fix unintended sizeof(pointer) introduced in rev 1.41
found with noderef.cocci, ok deraadt@

When getpwnam(3) reaches out to YP, it calls clntudp_create(3) with a
pre-initialized ypconnect(2) socket. That calls clntudp_bufcreate(),
which contains code checking if the socket and address are configured..
If not, socket(2) is called, or an address allocation is performed via
the portmapper (which calls a whole lot more code).

Split clnt_udp.c into two .c files (which will compile as seperate .o
files), and create a new libc-private clntudp_bufcreate_simple() function
which skips the socket and address work.

Result: In most static binaries, this reduces the text segment by
~100K, and removes 5-7 system call stubs -- which might matter for
non-pledged binaries with otherwise lack socket(2).
ok millert jmatthew

if the socket is connected, we can use recv() instead of recvfrom()
ok jmatthew

Add a new clnt*_control CLSET_CONNECTED, which says the socket has already
been connected. In the udp case, this means to use send(), not sendto()
ok jmatthew, claudio, miod

Mark all the rpc 'ops' vectors, for auth, client, service, and xdr,
as const, moving them from .data to .data.rel.ro. The other BSDs
did this a long time ago; NetBSD did a chunk in 1998, which is long
enough I didn't bother to get exact dates for others.

ok deraadt@ millert@

Add support for timeconting in userland.

This diff exposes parts of clock_gettime(2) and gettimeofday(2) to
userland via libc eliberating processes from the need for a context
switch everytime they want to count the passage of time.

If a timecounter clock can be exposed to userland than it needs to set
its tc_user member to a non-zero value. Tested with one or multiple
counters per architecture.

The timing data is shared through a pointer found in the new ELF
auxiliary vector AUX_openbsd_timekeep containing timehands information
that is frequently updated by the kernel.

Timing differences between the last kernel update and the current time
are adjusted in userland by the tc_get_timecount() function inside the
MD usertc.c file.

This permits a much more responsive environment, quite visible in
browsers, office programs and gaming (apparently one is are able to fly
in Minecraft now).

Tested by robert@, sthen@, naddy@, kmos@, phessler@, and many others!

OK from at least kettenis@, cheloha@, naddy@, sthen@

When system calls indicate an error they return -1, not some arbitrary
value < 0. errno is only updated in this case. Change all (most?)
callers of syscalls to follow this better, and let's see if this strictness
helps us in the future.

Remove dead gettimeofday calls and timevals.

ok jca@

Simplify and unify timespec variables used in those event loops

ok millert@

Use CLOCK_MONOTONIC for the rpc event loops

From Scott Cheloha, ok deraadt@ tb@

delete old lint ARGSUSED comments

Wrap <rpc/*.h> so that calls go direct and the symbols are all weak.
Hide __xprt_register() and _authenticate(); truncate <rpc/svc_auth.h>

ok deraadt@

Use SOCK_NONBLOCK in a couple more places

ok deraadt@

Remove all bogus writes to stderr. Only explicit requests should
go that way.
ok miod beck

Use fcntl() to set non-blocking-mode, rather ioctl(). This has a better
chance of working in systrace restricted environments.
ok guenther

Merge from NetBSD from 1999-03-25:"
* don't close the socket unless it was opened by the function
* note (in the comments) that the client is responsible for closing
the socket if they opened it, or they didn't use CLNT_DESTROY()

fixes a couple of unnecessary closing of already-closed sockets.
noted by: Matthias Drochner <M.Drochner@fz-juelich.de>"

tested by many in snaps
ok schwarze@

Oracle has re-licensed sunrpc under a three-clause BSD license.
Update our sources appropriately. OK deraadt@ jsg@

When authnone_create fails, do not leave behind a NULL cl_auth attribute,
because that one will typically derefenced later without any checking.
Instead, flag the error condition and bail out cleanly.
useful feedback from kjell@ and ok millert@

sprinkle a bit of ARGSUSED

zap remaining rcsid.

Kill old files that are no longer compiled.

okay theo

ansify. ok deraadt@ jaredy@

Bit of KNF - replace (<type> *)0 and (<type> *)NULL with NULL. Pointed
out by mickey@, seconded by millert@.

ok deraadt@.

Implement svc_getreq_poll(3) and friends and use poll(2) instead of select(2)
in the libc rpc code. The main difference between this and the previous
version is the use of a simple free list that simplifies the logic when
adding a socket to svc_pollfd. I've also added code to pack svc_pollfd
when the free list gets too big. The idea general idea is to keep
svc_pollfd as tightly packed as possible to make poll(2) efficient.
Tested by many people and OK deraadt@

use socklen_t where needed; henning pvalchev ok

compatiblity -> compatibility
decriptor -> descriptor
authentciated -> authenticated
transmition -> transmission

prototype cleanup

uninit variables

match xid nicer; plus notes for later work

if xdr_replymsg() fails, it can leave memory still allocated. thus we
need to go XDR_FREE it; wpaul@freebsd. clnt_raw.c also appears to have
the same problem, and there is precedent for the solution in various
other rpc code.

Some -Wall.

something is wrong with this level of randomness; let 2.1 go out with this "problem"

re-randomize xid on reuse of a CLIENT *; pointed out by adam@math.tau.ac.il

use arc4random for xid generation

Missed one cast.

perturb xid further. confuse the world

memset 0; also fix for byte order botch in __svc_fdset handling; thanks to mw@openbsd.org

Fix RCS ids
Make sure everything uses {SYS,}LIBC_SCCS properly

support descriptors > FD_SETSIZE and correct timeout handling

bcopy->memcpy & tag

typo

Pull prototypes into scope
Remove graticious casts

branches: 1.1.1;
Initial revision

When getpwnam(3) reaches out to YP, it calls clntudp_create(3) with a
pre-initialized ypconnect(2) socket. That calls clntudp_bufcreate(),
which contains code checking if the socket and address are configured..
If not, socket(2) is called, or an address allocation is performed via
the portmapper (which calls a whole lot more code).

Split clnt_udp.c into two .c files (which will compile as seperate .o
files), and create a new libc-private clntudp_bufcreate_simple() function
which skips the socket and address work.

Result: In most static binaries, this reduces the text segment by
~100K, and removes 5-7 system call stubs -- which might matter for
non-pledged binaries with otherwise lack socket(2).
ok millert jmatthew

if the socket is connected, we can use recv() instead of recvfrom()
ok jmatthew

Add a new clnt*_control CLSET_CONNECTED, which says the socket has already
been connected. In the udp case, this means to use send(), not sendto()
ok jmatthew, claudio, miod

Mark all the rpc 'ops' vectors, for auth, client, service, and xdr,
as const, moving them from .data to .data.rel.ro. The other BSDs
did this a long time ago; NetBSD did a chunk in 1998, which is long
enough I didn't bother to get exact dates for others.

ok deraadt@ millert@

Add support for timeconting in userland.

This diff exposes parts of clock_gettime(2) and gettimeofday(2) to
userland via libc eliberating processes from the need for a context
switch everytime they want to count the passage of time.

If a timecounter clock can be exposed to userland than it needs to set
its tc_user member to a non-zero value. Tested with one or multiple
counters per architecture.

The timing data is shared through a pointer found in the new ELF
auxiliary vector AUX_openbsd_timekeep containing timehands information
that is frequently updated by the kernel.

Timing differences between the last kernel update and the current time
are adjusted in userland by the tc_get_timecount() function inside the
MD usertc.c file.

This permits a much more responsive environment, quite visible in
browsers, office programs and gaming (apparently one is are able to fly
in Minecraft now).

Tested by robert@, sthen@, naddy@, kmos@, phessler@, and many others!

OK from at least kettenis@, cheloha@, naddy@, sthen@

When system calls indicate an error they return -1, not some arbitrary
value < 0. errno is only updated in this case. Change all (most?)
callers of syscalls to follow this better, and let's see if this strictness
helps us in the future.

Remove dead gettimeofday calls and timevals.

ok jca@

Simplify and unify timespec variables used in those event loops

ok millert@

Use CLOCK_MONOTONIC for the rpc event loops

From Scott Cheloha, ok deraadt@ tb@

delete old lint ARGSUSED comments

Wrap <rpc/*.h> so that calls go direct and the symbols are all weak.
Hide __xprt_register() and _authenticate(); truncate <rpc/svc_auth.h>

ok deraadt@

Use SOCK_NONBLOCK in a couple more places

ok deraadt@

Remove all bogus writes to stderr. Only explicit requests should
go that way.
ok miod beck

Use fcntl() to set non-blocking-mode, rather ioctl(). This has a better
chance of working in systrace restricted environments.
ok guenther

Merge from NetBSD from 1999-03-25:"
* don't close the socket unless it was opened by the function
* note (in the comments) that the client is responsible for closing
the socket if they opened it, or they didn't use CLNT_DESTROY()

fixes a couple of unnecessary closing of already-closed sockets.
noted by: Matthias Drochner <M.Drochner@fz-juelich.de>"

tested by many in snaps
ok schwarze@

Oracle has re-licensed sunrpc under a three-clause BSD license.
Update our sources appropriately. OK deraadt@ jsg@

When authnone_create fails, do not leave behind a NULL cl_auth attribute,
because that one will typically derefenced later without any checking.
Instead, flag the error condition and bail out cleanly.
useful feedback from kjell@ and ok millert@

sprinkle a bit of ARGSUSED

zap remaining rcsid.

Kill old files that are no longer compiled.

okay theo

ansify. ok deraadt@ jaredy@

Bit of KNF - replace (<type> *)0 and (<type> *)NULL with NULL. Pointed
out by mickey@, seconded by millert@.

ok deraadt@.

Implement svc_getreq_poll(3) and friends and use poll(2) instead of select(2)
in the libc rpc code. The main difference between this and the previous
version is the use of a simple free list that simplifies the logic when
adding a socket to svc_pollfd. I've also added code to pack svc_pollfd
when the free list gets too big. The idea general idea is to keep
svc_pollfd as tightly packed as possible to make poll(2) efficient.
Tested by many people and OK deraadt@

use socklen_t where needed; henning pvalchev ok

compatiblity -> compatibility
decriptor -> descriptor
authentciated -> authenticated
transmition -> transmission

prototype cleanup

uninit variables

match xid nicer; plus notes for later work

if xdr_replymsg() fails, it can leave memory still allocated. thus we
need to go XDR_FREE it; wpaul@freebsd. clnt_raw.c also appears to have
the same problem, and there is precedent for the solution in various
other rpc code.

Some -Wall.

something is wrong with this level of randomness; let 2.1 go out with this "problem"

re-randomize xid on reuse of a CLIENT *; pointed out by adam@math.tau.ac.il

use arc4random for xid generation

Missed one cast.

perturb xid further. confuse the world

memset 0; also fix for byte order botch in __svc_fdset handling; thanks to mw@openbsd.org

Fix RCS ids
Make sure everything uses {SYS,}LIBC_SCCS properly

support descriptors > FD_SETSIZE and correct timeout handling

bcopy->memcpy & tag

typo

Pull prototypes into scope
Remove graticious casts

branches: 1.1.1;
Initial revision

if the socket is connected, we can use recv() instead of recvfrom()
ok jmatthew

Add a new clnt*_control CLSET_CONNECTED, which says the socket has already
been connected. In the udp case, this means to use send(), not sendto()
ok jmatthew, claudio, miod

Mark all the rpc 'ops' vectors, for auth, client, service, and xdr,
as const, moving them from .data to .data.rel.ro. The other BSDs
did this a long time ago; NetBSD did a chunk in 1998, which is long
enough I didn't bother to get exact dates for others.

ok deraadt@ millert@

Add support for timeconting in userland.

This diff exposes parts of clock_gettime(2) and gettimeofday(2) to
userland via libc eliberating processes from the need for a context
switch everytime they want to count the passage of time.

If a timecounter clock can be exposed to userland than it needs to set
its tc_user member to a non-zero value. Tested with one or multiple
counters per architecture.

The timing data is shared through a pointer found in the new ELF
auxiliary vector AUX_openbsd_timekeep containing timehands information
that is frequently updated by the kernel.

Timing differences between the last kernel update and the current time
are adjusted in userland by the tc_get_timecount() function inside the
MD usertc.c file.

This permits a much more responsive environment, quite visible in
browsers, office programs and gaming (apparently one is are able to fly
in Minecraft now).

Tested by robert@, sthen@, naddy@, kmos@, phessler@, and many others!

OK from at least kettenis@, cheloha@, naddy@, sthen@

When system calls indicate an error they return -1, not some arbitrary
value < 0. errno is only updated in this case. Change all (most?)
callers of syscalls to follow this better, and let's see if this strictness
helps us in the future.

Remove dead gettimeofday calls and timevals.

ok jca@

Simplify and unify timespec variables used in those event loops

ok millert@

Use CLOCK_MONOTONIC for the rpc event loops

From Scott Cheloha, ok deraadt@ tb@

delete old lint ARGSUSED comments

Wrap <rpc/*.h> so that calls go direct and the symbols are all weak.
Hide __xprt_register() and _authenticate(); truncate <rpc/svc_auth.h>

ok deraadt@

Use SOCK_NONBLOCK in a couple more places

ok deraadt@

Remove all bogus writes to stderr. Only explicit requests should
go that way.
ok miod beck

Use fcntl() to set non-blocking-mode, rather ioctl(). This has a better
chance of working in systrace restricted environments.
ok guenther

Merge from NetBSD from 1999-03-25:"
* don't close the socket unless it was opened by the function
* note (in the comments) that the client is responsible for closing
the socket if they opened it, or they didn't use CLNT_DESTROY()

fixes a couple of unnecessary closing of already-closed sockets.
noted by: Matthias Drochner <M.Drochner@fz-juelich.de>"

tested by many in snaps
ok schwarze@

Oracle has re-licensed sunrpc under a three-clause BSD license.
Update our sources appropriately. OK deraadt@ jsg@

When authnone_create fails, do not leave behind a NULL cl_auth attribute,
because that one will typically derefenced later without any checking.
Instead, flag the error condition and bail out cleanly.
useful feedback from kjell@ and ok millert@

sprinkle a bit of ARGSUSED

zap remaining rcsid.

Kill old files that are no longer compiled.

okay theo

ansify. ok deraadt@ jaredy@

Bit of KNF - replace (<type> *)0 and (<type> *)NULL with NULL. Pointed
out by mickey@, seconded by millert@.

ok deraadt@.

Implement svc_getreq_poll(3) and friends and use poll(2) instead of select(2)
in the libc rpc code. The main difference between this and the previous
version is the use of a simple free list that simplifies the logic when
adding a socket to svc_pollfd. I've also added code to pack svc_pollfd
when the free list gets too big. The idea general idea is to keep
svc_pollfd as tightly packed as possible to make poll(2) efficient.
Tested by many people and OK deraadt@

use socklen_t where needed; henning pvalchev ok

compatiblity -> compatibility
decriptor -> descriptor
authentciated -> authenticated
transmition -> transmission

prototype cleanup

uninit variables

match xid nicer; plus notes for later work

if xdr_replymsg() fails, it can leave memory still allocated. thus we
need to go XDR_FREE it; wpaul@freebsd. clnt_raw.c also appears to have
the same problem, and there is precedent for the solution in various
other rpc code.

Some -Wall.

something is wrong with this level of randomness; let 2.1 go out with this "problem"

re-randomize xid on reuse of a CLIENT *; pointed out by adam@math.tau.ac.il

use arc4random for xid generation

Missed one cast.

perturb xid further. confuse the world

memset 0; also fix for byte order botch in __svc_fdset handling; thanks to mw@openbsd.org

Fix RCS ids
Make sure everything uses {SYS,}LIBC_SCCS properly

support descriptors > FD_SETSIZE and correct timeout handling

bcopy->memcpy & tag

typo

Pull prototypes into scope
Remove graticious casts

branches: 1.1.1;
Initial revision

Add a new clnt*_control CLSET_CONNECTED, which says the socket has already
been connected. In the udp case, this means to use send(), not sendto()
ok jmatthew, claudio, miod

Mark all the rpc 'ops' vectors, for auth, client, service, and xdr,
as const, moving them from .data to .data.rel.ro. The other BSDs
did this a long time ago; NetBSD did a chunk in 1998, which is long
enough I didn't bother to get exact dates for others.

ok deraadt@ millert@

Add support for timeconting in userland.

This diff exposes parts of clock_gettime(2) and gettimeofday(2) to
userland via libc eliberating processes from the need for a context
switch everytime they want to count the passage of time.

If a timecounter clock can be exposed to userland than it needs to set
its tc_user member to a non-zero value. Tested with one or multiple
counters per architecture.

The timing data is shared through a pointer found in the new ELF
auxiliary vector AUX_openbsd_timekeep containing timehands information
that is frequently updated by the kernel.

Timing differences between the last kernel update and the current time
are adjusted in userland by the tc_get_timecount() function inside the
MD usertc.c file.

This permits a much more responsive environment, quite visible in
browsers, office programs and gaming (apparently one is are able to fly
in Minecraft now).

Tested by robert@, sthen@, naddy@, kmos@, phessler@, and many others!

OK from at least kettenis@, cheloha@, naddy@, sthen@

When system calls indicate an error they return -1, not some arbitrary
value < 0. errno is only updated in this case. Change all (most?)
callers of syscalls to follow this better, and let's see if this strictness
helps us in the future.

Remove dead gettimeofday calls and timevals.

ok jca@

Simplify and unify timespec variables used in those event loops

ok millert@

Use CLOCK_MONOTONIC for the rpc event loops

From Scott Cheloha, ok deraadt@ tb@

delete old lint ARGSUSED comments

Wrap <rpc/*.h> so that calls go direct and the symbols are all weak.
Hide __xprt_register() and _authenticate(); truncate <rpc/svc_auth.h>

ok deraadt@

Use SOCK_NONBLOCK in a couple more places

ok deraadt@

Remove all bogus writes to stderr. Only explicit requests should
go that way.
ok miod beck

Use fcntl() to set non-blocking-mode, rather ioctl(). This has a better
chance of working in systrace restricted environments.
ok guenther

Merge from NetBSD from 1999-03-25:"
* don't close the socket unless it was opened by the function
* note (in the comments) that the client is responsible for closing
the socket if they opened it, or they didn't use CLNT_DESTROY()

fixes a couple of unnecessary closing of already-closed sockets.
noted by: Matthias Drochner <M.Drochner@fz-juelich.de>"

tested by many in snaps
ok schwarze@

Oracle has re-licensed sunrpc under a three-clause BSD license.
Update our sources appropriately. OK deraadt@ jsg@

When authnone_create fails, do not leave behind a NULL cl_auth attribute,
because that one will typically derefenced later without any checking.
Instead, flag the error condition and bail out cleanly.
useful feedback from kjell@ and ok millert@

sprinkle a bit of ARGSUSED

zap remaining rcsid.

Kill old files that are no longer compiled.

okay theo

ansify. ok deraadt@ jaredy@

Bit of KNF - replace (<type> *)0 and (<type> *)NULL with NULL. Pointed
out by mickey@, seconded by millert@.

ok deraadt@.

Implement svc_getreq_poll(3) and friends and use poll(2) instead of select(2)
in the libc rpc code. The main difference between this and the previous
version is the use of a simple free list that simplifies the logic when
adding a socket to svc_pollfd. I've also added code to pack svc_pollfd
when the free list gets too big. The idea general idea is to keep
svc_pollfd as tightly packed as possible to make poll(2) efficient.
Tested by many people and OK deraadt@

use socklen_t where needed; henning pvalchev ok

compatiblity -> compatibility
decriptor -> descriptor
authentciated -> authenticated
transmition -> transmission

prototype cleanup

uninit variables

match xid nicer; plus notes for later work

if xdr_replymsg() fails, it can leave memory still allocated. thus we
need to go XDR_FREE it; wpaul@freebsd. clnt_raw.c also appears to have
the same problem, and there is precedent for the solution in various
other rpc code.

Some -Wall.

something is wrong with this level of randomness; let 2.1 go out with this "problem"

re-randomize xid on reuse of a CLIENT *; pointed out by adam@math.tau.ac.il

use arc4random for xid generation

Missed one cast.

perturb xid further. confuse the world

memset 0; also fix for byte order botch in __svc_fdset handling; thanks to mw@openbsd.org

Fix RCS ids
Make sure everything uses {SYS,}LIBC_SCCS properly

support descriptors > FD_SETSIZE and correct timeout handling

bcopy->memcpy & tag

typo

Pull prototypes into scope
Remove graticious casts

branches: 1.1.1;
Initial revision

Mark all the rpc 'ops' vectors, for auth, client, service, and xdr,
as const, moving them from .data to .data.rel.ro. The other BSDs
did this a long time ago; NetBSD did a chunk in 1998, which is long
enough I didn't bother to get exact dates for others.

ok deraadt@ millert@

Add support for timeconting in userland.

This diff exposes parts of clock_gettime(2) and gettimeofday(2) to
userland via libc eliberating processes from the need for a context
switch everytime they want to count the passage of time.

If a timecounter clock can be exposed to userland than it needs to set
its tc_user member to a non-zero value. Tested with one or multiple
counters per architecture.

The timing data is shared through a pointer found in the new ELF
auxiliary vector AUX_openbsd_timekeep containing timehands information
that is frequently updated by the kernel.

Timing differences between the last kernel update and the current time
are adjusted in userland by the tc_get_timecount() function inside the
MD usertc.c file.

This permits a much more responsive environment, quite visible in
browsers, office programs and gaming (apparently one is are able to fly
in Minecraft now).

Tested by robert@, sthen@, naddy@, kmos@, phessler@, and many others!

OK from at least kettenis@, cheloha@, naddy@, sthen@

When system calls indicate an error they return -1, not some arbitrary
value < 0. errno is only updated in this case. Change all (most?)
callers of syscalls to follow this better, and let's see if this strictness
helps us in the future.

Remove dead gettimeofday calls and timevals.

ok jca@

Simplify and unify timespec variables used in those event loops

ok millert@

Use CLOCK_MONOTONIC for the rpc event loops

From Scott Cheloha, ok deraadt@ tb@

delete old lint ARGSUSED comments

Wrap <rpc/*.h> so that calls go direct and the symbols are all weak.
Hide __xprt_register() and _authenticate(); truncate <rpc/svc_auth.h>

ok deraadt@

Use SOCK_NONBLOCK in a couple more places

ok deraadt@

Remove all bogus writes to stderr. Only explicit requests should
go that way.
ok miod beck

Use fcntl() to set non-blocking-mode, rather ioctl(). This has a better
chance of working in systrace restricted environments.
ok guenther

Merge from NetBSD from 1999-03-25:"
* don't close the socket unless it was opened by the function
* note (in the comments) that the client is responsible for closing
the socket if they opened it, or they didn't use CLNT_DESTROY()

fixes a couple of unnecessary closing of already-closed sockets.
noted by: Matthias Drochner <M.Drochner@fz-juelich.de>"

tested by many in snaps
ok schwarze@

Oracle has re-licensed sunrpc under a three-clause BSD license.
Update our sources appropriately. OK deraadt@ jsg@

When authnone_create fails, do not leave behind a NULL cl_auth attribute,
because that one will typically derefenced later without any checking.
Instead, flag the error condition and bail out cleanly.
useful feedback from kjell@ and ok millert@

sprinkle a bit of ARGSUSED

zap remaining rcsid.

Kill old files that are no longer compiled.

okay theo

ansify. ok deraadt@ jaredy@

Bit of KNF - replace (<type> *)0 and (<type> *)NULL with NULL. Pointed
out by mickey@, seconded by millert@.

ok deraadt@.

Implement svc_getreq_poll(3) and friends and use poll(2) instead of select(2)
in the libc rpc code. The main difference between this and the previous
version is the use of a simple free list that simplifies the logic when
adding a socket to svc_pollfd. I've also added code to pack svc_pollfd
when the free list gets too big. The idea general idea is to keep
svc_pollfd as tightly packed as possible to make poll(2) efficient.
Tested by many people and OK deraadt@

use socklen_t where needed; henning pvalchev ok

compatiblity -> compatibility
decriptor -> descriptor
authentciated -> authenticated
transmition -> transmission

prototype cleanup

uninit variables

match xid nicer; plus notes for later work

if xdr_replymsg() fails, it can leave memory still allocated. thus we
need to go XDR_FREE it; wpaul@freebsd. clnt_raw.c also appears to have
the same problem, and there is precedent for the solution in various
other rpc code.

Some -Wall.

something is wrong with this level of randomness; let 2.1 go out with this "problem"

re-randomize xid on reuse of a CLIENT *; pointed out by adam@math.tau.ac.il

use arc4random for xid generation

Missed one cast.

perturb xid further. confuse the world

memset 0; also fix for byte order botch in __svc_fdset handling; thanks to mw@openbsd.org

Fix RCS ids
Make sure everything uses {SYS,}LIBC_SCCS properly

support descriptors > FD_SETSIZE and correct timeout handling

bcopy->memcpy & tag

typo

Pull prototypes into scope
Remove graticious casts

branches: 1.1.1;
Initial revision

Add support for timeconting in userland.

This diff exposes parts of clock_gettime(2) and gettimeofday(2) to
userland via libc eliberating processes from the need for a context
switch everytime they want to count the passage of time.

If a timecounter clock can be exposed to userland than it needs to set
its tc_user member to a non-zero value. Tested with one or multiple
counters per architecture.

The timing data is shared through a pointer found in the new ELF
auxiliary vector AUX_openbsd_timekeep containing timehands information
that is frequently updated by the kernel.

Timing differences between the last kernel update and the current time
are adjusted in userland by the tc_get_timecount() function inside the
MD usertc.c file.

This permits a much more responsive environment, quite visible in
browsers, office programs and gaming (apparently one is are able to fly
in Minecraft now).

Tested by robert@, sthen@, naddy@, kmos@, phessler@, and many others!

OK from at least kettenis@, cheloha@, naddy@, sthen@

When system calls indicate an error they return -1, not some arbitrary
value < 0. errno is only updated in this case. Change all (most?)
callers of syscalls to follow this better, and let's see if this strictness
helps us in the future.

Remove dead gettimeofday calls and timevals.

ok jca@

Simplify and unify timespec variables used in those event loops

ok millert@

Use CLOCK_MONOTONIC for the rpc event loops

From Scott Cheloha, ok deraadt@ tb@

delete old lint ARGSUSED comments

Wrap <rpc/*.h> so that calls go direct and the symbols are all weak.
Hide __xprt_register() and _authenticate(); truncate <rpc/svc_auth.h>

ok deraadt@

Use SOCK_NONBLOCK in a couple more places

ok deraadt@

Remove all bogus writes to stderr. Only explicit requests should
go that way.
ok miod beck

Use fcntl() to set non-blocking-mode, rather ioctl(). This has a better
chance of working in systrace restricted environments.
ok guenther

Merge from NetBSD from 1999-03-25:"
* don't close the socket unless it was opened by the function
* note (in the comments) that the client is responsible for closing
the socket if they opened it, or they didn't use CLNT_DESTROY()

fixes a couple of unnecessary closing of already-closed sockets.
noted by: Matthias Drochner <M.Drochner@fz-juelich.de>"

tested by many in snaps
ok schwarze@

Oracle has re-licensed sunrpc under a three-clause BSD license.
Update our sources appropriately. OK deraadt@ jsg@

When authnone_create fails, do not leave behind a NULL cl_auth attribute,
because that one will typically derefenced later without any checking.
Instead, flag the error condition and bail out cleanly.
useful feedback from kjell@ and ok millert@

sprinkle a bit of ARGSUSED

zap remaining rcsid.

Kill old files that are no longer compiled.

okay theo

ansify. ok deraadt@ jaredy@

Bit of KNF - replace (<type> *)0 and (<type> *)NULL with NULL. Pointed
out by mickey@, seconded by millert@.

ok deraadt@.

Implement svc_getreq_poll(3) and friends and use poll(2) instead of select(2)
in the libc rpc code. The main difference between this and the previous
version is the use of a simple free list that simplifies the logic when
adding a socket to svc_pollfd. I've also added code to pack svc_pollfd
when the free list gets too big. The idea general idea is to keep
svc_pollfd as tightly packed as possible to make poll(2) efficient.
Tested by many people and OK deraadt@

use socklen_t where needed; henning pvalchev ok

compatiblity -> compatibility
decriptor -> descriptor
authentciated -> authenticated
transmition -> transmission

prototype cleanup

uninit variables

match xid nicer; plus notes for later work

if xdr_replymsg() fails, it can leave memory still allocated. thus we
need to go XDR_FREE it; wpaul@freebsd. clnt_raw.c also appears to have
the same problem, and there is precedent for the solution in various
other rpc code.

Some -Wall.

something is wrong with this level of randomness; let 2.1 go out with this "problem"

re-randomize xid on reuse of a CLIENT *; pointed out by adam@math.tau.ac.il

use arc4random for xid generation

Missed one cast.

perturb xid further. confuse the world

memset 0; also fix for byte order botch in __svc_fdset handling; thanks to mw@openbsd.org

Fix RCS ids
Make sure everything uses {SYS,}LIBC_SCCS properly

support descriptors > FD_SETSIZE and correct timeout handling

bcopy->memcpy & tag

typo

Pull prototypes into scope
Remove graticious casts

branches: 1.1.1;
Initial revision

When system calls indicate an error they return -1, not some arbitrary
value < 0. errno is only updated in this case. Change all (most?)
callers of syscalls to follow this better, and let's see if this strictness
helps us in the future.

Remove dead gettimeofday calls and timevals.

ok jca@

Simplify and unify timespec variables used in those event loops

ok millert@

Use CLOCK_MONOTONIC for the rpc event loops

From Scott Cheloha, ok deraadt@ tb@

delete old lint ARGSUSED comments

Wrap <rpc/*.h> so that calls go direct and the symbols are all weak.
Hide __xprt_register() and _authenticate(); truncate <rpc/svc_auth.h>

ok deraadt@

Use SOCK_NONBLOCK in a couple more places

ok deraadt@

Remove all bogus writes to stderr. Only explicit requests should
go that way.
ok miod beck

Use fcntl() to set non-blocking-mode, rather ioctl(). This has a better
chance of working in systrace restricted environments.
ok guenther

Merge from NetBSD from 1999-03-25:"
* don't close the socket unless it was opened by the function
* note (in the comments) that the client is responsible for closing
the socket if they opened it, or they didn't use CLNT_DESTROY()

fixes a couple of unnecessary closing of already-closed sockets.
noted by: Matthias Drochner <M.Drochner@fz-juelich.de>"

tested by many in snaps
ok schwarze@

Oracle has re-licensed sunrpc under a three-clause BSD license.
Update our sources appropriately. OK deraadt@ jsg@

When authnone_create fails, do not leave behind a NULL cl_auth attribute,
because that one will typically derefenced later without any checking.
Instead, flag the error condition and bail out cleanly.
useful feedback from kjell@ and ok millert@

sprinkle a bit of ARGSUSED

zap remaining rcsid.

Kill old files that are no longer compiled.

okay theo

ansify. ok deraadt@ jaredy@

Bit of KNF - replace (<type> *)0 and (<type> *)NULL with NULL. Pointed
out by mickey@, seconded by millert@.

ok deraadt@.

Implement svc_getreq_poll(3) and friends and use poll(2) instead of select(2)
in the libc rpc code. The main difference between this and the previous
version is the use of a simple free list that simplifies the logic when
adding a socket to svc_pollfd. I've also added code to pack svc_pollfd
when the free list gets too big. The idea general idea is to keep
svc_pollfd as tightly packed as possible to make poll(2) efficient.
Tested by many people and OK deraadt@

use socklen_t where needed; henning pvalchev ok

compatiblity -> compatibility
decriptor -> descriptor
authentciated -> authenticated
transmition -> transmission

prototype cleanup

uninit variables

match xid nicer; plus notes for later work

if xdr_replymsg() fails, it can leave memory still allocated. thus we
need to go XDR_FREE it; wpaul@freebsd. clnt_raw.c also appears to have
the same problem, and there is precedent for the solution in various
other rpc code.

Some -Wall.

something is wrong with this level of randomness; let 2.1 go out with this "problem"

re-randomize xid on reuse of a CLIENT *; pointed out by adam@math.tau.ac.il

use arc4random for xid generation

Missed one cast.

perturb xid further. confuse the world

memset 0; also fix for byte order botch in __svc_fdset handling; thanks to mw@openbsd.org

Fix RCS ids
Make sure everything uses {SYS,}LIBC_SCCS properly

support descriptors > FD_SETSIZE and correct timeout handling

bcopy->memcpy & tag

typo

Pull prototypes into scope
Remove graticious casts

branches: 1.1.1;
Initial revision

Remove dead gettimeofday calls and timevals.

ok jca@

Simplify and unify timespec variables used in those event loops

ok millert@

Use CLOCK_MONOTONIC for the rpc event loops

From Scott Cheloha, ok deraadt@ tb@

delete old lint ARGSUSED comments

Wrap <rpc/*.h> so that calls go direct and the symbols are all weak.
Hide __xprt_register() and _authenticate(); truncate <rpc/svc_auth.h>

ok deraadt@

Use SOCK_NONBLOCK in a couple more places

ok deraadt@

Remove all bogus writes to stderr. Only explicit requests should
go that way.
ok miod beck

Use fcntl() to set non-blocking-mode, rather ioctl(). This has a better
chance of working in systrace restricted environments.
ok guenther

Merge from NetBSD from 1999-03-25:"
* don't close the socket unless it was opened by the function
* note (in the comments) that the client is responsible for closing
the socket if they opened it, or they didn't use CLNT_DESTROY()

fixes a couple of unnecessary closing of already-closed sockets.
noted by: Matthias Drochner <M.Drochner@fz-juelich.de>"

tested by many in snaps
ok schwarze@

Oracle has re-licensed sunrpc under a three-clause BSD license.
Update our sources appropriately. OK deraadt@ jsg@

When authnone_create fails, do not leave behind a NULL cl_auth attribute,
because that one will typically derefenced later without any checking.
Instead, flag the error condition and bail out cleanly.
useful feedback from kjell@ and ok millert@

sprinkle a bit of ARGSUSED

zap remaining rcsid.

Kill old files that are no longer compiled.

okay theo

ansify. ok deraadt@ jaredy@

Bit of KNF - replace (<type> *)0 and (<type> *)NULL with NULL. Pointed
out by mickey@, seconded by millert@.

ok deraadt@.

Implement svc_getreq_poll(3) and friends and use poll(2) instead of select(2)
in the libc rpc code. The main difference between this and the previous
version is the use of a simple free list that simplifies the logic when
adding a socket to svc_pollfd. I've also added code to pack svc_pollfd
when the free list gets too big. The idea general idea is to keep
svc_pollfd as tightly packed as possible to make poll(2) efficient.
Tested by many people and OK deraadt@

use socklen_t where needed; henning pvalchev ok

compatiblity -> compatibility
decriptor -> descriptor
authentciated -> authenticated
transmition -> transmission

prototype cleanup

uninit variables

match xid nicer; plus notes for later work

if xdr_replymsg() fails, it can leave memory still allocated. thus we
need to go XDR_FREE it; wpaul@freebsd. clnt_raw.c also appears to have
the same problem, and there is precedent for the solution in various
other rpc code.

Some -Wall.

something is wrong with this level of randomness; let 2.1 go out with this "problem"

re-randomize xid on reuse of a CLIENT *; pointed out by adam@math.tau.ac.il

use arc4random for xid generation

Missed one cast.

perturb xid further. confuse the world

memset 0; also fix for byte order botch in __svc_fdset handling; thanks to mw@openbsd.org

Fix RCS ids
Make sure everything uses {SYS,}LIBC_SCCS properly

support descriptors > FD_SETSIZE and correct timeout handling

bcopy->memcpy & tag

typo

Pull prototypes into scope
Remove graticious casts

branches: 1.1.1;
Initial revision