[ofa-general] difference between flint and mstflint
jean-vincent.ficet at bull.net
Fri Jan 16 01:58:31 PST 2009
While using mstflint to read the characteristics of an IS4 switch, it
fails as follows:
[user at host ~ ] mstflint -d "lid-11" q
Unable to parse device name lid-11
*** ERROR *** Can not open lid-11: Invalid argument MFE_CR_ERROR
However, when I use flint, it works fine:
[user at host ~ ] flint -d "lid-11" q
Image type: FS2
FW Version: 7.1.0
Device ID: 48438
Chip Revision: A0
Description: Node Sys image
GUIDs: 0002c90200404798 0002c9020040479b
Board ID: (MT_0C20110003)
Looking at the flint code (MFT 2.5) and the mstflint code (git head),
there is no difference apart from the version ID for the following files:
Digging deeper into the issue I found that difference was that:
- mstflint uses the mopen() function implemented in mtcr.h. This
function does not support the 'lid-' syntax
- flint uses the mopend() function implemented in a separate library
(libmtcr.a) and supports the 'lid-' syntax.
This library is built using the files mtcr.c mtcr_ib.c mtcr_i2c.c
usbioctl.c located in /usr/mst/src.
Why doesn't mstflint use the same library as flint (hence providing
better functionnality for inband analysis), knowing that both flint and
mstflint already share a fair amount of common code ?
Is there any plan to push the library and its corresponding source files
into the mstflint git repository ?
Thanks for your help,
More information about the general