| /dragonfly/sys/dev/disk/sili/ |
| H A D | sili_dragonfly.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| /dragonfly/sys/kern/ |
| H A D | subr_diskgpt.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | subr_disklabel32.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | subr_disklabel64.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | lwkt_serialize.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | kern_umtx.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | kern_device.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | lwkt_ipiq.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | lwkt_thread.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | kern_lock.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | kern_physio.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | subr_diskmbr.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | sys_pipe.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| /dragonfly/sys/dev/drm/ |
| H A D | drm_lock.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | drm_drv.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| /dragonfly/sys/dev/disk/ahci/ |
| H A D | ahci_dragonfly.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| /dragonfly/sys/dev/disk/nata/ |
| H A D | ata-raid.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| /dragonfly/sys/sys/ |
| H A D | bio.h | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | buf2.h | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | thread.h | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| /dragonfly/sys/dev/netif/iwi/ |
| H A D | if_iwi.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| /dragonfly/sys/vfs/hammer/ |
| H A D | hammer_subs.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | hammer_io.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| /dragonfly/sys/bus/cam/ |
| H A D | cam_sim.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|
| H A D | cam_xpt.c | diff ae8e83e6 Wed Jul 15 02:31:18 GMT 2009 Matthew Dillon <dillon@apollo.backplane.com> MPSAFE - tsleep_interlock, BUF/BIO, cluster, swap_pager.
* tsleep_interlock()/tsleep() could miss wakeups during periods of
heavy cpu activity. What would happen is code inbetween the two
calls would try to send an IPI (say, issue a wakeup()), but while
sending the IPI the kernel would be forced to process incoming IPIs
synchronous to avoid a deadlock.
The new tsleep_interlock()/tsleep() code adds another TAILQ_ENTRY to
the thread structure allowing tsleep_interlock() to formally place
the thread on the appropriate sleep queue without having to deschedule
the thread. Any wakeup which occurs between the interlock and the
real tsleep() call will remove the thread from the queue and the
later tsleep() call will recognize this and simply return without sleeping.
The new tsleep() call requires PINTERLOCKED to be passed to tsleep
so tsleep() knows that the thread has already been placed on a sleep
queue.
* Continue making BUF/BIO MPSAFE. Remove B_ASYNC and B_WANT from buf->b_flag
and add a new bio->bio_flags field to the bio. Add BIO_SYNC, BIO_WANT,
and BIO_DONE. Use atomic_cmpset_int() (aka cmpxchg) to interlock
biodone() against biowait().
vn_strategy() and dev_dstrategy() call semantics now require that
synchronous BIO's install a bio_done function and set BIO_SYNC in
the bio.
* Clean up the cluster code a bit.
* Redo the swap_pager code. Instead of issuing I/O during the collection,
which depended on critical sections to avoid races in the cluster append,
we now build the entire collection first and then dispatch the I/O.
This allows us to use only async completion for the BIOs, instead of
a hybrid sync-or-async completion.
|