Fix comment about skipped interfaces

After r1.44 "Start on DOWN interfaces" this sentence makes no sense
and just repeats the obvious conditions, so zap it.

With/OK stsp

Start on DOWN interfaces

Do not skip interfaces during startup because they're not UP.

dhcpd(8) can start listening on such interfaces just fine (given suitable
IPs on them) and interfaces may come UP later in time.

Otherwise dhcpd would fail to start completely if it didn't find any
interface that would be UP early on and administrators had to restart dhcpd.

Other daemons and nc(1)'s `-l' listen just fine on DOWN interfaces as well.

Noticed while switching a bridge(4)/vether(4) setup to veb(4)/vport(4) and
hostname.vport without explicit "up" (vport does not implicitly pulls itself
UP when configuring an address).

OK dlg florian

One 'foo' per function is enough. Rename one to eliminate shadowing.

Eliminate pointless'%m' (a.k.a. hand rolled strerror()) by using fatal() and
log_warn(). Zap a couple of explicit 'syslog()' calls.

Adjust some long lines.

Do the strerror() elimination dance with log_warnx() -> log_warn(),
fatalx() -> fatal() and even a couple of fprintf(stderr) -> log_warn().

Switch from old errwarn.c logging to shiny new log.[ch].

ok benno@

Replace bzero(3) with memset(3)

"Looks good" to deraadt@

Pull in <time.h> for one or more of gmtime, strftime, strptime, time,
timegm, and tzset

ok deraadt@

Eliminate #include inside *.h files and include only needed headers in
each *.c file.

Inspired by mention of header silliness by Edgar Pettijohn and mmcc@
on tech@.

stdlib.h is in scope; do not cast malloc/calloc/realloc*
ok millert krw

use reallocarray() to cope with multiplicative integer overflow; ok doug

Don't call the BOOTP handler indirectly.

ok krw

Scattering a few bcopy()'s around source using mostly memcpy() is just
asking for confusion. Replace bcopy()'s with memcpy()'s.

poll(2) might have slept for an arbitrary amount of time. We should
update 'cur_time' after poll(2) returns.

ok otto

Log (at LOG_INFO) which interfaces dhcpd listens to and (as suggested by
krw) their address. As dhcpd uses BPF, it's not possible to find this
information out from fstat/netstat. ok krw@

Zero option processing stack storage to prevent stack garbage that
happens to look like options from being processed as options when
truncated packets received.

Potential problem pointed out by mikeb@, from a couple of NIST
reports (CVE-2011-2748 and CVE-2011-2749).

Prevent dhcpd(8) from trying to listen on interfaces that don't have a
valid broadcast (e.g. pflog0).

"no objection" krw@, ok claudio@

Fix the naming of interfaces and variables for rdomains and rtables
and make it possible to bind sockets (including listening sockets!)
to rtables and not just rdomains. This changes the name of the
system calls, socket option, and ioctl. After building with this
you should remove the files /usr/share/man/cat2/[gs]etrdomain.0.

Since this removes the existing [gs]etrdomain() system calls, the
libc major is bumped.

Written by claudio@, criticized^Wcritiqued by me

If interfaces are specified, get their rdomain and bind dhcpd into the same
domain with setrdomain(). This allows to run dhcpd on multiple rdomains.
OK krw@

Eliminate all uses of dmalloc() where the returned pointer
is checked for NULL and a specific error/warning issued. Add
two such manual warning/error checks and kill those dmalloc
calls. And then there were none, so kill dmalloc(). Whew.

When checking if a syscall like open(), ioctl() or writev() failed compare
directly against -1 and do not use a < 0 test. OK henning@

Add synchronisation support for dhcpd - this allows for two dhcpd's
with the same configuration to be run on the same net and they will
keep their lease files/state in synch, and therefore allowing you to
run redundant dhcpd's. Synchronization code stolen from spamd, uses
an hmac key in /var/db/dhcpd.key if it exists.
ok krw@ deraadt@

remove unused #defines; no binary change

This patch renames dhcpd's "struct timeout" to "struct dhcpd_timeout"
so as not to conflict with "struct timeout" from <sys/timeout.h>.
ok henning@

Don't exit just because an interface has no subnet configured in
dhcpd.conf. Just don't listen on that interface.

Make warning messages textually consistant.

tested & ok beck@

When an interface is ignored because it has no IP address, say so.
Don't just say 'not found". And always complain, not just when
specific interfaces were requested.

requested by and ok uwe@

If a list of interfaces is supplied via the command line or
dhcpd.interfaces then

a) don't bother looking up information on interfaces that were not
requested;

b) don't exit if a requested interface is not found, just issue a
warning message;

c) exit if none of the interfaces were found.

The command line for dhcpd shown in ps will continue to show requested
but ignored interfaces.

As usual with dhc* code, whack a bunch of unused states, constants,
flags, etc. Since we only invoke discover_interfaces() with
DISCOVER_SERVER, there is no need to keep track of other
possibilities.

KNF. No binary change.

Don't malloc fds each time, just realloc() as needed.
There's no need to clear revents -- poll() does that for us.
Move setting of cur_time to the top of the loop for better accuracy.
When poll returns 0, don't check revents since we know none were set.

change the undocumented -t flag to -n used for configuration testing
and document it. cleanup some unused/unneeded variables while there.
idea & ok henning@

spelling fixes; ok jmc@ henning@

avoid aliasing with libc functions; ok henning

Remove the not initialized fallback_interface code and use the normal bpf
interface for that. Also store the hardware (ethernet) address in the state
so that relayed dhcp request are sent to the correct destination.
OK henning@ requested by deraadt@ tested otto@

size_t vs ssize_t confusion

handle interface removals gracefully. reported by Oliver J. Morais
ok henning@, tedu@

remove things not used, spotted by lint mostly; ok henning

$OpenBSD$

more KNF

various knf; henning ok

a bunch of knf, ok henning

spaces

more crap bites the dust

time_t and kill a few dumb defines

may the whacking begin

Start on DOWN interfaces

Do not skip interfaces during startup because they're not UP.

dhcpd(8) can start listening on such interfaces just fine (given suitable
IPs on them) and interfaces may come UP later in time.

Otherwise dhcpd would fail to start completely if it didn't find any
interface that would be UP early on and administrators had to restart dhcpd.

Other daemons and nc(1)'s `-l' listen just fine on DOWN interfaces as well.

Noticed while switching a bridge(4)/vether(4) setup to veb(4)/vport(4) and
hostname.vport without explicit "up" (vport does not implicitly pulls itself
UP when configuring an address).

OK dlg florian

One 'foo' per function is enough. Rename one to eliminate shadowing.

Eliminate pointless'%m' (a.k.a. hand rolled strerror()) by using fatal() and
log_warn(). Zap a couple of explicit 'syslog()' calls.

Adjust some long lines.

Do the strerror() elimination dance with log_warnx() -> log_warn(),
fatalx() -> fatal() and even a couple of fprintf(stderr) -> log_warn().

Switch from old errwarn.c logging to shiny new log.[ch].

ok benno@

Replace bzero(3) with memset(3)

"Looks good" to deraadt@

Pull in <time.h> for one or more of gmtime, strftime, strptime, time,
timegm, and tzset

ok deraadt@

Eliminate #include inside *.h files and include only needed headers in
each *.c file.

Inspired by mention of header silliness by Edgar Pettijohn and mmcc@
on tech@.

stdlib.h is in scope; do not cast malloc/calloc/realloc*
ok millert krw

use reallocarray() to cope with multiplicative integer overflow; ok doug

Don't call the BOOTP handler indirectly.

ok krw

Scattering a few bcopy()'s around source using mostly memcpy() is just
asking for confusion. Replace bcopy()'s with memcpy()'s.

poll(2) might have slept for an arbitrary amount of time. We should
update 'cur_time' after poll(2) returns.

ok otto

Log (at LOG_INFO) which interfaces dhcpd listens to and (as suggested by
krw) their address. As dhcpd uses BPF, it's not possible to find this
information out from fstat/netstat. ok krw@

Zero option processing stack storage to prevent stack garbage that
happens to look like options from being processed as options when
truncated packets received.

Potential problem pointed out by mikeb@, from a couple of NIST
reports (CVE-2011-2748 and CVE-2011-2749).

Prevent dhcpd(8) from trying to listen on interfaces that don't have a
valid broadcast (e.g. pflog0).

"no objection" krw@, ok claudio@

Fix the naming of interfaces and variables for rdomains and rtables
and make it possible to bind sockets (including listening sockets!)
to rtables and not just rdomains. This changes the name of the
system calls, socket option, and ioctl. After building with this
you should remove the files /usr/share/man/cat2/[gs]etrdomain.0.

Since this removes the existing [gs]etrdomain() system calls, the
libc major is bumped.

Written by claudio@, criticized^Wcritiqued by me

If interfaces are specified, get their rdomain and bind dhcpd into the same
domain with setrdomain(). This allows to run dhcpd on multiple rdomains.
OK krw@

Eliminate all uses of dmalloc() where the returned pointer
is checked for NULL and a specific error/warning issued. Add
two such manual warning/error checks and kill those dmalloc
calls. And then there were none, so kill dmalloc(). Whew.

When checking if a syscall like open(), ioctl() or writev() failed compare
directly against -1 and do not use a < 0 test. OK henning@

Add synchronisation support for dhcpd - this allows for two dhcpd's
with the same configuration to be run on the same net and they will
keep their lease files/state in synch, and therefore allowing you to
run redundant dhcpd's. Synchronization code stolen from spamd, uses
an hmac key in /var/db/dhcpd.key if it exists.
ok krw@ deraadt@

remove unused #defines; no binary change

This patch renames dhcpd's "struct timeout" to "struct dhcpd_timeout"
so as not to conflict with "struct timeout" from <sys/timeout.h>.
ok henning@

Don't exit just because an interface has no subnet configured in
dhcpd.conf. Just don't listen on that interface.

Make warning messages textually consistant.

tested & ok beck@

When an interface is ignored because it has no IP address, say so.
Don't just say 'not found". And always complain, not just when
specific interfaces were requested.

requested by and ok uwe@

If a list of interfaces is supplied via the command line or
dhcpd.interfaces then

a) don't bother looking up information on interfaces that were not
requested;

b) don't exit if a requested interface is not found, just issue a
warning message;

c) exit if none of the interfaces were found.

The command line for dhcpd shown in ps will continue to show requested
but ignored interfaces.

As usual with dhc* code, whack a bunch of unused states, constants,
flags, etc. Since we only invoke discover_interfaces() with
DISCOVER_SERVER, there is no need to keep track of other
possibilities.

KNF. No binary change.

Don't malloc fds each time, just realloc() as needed.
There's no need to clear revents -- poll() does that for us.
Move setting of cur_time to the top of the loop for better accuracy.
When poll returns 0, don't check revents since we know none were set.

change the undocumented -t flag to -n used for configuration testing
and document it. cleanup some unused/unneeded variables while there.
idea & ok henning@

spelling fixes; ok jmc@ henning@

avoid aliasing with libc functions; ok henning

Remove the not initialized fallback_interface code and use the normal bpf
interface for that. Also store the hardware (ethernet) address in the state
so that relayed dhcp request are sent to the correct destination.
OK henning@ requested by deraadt@ tested otto@

size_t vs ssize_t confusion

handle interface removals gracefully. reported by Oliver J. Morais
ok henning@, tedu@

remove things not used, spotted by lint mostly; ok henning

$OpenBSD$

more KNF

various knf; henning ok

a bunch of knf, ok henning

spaces

more crap bites the dust

time_t and kill a few dumb defines

may the whacking begin

One 'foo' per function is enough. Rename one to eliminate shadowing.

Eliminate pointless'%m' (a.k.a. hand rolled strerror()) by using fatal() and
log_warn(). Zap a couple of explicit 'syslog()' calls.

Adjust some long lines.

Do the strerror() elimination dance with log_warnx() -> log_warn(),
fatalx() -> fatal() and even a couple of fprintf(stderr) -> log_warn().

Switch from old errwarn.c logging to shiny new log.[ch].

ok benno@

Replace bzero(3) with memset(3)

"Looks good" to deraadt@

Pull in <time.h> for one or more of gmtime, strftime, strptime, time,
timegm, and tzset

ok deraadt@

Eliminate #include inside *.h files and include only needed headers in
each *.c file.

Inspired by mention of header silliness by Edgar Pettijohn and mmcc@
on tech@.

stdlib.h is in scope; do not cast malloc/calloc/realloc*
ok millert krw

use reallocarray() to cope with multiplicative integer overflow; ok doug

Don't call the BOOTP handler indirectly.

ok krw

Scattering a few bcopy()'s around source using mostly memcpy() is just
asking for confusion. Replace bcopy()'s with memcpy()'s.

poll(2) might have slept for an arbitrary amount of time. We should
update 'cur_time' after poll(2) returns.

ok otto

Log (at LOG_INFO) which interfaces dhcpd listens to and (as suggested by
krw) their address. As dhcpd uses BPF, it's not possible to find this
information out from fstat/netstat. ok krw@

Zero option processing stack storage to prevent stack garbage that
happens to look like options from being processed as options when
truncated packets received.

Potential problem pointed out by mikeb@, from a couple of NIST
reports (CVE-2011-2748 and CVE-2011-2749).

Prevent dhcpd(8) from trying to listen on interfaces that don't have a
valid broadcast (e.g. pflog0).

"no objection" krw@, ok claudio@

Fix the naming of interfaces and variables for rdomains and rtables
and make it possible to bind sockets (including listening sockets!)
to rtables and not just rdomains. This changes the name of the
system calls, socket option, and ioctl. After building with this
you should remove the files /usr/share/man/cat2/[gs]etrdomain.0.

Since this removes the existing [gs]etrdomain() system calls, the
libc major is bumped.

Written by claudio@, criticized^Wcritiqued by me

If interfaces are specified, get their rdomain and bind dhcpd into the same
domain with setrdomain(). This allows to run dhcpd on multiple rdomains.
OK krw@

Eliminate all uses of dmalloc() where the returned pointer
is checked for NULL and a specific error/warning issued. Add
two such manual warning/error checks and kill those dmalloc
calls. And then there were none, so kill dmalloc(). Whew.

When checking if a syscall like open(), ioctl() or writev() failed compare
directly against -1 and do not use a < 0 test. OK henning@

Add synchronisation support for dhcpd - this allows for two dhcpd's
with the same configuration to be run on the same net and they will
keep their lease files/state in synch, and therefore allowing you to
run redundant dhcpd's. Synchronization code stolen from spamd, uses
an hmac key in /var/db/dhcpd.key if it exists.
ok krw@ deraadt@

remove unused #defines; no binary change

This patch renames dhcpd's "struct timeout" to "struct dhcpd_timeout"
so as not to conflict with "struct timeout" from <sys/timeout.h>.
ok henning@

Don't exit just because an interface has no subnet configured in
dhcpd.conf. Just don't listen on that interface.

Make warning messages textually consistant.

tested & ok beck@

When an interface is ignored because it has no IP address, say so.
Don't just say 'not found". And always complain, not just when
specific interfaces were requested.

requested by and ok uwe@

If a list of interfaces is supplied via the command line or
dhcpd.interfaces then

a) don't bother looking up information on interfaces that were not
requested;

b) don't exit if a requested interface is not found, just issue a
warning message;

c) exit if none of the interfaces were found.

The command line for dhcpd shown in ps will continue to show requested
but ignored interfaces.

As usual with dhc* code, whack a bunch of unused states, constants,
flags, etc. Since we only invoke discover_interfaces() with
DISCOVER_SERVER, there is no need to keep track of other
possibilities.

KNF. No binary change.

Don't malloc fds each time, just realloc() as needed.
There's no need to clear revents -- poll() does that for us.
Move setting of cur_time to the top of the loop for better accuracy.
When poll returns 0, don't check revents since we know none were set.

change the undocumented -t flag to -n used for configuration testing
and document it. cleanup some unused/unneeded variables while there.
idea & ok henning@

spelling fixes; ok jmc@ henning@

avoid aliasing with libc functions; ok henning

Remove the not initialized fallback_interface code and use the normal bpf
interface for that. Also store the hardware (ethernet) address in the state
so that relayed dhcp request are sent to the correct destination.
OK henning@ requested by deraadt@ tested otto@

size_t vs ssize_t confusion

handle interface removals gracefully. reported by Oliver J. Morais
ok henning@, tedu@

remove things not used, spotted by lint mostly; ok henning

$OpenBSD$

more KNF

various knf; henning ok

a bunch of knf, ok henning

spaces

more crap bites the dust

time_t and kill a few dumb defines

may the whacking begin