1) Files weren't properly synced on filesystems other than UFS. In some
cases, this lead to lost data. Most likely would be noticed on NFS.
The fix is to make the VM page sync/object_clean general rather than
in each filesystem.
2) Mixing regular and mmaped file I/O on NFS was very broken. It caused
chunks of files to end up as zeroes rather than the intended contents.
The fix was to fix several race conditions and to kludge up the
"b_dirtyoff" and "b_dirtyend" that NFS relies upon - paying attention
to page modifications that occurred via the mmapping.
Reviewed by: David Greenman
Submitted by: John Dyson
These changes solve the problem in a general way by moving the
initialization out of the individual fs_mountroot's and into swaponvp().
Submitted by: Poul-Henning Kamp
changes. The check for nswap was bogus, but the code was so convoluted
that it was difficult to tell. It's better now. :-)
Reviewed by: David Greenman (extensively), and John Dyson
Submitted by: Poul-Henning Kamp, w/tweaks by me.
inconsistencies in the VM system that eventually lead to a panic. These
changes fix the behavior to conform to the behavior in SunOS, which is
to deny faults to pages beyond the EOF (returning SIGBUS). Internally,
this is implemented by requiring faults to be within the object size
boundaries. These changes exposed another bug, namely that passing in
an offset to mmap when trying to map an unnamed anonymous region also
results in internal inconsistencies. In this case, the offset is forced
to zero.
Reviewed by: John Dyson and others
serial_putchar() always hung if it was called and the serial port existed,
so booting with -h hung when the above bug was fixed. Previously, setting
-h did nothing but -h was sometimes the default due to the stack garbage
bug.
Submitted by: DI. Christian Gusenbauer <cg@scotty.edvz.uni-linz.ac.at>
The `howto' arg to boot() was not supplied, so it was stack garbage (actually
the return address in the boot program). I didn't use the submitted fix.
1) If a target initiated a sync negotiation with us and happened to chose a
value above 15, the old code inadvertantly truncated it with an "& 0x0f".
If the periferal picked something really bad like 0x32, you'd end up with
an offset of 2 which would hang the drive since it didn't expect to ever
get something so low. We now do a MIN(maxoffset, given_offset).
2) In the case of Wide cards, we were turning on sync transfers after a
sucessfull wide negotiation. Now we leave the offset alone in the per
target scratch space (which implies asyncronous transfers since we initialize
it that way) until a syncronous negotation occurs.
3) We were advertizing a max offset of 15 instead of 8 for wide devices.
4) If the upper level SCSI code sent down a "SCSI_RESET", it would hang the
system because we would end up sending a null command to the sequencer. Now
we handle SCSI_RESET correctly by having the sequencer interrupt us when it
is about to fill the message buffer so that we can fill it in ourselves.
The sequencer will also "simulate" a command complete for these "message only"
SCBs so that the kernel driver can finish up properly. The cdplay utility
will send a "SCSI_REST" to the cdplayer if you use the reset command.
5) The code that handles SCSIINTs was broken in that if more than one type
of error was true at once, we'd do outbs without the card being paused.
The else clause after the busfree case was also an accident waiting to
happen. I've now turned this into an if, else if, else type of thing, since
in most cases when we handle one type of error, it should be okay to ignore
the rest (ie if we have a SELTO, who cares if there was a parity error on
the transaction?), but the section should really be rewritten after 2.0.5.
This fix was the least obtrusive way to patch the problem.
6) Only tag either SDTR or WDTR negotiation on an SCB. The real problem is
that I don't account for the case when an SCB that is tagged to do a particular
type of negotiation completes or SELTOs (selection timeout) without the
negotiation taking place, so the accounting of sdtrpending and wdtrpending
gets screwed up. In the wide case, if we tag it to do both wdtr and sdtr,
it only performs wdtr (since wdtr must occur first and we spread out the
negotiation over two commands) so we always have sdtrpending set for that
target and we never do a real SDTR. I fill properly fix the accounting
after 2.0.5 goes out the door, but this works (as confirmed by Dan) on
wide targets.
Other stuff that is also included:
1) Don't do a bzero when recycling SCBs. The only thing that must explicitly
be set to zero is the scb control byte which is done in ahc_get_scb. We also
need to set the SG_list_pointer and SG_list_count to 0 for commands that do
not transfer data.
2) Mask the interrupt type printout for the aic7870 case. The bit we were
using to determine interrupt type is only valid for the aic7770.
Submitted by: Justin Gibbs
IGMPv2 spec. This fixes the following bugs:
o ntohs() on a char provides silly results
o timer needs to be scaled to units of PR_FASTHZ; this was being done
inconsistenly so now it gets done when it is initialized.
Reviewed by: Garrett Wollman
Submitted by: Bill Fenner <fenner@parc.xerox.com>
when the single user shell was terminated. These changes disallow mounting
or R/W upgrading filesystems that are dirty unless "-f" (force) option
is used with mount. /etc/rc has been modified to abort the startup if
one or more non-nfs partitions fail to mount.
Reviewed by: Poul-Henning Kamp, Rod Grimes
I ran into another manifestation of the problem reported in PR 211 and
fixed it. Try this:
as non-root:
cd /tmp; mkdir x y x/z
as root:
chown root /tmp/x/z
as non-root:
cd /tmp/x; mv z ../y # EACCES as expected
as root:
cd /tmp/x; mv z ../y # EINVAL NOT as expected
This is because ufs_rename() sets IN_RENAME and fails to clear it.
Reviewed by: davidg
Submitted by: bde
the 802.3 frames generated by the DC21040 (which does automatic padding
of less-than-minimum frames) and the frames generated by the 'ed'
driver, I've found that there is indeed a bug in the size of "ETHER_MIN_LEN"
as reported by several people, John Hay being the most recent. The driver
was actually setting the length to 6+6+2+50 (64 bytes), which when adding
in the CRC (which is automatically appended to the frame and not included
in the length), the minimum frame is 4 bytes larger than it is supposed to
be. All of this is confirmed by tcpdump showing 50 bytes of data for
minimum frames from the 'ed' cards and 46 bytes from 'de' cards. This
analysis has also revealed that there is garbage in the un-filled in
portion at the end of the minimum frames from the 'ed' driver; I don't
plan to fix this.
require specific partitions be mentioned in the kernel config
file ("swap on foo" is now obsolete).
From Poul-Henning:
The visible effect is this:
As default, unless
options "NSWAPDEV=23"
is in your config, you will have four swap-devices.
You can swapon(2) any block device you feel like, it doesn't have
to be in the kernel config.
There is a performance/resource win available by getting the NSWAPDEV right
(but only if you have just one swap-device ??), but using that as default
would be too restrictive.
The invisible effect is that:
Swap-handling disappears from the $arch part of the kernel.
It gets a lot simpler (-145 lines) and cleaner.
Reviewed by: John Dyson, David Greenman
Submitted by: Poul-Henning Kamp, with minor changes by me.
A phone call from Manfred quickly pointed up the fact that I got the conflict
check backwards. NOW we implement the conflict checking correctly! Wheesh!
- Do the right thing when booting in NFS diskless mode, which is nothing.
Make the default unconfigured entries for swdevt[0] and dumplo something
that swapconf() will ignore and not choke on (the swap setup is done
in nfs_vfsops.c when booting diskless).
is more representative of worst case situations of 4 files/directory. (If
that last sentence doesn't make any sense, I'm not surprised. It's rather
compilcated how this all fits together....).
This should fix a problem that Ed Hudson has been complaining about where
directories with lots of symlinks could cause excessive disk I/O.
with davidg about it, I hereby kill two undocumented misfeatures:
The code to skip a miniroot in the swapdev is not particular useful, and
if we need it we need it to be done properly, ie size the fs and skip all
of it not some hardcoded size, and subtract what we skip from the length
in the first place.
The SEQSWAP dies too. It's not the way to do it, it doesn't work, and
nobody have expressed any great desire for it to work. The way to
implement it correctly would be a second argument to swapon(2) to give
a priority/policy information. Low priority swapdevs can be made so
by adding them at a far offset (0x80000000 kind of thing), with almost no
modification to the strategy routine (in particular a offset per swapdev).
But until the need is obvious, it will not be done.
to access it. setdelayed() actually ORs the bits in `idelayed' into
`ipending' and clears `idelayed'.
Call setdelayed() every (normal) clock tick to convert delayed
interrupts into pending ones.
Drivers can set bits in `idelayed' at any time to schedule an interrupt
at the next clock tick. This is more efficient than calling timeout().
Currently only software interrupts can be scheduled.
boot diskless with it, you get a panic because setconf() is only
called for mountroot == ffs_mountroot. It really needs to be called
no matter what manner of rootfs we have. I can't really say if
swapgeneric will work with a CD-ROM though. (I get the feeling I'm
the only one who uses swapgeneric these days anyway.)
currently considering reducing the TCP fasttimo to 100ms to help improve
things, but this would be done as a seperate step at some point in the
future.
This was done because it was causing some sometimes serious performance
problems with T/TCP.
there may even be LKMs.) Also, change the internal name of `unixdomain'
to `localdomain' since AF_LOCAL is now the preferred name of this family.
Declare netisr correctly and in the right place.
msdosfs_lookup() did no validation to see if the caller was validated
to delete/rename/create files. msdosfs_setattr() did no validation
to see if the caller was allowed to change the file permissions (turn
on/off the write bit) or update the file modification time (utimes).
The routines were fixed to validate the calls just like ufs does.
On Tue, 09 May 1995 04:35:27 PDT, Richard Stevens wrote:
> In tcp_dooptions() under the case TCPOPT_CC there is an assignment
>
> to->to_flag |= TCPOPT_CC;
>
> that should be
>
> to->to_flag |= TOF_CC;
>
> I haven't thought through the ramifications of what's been happening ...
>
> Rich Stevens
Submitted by: rstevens@noao.edu (Richard Stevens)
It closed the wrong device (usually the B partition instead of the C
partition).
It closed a device without having opened it.
It didn't open a device often enough. This caused swap partitions on
slices other than the first slice looked at to be unavailable for swapping.
It didn't check the device number sufficiently.
Reopen the bdev for the raw partition and not the cdev if only the bdev
was open.
Don't use a bogus limit for the number of partitions to possibly reopen
(bug found by Julian).
Add function dssize() to help fix wdsize() and sdsize(). The slice
layer knows more about (un)open partitions and partition sizes than
the driver layer.
Remove silly "Naffy, the Wonder Porpoise" attribution and add more
justifiable (and overdue) attribution to Bruce Evans. Look at it
as a delete and add operation batched together, not a substitution. :-)
notice, performed all of the structural changes necessary to get this thing
to work with the unidirectional-DMA version of voxware.
This work is -not- complete, but it's in far better shape than it was, and
I may not touch it again for another few months.
in read() and write(). FNONBLOCK is valid in ioctl() and close().
The bug caused hung ptys when a process talked to itself using nonblocking
i/o and exited while the slave pty had output to flush. ttywait() was
called and hung. Signals didn't work because the process was exiting.
`comcontrol /dev/ttyp0 drainwait 1' worked to terminate the wait. This
shows that comcontrol is not limited to hardware control. It has no i386
or driver dependencies and doesn't belong in src/sbin/i386.
Bruce
to emit spurious page outside of object type messages. It is not
a fatal condition anyway, so the message will be omitted for
release. Also, the code that "clips" the allocation size, associated
with the above problem, was fixed.
The ``flags 1'' in the fdc line is now only needed for owners of an
Insight tape (perhaps there aren't any? Mine is disfunctional). All
other probes are safe wrt. to the motor-control line of floppy disk
drives. Document the flag in LINT finally.
fragmented.
Added support for Cogent em100 boards.
Fixed bug that caused BPF to toggle the card to UTP.
Various other improvements.
Submitted by: Matt Thomas and David Greenman
Change IPTOS_PREC_ROUTINE to 0 (was conflict with IPTOS_LOWDELAY) according
to RFC 791 (unchanged since it) and BSDI 2.0 style
Submitted by: Igor Sviridov <siac@ua.net>
initializing proc0's frame base, too, using cpu_set_init_frame(). It's
a kludge because that macro is intended to be used only for init, but
does what we want nonetheless.
loading kernel
worm.o: Undefined symbol `_STUNIT' referenced from text segment
I copied the STUNIT definition from the old scsiconf.c into this file to
work around this problem.
if the 'time on a node is 0,..
tell the world it is the same as 'boottime'.
This is becasue 'time' is not set up when we create the nodes,
so we can't set them then.
Added CONTROL device that only does user-ioctl and nothing else;
Added protection so user-ioctl requires write access;
Clean up scsiconf.h a little. It needs more work.
the lookup fails. Updated callers to deal with this. Call in_pcblookuphash
instead of in_pcblookup() in in_pcbconnect; this improves performance of
UDP output by about 17% in the standard case.
space for the hash list buckets and is a little faster. The features
of tailq aren't needed. Increased the size of the object hash table
to improve performance. In the future, this will be changed so that
the table is sized dynamically.
as side-effect, removed some legacy code that was necessary
when we called vm_fault inside of vm_fault_quick instead of using
the kernel/user space byte move routines.
(2GB). If this limit is not imposed, then filesystem corruption will
ensue when files larger than 2GB are created. This is temporary,
and the underlying limitation will be removed later.
algorithm is used on aic7770 Rev E or higher chips to improve perfomance.
This required a hardware change but we don't know exactly what (most
likely some special register to do fast SCB indexing into host memory),
and we are not at all sure that there are more than 4 SCBs on these
chips. This probe will still classify the revision of the aic7xxx, but
we now default to 4 SCBs (at least until we know more of what was done).
This also fixes a bug in the timeout routine where we cleared a flag
too soon making it imposible to enter one section of the routine.
Submitted by: Timeout bug - Dan Eischen <deischen@iworks.InterWorks.org>
interrupts to zero to start with. This is the value we return in
most cases. The sense code then outbs the apropriate value to cause
sense retrieval. The return value was uninitialized before this
change (something that was an okay thing with the old sequencer code,
but not with the semantics of the new).
This problem caused us to always retrieve sense upon recieving a non
zero status byte. This is exactly what was happening for the "target
busy" status returned by exabyte tape drives when they rewind or
power on with a tape installed. The request sense proved fatal.
This should fix the tape problems.
but it outlines what I'm GOING to do to this file.
It's sort of an unignorable notification of coming changes..
This is a bit rude I understand.. but I can't afford to haqve the
diskslice code drifting off too much further from a workable system
and I think I need to jump in now to make it obvious what has to be done
before it's too late.
appologies to bruce in advance.
bin/stty. Define alias CCAR_OFLOW for MDMBUF.
Declare speeds as having type speed_t instead of long. speed_t is
long, which is wrong (POSIX specifies it to be unsigned integral),
but fixing it might introduce more serious bugs.
user-level part has already been commited.)
Note that i've lost the "official" code for this; it went into the
system after 1.1.5.1. The commited code is my own version, but it has
proven to work for me for more than a year now.
if an invalid ioctl is done on /dev/klog. logioctl() needs to return
an errno instead of -1 on a failed ioctl.
Submitted by: Mike Pritchard <mpp@mpp.com>
This submission was done by hand-applying FreeBSD local modifications on
top of the UCB code, rather than trying to patch the UCB code in on top
of the FreeBSD code due to the extensive changes.
Reviewed by: pst (been handling 30k routes for 4+ months)
Obtained from: Sklower/Woody/Honing/Traina (8.4 UCB release)
It is the kernel driver's responsibility to do the list manipulation whenever
a selection timeout or a request sense occurs.
Print out the interrupt type that the device has been set to. It seems that
one of the Asus motherboards botches this and David thought a diagnostic would
be nice.
Fix a bug in my diagnostic code that David found.
Reviewed by: Wcarchive and David Greenman
the adapter's selections. Many fast periferals were getting upset when
the sequencer decided to rearbitrate after the device had already won
arbitration. This also forced the creation of a list threaded through
the SCBs (since we don't have enough space anywhere else) of commands that
are awaiting reselection. This list is run down before any new transactions
from the input queue are allowed. The list is appened to whenever we begin
a selection (simple case since the selecting device is always at the head)
and by the kernel driver whenever a request sense occurs. In the common
case, the list is only one element long, but when a reselection wins out
over a selection and that reselection generates a request sense, the
outstanding selection required for the retreval of the sense code grows
the list. On machines with many targets, this might cause the list to grow
large, so this solution, which will allow up to the maximum number of I/O
requests capible of the card elements in the list, was chosen. The list
manipulation is trivial and adds three sequencer instructions of overhead
to the selection phase.
This fixes the "target busy" errors from micropolis drives and the bursty
I/O problem when performing I/O between a Quantum Grand Prix and any other
device. I anticipate that this will correct many of the problems that
have been reported with this driver.
Reviewed by: Wcarchive and David Greenman
