No longer truncate passwords (16 < p <= 20) silently, instead attempt
to set a 20-char password when such a password is given.
Fail if an explicit length is exceeded, and any time the upper limit
is exceeded.
Remove all direct comparisons to 'NULL' for pointers.
Replace them with boolean-like 'if (ptr)' and 'if (!ptr)'.
This makes conditions shorter and easier to read.
Signed-off-by: Alexander Amelkin <alexander@amelkin.msk.ru>
Make code better readable by replacing `if (rsp->ccode > 0)`
and 'if (rsp->ccode != 0)' with just `if (rsp->ccode)` as
rsp->ccode is anyway an unsigned byte and can't be negative.
Also replace 'if (rsp->ccode == 0)' with 'if (!rsp->ccode)'.
All these changes make lines shorter and easier to read as
a non-zero ccode is an indication of some failure, and so !ccode
naturally reads as 'no error'.
Signed-off-by: Alexander Amelkin <alexander@amelkin.msk.ru>
There are lots of feature test macros like _POSIX_SOURCE,
_BSD_SOURCE, etc. chaotically placed at the beginning of
multiple files without apparent reason, sometime purely
erroneously.
This commit removes them all for the sake of cleaner code.
No new warnings are added due to that at least for gcc 5.4
on Linux with glibc 2.23.
In multiple places throughout ipmi_user.c a user id mask
was used as a magic number, in some places the mask was wrong.
This commit replaces all those magic numbers with a single
IPMI_UID() macro.
Resolvesipmitool/ipmitool#6
Commit adds macros and #include in order to bring down the number of `warning:
implicit declaration of function 'X' [-Wimplicit-function-declaration]`
warnings.
Commit adds a work-around for some BMCs which return ccode 0xCC when user is
disabled. However, this isn't reason to stop listing users as this ccode is
perceived as being "normal".
When 0xCC is returned, empty user name will be printed instead of bailing out.
Commit hooks ipmi_user_priv() to _ipmi_set_user_access(). Later got extended for
'change_priv_limit_only', because of the former.
Commit removes ipmi_user_set_userpriv(), because it's not used anywhere.
Commit fixes bug in eval logic of user input sanitization, resp. && -> ||. Also,
priv_level and user_id variables are initialized to 0 now(CID#1261318).
Commit adds option to 'user set password' to choose whether password should be
stored as 16 byte or 20 byte. From now on, it's possible to store even 5 bytes
long password as 20 byte password which should make it useless for IPMI v1.5
auth/sessions.
Despite the fact we could leave this up to IPMI stack, we won't do that.
Therefore, is_ipmi_user_priv_limit() is added and user provided value checked.
