If I have say 10 different containers, I have to specify 10 different key_addr, right?
Are there any limitations when I randomly specify 10 key_addr?
One obvious limitation is to make sure the difference between 2 key_addr is larger than the potential container size so that 2 containers will not overlap. right?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
No, its easier than that. Each combination of templates represents a new type, so you should reuse the same key_addr and just change the key_val. The template type is
specified by the combination of parameters. So you can reuse the same key_addr and put the various containers all
in the same shared memory segment.
I think mmap will take care of the key_addr values, but yes, different segments should be separated by at least the segment size. The system was designed to place multiple containers in the same segment and has been sucessfully tested doing so. Most systems limit the number of open shared memory segments, so you would be quickly limited if
forced to keep each container in a separate container.
Marc
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
If I have say 10 different containers, I have to specify 10 different key_addr, right?
Are there any limitations when I randomly specify 10 key_addr?
One obvious limitation is to make sure the difference between 2 key_addr is larger than the potential container size so that 2 containers will not overlap. right?
No, its easier than that. Each combination of templates represents a new type, so you should reuse the same key_addr and just change the key_val. The template type is
specified by the combination of parameters. So you can reuse the same key_addr and put the various containers all
in the same shared memory segment.
I think mmap will take care of the key_addr values, but yes, different segments should be separated by at least the segment size. The system was designed to place multiple containers in the same segment and has been sucessfully tested doing so. Most systems limit the number of open shared memory segments, so you would be quickly limited if
forced to keep each container in a separate container.
Marc