KingCobra Code-svn

KingCobra usermode Cloud Perfect Forwarding invocation:
-------------------------------------------------------
- kingcobra_commands.c has been updated with a clause for AsFer Cloud Perfect Forwarding to exec()
the client-server binaries via a new script - kingcobra_cloud_perfect_forwarding.sh
- kern.log with a telnet request-reply to VIRGO Queue+KingCobra has been added to testlogs
- paths have been updated

Updated KingCobra Design Document

Updated KingCobra Design Notes and ProductOwnerProfile

Updated ProductOwnerProfile

Updated KingCobra Design Document

GitHub commits:
https://github.com/shrinivaasanka/kingcobra-github-code/commit/b45b2bb7cc966369f92a43d37eb612097c27ae08
https://github.com/shrinivaasanka/kingcobra-github-code/commit/6693e433f25ac6250cc342ea90b7125edf9a1441

KingCobra + VIRGO Queuing port of Linux Kernel 4.1.5 :
- changed the REQUEST_REPLY.queue disk persisted queue path to /var/log/kingcobra/REQUEST_REPLY.queue
- kernel built sources, object files
- kern.log with logs for telnet request sent to VIRGO queue driver, queued in kernel work queue and handler invocation for the
KingCobra service request kernel function for the popped request; disk persisted /var/log/kingcobra/REQUEST_REPLY.queue

Updated ProductOwnerProfile

Updated KingCobra Design Document

KingCobra Linux Kernel 4.1.5 port

KingCobra Module Linux Kernel 4.0.5 port
-----------------------------------------------------
- vfs_write() has a problem in 4.0.5
- the filp_open() args and flags which were working in 3.15.5 cause a
kernel panic implicitly and nothing was written to logs
- It took a very long time to figure out the reason to be vfs_write and filp_open
- O_CREAT, O_RDWR and O_LARGEFILE cause the panic and only O_APPEND is working, but
does not do vfs_write(). All other VIRGO Queue + KingCobra functionalities work viz.,
enqueueing, workqueue handler invocation, dequeueing, invoking kingcobra kernelspace service
request function from VIRGO queue handler, timestamp, timestamp and IP parser, reply_to_publisher etc.,
- As mentioned in Greg Kroah Hartman's "Driving me nuts", persistence in Kernel space is
a bad idea but still seems to be a necessary stuff - yet only vfs calls are used which have to be safe
- Thus KingCobra has to be in-memory only in 4.0.5 if vfs_write() doesn't work
- Intriguingly cloudfs filesystems primitives - virgo_cloud_open, virgo_cloud_read, virgo_cloud_write etc.,
work perfectly and append to a file.
- kern.logs for these have been added to testlogs
- Module.symvers has been recreated for 4.0.5
- updated .ko and .mod.c

(Commit backlog due to repo outage which was resolved today - GitHub KingCobra already has the above changes)