track of the number of dirty buffers held by a vnode. When a
bdwrite is done on a buffer, check the existing number of dirty
buffers associated with its vnode. If the number rises above
vfs.dirtybufthresh (currently 90% of vfs.hidirtybuffers), one
of the other (hopefully older) dirty buffers associated with
the vnode is written (using bawrite). In the event that this
approach fails to curb the growth in it the vnode's number of
dirty buffers (due to soft updates rollback dependencies),
the more drastic approach of doing a VOP_FSYNC on the vnode
is used. This code primarily affects very large and actively
written files such as snapshots. This change should eliminate
hanging when taking snapshots or doing background fsck on
very large filesystems.
Hopefully, one day it will be possible to cache filesystem
metadata in the VM cache as is done with file data. As it
stands, only the buffer cache can be used which limits total
metadata storage to about 20Mb no matter how much memory is
available on the system. This rather small memory gets badly
thrashed causing a lot of extra I/O. For example, taking a
snapshot of a 1Tb filesystem minimally requires about 35,000
write operations, but because of the cache thrashing (we only
have about 350 buffers at our disposal) ends up doing about
237,540 I/O's thus taking twenty-five minutes instead of four
if it could run entirely in the cache.
Reported by: Attila Nagy <bra@fsn.hu>
Sponsored by: DARPA & NAI Labs.
- Remove the buftimelock mutex and acquire the buf's interlock to protect
these fields instead.
- Hold the vnode interlock while locking bufs on the clean/dirty queues.
This reduces some cases from one BUF_LOCK with a LK_NOWAIT and another
BUF_LOCK with a LK_TIMEFAIL to a single lock.
Reviewed by: arch, mckusick
- Get rid of the useless atop() / pmap_phys_address() detour. The
device mmap handlers must now give back the physical address
without atop()'ing it.
- Don't borrow the physical address of the mapping in the returned
int. Now we properly pass a vm_offset_t * and expect it to be
filled by the mmap handler when the mapping was successful. The
mmap handler must now return 0 when successful, any other value
is considered as an error. Previously, returning -1 was the only
way to fail. This change thus accidentally fixes some devices
which were bogusly returning errno constants which would have been
considered as addresses by the device pager.
- Garbage collect the poorly named pmap_phys_address() now that it's
no longer used.
- Convert all the d_mmap_t consumers to the new API.
I'm still not sure wheter we need a __FreeBSD_version bump for this,
since and we didn't guarantee API/ABI stability until 5.1-RELEASE.
Discussed with: alc, phk, jake
Reviewed by: peter
Compile-tested on: LINT (i386), GENERIC (alpha and sparc64)
Runtime-tested on: i386
time and there's no indication that it will improve anytime soon.
By removing support for SimOS it is possible to build LINT on
Alpha, which is considered more important at the moment.
Not objected to on: alpha@
- Changed VM_MAXUSER_ADDRESS to be defined in terms of PTDPTDI. In order for
assumptions about the recursive page table map to work it must be the base
of the recursive map. Any pte offset that's not NPTEPG will break these
assumptions.
Sponsored by: DARPA, Network Associates Laboratories
flag that can be marked on each symmetric op
o eliminate hw.ubsec.maxbatch and hw.ubsec.maxaggr since they are not
needed anymore
o change ubsec_feed to return void instead of int since zero is always
returned and noone ever looked at the return value
instead of allocating another page of kva and mapping it in again. This was
likely an oversight in revision 1.174 (cut and paste from pmap_pinit).
Discussed with: peter, tegge
Sponsored by: DARPA, Network Associates Laboratories
Reading the PCI config space with the wrong (larger) size is not
a problem in this case, but writing can be as it clobbers unrelated
registers. In this case the clobbering is for reserved fields, which
too is mostly harmless... for now. Hence, this change is mostly
preventive in nature.
page directory.
- Use these instead of the magic constants 1 or PAGE_SIZE where appropriate.
There are still numerous assumptions that the page directory is exactly
1 page.
Sponsored by: DARPA, Network Associates Laboratories
without waiting, since they are called from a system-call context only.
This appears to fix all sorts of problems with open("/dev/dsp", O_WRONLY)
randomly returning ENXIO.
Found by: cognet
Security improvements:
- Increase the size of each syncookie secret from 32 to 128 bits
in order to make brute force attacks on the secrets much more
difficult.
- Always return the lowest order dword from the MD5 hash; this
allows us to expose 2 more bits of the cookie and makes ACK
floods which seek to guess the cookie value more difficult.
Performance improvements:
- Increase the lifetime of each syncookie from 4 seconds to 16
seconds. This increases the usefulness of syncookies during
an attack.
- From Yahoo!: Reduce the number of calls to MD5Update; this
results in a ~17% increase in cookie generation time here.
Reviewed by: hsu, jayanth, jlemon, nectar
MFC After: 15 seconds
in massive locking issues on diskless systems.
It is also not clear that this sysctl is non-dangerous in its
requirements for locked down memory on large RAM systems.
type M_STRING, now defined in malloc.h. Useful when string parsing
must occur using the kernel strsep() and we want to avoid toasting
the source string.
Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories
should be done in crypto_done rather than in the callback thread
o use this flag to mark operations from /dev/crypto since the callback
routine just does a wakeup; this eliminates the last unneeded ctx switch
o change CRYPTO_F_NODELAY to CRYPTO_F_BATCH with an inverted meaning
so "0" becomes the default/desired setting (needed for user-mode
compatibility with openbsd)
o change crypto_dispatch to honor CRYPTO_F_BATCH instead of always
dispatching immediately
o remove uses of CRYPTO_F_NODELAY
o define COP_F_BATCH for ops submitted through /dev/crypto and pass
this on to the op that is submitted
Similar changes and more eventually coming for asymmetric ops.
MFC if re gives approval.
branch targets that are too far apart for the BRADDR relocation.
This is caused by the branch prediction optimizationi in the atomic
inlines here, because they jump across sections.
The workaround is to suppress jumping to a different section when
compiling LINT. To generate correct code in that case, the section
directives are replaced by a branch and a label to deal with the
fall-through case. Reasonably good C compilers will optimize this
away anyway, so the end result isn't really that bad.
packets coming out of a GIF tunnel are re-processed by ipfw, et. al.
By default they are not reprocessed. With the option they are.
This reverts 1.214. Prior to that change packets were not re-processed.
After they were which caused problems because packets do not have
distinguishing characteristics (like a special network if) that allows
them to be filtered specially.
This is really a stopgap measure designed for immediate MFC so that
4.8 has consistent handling to what was in 4.7.
PR: 48159
Reviewed by: Guido van Rooij <guido@gvr.org>
MFC after: 1 day
o Make DXS3 the primary playback channel. It may be the only
universally supported channel with the assorted revisions of this
chipset.
o Add sysctl and handler for enabling s/pdif output from DXS3.
as opposed to one after the other. This is faster in both -CURRENT
and -STABLE. Additionally, there is less code duplication for
error-checking.
One thing to note is that this code seems to return(1) when no buffers
are available; perhaps ENOBUFS should be the correct return value?
Partially submitted & tested by: Hiten Pandya <hiten@unixdaemons.com>
MFC after: 1 week
and enable it by default, with a limit of 16.
At the same time, tweak maxfragpackets downward so that in the worst
possible case, IP reassembly can use only 1/2 of all mbuf clusters.
MFC after: 3 days
Reviewed by: hsu
Liked by: bmah
a snapshot. As part of taking a snapshot of a filesystem, the kernel
builds up a list of the filesystem metadata (such as the cylinder
group bitmaps) that are contained in the snapshot. When doing a
copy-on-write check, the list is first consulted. If the block being
written is found on the list, then the full snapshot lookup can be
avoided. Besides providing an important performance speedup this
check also avoids a potential deadlock between the code creating
the snapshot and the bufdaemon trying to cleanup snapshot related
buffers. This fix creates a temporary list containing the key
metadata blocks that can cause the deadlock. This temporary list
is used between the time that the snapshot is first enabled and the
time that the fully complete list is built.
Reported by: Attila Nagy <bra@fsn.hu>
Sponsored by: DARPA & NAI Labs.
is being taken from panicing with either "freeing free block" or
"freeing free inode". The problem arises when the snapshot code
is scanning the filesystem looking for inodes with a reference
count of zero (e.g., unlinked but still open) so that it can
expunge them from its view. If it encounters a reclaimed vnode
and has to restart its scan, then it will panic if it encounters
and tries to free an inode that it has already processed. The fix
is to check each candidate inode to see if it has already been
processed before trying to delete it from the snapshot image.
Sponsored by: DARPA & NAI Labs.
that they convert to 64-bit values before shifting rather than
afterwards. Once fixed, they can be used rather than inline expanded.
Sponsored by: DARPA & NAI Labs.
Retire the "d_dump_t" and use the "dumper_t" type instead.
Dumper_t takes a void * as first arg which is more general than the
dev_t taken by d_dump_t. (Remember: we could have net-dumpers if
somebody wrote us one!)
Define the convention for GEOM controlled disk devices to be that the
first argument to the dumper function is the struct disk pointer.
Change device drivers accordingly.
Change the argument to disk_destroy() to be the same struct disk * as
disk_create() takes.
This enables drivers to ignore the (now) bogus dev_t which disk_create()
returns.
a number of related problems along the way.
- Automatically detect CDROM drives that can't handle 6 byte mode
sense and mode select, and adjust our command size accordingly.
We have to handle this in the cd(4) driver (where the buffers are
allocated), since the parameter list length is different for the
6 and 10 byte mode sense commands.
- Remove MODE_SENSE and MODE_SELECT translation removed in ATAPICAM
and in the umass(4) driver, since there's no way for that to work
properly.
- Add a quirk entry for CDROM drives that just hang when they get a 6
byte mode sense or mode select. The reason for the quirk must be
documented in a PR, and all quirks must be approved by
ken@FreeBSD.org. This is to make sure that we fully understand why
each quirk is needed. Once the CAM_NEW_TRAN_CODE is finished, we
should be able to remove any such quirks, since we'll know what
protocol the drive speaks (SCSI, ATAPI, etc.) and therefore whether
we should use 6 or 10 byte mode sense/select commands.
- Change the way the da(4) handles the no_6_byte sysctl. There is
now a per-drive sysctl to set the minimum command size for that
particular disk. (Since you could have multiple disks with
multiple requirements in one system.)
- Loader tunable support for all the sysctls in the da(4) and cd(4)
drivers.
- Add a CDIOCCLOSE ioctl for cd(4) (bde pointed this out a long
time ago).
- Add a media validation routine (cdcheckmedia()) to the cd(4)
driver, to fix some problems bde pointed out a long time ago. We
now allow open() to succeed no matter what, but if we don't detect
valid media, the user can only issue CDIOCCLOSE or CDIOCEJECT
ioctls.
- The media validation routine also reads the table of contents off
the drive. We use the table of contents to implement the
CDIOCPLAYTRACKS ioctl using the PLAY AUDIO MSF command. The
PLAY AUDIO TRACK INDEX command that we previously used was
deprecated after SCSI-2. It works in every SCSI CDROM I've tried,
but doesn't seem to work on ATAPI CDROM drives. We still use the
play audio track index command if we don't have a valid TOC, but
I suppose it'll fail anyway in that case.
- Add _len() versions of scsi_mode_sense() and scsi_mode_select() so
that we can specify the minimum command length.
- Fix a couple of formatting problems in the sense printing code.
MFC after: 4 weeks
OSes has probably caused more problems than it ever solved. Allow the
user to retire the old behavior by specifying their own privileged
range with,
net.inet.ip.portrange.reservedhigh default = IPPORT_RESERVED - 1
net.inet.ip.portrange.reservedlo default = 0
Now you can run that webserver without ever needing root at all. Or
just imagine, an ftpd that can really drop privileges, rather than
just set the euid, and still do PORT data transfers from 20/tcp.
Two edge cases to note,
# sysctl net.inet.ip.portrange.reservedhigh=0
Opens all ports to everyone, and,
# sysctl net.inet.ip.portrange.reservedhigh=65535
Locks all network activity to root only (which could actually have
been achieved before with ipfw(8), but is somewhat more
complicated).
For those who stick to the old religion that 0-1023 belong to root and
root alone, don't touch the knobs (or even lock them by raising
securelevel(8)), and nothing changes.
Some drives seem to be confused by simultaneous probes.
Tested by: marcel
As a side effect, logical units whose lun is greater than 0 might not be
probed correctly if the lun of 0 doesn't exist in the target because
CAM doesn't scan such luns.
I have a SCSI-FireWire bridge which maps SCSI-ID to LUN and it is an
example of such targets.
dev_t to the method functions.
The dev_t can still be found at struct consdev *->cn_dev.
Add a void *cn_arg element to struct consdev which the drivers can use
for retrieving their softc.
This moves all chipset specific code to a new file 'ata-chipset.c'.
Extensive use of tables and pointers to avoid having the same switch
on chipset type in several places, and to allow substituting various
functions for different HW arch needs.
Added PIO mode setup and all DMA modes.
Support for all known SiS chipsets. Thanks to Christoph Kukulies for
sponsoring a nice ASUS P4S8X SiS648 based board for this work!
Tested on: i386, PC98, alpha and sparc64
This moves all chipset specific code to a new file 'ata-chipset.c'.
Extensive use of tables and pointers to avoid having the same switch
on chipset type in several places, and to allow substituting various
functions for different HW arch needs.
Added PIO mode setup and all DMA modes.
Support for all known SiS chipsets. Thanks to Christoph Kukulies for
sponsoring a nice ASUS P4S8X SiS648 based board for this work!
Tested on: i386, PC98, alpha and sparc64
In devsw() return dead_cdevsw instead of NULL in case the dev_t does not
have a si_devsw.
This may improve our survival chances with devices which go away unexpectedly.
compile-time constants). That is, a "bucket" now is not necessarily
a page-worth of mbufs or clusters, but it is MBUF_BUCK_SZ, CLUS_BUCK_SZ
worth of mbufs, clusters.
o Rename {mbuf,clust}_limit to {mbuf,clust}_hiwm and introduce
{mbuf,clust}_lowm, which currently has no effect but will be used
to set the low watermarks.
o Fix netstat so that it can deal with the differently-sized buckets
and teach it about the low watermarks too.
o Make sure the per-cpu stats for an absent CPU has mb_active set to 0,
explicitly.
o Get rid of the allocate refcounts from mbuf map mess. Instead,
just malloc() the refcounts in one shot from mbuf_init()
o Clean up / update comments in subr_mbuf.c
used to share resource limits between rfork threads, but never was.
Removing it makes resource limit locking much simpler -- only the current
process can change the contents of the structure that p_limit points to.
support for Tekram DC395U2W cards.
Add a fix submitted by joerg@ to correctly report some errors to CAM.
Use bus_dma instead of the remaining vtophys().
reference counter array for mbuf clusters. I don't know
how this got past early testing nor how it survived so long
without getting caught. If anyone was seeing really really
bizarre memory corruption in a few mbufs this would be why.
field for holding driver-dependant data. Instead of putting the pointer
to the driver command struct in there, take advantage of these structs
being a (virtually) contiguous array and just put the array index in the
field.
control block. Allow the socket and tcpcb structures to be freed
earlier than inpcb. Update code to understand an inp w/o a socket.
Reviewed by: hsu, silby, jayanth
Sponsored by: DARPA, NAI Labs
routine does not require a tcpcb to operate. Since we no longer keep
template mbufs around, move pseudo checksum out of this routine, and
merge it with the length update.
Sponsored by: DARPA, NAI Labs
that a command completion happened, all further processing is deferred to
a taskqueue. The taskqueue itself runs implicetely under Giant, but we
already used a taskqueue for the biodone() processing, so this at least
saves the contesting of Giant in the interrupt handler.
retain symetry with aac_alloc_commans(). Since aac_alloc_commands()
allocates fib maps and places them onto the fib lists, aac_free_commands()
should reverse those operations.
o Combine two ifs with the same body with an ||.
o Switch from uintptr_t to uint32_t for fib map load operations.
The target is a uint32_t so using this type for the map load call
avoids an extra cast. uintptr_t should only be used when you need
an "int sized the same as the machine's poitner size" which is not
the case here.
o Removed the commented out M_WAITOK flag in the allocation in
aac_alloc_commands(). The kernel will only block in the allocator
if it can grow the size of the kernel. This usually results in a
page-out which could involve this aac device. Thus, sleeping here
could deadlock the machine. Assuming this operation is occurring outside
of attach time, we have enough fibs to operate anyway, so waiting for
fibs to free up is okay if not optimal.
o In aac_alloc_commands(), if we cannot dmamem_alloc additional fib
space, free the fib map.
o In aac_alloc_commands(), if we cannot create per-command dmamaps, don't
lose track of the fib map that is mapping all of the commands that we
have already released into the free pool. Instead, just cut out of
the loop and modify aac_free_commands to not attempt to free maps that
have not been allocated.
o Don't use a magic number when pre-allocating fibs.
o Use PAGE_SIZE to allocate in page sized chunks instead of an
architecture specific constant.
Submitted by: gibbs
- delay acks for T/TCP regardless of delack setting
- fix bug where a single pass through tcp_input might not delay acks
- use callout_active() instead of callout_pending()
Sponsored by: DARPA, NAI Labs
wish the busdma APIs were more consistent accross architectures.
We should probably move all the other DMA map creations in
xl_attach() where we can really handle them failing, since
xl_init() is void and shouldn't fail.
Pointy hat to: mux
Tested by: Anders Andersson <anders@hack.org>
One of the vnodes is on different mount and is possibly on a different
kind of filesystem; treating it as an smbfs vnode then writing to it
will probably corrupt it.
PR: 48381
MFC after: 1 month
group number properly based on the board id. Perform dummy reads of
registers after writing to flush the hardware write buffers.
This gets the soon to be committed zs attachment working.
Minor CIS resource allocation code cleanup
Remove some fairly useless debug writes.
This finishes the work to move as much cardbus code as possible into
pci. We wind up removing 800-odd lines from cardbus.c: we go from
1285 to 400 lines.
Reviewed by: mdodd
bus_dmamap_load() it.
- Make it so reusing mbufs when we can't allocate (or map) new ones
actually works. We were previously trying to reuse a mbuf which
was already bus_dmamap_unload()'ed.
Reviewed by: silby
- Fix memory leak in detaching.
- Initialize fc->status to other than FWBUSREST.
* fwohci.c
- Ignore BUS reset events while BUS reset phase. We can't clear that flag
during bus reset phase.
UltraSPARCs, and an eeprom attachment for fhc, which allows the date
to be set properly on these machines. Central is a wierd bus which
seems to only ever have 1 fhc attached to it. FHC (FireHose Controller)
is another wierd bus with various things on it depending where its attached.
The fhc attached to central has eeprom and zs, and the fhcs which attach
directly to nexus have simm-status, environment and other nodes, none of
which I'll probably ever have documentation for.
Thanks to Ade Lovett for providing access to an 8 cpu e4500.
#if'ed out for a while. Complete the deed and tidy up some other bits.
We need to be able to call this stuff from outer edges of interrupt
handlers for devices that have the ISR bits in pci config space. Making
the bios code mpsafe was just too hairy. We had also stubbed it out some
time ago due to there simply being too much brokenness in too many systems.
This adds a leaf lock so that it is safe to use pci_read_config() and
pci_write_config() from interrupt handlers. We still will use pcibios
to do interrupt routing if there is no acpi.. [yes, I tested this]
Briefly glanced at by: imp
is encoded in the PCI BAR. The latter is more reliable.
This allows the sio/modem function of the Xircom RealPort ethernet+modem
card to work. Note that there still seem to be issues with sio_pci not
releasing resources on detach.
pci busses implement this.
Also minor comment smithing in cardbus. Fix copyright to this year
with my name on it since I've been doing a lot to this file.
Reviewed by: jhb
on the fly and read into userland one at a time, this costs very
little total memory. The pnpinfo sizes of pccard is more than 64
bytes due to the length of the strings that man cards have in their
CIS.
- Don't initiate bus reset even if probe failed for some nodes to prevent
infinite bus reset loop.
Problem Reported by: Pierre Beyssac <pb@fasterix.frmug.org>
- Protect timeout routine with splfw() for 4-stable.
* sbp.c
- Make sure to release devq when start request.
cr_uid.
Note: we do not have socheckuid() in RELENG_4, ip_fw2.c uses its
own macro for a similar purpose that is why ipfw2 in RELENG_4 processes
uid rules correctly. I will MFC the diff for code consistency.
Reported by: Oleg Baranov <ol@csa.ru>
Reviewed by: luigi
MFC after: 1 month
sched_lock around accesses to p_stats->p_timer[] to avoid a potential
race with hardclock. getitimer(), setitimer() and the realitexpire()
callout are now Giant-free.
