]
  • [Gems-users] workload checkpoints problem, Krishnaiah Gummidipudi <=