| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|
|
|
| |
Submitted by: truckman
|
|
|
|
| |
busdma and has extensively been tested on i386 and sparc64.
|
|
|
|
| |
Shamed into fixing by: John Cagle <john.cagle@hp.com>
|
|
|
|
|
|
|
|
| |
smbfs_close(). This fixes paging to and from mmap()'d regions of smbfs
files after the descriptor has been closed, and makes thttpd, GNU ld,
and perhaps more things work that depend on being able to do this.
PR: 48291
|
|
|
|
|
|
| |
product ID, so we have to use the CIS strings to ID it.
dmesg by: Christian Laursen
|
| |
|
| |
|
|
|
|
| |
and close VCCs.
|
|
|
|
|
|
|
| |
because the run time exceeds the largest value a signed int can hold.
The real solution involves calculating how far we are over the limit.
To quickly solve this problem we loop removing 1/5th of the current value
until it falls below the limit. The common case requires no passes.
|
|
|
|
|
| |
with other synthetic filesystems, which have f_mntfromname the same
as f_fstypename. Noticed by Sean Kelly on -current.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Emulate lock draining (LK_DRAIN) in null_lock() to avoid deadlocks
when the vnode is being recycled.
- Don't allow null_nodeget() to return a nullfs vnode from the wrong
mount when multiple nullfs's are mounted. It's unclear why these checks
were removed in null_subr.c 1.35, but they are definitely necessary.
Without the checks, trying to unmount a nullfs mount will erroneously
return EBUSY, and forcibly unmounting with -f will cause a panic.
- Bump LOG2_SIZEVNODE up to 8, since vnodes are >256 bytes now. The old
value (7) didn't cause any problems, but made the hash algorithm
suboptimal.
These changes fix nullfs enough that a parallel buildworld succeeds.
Submitted by: tegge (partially; LK_DRAIN)
Tested by: kris
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
and run time.
- Scale the sleep and run time back via sched_interact_update() in more
places. This is to keep the statistic more accurate.
- Charge a parent one tick for forking a child.
- Add only the run time and not the sleep time to the parents kg when a
thread exits. This allows us to give a penalty for having an expensive
thread exit but does not give a bonus for having an interactive thread
exit.
- Change the SLP_RUN_THROTTLE to limit us to 4/5th and not 1/2.
- Change the SLP_RUN_MAX to two seconds. This keeps bursty interactive
applications like mozilla and openoffice in the interactive range even
through expensive tasks.
- Recalculate the slice after every sleep. This ensures that once a task
has been marked interactive it only has a slice of 1 at the risk of
giving tasks that sleep for a very brief period a longer time slice.
|
|
|
|
|
|
|
|
|
| |
this makes connect act more sensibly in these cases.
PR: 50839
Submitted by: Barney Wolff <barney@pit.databus.com>
Patch delayed by laziness of: silby
MFC after: 1 week
|
| |
|
|
|
|
|
|
|
|
| |
to be no need to see them in the general case (and they aren't
smp-safe anyway.)
Suggested by: hmp
MFC after: 1 week
|
|
|
|
|
|
|
|
|
|
| |
This version of the driver code is compatible with near-release FreeBSD 5.1
kernel/driver interfaces.
modules/Makefile, man page and other bindings to follow shortly, once I get
this part of the check-in right.
Approved by: markm(mentor)
|
|
|
|
|
|
| |
state, as inp_socket will then be NULL. This fixes a panic that occurs when one
tries to bind a port that was previously binded with remaining TIME_WAIT
sockets.
|
|
|
|
|
|
| |
group again except first kse_create syscall.
Noticed by: julian
|
|
|
|
|
|
|
|
|
|
| |
the terminating '\0'. Since the initialisation of rootpath in
libstand/bootp.c may copy junk into the rest of the buffer, it was
possible for the code to find a ':' after the '\0' and do the wrong
thing.
Reviewed by: ps
MFC after: 1 week
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
resource deallocation back to fifo_close(). This eliminates any
stale data that might be stuck in the socket buffers after all the
readers and writers have closed the fifo.
Tested by: Thorsten Schroeder <ths@katjusha.de>
|
|
|
|
|
| |
so that all of the driver sources are in one place. Adjust the configuration
files and the module build.
|
|
|
|
|
|
|
|
| |
on friday 13th and without making a universe). This adds struct and
constant definitions for ATM traffic parameters and re-enables the
build of the midway driver.
Tested by: make universe
|
|
|
|
|
|
| |
thread in such group.
Change message text from kse_rel to kserel, it is better displayed
in top.
|
|
|
|
|
| |
Remove pointless initialization of b_spc field, which now no longer
exists.
|
| |
|
|
|
|
|
|
|
|
|
| |
Previously, any normal I/O on an fdc(4) device would fail with ENXIO
if the device had been opened in non-blocking mode and then closed
prior to the conventional access; that would last until the floppy
disk was ejected and re-inserted to raise the unit attention condition.
Add a clarifying comment.
|
| |
|
|
|
|
|
|
|
| |
on an I/O attempt. This is needed for consistency with
the concept of the half-opened state of fdc(4).
PR: kern/52338
|
| |
|
|
|
|
| |
available caller private field.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
populated. Apparently, if you use an ehci controller, it's not.
Use usbd_device2interface_handle() to retrieve the interface handle.
NOTE: uaa->iface is populated in the probe routine, so I suspect the
fact that it's NULL in the attach routine is a bug in the ehci driver.
Also, don't depend on the PHY addresses returned by the AXE_CMD_READ_PHYID
command. The address is correct for my LinkSys NIC, but a user has
reported that with a D-Link NIC, the PHYID command returns address 4
while the attached Broadcom PHY is in fact strapped for address 0.
Instead, latch onto the first PHY address that returns valid data
during a readreg operation.
|
|
|
|
|
| |
to assertion of ownership. Serves me right for not booting a witness
kernel.
|
|
|
|
|
| |
Retain b_bio as the first element of struct buf for now in case some code
somewhere still do the evil cast thing.
|
|
|
|
| |
the device.
|
|
|
|
|
|
|
|
|
| |
- Add vm page queue locking in certain places that are only needed on
sparc64.
This should make pmap_qenter and pmap_qremove MP-safe.
Discussed with: alc
|
|
|
|
|
| |
gentle versions of the vop_strategy()/vop_specstrategy() mismatch methods
and use vop_panic() instead.
|
|
|
|
| |
Reviewed by: tmm
|
| |
|
|
|
|
| |
vp == bp->b_vp in specfs, that was the entire point of VOP_SPECSTRATEGY().
|
|
|
|
| |
to check that the buffer points to the correct vnode.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
as should every block device strategy routine.
There was at least one evil consequence of not doing so:
Some errors returned by fdstrategy() could be lost (EAGAIN,
in particular.)
PR: kern/52338 (in the audit-trail)
Discussed with: bde
|
|
|
|
|
|
|
|
|
|
|
|
| |
to access floppy parameters through it.
Note: The DIOCGSECTORSIZE and DIOCGMEDIASIZE handlers withing
fdioctl() couldn't be just moved to below the existing check
for blocking mode because fd->ft can be non-NULL while still
in non-blocking mode (fd->ft can be set with the FD_STYPE ioctl.)
PR: kern/52338
No MFC: Not applicable to STABLE
|
|
|
|
| |
code is locked down.
|
|
|
|
|
|
|
| |
schedules an upcall. Signal delivering to a bound thread is same as
non-threaded process. This is intended to be used by libpthread to
implement PTHREAD_SCOPE_SYSTEM thread.
2. Simplify kse_release() a bit, remove sleep loop.
|
|
|
|
|
|
|
|
|
| |
ulpt_status() afterwards. This fixes a crash that can occur if a
USB printer is power-cycled when printing is just starting. The
problem is similar to that fixed in revision 1.33, but it is much
less likely to occur.
MFC after: 1 week
|