NAME

hps_dloop, hps_rloop − MVME338 in-service diagnostic loopback programs

SYNOPSIS

hps_dloop [ −crq ] [−d device ] [ −l limit ] [ −n bcount ] [ −s bsize ] [ −t time ] hps_rloop [ −crq ] −a addr  [ −d device ] [ −l limit ] [ −n bcount ] [ −s bsize ] [ −t time ]

DESCRIPTION

Hps_dloop and hps_rloop interface with the downloaded MVME338 Diagnostic Application to perform DMA loopback and remote cluster controller loopback tests, respectively.  The basic function of the tests are identical: they send a block to the board,  then read it back in, verifying that its contents are unchanged.  The DMA loopback test ( hps_dloop ) simply moves the block (via DMA operations or dual port RAM as appropriate) to the board and back.  The remote loopback test ( hps_rloop ) moves the block to the board, then transmits it to the specified cluster controller and back again to the host. 

OPTIONS

One or more of the following options must be selected:

−aSet cluster controller address.  With the hps_rloop program, this option selects which cluster controller address to run the diagnostic loopback through.  The cluster controller address is specified as a decimal value in the addr parameter on the command line.  This option must be used with the hps_rloop program, but it may not be used with the hps_dloop program. 

−cDisable checking results.  This option tells the program not to check the results of the DMA or remote loopback to verify if the block read back in matches the block sent out.  The default is to check blocks. 

−dSet MVME338 raw device file name.  This option allows the user to override the default MVME338 device name.  If this option is selected, the overriding device name must be specified in the device parameter on the command line. 

−lSet throughput limit.  This option allows a throttling of the test to prevent too much of the board interface bandwidth to be consumed.  If this option is selected, the maximum throughput rate (in characters per second) must be specified in the limit parameter on the command line.  The default throughput limit is 40000. 

−nSet block count.  This option tells the program how many loopback blocks to send before terminating.  The default is to send an unlimited number of blocks.  If this option is selected, the number of blocks to send must be specified in the bcount parameter on the command line.  Setting this option disables the −t option, if entered before the −t option. 

−rUse random sized blocks.  This option specifies that the program randomly varies the size of the blocks used for the test.  The block sizes will range from 1 to 512 for hps_dloop, and 1 to 490 for hps_rloop.  The default is to not use random sized blocks.  Setting this option disables the −s option, if entered before the −s option. 

−sSet fixed block size.  This option specifies that fixed size blocks be used for the test.  Note that this option overrides the random option.  If this option is selected, fixed block size must be specified in the bsize parameter on the command line.  The maximum block size is 512 for hps_dloop, and 490 for hps_rloop.  Setting this option disables the −r option, if entered before the −r option. 

−tSet running time.  This option tells the program how long to run before terminating.  The default is to run for 30 seconds.  If this option is selected, the number of seconds to run must be specified in the time parameter on the command line.  Setting this option disables the −n option, if entered before the −n option. 

−qQuiet operation.  Normally the program reports the test results to standard output.  If this option is specified, the program will run silently unless any errors occur. 

EXAMPLES

To run the DMA loopback to the default device for 3 minutes, using random block sizes, and display the results on standard output:

 hps_dloop -r -t 180
 

To run the remote loopback through the second host adapter to the cluster controller at address 3 for 100 fixed size (256 byte) blocks, and display the results on standard output:

 hps_rloop ns -a 3 -d /dev/xedt/mvme338_c1 -n 100 -s 256
 

FILES

Both hps_rloop and hps_dloop are linked to the executable file hps_loop .  The program distinquishes the function requested by the name of the invoked file. 

ERRORS

The following errors can be reported by the MVME338:

E_OFFLINE
occurs when the Network Manager attempts but fails to transmit a packet of data to a cluster controller because the host adapter is not attached to the network.

E_NORESP
occurs when an attempt is made to access a cluster controller that does not respond.

E_BADRESP
occurs when a request for data is made from a cluster controller and incorrect data returned.

E_BUSY
occurs when a request for a diagnostic connection has been made and there are no more connections available.  There is a maximum of 8 instances of both hps_rloop and hps_dloop at any time. 

SEE ALSO

mvme338(7)

WARNINGS

You are only allowed to kill the process with a kill [process_id] command or with an interrupt if the job is running in the foreground.  Do not use a kill -9 command to kill this program because that will cause the program to "die," but will not close the channel properly. 
 
Revision 1.1  92/09/01  14:52:49  annette Initial revision