gem5-users@gem5.org

The gem5 Users mailing list

View all threads

Problems in Generating Checkpoint of the Booting process of Full System Mode

黄泳杰
Sat, Apr 6, 2024 3:34 PM

Dear friends,

I want to record a checkpoint as soon as the operating system is booted succesfully.  My command line is bellow:
nohup /mydata/gem5/build/X86/gem5.opt --outdir=/mydata/fullSystemNPBSplash -r -e /mydata/gem5/configs/example/fs.py --disk-image=/mydata/gem5/fs_image_and_kernel/disks/base.img --kernel=/mydata/gem5/fs_image_and_kernel/binaries/vmlinux-5.4.49 --num-cpus=1 --cpu-type=AtomicSimpleCPU --checkpoint-dir=/mydata/gem5/checkpoint_fullSystemNPBSplash --script=/mydata/gem5/configs/boot/hack_back_ckpt.rcS &

But, the process runs ceaselessly, though there is 'root@gem5-host:/#'  in system.pc.com_1.device. And I can type ls, top and so forth with m5term.   Both means that the system is running, but it just does not generate the checkpoint and exit.
Besides, I can manually write a checkpoint with 'm5 checkpoint' in m5term.   
(I alse try another two command line-
1)  /mydata/gem5/build/X86/gem5.opt -d /mydata/gem5/checkpoint_fullSystemNPBSplash /mydata/gem5/configs/example/fs.py --disk-image=/mydata/gem5/fs_image_and_kernel/disks/base.img --kernel=/mydata/gem5/fs_image_and_kernel/binaries/vmlinux-5.4.49 --num-cpus=1 --cpu-type=AtomicSimpleCPU  [I type m5 checkpoint in m5term manually]
2)nohup /mydata/gem5/build/X86/gem5.opt -d /mydata/gem5/checkpoint_fullSystemNPBSplash2  /mydata/gem5/configs/example/fs.py  --disk-image=/mydata/gem5/fs_image_and_kernel/disks/base.img --kernel=/mydata/gem5/fs_image_and_kernel/binaries/vmlinux-5.4.49 --num-cpus=1 --cpu-type=AtomicSimpleCPU --script=/mydata/gem5/configs/boot/hack_back_ckpt.rcS &
All of them did not write a checkpoint and exit, instead the  gem5 process run permanently.
)

However, When I restore that checkpoint with command line- /mydata/gem5/build/X86/gem5.opt --outdir=/mydata/fullSystemNPBSplash -r 1 /mydata/gem5/configs/example/fs.py --disk-image=/mydata/gem5/fs_image_and_kernel/disks/base.img --kernel=/mydata/gem5/fs_image_and_kernel/binaries/vmlinux-5.4.49 --num-cpus=1 --restore-with-cpu=DerivO3CPU --cpu-type=DerivO3CPU --caches --l2cache --checkpoint-dir=/mydata/gem5/checkpoint_fullSystemNPBSplash
It does nothing but boot the system from scratch  as if it is run without a checkpoint.

What's wrong with these?

Thanks,
Yongjie

Dear friends, I want to record a checkpoint as soon as the operating system is booted succesfully.  My command line is bellow: nohup /mydata/gem5/build/X86/gem5.opt --outdir=/mydata/fullSystemNPBSplash -r -e /mydata/gem5/configs/example/fs.py --disk-image=/mydata/gem5/fs_image_and_kernel/disks/base.img --kernel=/mydata/gem5/fs_image_and_kernel/binaries/vmlinux-5.4.49 --num-cpus=1 --cpu-type=AtomicSimpleCPU --checkpoint-dir=/mydata/gem5/checkpoint_fullSystemNPBSplash --script=/mydata/gem5/configs/boot/hack_back_ckpt.rcS & But, the process runs ceaselessly, though there is 'root@gem5-host:/#'  in system.pc.com_1.device. And I can type ls, top and so forth with m5term.   Both means that the system is running, but it just does not generate the checkpoint and exit. Besides, I can manually write a checkpoint with 'm5 checkpoint' in m5term.    (I alse try another two command line- 1)  /mydata/gem5/build/X86/gem5.opt -d /mydata/gem5/checkpoint_fullSystemNPBSplash /mydata/gem5/configs/example/fs.py --disk-image=/mydata/gem5/fs_image_and_kernel/disks/base.img --kernel=/mydata/gem5/fs_image_and_kernel/binaries/vmlinux-5.4.49 --num-cpus=1 --cpu-type=AtomicSimpleCPU  [I type m5 checkpoint in m5term manually] 2)nohup /mydata/gem5/build/X86/gem5.opt -d /mydata/gem5/checkpoint_fullSystemNPBSplash2  /mydata/gem5/configs/example/fs.py  --disk-image=/mydata/gem5/fs_image_and_kernel/disks/base.img --kernel=/mydata/gem5/fs_image_and_kernel/binaries/vmlinux-5.4.49 --num-cpus=1 --cpu-type=AtomicSimpleCPU --script=/mydata/gem5/configs/boot/hack_back_ckpt.rcS & All of them did not write a checkpoint and exit, instead the  gem5 process run permanently. ) However, When I restore that checkpoint with command line- /mydata/gem5/build/X86/gem5.opt --outdir=/mydata/fullSystemNPBSplash -r 1 /mydata/gem5/configs/example/fs.py --disk-image=/mydata/gem5/fs_image_and_kernel/disks/base.img --kernel=/mydata/gem5/fs_image_and_kernel/binaries/vmlinux-5.4.49 --num-cpus=1 --restore-with-cpu=DerivO3CPU --cpu-type=DerivO3CPU --caches --l2cache --checkpoint-dir=/mydata/gem5/checkpoint_fullSystemNPBSplash It does nothing but boot the system from scratch  as if it is run without a checkpoint. What's wrong with these? Thanks, Yongjie