Signed-off-by: Alexander Amelkin <alexander@amelkin.msk.ru>

The endian.h header is not used for anything and was earlier
added by mistake. It however hampers building on some systems
where it doesn't exist.

Resolves ipmitool/ipmitool#101
Signed-off-by: Alexander Amelkin <alexander@amelkin.msk.ru>

Signed-off-by: George Keishing <gkeishin@in.ibm.com>

Swap calls to free() with calls to free_n() to leverage helper method
and handle clearing pointers after freeing in one step.

Signed-off-by: Patrick Venture <venture@google.com>

Check the received IPMI response message id against
the id expected given the IPMI request.  They need to
match.  It is possible that request A times out,
request B is sent, and then request A responds to the
request B.  The value for request B may be behind it
in the queue.

Note: This may only be possible if the file is kept
open between multiple IPMI messages (a common
occurrence).

Resolves: #82

Signed-off-by: Patrick Venture <venture@google.com>

Apply whitespace fixes via clang-format-6.0 to src/plugins/open/open.c
to cleanup spaces and fixup line lengths, etc.

Signed-off-by: Patrick Venture <venture@google.com>

Make the argument to free_n() compatible with any pointers,
thus reducing the number of compilation warnings.

End-user-impact: None
Signed-off-by: Alexander Amelkin <alexander@amelkin.msk.ru>

Swap calls to free() with calls to free_n() to leverage helper method
and handle clearing pointers after freeing in one step.

Signed-off-by: Patrick Venture <venture@google.com>

Check against FRU_AREA_MAXIMUM_BLOCK_SZ instead of FRU_BLOCK_SZ
when checking if the write chunk needs to be reduced.
Apparently, that was the original intention, and then there
was just a typo. In other places the same check is done properly.

Signed-off-by: Patrick Venture <venture@google.com>

Use two macros defining the FRU block sizes instead of hard-coded magic
values.

Signed-off-by: Patrick Venture <venture@google.com>