Tor created a while ago, removes the raw I/O piece (that has cache coherency
problems), and adds a buffer cache / VM freeing piece.
Essentially this patch causes O_DIRECT I/O to not be left in the cache, but
does not prevent it from going through the cache, hence the 80%. For
the last 20% we need a method by which the I/O can be issued directly to
buffer supplied by the user process and bypass the buffer cache entirely,
but still maintain cache coherency.
I also have the code working under -stable but the changes made to sys/file.h
may not be MFCable, so an MFC is not on the table yet.
Submitted by: tegge, dillon
o If the class is PCIC_BRIDGE, subclass is PCIS_BRIDGE_PCMCIA and
programming interface is 0, assume that it is a generic PCMCIA PCI
chip we can program. I don't think there are any of these that
we don't know about, but you never know.
o If the class is PCIC_BRIDGE, subclass is PCIS_BRIDGE_CARDBUS and
programming interface is 0, assume that it is a YENTA cardbus bridge
that we know how to cope with. There are likely some cardbus bridges
that haven't it made it in here yet.
pcic_{get,put}b_io. There are some pci bridges (the CL-PD6729 and
maybe others) that do not have memory mapped registers, so we'll need
these in both places. Declare them in pcicvar.h.
- Assert Giant in vm_pageout_scan() for the vnode hacking that it does.
- Don't hold vm_mtx around vget() or vput().
- Lock Giant when calling vm_pageout_scan() from the pagedaemon. Also,
lock curproc while setting the P_BUFEXHAUST flag.
- For now we still hold Giant for all of the vm_daemon. When process
limits are locked we will be only need Giant for swapout_procs().
- Restore the previous order of setting up a new vm_object. The previous
had a small bug where we zero'd out the flags after we set the
OBJ_ONEMAPPING flag.
- Add several asserts of vm_mtx.
- Assert Giant is held rather than locking and unlocking it in a few
places.
- Add in some #ifdef objlocks code to lock individual vm objects when
vm objects each have their own lock someday.
- Don't bother acquiring the allproc lock for a ddb command. If DDB
blocked on the lock, that would be worse than having an inconsistent
allproc list.
- Add a few KTR tracepoints to track the addition and removal of
vm_map_entry's and the creation adn free'ing of vmspace's.
- Adjust a few portions of code so that we update the process' vmspace
pointer to its new vmspace before freeing the old vmspace.
- Don't lock Giant in the scheduler() function except for when calling
faultin().
- In swapout_procs(), lock the VM before the proccess to avoid a lock order
violation.
- In swapout_procs(), release the allproc lock before calling swapout().
We restart the process scan after swapping out a process.
- In swapout_procs(), un #if 0 the code to bump the vmspace reference count
and lock the process' vm structures. This bug was introduced by me and
could result in the vmspace being free'd out from under a running
process.
- Fix an old bug where the vmspace reference count was not free'd if we
failed the swap_idle_threshold2 test.
acquired.
- Assert Giant is held in the strategy, getpages, and putpages methods and
the getchainbuf, flushchainbuf, and waitchainbuf functions.
- Always call flushchainbuf() w/o the VM lock.
- Always call vfs_setdirty() with vm_mtx held.
- Fix an old comment: vm_hold_unload_pages is called vm_hold_free_pages()
nowadays.
- Always call vm_hold_free_pages() w/o vm_mtx held.
vnodes.
- Fix an old bug that would leak a reference to a fd if the vnode being
mmap'd wasn't of type VREG or VCHR.
- Lock Giant in vm_mmap() around calls into the VM that can call into
pager routines that need Giant or into other VM routines that need
Giant.
- Replace code that used a goto to jump around the else branch of a test
to use an else branch instead.
the built-in 1000baseX interface in the Level 1 LXT1001 chip. The Level 1
PHY comes up with the isolate bit in the control register set by default,
but it also has the autonegotiate bit set. When you tell the xmphy driver
to select IFM_AUTO mode, it sees that the autoneg bit is already on, and
thus doesn't bother updating the control register. However this means that
the isolate bit is never turned off (unless you manually select 1000baseSX
full or half duplex mode, which does result in the control register being
modified and the ISO bit being turned off).
This subtle and unusual behavioral difference stopped me from being able
to receive packets on the SMC9462TX card for several days, since isolating
the PHY disconnects it from the MAC's data interface. The fix is to omit
the 'is the autoneg big set?' test, since it doesn't really provide much
of an optimization anyway.
This commit also updates the xmphy driver to support the Jato/Level 1
internal PHY. (I'm not sure how Jato Technologies is related to Level 1:
all I know is the OUI from the PHY ID registers maps to Jato in the OUI
database.) This will be used once I add the if_lge driver to support
the LXT10010 chip.
- Don't bother releasing Giant while doing a lookup on the vm_map of
initproc while starting up init. We have to grab it again right after
the lookup anyways.
the chipset. This is already how the multi-hose systems handle resource
allocation and it fixes a bug where dense and bwx memory allocations were
not handled properly.
Reviewed by: gallatin
at the top of the minute, whichever comes first. It seems
logtimeout() is only called once after the kernel log is opened
and then never again after that. So I guess syslogd only gets
kernel log messages by virtue of syncer(4)'s flushes ...?
PR: 27361
Submitted by: pkern@utcc.utoronto.ca
MFC after: 1 week
This fixes a number of warnings relating to removed cloned devices.
It also makes it possible to recreate deleted devices with
mknod(2). The major/minor arguments are ignored.
systems were repo-copied from sys/miscfs to sys/fs.
- Renamed the following file systems and their modules:
fdesc -> fdescfs, portal -> portalfs, union -> unionfs.
- Renamed corresponding kernel options:
FDESC -> FDESCFS, PORTAL -> PORTALFS, UNION -> UNIONFS.
- Install header files for the above file systems.
- Removed bogus -I${.CURDIR}/../../sys CFLAGS from userland
Makefiles.
requiring fewer header files for userland programs.
Remove the gross debug device/non-debug device hack used to recognize
whether the kernel module was in sync with the userland module.
compiled with debug support. This can be used by userland programs to
recognize which ioctls the module supports.
As a result, remove the gross debug device/non-debug device hack used
to recognize whether the kernel module was in sync with the userland
module.
Replace explicit references to major/minor numbers of vinum
superdevice with the VINUM_SUPERDEV macro written for that purpose.
have a slightly different 3.3V support than the other clones, so
compensate as best we can. Note: 3.3V support is untested since I do
not have any 3.3V cards that I know of to test it with.
needs instead of relying on idiosyncratic hacks in the tty subsystem.
Also add module code since this can now be compiled as a module.
Silence by: -hackers, -audit
simpler for npx exceptions that start as traps (no assembly required...)
and works better for npx exceptions that start as interrupts (there is
no longer a problem for nested interrupts).
Submitted by: original (pre-SMPng) version by luoqi
shm_deallocate_segment because shmexit_myhook calls it, and the latter
should always be called with it already held.
Submitted by: dwmalone, dd
Approved by: alfred
gets incremented every time the kernel-userland interface changes.
This enables vinum(8) to check for the correct kernel version and to
produce a useful message if it doesn't match.
Requested by: Too many to count.
Move the definitions of struct drive, sd, plex and volume to
vinumobj.h.
Add a new debug flag, DEBUG_LOCKREQS, which logs only lock requests.
with more than one plex, the data will be accessed
multiple times. During this time, userland code could
potentially modify the buffer, thus causing data
corruption. In the case of a multi-plexed volume this
might be cosmetic, but in the case of a RAID-[45] plex it
can cause severe data corruption which only becomes
evident after a drive failure. Avoid this situation by
making a copy of the data buffer before using it.
Note that this solution does not guarantee any particular
content of the buffer, just that it remains unchanged for
the duration of the request.
Suggested by: alfred
Use this instead of DEBUG_LASTREQS to decide whether to log lock
requests.
MFS:
vinumlock: Catch a potential race condition where one process is
waiting for a lock, and between the time it is woken and
it retries the lock, another process gets it and places it
in the first entry in the table.
This problem has not been observed, but it's possible, and
it's easy enough to fix.
Submitted by: tegge
vinumunlock: Catch a real bug capable of hanging a system. When
releasing a lock, vinumunlock() called wakeup_one. This
caused wakeups to sometimes get lost. After due
consideration, we think that this is due to the fact that
you can't guarantee that some other process is also
waiting on the same address. This makes wakeup_one a
very dangerous function to use.
Requested by: bde
Add retryerrors keyword.
vinum_scandisk: Print a different message if an inadvertent start
command did not find any additional drives. The previous message "no
drives found" confused and worried many people.
MFS:
vinum_open: Recognize Mylex devices as storage devices.
In case of error, check the VF_RETRYERRORS flag in the subdisk and
don't take the subdisk down if it's set, just retry the I/O.
Requested by: peter
If the buffer has been copied (XFR_COPYBUF), release the copied
buffer when the I/O completes.
Suggested by: alfred
Desired by: bde
This commit is the first of a general cleanup of the header files..
It won't be enough to make bde happy.
Move debug definitions from vinumhdr.h.
Create a new struct rangelockinfo. In revision 1.21 of vinumlock.c,
the plex info was removed from struct rangelock, since it wasn't
needed there. It *is* needed for trace information, however, so use
struct rangelockinfo for that.
- Don't release the vm mutex early in pipespace() but instead hold it
across vm_object_deallocate() if vm_map_find() returns an error and
across pipe_free_kmem() if vm_map_find() succeeds.
- Add a XXX above a zfree() since zalloc already has its own locking,
one would hope that zfree() wouldn't need the vm lock.
flags if it is safe to do so, otherwise it will just alter the pmap state
(eg, clear the appropriate PG_FOx bits).
This gets alpha booting in the face of the vm_mtx introduction.
Reviewed by: dfr
- pc98_getmemsize() function returns available memory size under 16MB.
- getmemsize() function is merged from PC-AT's one.
Submitted by: chi@bd.mbn.or.jp (Chiharu Shibata) and
NOKUBI Hirotaka <nokubi@ff.iij4u.or.jp>
Reviewed by: hm
Bug in i4btel driver read routine corrected. The conditions in the
while() clause caused the receive queue to be referenced before checking
if a channel is connected, leading to kernel panic (do a 'dd
if=/dev/i4btel0 of=/dev/null' on an unconnected tel device, panic will
follow). Correction was to reorder the while clause conditions to check
for connectedness first.
Work through the various power commands and convert them from a "is
this a foo controller or a foo' controller or a foo''' controller" to
a cabability based scheme. We have bits in the softc that tell us
what kind of power control scheme the controller uses, rather than
relying on being able to enumerate them all. Cardbus bridges are
numerous, but nearly all implement the i82365sl-DF scheme (well, a few
implement cirrus CL-PD67xx, but those were made by Cirrus Logic!).
Add a pointer back to the softc in each pcic_slot so we can access
these flags.
Add comments that talk about the issues here. Also note in passing
that there are two differ Vpp schemes in use and that we may need to
adjust the code to deal with both of them. Note why it usually works
now.
We have 5 power management modes right now: KING, AB, DF, PD and VG.
AB is for the i82365 stpes A, B and C. DF is for step DF. PD is the
cirrus logic extensions for 3.3V while VG is the VADEM extensions for
3.3V. KING is for the IBM KING controller found on some old cards.
# I'm looking for one of those old cards or a laptop that has the KING
# bridge in it.
We have to still cheat and treat the AB parts like the DF parts
because pci isn't here yet. As far as I can tell, this is harmless
for actual old parts and necessary to work with 3.3V cards in some
laptops.
This almost eliminates all tests for controller in the code. There
are still a few unrelated to power that need taming as well.
o Introduce flags word to the softc. This will be used to control various
aspects of the driver. Right now there are two bits defined, PCIC_IO_MAPPED
and PCIC_MEM_MAPPED. One for ISA cards that are I/O mapped, the other is
for PCI cards that are memory mapped. Only the ISA side is implemented
with this commit.
o Introduce a pcic_dealloc which will cleanly dealloc resources used. Right
now it is only supported when called from probe/attach.
o Keep track of resources allocated in the pcic_softc.
o move pcictimeout_ch to the softc so we can support multiple devices
in polling mode.
o In ISA probe, set PCIC_IO_MAPPED.
o Introduce and compute the slot mask. This will be used later when
we expand the number of slots on ISA from 2 to 4. In such a case, we
appear to have to use polling mode otherwise we get two different cards
trying to drive the same interrupt line. I don't have hardware to
test this configuration, so I'll stop here.
o Add defines for the VS[12]# bits in register 0x16.
o Add comment about what we're doing reading register 0x16 (PCIC_CDGC)
in the DF case.
o Check bit VS1# rather than a random bit I was checking due to a bogus
transcrition on my part from nakagawa-san's article.
o Add note about IBM KING and 3.3V operation from information larned from
wildboard.
npxsave() went to great lengths to excecute fnsave with interrupts
enabled in case executing it froze the CPU. This case can't happen,
at least for Intel CPU/NPX's. Spurious IRQ13's don't imply spurious
freezes. Anyway, the complications were usually no-ops because IRQ13
is not used on i486's and newer CPUs, and because SMPng broke them in
rev.1.84. Forcible enabling of interrupts was changed to
write_eflags(old_eflags), but since SMPng usually calls npxsave() from
cpu_switch() with interrupts disabled, write_eflags() usually just
kept interrupts disabled.
npxinit() didn't have the usual race because it doesn't save to curpcb,
but it may have had a worse form of it since it uses the npx when it
doesn't "own" it. I'm not sure if locking prevented this. npxinit()
is normally caled with the proc lock but not sched_lock.
Use a critical region to protect pushing of curproc's npx state to
curpcb in npxexit(). Not doing so was harmless since it at worst
saved a wrong state to a dieing pcb.
Not doing this was fairly harmless because savectx() is only called
for panic dumps and the bug could at worse reset the state.
savectx() is still missing saving of (volatile) debug registers, and
still isn't called for core dumps.
machines. The code formerly read:
long val;
if (val < (long)-0x80000000 || ...)
return EINVAL;
The constant 0x80000000 has type unsigned int. The unary `-'
operator does not change the type (or the value, in this case).
Therefore the promotion to long is done by 0-extension, giving
0x0000000080000000 instead of the desired 0xffffffff80000000. I
got rid of the `-' and changed the cast to (int32_t) to give proper
sign-extension on all architectures and to better reflect the fact
that we are range-checking a 32-bit value.
This commit also makes the analogous changes to ng_int{8,16}_parse
for consistency.
MFC after: 3 days
committed to disk before clearing them. More specifically, when
free_newdirblk is called, we know that the inode claims the new
directory block. However, if the associated pagedep is still linked
onto the directory buffer dependency chain, then some of the entries
on the pd_pendinghd list may not be committed to disk yet. In this
case, we will simply note that the inode claims the block and let
the pd_pendinghd list be processed when the pagedep is next written.
If the pagedep is no longer on the buffer dependency chain, then
all the entries on the pd_pending list are committed to disk and
we can free them in free_newdirblk. This corrects a window of
vulnerability introduced in the code added in version 1.95.
things to get 3.3V. It appears that some cardbus chipsets have id
registers that say they are C step parts, but they really support the
DF step 3.3V functionality.
# Need to verify that IBM KING is handled properly since the MISC1
# register is really a cirrus logic only register.
82C146. The Intel i82365SL-DF supports 3.3V cards. The Step A/B/C
parts do not appear to support this. This is hard to know for sure
since it was deduced from "compatible" parts' data sheets and the
article mentioned below.
Rework the VLSI detection to be a little nicer and not depend on
scanning cards twice. This would allow bad VLSI cards to coexist with
a good intel card, for example. We now detect i82365SL-DF cards where
before we'd detect a VLSI. For the most part, this is good, but we
run a small chance of detecting a single slot 82C146 as a i82365SL-DF.
Since I can't find a datasheet for the 82c146, I don't know if this is
a problem or not.
This work is based on an excellent article, in Japanese, by NAKAGAWA,
Yoshihisa-san that appeared in FreeBSD Press Number 4. He provided a
patch against PAO3 in his article. Since the pcic.c code has changed
some since then, I've gone ahead and cleaned up his patch somewhat and
changed how the code detects the buggy '146 cards.
I also removed the comment asking if there were other cards that
matched the 82C146 since we found one and additional information isn't
necessary.
vm_mtx does not recurse and is required for most low level
vm operations.
faults can not be taken without holding Giant.
Memory subsystems can now call the base page allocators safely.
Almost all atomic ops were removed as they are covered under the
vm mutex.
Alpha and ia64 now need to catch up to i386's trap handlers.
FFS and NFS have been tested, other filesystems will need minor
changes (grabbing the vm lock when twiddling page properties).
Reviewed (partially) by: jake, jhb
copies out the current contents of the video buffer for a syscons terminal,
providing a snapshot of the text and attributes.
Based heavily on work originally submitted by Joel Holveck <joelh@gnu.org>
for 2.2.x almost 30 months ago, which I cleaned up a little, and forward
ported to -current.
See also the usr.bin/scrshot utility.
vn_start_write() on the given vnode will be successful. VOP_LEASE() may
help to solve this problem, but its return value ignored nearly everywhere.
For now just assume that the missing upper layer on write means insufficient
access rights (which is correct for most cases).
wakeup proc0 by hand to enforce the timeout.
- When swapping out a process, keep the process locked via the proc lock
from the first checks up until we clear PS_INMEM and set PS_SWAPPING in
swapout(). The swapout() function now must be called with the proc lock
held and releases it before returning.
- Comment out the code to attempt to lock a process' VM structures before
swapping out. It is broken in that it releases the lock after obtaining
it. If it does grab the lock, it needs to hand it off to swapout()
instead of releasing it. This can be revisisted when the VM is locked
as this is a valid test to perform. It also causes a lock order reversal
for the time being, which is the immediate cause for temporarily
disabling it.
the process in question locked as soon as we find it and determine it to
be eligible until we actually kill it. To avoid deadlock, we don't block
on the process lock but skip any process that is already locked during our
search.
lock. Since we won't actually block on a try lock operation, it's not
a problem. Add a comment explaining why it is safe to skip lock order
checking with try locks.
- Remove the ithread list lock spin lock from the order list.
sleep locks.
- Delay returning from ithread_remove_handler() until we are certain that
the interrupt handler being removed has in fact been removed from the
ithread.
- XXX: There is still a problem in that nothing protects the kernel from
adding a new handler while the ithread is running, though with our
current architectures this is not a problem.
Requested by: gibbs (2)
- Attach a writable sysctl to bootverbose (debug.bootverbose) so it can be
toggled after boot.
- Move the printf of the version string to a SI_SUB_COPYRIGHT SYSINIT just
afer the display of the copyright message instead of doing it by hand in
three MD places.
Pro and Raylink cards with version 5 firmware. Only infra-structure
mode has been tested. Specific changes for this feature are:
o Add RFC1042 encapsulation of IP datagrams
o Add authentication and association
o Decode of the beacon (although not used)
Other changes have been made:
o Pass command completion status to *_done (in place for
adding proper error recovery)
o Move a couple of state variables into the current
network parameters structure. This is in prep. for
dealing with roaming.
MFC after: 1 week
follows: the effective uid of p1 (subject) must equal the real, saved,
and effective uids of p2 (object), p2 must not have undergone a
credential downgrade. A subject with appropriate privilege may override
these protections.
In the future, we will extend these checks to require that p1 effective
group membership must be a superset of p2 effective group membership.
Obtained from: TrustedBSD Project
Remove comment about setting error for reads on EOF, read returns 0 on
EOF so the code should be ok.
Remove non-effective priority boost, PRIO+1 doesn't do anything
(according to McKusick), if a real priority boost is needed it should
have been +4.
Style fixes:
.) return foo -> return (foo)
.) FLAG1|FlAG2 -> FLAG1 | FlAG2
.) wrap long lines
.) unwrap short lines
.) for(i=0;i=foo;i++) -> for (i = 0; i=foo; i++)
.) remove braces for some conditionals with a single statement
.) fix continuation lines.
md5 couldn't verify the binary because some code had to
be shuffled around to address the style issues.
whose name is within that block must ensure not only that the block
containing the file name has been written, but also that the on-disk
directory inode references that block. When a new directory block
is created, we allocate a newdirblk structure which is linked to
the associated allocdirect (on its ad_newdirblk list). When the
allocdirect has been satisfied, the newdirblk structure is moved
to the inodedep id_bufwait list of its directory to await the inode
being written. When the inode is written, the directory entries
are fully committed and can be deleted from their pagedep->id_pendinghd
and inodedep->id_pendinghd lists.
the number of references on the filesystem root vnode to be both
expected and released. Many filesystems hold an extra reference on
the filesystem root vnode, which must be accounted for when
determining if the filesystem is busy and then released if it isn't
busy. The old `skipvp' approach required individual filesystem
xxx_unmount functions to re-implement much of vflush()'s logic to
deal with the root vnode.
All 9 filesystems that hold an extra reference on the root vnode
got the logic wrong in the case of forced unmounts, so `umount -f'
would always fail if there were any extra root vnode references.
Fix this issue centrally in vflush(), now that we can.
This commit also fixes a vnode reference leak in devfs, which could
result in idle devfs filesystems that refuse to unmount.
Reviewed by: phk, bp
soon attach directly to pcic rather than the kludge pci-pcic device we
have now.
In some ways, this is similar to the work PAO3 did to try to support
cardbus bridges. In some ways different. This and future commits
will be taking from the spirit of many of those changes. pcicvar.h is
completely different from the pcicvar.h that appeared in PAO3, but
similar in concept.
registers better. Hold sched_lock not only for checking the flag but
also while performing the actual operation to ensure the process doesn't
get swapped out by another CPU while we the operation is being performed.
controller found in many of the early NOTE98 machines that were
produced. This controller is completely unlike the intel 82365, so
I've separated it out from the main pcic driver.
- Require the proc lock be held for killproc() to allow for the vmdaemon to
kill a process when memory is exhausted while holding the lock of the
process to kill.
