print all the rings on a nic without limiting it to the number of mclpools

"makes sense" jmatthew@

print the name of the rings in systat mbuf output too.

useful if you're looking at a nic with a lot of different rings.

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.

Add sys/queue.h where it's needed. Unbreak userland following recent
removal from mbuf.h. ok mpi@

Avoid calling freeifaddrs() with an uninitialised pointer in an
error path.

Adjust <sys/param.h> comments regarding use of use of MSIZE, or
delete <sys/param.h> if now possible
ok guenther

first batch of programs adapting to the namespace cleanup
(pfvar.h nameser.h proc.h ucred.h)
ok guenther millert, and some review from doug as well.

i renamed the mbuf pool.

fix if (statement);
ok dlg@

cut things that relied on mclgeti for rx ring accounting/restriction over
to using if_rxr.

cut the reporting systat did over to the rxr ioctl.

tested as much as i can on alpha, amd64, and sparc64.
mpi@ has run it on macppc.
ok mpi@

info about pools is currently given to userland by copying each
pools struct out. however, struct pool in the kernel contains lots
of things that userland probably isnt interested in, like actual
mutexes, and probably shouldnt get easy access to, like pointers
to kernel memory via all the lists/trees.

this implements a kinfo_pool structure that has only the data that
userland needs to know about. it cuts the sysctl code over to
building it from struct pool as required and copying that out
instead, and cuts userland over to only handling kinfo_pool.

the only problem with this is vmstat, which can read kernel images
via kvm, which needs some understanding of struct pool. to cope,
the struct pool definition is guarded by if defined(_KERNEL) ||
defined(_LIBKVM) as inspired by sysctl which needs to do the same
thing sometimes. struct pool itself is generally not visible to
userland though, which is good.

matthew@ suggested struct kinfo_pool instead of struct pool_info.
the kinfo prefix has precedent.
lots of people liked this.

Now that the water marks are updated on a per-pool basis indicate in
the LIVELOCKS column if there is a pending (deferred) update.

ok claudio@

- use a common FIELD_ADDR macro, instead of rolling 78 identical copies.

ok canacar@

Print relativ number of livelocks that happend between two systat
updates instead of showing absolut figures. Makes it easier to
estimate the "livelockyness" of a system.
OK dlg@

expose the kernels network livelock counter

tweak the mclgeti algorithm to behave better under load.

instead of letting hardware rings grow on every interrupt, restrict
it so it can only grow once per softclock tick. we can only punish
the rings on softclock ticks, so it make sense to only grow on
softclock tick boundaries too.

the rings are now punished after >1 lost softclock tick rather than
>2. mclgeti is now more aggressive at detecting livelock.

the rings get punished by an 8th, rather than by half.

we now allow the rings to be punished again even if the system is
already considered in livelock.

without this diff a livelocked system will have its rx ring sizes
scale up and down very rapidly, while holding the rings low for too
long. this affected throughput significantly.

discussed and tested heavily at j2k10. there are still some games
with softnet we can play, but this is a good first step.

"put it in" and ok deraadt@
ok claudio@ krw@ henning@ mcbride@

if we find out that it sucks we can pull it out again later. till then
we'll run with it and see how it goes.

don't rely on mbuf.h to pull in pool.h here either

ok miod@

fix use of uninitialized variable.

reported by gcc.

ok canacar@

args, not that file

no more natpass

remove unneeded argument to warnx()
remove unused variable
kill some extra blank lines

ok canacar@

Make sure systat mbufs prints the interface name even if an interface only
allocates clusters larger than 2k.

ok deraadt@

fix the mbuf view to cope with the recent change to the mclgeti allocator.

from canacar@

Skip missing pool indices instead of printing errors.

Print the interface livelocks if non-zero. From deraadt@

Rewrite the 'mbuf' page to display mbuf usage information from system pools
and mbuf cluster statistics collected for each interface. More useful than
earlier bar graphs about mbuf types. Requested by and ok deraadt@

New display engine for systat, based on pftop. Adds new views for pf
(status, state, rule, queue). While all displays work, some keyboard
comands are not implemented yet. Other features include better handling
of display resize and scrolling for long views. Committing now to fix
the remaining issues in the tree.
Testing and comments by otto@ and harding@, ok deraadt@

use the 1 line load line from 'systat vm' on all the other displays, giving
them more room. much cleaner, and it is less code too; ok deanna

fairly simple cleanup, lots of testing done

post-mortem stuff can go away here too

move sysctl comment on namelist[] entry

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

mark which nlist[] kmem entries are not needed in the sysctl case, and which are

ANSI

Use sysctl for mbstat

move back. pvalchev -- the bug was quite obscure

back out latest changes which cause problems on alpha/sparc64 until
they are fixed

fix signal races. use sysctl() instead of kvm where possible, and then KNF
the hell out of it; checked by miod

kill more registers

millert@ ok

handle kvm_nlist() failing, from pr#1798.
Patch similar to the one submitted by <peterw@documenta.com.au>

bye bye sprintf

rcsid

it compiles w/ new curses

branches: 1.1.1;
Initial revision

print the name of the rings in systat mbuf output too.

useful if you're looking at a nic with a lot of different rings.

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.

Add sys/queue.h where it's needed. Unbreak userland following recent
removal from mbuf.h. ok mpi@

Avoid calling freeifaddrs() with an uninitialised pointer in an
error path.

Adjust <sys/param.h> comments regarding use of use of MSIZE, or
delete <sys/param.h> if now possible
ok guenther

first batch of programs adapting to the namespace cleanup
(pfvar.h nameser.h proc.h ucred.h)
ok guenther millert, and some review from doug as well.

i renamed the mbuf pool.

fix if (statement);
ok dlg@

cut things that relied on mclgeti for rx ring accounting/restriction over
to using if_rxr.

cut the reporting systat did over to the rxr ioctl.

tested as much as i can on alpha, amd64, and sparc64.
mpi@ has run it on macppc.
ok mpi@

info about pools is currently given to userland by copying each
pools struct out. however, struct pool in the kernel contains lots
of things that userland probably isnt interested in, like actual
mutexes, and probably shouldnt get easy access to, like pointers
to kernel memory via all the lists/trees.

this implements a kinfo_pool structure that has only the data that
userland needs to know about. it cuts the sysctl code over to
building it from struct pool as required and copying that out
instead, and cuts userland over to only handling kinfo_pool.

the only problem with this is vmstat, which can read kernel images
via kvm, which needs some understanding of struct pool. to cope,
the struct pool definition is guarded by if defined(_KERNEL) ||
defined(_LIBKVM) as inspired by sysctl which needs to do the same
thing sometimes. struct pool itself is generally not visible to
userland though, which is good.

matthew@ suggested struct kinfo_pool instead of struct pool_info.
the kinfo prefix has precedent.
lots of people liked this.

Now that the water marks are updated on a per-pool basis indicate in
the LIVELOCKS column if there is a pending (deferred) update.

ok claudio@

- use a common FIELD_ADDR macro, instead of rolling 78 identical copies.

ok canacar@

Print relativ number of livelocks that happend between two systat
updates instead of showing absolut figures. Makes it easier to
estimate the "livelockyness" of a system.
OK dlg@

expose the kernels network livelock counter

tweak the mclgeti algorithm to behave better under load.

instead of letting hardware rings grow on every interrupt, restrict
it so it can only grow once per softclock tick. we can only punish
the rings on softclock ticks, so it make sense to only grow on
softclock tick boundaries too.

the rings are now punished after >1 lost softclock tick rather than
>2. mclgeti is now more aggressive at detecting livelock.

the rings get punished by an 8th, rather than by half.

we now allow the rings to be punished again even if the system is
already considered in livelock.

without this diff a livelocked system will have its rx ring sizes
scale up and down very rapidly, while holding the rings low for too
long. this affected throughput significantly.

discussed and tested heavily at j2k10. there are still some games
with softnet we can play, but this is a good first step.

"put it in" and ok deraadt@
ok claudio@ krw@ henning@ mcbride@

if we find out that it sucks we can pull it out again later. till then
we'll run with it and see how it goes.

don't rely on mbuf.h to pull in pool.h here either

ok miod@

fix use of uninitialized variable.

reported by gcc.

ok canacar@

args, not that file

no more natpass

remove unneeded argument to warnx()
remove unused variable
kill some extra blank lines

ok canacar@

Make sure systat mbufs prints the interface name even if an interface only
allocates clusters larger than 2k.

ok deraadt@

fix the mbuf view to cope with the recent change to the mclgeti allocator.

from canacar@

Skip missing pool indices instead of printing errors.

Print the interface livelocks if non-zero. From deraadt@

Rewrite the 'mbuf' page to display mbuf usage information from system pools
and mbuf cluster statistics collected for each interface. More useful than
earlier bar graphs about mbuf types. Requested by and ok deraadt@

New display engine for systat, based on pftop. Adds new views for pf
(status, state, rule, queue). While all displays work, some keyboard
comands are not implemented yet. Other features include better handling
of display resize and scrolling for long views. Committing now to fix
the remaining issues in the tree.
Testing and comments by otto@ and harding@, ok deraadt@

use the 1 line load line from 'systat vm' on all the other displays, giving
them more room. much cleaner, and it is less code too; ok deanna

fairly simple cleanup, lots of testing done

post-mortem stuff can go away here too

move sysctl comment on namelist[] entry

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

mark which nlist[] kmem entries are not needed in the sysctl case, and which are

ANSI

Use sysctl for mbstat

move back. pvalchev -- the bug was quite obscure

back out latest changes which cause problems on alpha/sparc64 until
they are fixed

fix signal races. use sysctl() instead of kvm where possible, and then KNF
the hell out of it; checked by miod

kill more registers

millert@ ok

handle kvm_nlist() failing, from pr#1798.
Patch similar to the one submitted by <peterw@documenta.com.au>

bye bye sprintf

rcsid

it compiles w/ new curses

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.

Add sys/queue.h where it's needed. Unbreak userland following recent
removal from mbuf.h. ok mpi@

Avoid calling freeifaddrs() with an uninitialised pointer in an
error path.

Adjust <sys/param.h> comments regarding use of use of MSIZE, or
delete <sys/param.h> if now possible
ok guenther

first batch of programs adapting to the namespace cleanup
(pfvar.h nameser.h proc.h ucred.h)
ok guenther millert, and some review from doug as well.

i renamed the mbuf pool.

fix if (statement);
ok dlg@

cut things that relied on mclgeti for rx ring accounting/restriction over
to using if_rxr.

cut the reporting systat did over to the rxr ioctl.

tested as much as i can on alpha, amd64, and sparc64.
mpi@ has run it on macppc.
ok mpi@

info about pools is currently given to userland by copying each
pools struct out. however, struct pool in the kernel contains lots
of things that userland probably isnt interested in, like actual
mutexes, and probably shouldnt get easy access to, like pointers
to kernel memory via all the lists/trees.

this implements a kinfo_pool structure that has only the data that
userland needs to know about. it cuts the sysctl code over to
building it from struct pool as required and copying that out
instead, and cuts userland over to only handling kinfo_pool.

the only problem with this is vmstat, which can read kernel images
via kvm, which needs some understanding of struct pool. to cope,
the struct pool definition is guarded by if defined(_KERNEL) ||
defined(_LIBKVM) as inspired by sysctl which needs to do the same
thing sometimes. struct pool itself is generally not visible to
userland though, which is good.

matthew@ suggested struct kinfo_pool instead of struct pool_info.
the kinfo prefix has precedent.
lots of people liked this.

Now that the water marks are updated on a per-pool basis indicate in
the LIVELOCKS column if there is a pending (deferred) update.

ok claudio@

- use a common FIELD_ADDR macro, instead of rolling 78 identical copies.

ok canacar@

Print relativ number of livelocks that happend between two systat
updates instead of showing absolut figures. Makes it easier to
estimate the "livelockyness" of a system.
OK dlg@

expose the kernels network livelock counter

tweak the mclgeti algorithm to behave better under load.

instead of letting hardware rings grow on every interrupt, restrict
it so it can only grow once per softclock tick. we can only punish
the rings on softclock ticks, so it make sense to only grow on
softclock tick boundaries too.

the rings are now punished after >1 lost softclock tick rather than
>2. mclgeti is now more aggressive at detecting livelock.

the rings get punished by an 8th, rather than by half.

we now allow the rings to be punished again even if the system is
already considered in livelock.

without this diff a livelocked system will have its rx ring sizes
scale up and down very rapidly, while holding the rings low for too
long. this affected throughput significantly.

discussed and tested heavily at j2k10. there are still some games
with softnet we can play, but this is a good first step.

"put it in" and ok deraadt@
ok claudio@ krw@ henning@ mcbride@

if we find out that it sucks we can pull it out again later. till then
we'll run with it and see how it goes.

don't rely on mbuf.h to pull in pool.h here either

ok miod@

fix use of uninitialized variable.

reported by gcc.

ok canacar@

args, not that file

no more natpass

remove unneeded argument to warnx()
remove unused variable
kill some extra blank lines

ok canacar@

Make sure systat mbufs prints the interface name even if an interface only
allocates clusters larger than 2k.

ok deraadt@

fix the mbuf view to cope with the recent change to the mclgeti allocator.

from canacar@

Skip missing pool indices instead of printing errors.

Print the interface livelocks if non-zero. From deraadt@

Rewrite the 'mbuf' page to display mbuf usage information from system pools
and mbuf cluster statistics collected for each interface. More useful than
earlier bar graphs about mbuf types. Requested by and ok deraadt@

New display engine for systat, based on pftop. Adds new views for pf
(status, state, rule, queue). While all displays work, some keyboard
comands are not implemented yet. Other features include better handling
of display resize and scrolling for long views. Committing now to fix
the remaining issues in the tree.
Testing and comments by otto@ and harding@, ok deraadt@

use the 1 line load line from 'systat vm' on all the other displays, giving
them more room. much cleaner, and it is less code too; ok deanna

fairly simple cleanup, lots of testing done

post-mortem stuff can go away here too

move sysctl comment on namelist[] entry

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

mark which nlist[] kmem entries are not needed in the sysctl case, and which are

ANSI

Use sysctl for mbstat

move back. pvalchev -- the bug was quite obscure

back out latest changes which cause problems on alpha/sparc64 until
they are fixed

fix signal races. use sysctl() instead of kvm where possible, and then KNF
the hell out of it; checked by miod

kill more registers

millert@ ok

handle kvm_nlist() failing, from pr#1798.
Patch similar to the one submitted by <peterw@documenta.com.au>

bye bye sprintf

rcsid

it compiles w/ new curses

branches: 1.1.1;
Initial revision

Add sys/queue.h where it's needed. Unbreak userland following recent
removal from mbuf.h. ok mpi@

Avoid calling freeifaddrs() with an uninitialised pointer in an
error path.

Adjust <sys/param.h> comments regarding use of use of MSIZE, or
delete <sys/param.h> if now possible
ok guenther

first batch of programs adapting to the namespace cleanup
(pfvar.h nameser.h proc.h ucred.h)
ok guenther millert, and some review from doug as well.

i renamed the mbuf pool.

fix if (statement);
ok dlg@

cut things that relied on mclgeti for rx ring accounting/restriction over
to using if_rxr.

cut the reporting systat did over to the rxr ioctl.

tested as much as i can on alpha, amd64, and sparc64.
mpi@ has run it on macppc.
ok mpi@

info about pools is currently given to userland by copying each
pools struct out. however, struct pool in the kernel contains lots
of things that userland probably isnt interested in, like actual
mutexes, and probably shouldnt get easy access to, like pointers
to kernel memory via all the lists/trees.

this implements a kinfo_pool structure that has only the data that
userland needs to know about. it cuts the sysctl code over to
building it from struct pool as required and copying that out
instead, and cuts userland over to only handling kinfo_pool.

the only problem with this is vmstat, which can read kernel images
via kvm, which needs some understanding of struct pool. to cope,
the struct pool definition is guarded by if defined(_KERNEL) ||
defined(_LIBKVM) as inspired by sysctl which needs to do the same
thing sometimes. struct pool itself is generally not visible to
userland though, which is good.

matthew@ suggested struct kinfo_pool instead of struct pool_info.
the kinfo prefix has precedent.
lots of people liked this.

Now that the water marks are updated on a per-pool basis indicate in
the LIVELOCKS column if there is a pending (deferred) update.

ok claudio@

- use a common FIELD_ADDR macro, instead of rolling 78 identical copies.

ok canacar@

Print relativ number of livelocks that happend between two systat
updates instead of showing absolut figures. Makes it easier to
estimate the "livelockyness" of a system.
OK dlg@

expose the kernels network livelock counter

tweak the mclgeti algorithm to behave better under load.

instead of letting hardware rings grow on every interrupt, restrict
it so it can only grow once per softclock tick. we can only punish
the rings on softclock ticks, so it make sense to only grow on
softclock tick boundaries too.

the rings are now punished after >1 lost softclock tick rather than
>2. mclgeti is now more aggressive at detecting livelock.

the rings get punished by an 8th, rather than by half.

we now allow the rings to be punished again even if the system is
already considered in livelock.

without this diff a livelocked system will have its rx ring sizes
scale up and down very rapidly, while holding the rings low for too
long. this affected throughput significantly.

discussed and tested heavily at j2k10. there are still some games
with softnet we can play, but this is a good first step.

"put it in" and ok deraadt@
ok claudio@ krw@ henning@ mcbride@

if we find out that it sucks we can pull it out again later. till then
we'll run with it and see how it goes.

don't rely on mbuf.h to pull in pool.h here either

ok miod@

fix use of uninitialized variable.

reported by gcc.

ok canacar@

args, not that file

no more natpass

remove unneeded argument to warnx()
remove unused variable
kill some extra blank lines

ok canacar@

Make sure systat mbufs prints the interface name even if an interface only
allocates clusters larger than 2k.

ok deraadt@

fix the mbuf view to cope with the recent change to the mclgeti allocator.

from canacar@

Skip missing pool indices instead of printing errors.

Print the interface livelocks if non-zero. From deraadt@

Rewrite the 'mbuf' page to display mbuf usage information from system pools
and mbuf cluster statistics collected for each interface. More useful than
earlier bar graphs about mbuf types. Requested by and ok deraadt@

New display engine for systat, based on pftop. Adds new views for pf
(status, state, rule, queue). While all displays work, some keyboard
comands are not implemented yet. Other features include better handling
of display resize and scrolling for long views. Committing now to fix
the remaining issues in the tree.
Testing and comments by otto@ and harding@, ok deraadt@

use the 1 line load line from 'systat vm' on all the other displays, giving
them more room. much cleaner, and it is less code too; ok deanna

fairly simple cleanup, lots of testing done

post-mortem stuff can go away here too

move sysctl comment on namelist[] entry

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

mark which nlist[] kmem entries are not needed in the sysctl case, and which are

ANSI

Use sysctl for mbstat

move back. pvalchev -- the bug was quite obscure

back out latest changes which cause problems on alpha/sparc64 until
they are fixed

fix signal races. use sysctl() instead of kvm where possible, and then KNF
the hell out of it; checked by miod

kill more registers

millert@ ok

handle kvm_nlist() failing, from pr#1798.
Patch similar to the one submitted by <peterw@documenta.com.au>

bye bye sprintf

rcsid

it compiles w/ new curses

branches: 1.1.1;
Initial revision