(eg: above 0xffc00000). Programs using /dev/kmem are implicitly racing
the kernel, and can get right up high in memory. I've been running
these for some time now, but with printfs. It's saved two panics at
least that I can remember.
md_regs being struct trapframe *. Do a npxsave() if needed and copy the
pcb rather than use the increasingly defunct savectx(). Copy %edi and
%ebp explicitly.
Submitted by: bde
XXX npxproc could be declared in npx.h so the externs with smp fruit
are not needed.
Remove TF_REGP() macro and use. The original reason (address space
problems due to having UPAGES in mapped into user space) is gone. It
looks cleaner without it.
the patches in freefall:/home/dfr/ld.diffs to your ld sources and set
BINFORMAT to aoutkld when linking the kernel.
Library changes and userland utilities will appear in a later commit.
- doesn't break my system.
- NOT yet verified on the affected motherboard.
Stifle an annoying dma_start busy message for the sound cards.
Submitted by: "John S. Dyson" <toor@dyson.iquest.net>
the work is silly, and all this will be a moot point with Justin's CAM
driver, and there are still many ioctls missing -- but i need these
right now. It's silly to need to move the CD-R back to the other
drive in order to determine the session start info when burning multi-
session CD-Rs.
. It makes cd9660 root f/s working again.
. It makes CD9660 a new-style option.
. It adds support to mount an ISO9660 multi-session CD-ROM as the root
filesystem (the last session actually, but that's what is expected
behaviour).
Sigh. The CDIOREADTOCENTRYS did a copyout() of its own, and thus has
been unusable for me for this work. Too bad it didn't simply stuff
the max 100 entries into the struct ioc_read_toc_entry, but relied on
a user supplied data buffer instead. :-( I now had to reinvent the
wheel, and created a CDIOREADTOCENTRY ioctl command that can be used
in a kernel context.
While doing this, i noticed the following bogosities in existing CD-ROM
drivers:
wcd: This driver is likely to be totally bogus when someone tries
two succeeding CDIOREADTOCENTRYS (or now CDIOREADTOCENTRY)
commands with requesting MSF format, since it apparently
operates on an internal table.
scd: This driver apparently returns just a single TOC entry only for
the CDIOREADTOCENTRYS command.
I have only been able to test the CDIOREADTOCENTRY command with the
cd(4) driver. I hereby request the respective maintainers of the
other CD-ROM drivers to verify my code for their driver. When it
comes to merging this CD-ROM multisession stuff into RELENG_2_2 i will
only consider drivers where i've got a confirmation that it actually
works.
simplifies some assumptions and stops some code compile problems.
This should fix the compile hiccup in PR#3491, but smp kernel profiling
isn't likely to be fixed by this.
".." vnode. This is cheaper storagewise than keeping it in the
namecache, and it makes more sense since it's a 1:1 mapping.
2. Also handle the case of "." more intelligently rather than stuff
the namecache with pointless entries.
3. Add two lists to the vnode and hang namecache entries which go from
or to this vnode. When cleaning a vnode, delete all namecache
entries it invalidates.
4. Never reuse namecache enties, malloc new ones when we need it, free
old ones when they die. No longer a hard limit on how many we can
have.
5. Remove the upper limit on namelength of namecache entries.
6. Make a global list for negative namecache entries, limit their number
to a sysctl'able (debug.ncnegfactor) fraction of the total namecache.
Currently the default fraction is 1/16th. (Suggestions for better
default wanted!)
7. Assign v_id correctly in the face of 32bit rollover.
8. Remove the LRU list for namecache entries, not needed. Remove the
#ifdef NCH_STATISTICS stuff, it's not needed either.
9. Use the vnode freelist as a true LRU list, also for namecache accesses.
10. Reuse vnodes more aggresively but also more selectively, if we can't
reuse, malloc a new one. There is no longer a hard limit on their
number, they grow to the point where we don't reuse potentially
usable vnodes. A vnode will not get recycled if still has pages in
core or if it is the source of namecache entries (Yes, this does
indeed work :-) "." and ".." are not namecache entries any longer...)
11. Do not overload the v_id field in namecache entries with whiteout
information, use a char sized flags field instead, so we can get
rid of the vpid and v_id fields from the namecache struct. Since
we're linked to the vnodes and purged when they're cleaned, we don't
have to check the v_id any more.
12. NFS knew about the limitation on name length in the namecache, it
shouldn't and doesn't now.
Bugs:
The namecache statistics no longer includes the hits for ".."
and "." hits.
Performance impact:
Generally in the +/- 0.5% for "normal" workstations, but
I hope this will allow the system to be selftuning over a
bigger range of "special" applications. The case where
RAM is available but unused for cache because we don't have
any vnodes should be gone.
Future work:
Straighten out the namecache statistics.
"desiredvnodes" is still used to (bogusly ?) size hash
tables in the filesystems.
I have still to find a way to safely free unused vnodes
back so their number can shrink when not needed.
There is a few uses of the v_id field left in the filesystems,
scheduled for demolition at a later time.
Maybe a one slot cache for unused namecache entries should
be implemented to decrease the malloc/free frequency.
in savedvp variable and it is used for the argument of
MOUNTTOUNIONMOUNT(). I didn't realize ap->a_vp is modified before
MOUNTTOUNIONMOUNT(), so the change by revision 1.22 is incorrect.
- one-liners all become inline.
- multi-liners become functions.
- FAST_IPI defines go away.
re-worked APICIPI_BANDAID code.
- now refered to as DETECT_DEADLOCK.
- on by default.
the pageout daemon wasn't always being waken up appropriately when the
(cache + free) queues were depleted.
Submitted by: David S. Miller <davem@jenolan.rutgers.edu>
but now that we've widened the scope of the smp work to -current, it might
be an idea to warn new people that might not have read all the docs yet
that the SMP support needs to be activated via a sysctl.
Added [SR]RGBMASKs ioctl for byte swapping.
1.16 4/20/97 Randall Hopper <rhh@ct.picker.com>
Generalized RGBMASK ioctls for general pixel
format setting [SG]ACTPIXFMT, and added query API
to return driver-supported pix fmts GSUPPIXFMT.
1.17 4/21/97 hasty@rah.star-gate.com
Clipping support added.
1.18 4/23/97 Clean up after failed CAP_SINGLEs where bt
interrupt isn't delivered, and fixed fixing
CAP_SINGLEs that for ODD_ONLY fields.
Submitted by: individuals in above log messages.
Added [SR]RGBMASKs ioctl for byte swapping.
1.16 4/20/97 Randall Hopper <rhh@ct.picker.com>
Generalized RGBMASK ioctls for general pixel
format setting [SG]ACTPIXFMT, and added query API
to return driver-supported pix fmts GSUPPIXFMT.
1.17 4/21/97 hasty@rah.star-gate.com
Clipping support added.
1.18 4/23/97 Clean up after failed CAP_SINGLEs where bt
interrupt isn't delivered, and fixed fixing
CAP_SINGLEs that for ODD_ONLY fields.
Submitted by: individuals in above log messages.
opt_smp_invltlb.h
from:
SMPHDRS= opt_smp.h opt_smp_invltlb.h
SMP_INVLTLB is no longer a valid config option, the invalidation of the TLB
via inter-CPU IPIs is now standard when APIC_IO is used.
This code re-numbers PCI busses in the MP table to match PCI semantics
when the MP BIOS fails to do it properly.
Reviewed by: Peter Wemm <peter@spinner.DIALix.COM>
but searching the directory on something else than the default
location.
NB: this comprises an interface change to the mount_cd9660(8)
utility (commit will follow). You need to rebuild both.
I've got similar patches for RELENG_2_2, should i commit them too?
and FNM_LEADING_DIR were specified and the pattern ended with "*".
Example: pattern="src/usr.sbin/w*", string="src/usr.sbin/watch/watch.8,v".
This should match, but did not.
UN_KLOCK flag.
When UN_KLOCK is set, VOP_UNLOCK should keep uppervp locked and clear
UN_ULOCK flag. To do this, when UN_KLOCK is set, (1) union_unlock
clears UN_ULOCK and does not clear UN_KLOCK, (2) union_lock() does not
access uppervp and does not clear UN_KLOCK, and (3) callers of
vput/VOP_UNLOCK should clear UN_KLOCK. For example, vput becomes:
SETKLOCK(union_node);
vput(vnode);
CLEARKLOCK(union_node);
where SETKLOCK macro sets UN_KLOCK and CLEARKLOCK macro clears
UN_KLOCK.
replace invldebug with invltlb_ok for throttling smp_invltlb() during boot.
Reviewed by: informal discussion with Peter Wemm <peter@spinner.DIALix.COM>
make world fails when it tries to build LKMs because the files
opt_smp.h
opt_smp_invltbl.h
are missing.
This patch to /usr/src/share/mk/bsd.kmod.mk is a temporary workaround.
Note that LKMs built in this way may or may NOT work properly with an
SMP kernel.
Peter Wemm <peter@spinner.DIALix.COM>, Steve Passe <smp@csn.net>
removed all the IPI_INTS code.
made the XFAST_IPI32 code default, renaming Xfastipi32 to Xinvltlb.
cleanup of i386/isa/isa_device.h to eliminate SMP dependancies:
made the id_irq member of struct isa_device an u_int.
made the id_drq member of struct isa_device an int.
removed all other '#ifdefs' concerning SMP & APIC_IO.
removed SMP/APIC_IO dependancies from if_ze.c.
Peter Wemm <peter@spinner.DIALix.COM>, Steve Passe <smp@csn.net>
removed all the IPI_INTS code.
made the XFAST_IPI32 code default, renaming Xfastipi32 to Xinvltlb.
source files via a 'cc -include opt_global.h ...' type arrangement.
This means we can untangle certain header files.
options.i386 has a placeholder until it has a real member so we can avoid
having to teach config about it just yet.
Reviewed by: bde
This commit includes the following changes:
1) Old-style (pr_usrreq()) protocols are no longer supported, the compatibility
glue for them is deleted, and the kernel will panic on boot if any are compiled
in.
2) Certain protocol entry points are modified to take a process structure,
so they they can easily tell whether or not it is possible to sleep, and
also to access credentials.
3) SS_PRIV is no more, and with it goes the SO_PRIVSTATE setsockopt()
call. Protocols should use the process pointer they are now passed.
4) The PF_LOCAL and PF_ROUTE families have been updated to use the new
style, as has the `raw' skeleton family.
5) PF_LOCAL sockets now obey the process's umask when creating a socket
in the filesystem.
As a result, LINT is now broken. I'm hoping that some enterprising hacker
with a bit more time will either make the broken bits work (should be
easy for netipx) or dike them out.
1) i586_bcopy() problem
There have been a number of reports that the syscons doesn't work
properly if i586_bcopy() is enabled.
The problem prevented users from installing 2.2(.1)-RELEASE. The
symptom is that the system looks frozen during device probe or just
before the main installation menu. The workaround was to specify the
flag 0x01 to the npx device so that i586_bcopy() is disabled.
The patch forces the syscons to call generic_bcopy() when copying
to/from the video memory, even if CPU is Pentium and i586_bcopy() is
enabled. i586_bcopy() is still called for copy operations between
non-video memory regions.
PR: kern/2277, kern/3066, kern/3107, kern/3134
2) video mode parameter table problem
The syscons reads and uses the video mode parameter table provided by
the VGA BIOS to set VGA registers when changing video mode and
modifying font data. It appears that in some VGA BIOSes the table is
not ordered as the syscons expects, and this leads to screen
corruption.
The problem prevented users from installing 2.2(.1)-RELEASE. The
symptom is the corrupt screen or strange vertical lines soon after the
kernel is loaded into memory (just after the kernel decompression).
The patch performs simplistic test and if it fails, set video_mode_ptr
to NULL so that the video mode switching won't happen.
This is an interim kludge. There should be a better way to deal with
the problem.
PR: kern/2498, conf/2775, conf/3354
Reviewed by: sos
Tested by: PR originators (not all of them, though)
The SMP source was merged into 3.0-current last nite and this broke
the make of sbin/dset.
Should make things work for non-SMP case.
People making SMP kernels will need to edit i386/isa/isa_device.h, re-enabling
'#include "opt_smp.h"'. People making SMP worlds will have to punt for now...
I'm thinking about the real solution, but for now the goal is to NOT break
the world!
There are various options documented in i386/conf/LINT, there is more to
come over the next few days.
The kernel should run pretty much "as before" without the options to
activate SMP mode.
There are a handful of known "loose ends" that need to be fixed, but
have been put off since the SMP kernel is in a moderately good condition
at the moment.
This commit is the result of the tinkering and testing over the last 14
months by many people. A special thanks to Steve Passe for implementing
the APIC code!
When we request sense, don't allow disconnection. This closes a window
where we might allow an overlapped tagged and non-tagged transaction.
The correct fix is to freeze the queue for the target that requests sense
which is what will happen in the new CAM framework.
Our vput calls vm_object_deallocate() --> vm_object_terminate(). The
vm_object_terminate() calls vn_lock(), since UN_LOCKED has been
already cleared in union_unlock(). Then, union_lock locks upper vnode
when UN_ULOCK is not set. The upper vnode is not unlocked when
UN_KLOCK is set in union_unlock(), thus, union_lock tries to lock
locked vnode and we get panic.
if SCB Paging was enabled:
disconnect with more data to transfer
disconnected SCB gets paged out
target reconnects so we page SCB back in
target completes transfer so residual is 0
target disconnects
SCB gets reused but not paged out since the residual is 0 (optimization)
target reconnects so we page the SCB back in
we report a residual because of stale residual information.
The fix for this is to set a flag that forces the SCB to be paged back
up to the host if we page in an SCB with a residual
Pointed out by: Doug Ledford <dledford@dialnet.net>
Fix another bug: if argv[0] is NULL, garbadge args might be added for
shell script
Submitted by: Tor Egge <Tor.Egge@idi.ntnu.no> (with yet one fault detect from me)
for syscalls, so one frame was lost in backtraces from syscalls.
This is handled better in the kernel by using a different mcount
entry point for profiling before the frame pointer is set up.
Expand RCSID().
Use .p2align instead of the ambiguous .align.
Added idempotency ifdef.
Removed unused macros ALTENTRY(), ALTASENTRY(), ASENTRY(), _MID_ENTRY.
Cleaned up formatting.
Reviewed by: jdp reviewed an old version
Obtained from: parts from NetBSD
large enough to contain the ethernet header. There appears to be a
condition where the card can return "0" in some failure cases, and this
causes bad things to happen (a panic).
have successfully built, booted, and run a number of different ELF
kernel configurations, including GENERIC. LINT also builds and
links cleanly, though I have not tried to boot it.
The impact on developers is virtually nil, except for two things.
All linker sets that might possibly be present in the kernel must be
listed in "sys/i386/i386/setdefs.h". And all C symbols that are
also referenced from assembly language code must be listed in
"sys/i386/include/asnames.h". It so happens that failure to do
these things will have no impact on the a.out kernel. But it will
break the build of the ELF kernel.
The ELF bootloader works, but it is not ready to commit quite yet.
UN_ULOCK flag. This shows a locking violation but I couldn't find the
reason UN_ULOCK is not set or upper vnode is not unlocked. I added
the code that detect this case and adjust un_flags. DIAGNOSTIC kernel
doesn't adjust un_flags, but just panic here to help debug by kernel
hackers.
delay after we reset the card to allow the card to come back to life.
It appears the newer card takes longer to reset.
Submitted by: Samuel Lam <skl@ScalableNetwork.com>
type mismatches. There was no problem in practice (at least on 386's).
The timeout args still get bogusly cast from int to `void *' via
caddr_t and back to int.
type mismatches. mcd and scd were/are particularly bogus. They
used a general purpose function taking 2 args for the timeout
function and fudged varargs stuff to supply the second arg for the
timeout case. This broke `cc -mrtd'. Bounce through a timeout
function instead. The timeout arg still gets bogusly cast from
int to `void *' and back.
difference of approx 3mins in make world on my P6!!! This means
that vfork now has full address space sharing, so beware with
sloppy vfork programming. Also, you really do need to apply
the previously committed popen fix in libc.
type mismatches. There was no problem in practice (at least on 386's).
Don't cast NULL in calls to timeout functions. pcvt is fully prototyped
and doesn't support K&R.
Timeout functions take args of type `void *', so use magic numbers of
type `void *' for UPDATE_* to reduce the danger of wrong conversions.
Removed FreeBSD-pre-1.1-related TIMEOUT_FUNC_T macro. It was especially
bogus for the pre-1.1 case.
type mismatches. There was no problem in practice (at least on 386's).
Removed NetBSD-related TIMEOUT macro. NetBSD uses the same BSD4.4Lite
timeout interface as FreeBSD. As a concession to portability, declare
the timeout function without using the FreeBSD timeout_t typedef.
properly. Simply don't use the gcc macros if we're not gcc, and
declare prototypes for the byte-swapping functions in case the
macro versions are not used. The previous fix was wrong and broke
libpcap, which abuses -Dlint.
Don't pollute the namespace if _POSIX_SOURCE is defined. This is
broken - it makes almost everything in <machine/endian.h> inaccessible
if _POSIX_SOURCE is defined, yet <machine/endian.h> is not a POSIX
header. Other systems don't do it any better.
Removed always-false `BYTE_ORDER == BIG_ENDIAN' ifdef.
Obtained from: partly from Lite(2?) and NetBSD
(pt. unused) for TI PCI1130.
2. pccard.c: PCIC_RESUME_RESET is now (also) a sysctl. (Never make it
a #ifdef if it can be made a sysctl!)
3. pcic.c: make getb() and putb() proper member functions of struct
pcic_slot. Add a couple of missing casts.
wind up writing zeros instead of real data when the file is on an NFSv2
mounted directory.
While tracking this bug down, I noticed that nfs_asyncio was waking *all*
the iods when a block was written instead of just one per block. Fixing this
gives a 25% performance improvment for writes on v2 (less for v3).
Both are 2.2 candidates.
PR: kern/2774
# mount -t union (or null) dir1 dir2
# mount -t union (or null) dir2 dir1
The function namei in union_mount calls union_root. The upper vnode
has been already locked and vn_lock in union_root causes above panic.
Add printf's included in `#ifdef DIAGNOSTIC' for EDEADLK cases.
Clean up the unexpected busfree handler. We now look directly at the
message that might have caused the bus free to occur instead of looking
at an SCB flag. This makes the handling more robust and also allows for
recovery actions that might cause an "unexpected busfree" to be performed
even if an SCB is not availible to "tag". Most notably, this happens
when we don't find an SCB for a reconnecting target.
to fix a selection timeout problem.
If we can't find an SCB for the reconnecting target, issue a bus device
reset as the SCSI2 spec suggests.
Add a missing call to "add_scb_to_free_list" in the non paging case. In
the non-paging case, the SCBs don't really need to be on the free list,
but putting them there clears the tag field which is something the recovery
code depends on.
Zero the b_dirty{off,end} after cluster-comitting a group of buffers.
With these fixes, I was able to complete a 'make world' with remote src
and obj directories.
1) Dell Latitude XPi
This laptop has a strange, IMHO broken :-), keyboard controller which
wouldn't disable the keyboard interrupt. The kludge is to disable tty
intr. during set_keyboard(), used for changing LED and setting
typematic.
The patch also changes the function name:
set_keyboard() -> set_keyboard_param()
Although it is a static function, the name corrides with a routine in
`syscons' and is confusing when debugging the kernel which has both
`syscons' and `pcvt' with DDB. (Suggested by Bruce)
2) doreset() bug
doreset() failed to preserve some bits in the keyboard controller's
command byte during keyboard reset. This bug may put some keyboard
controllers in old motherboards (386 and 486) in a strange state,
resulting in complete keyboard lockup or random key input.
Reviewed by: Joerg
This patch fixes the problem of vic only capturing an even or odd frame plus
the my early patch for missing frames with resolutions higher than 320x240
in rgb mode.
The yuv422 patch introduces a minor bug in that a green line appears at the
bottom of the captured window . There is no easy work around for this right
now.
Reviewed by: various bt848 hackers
Submitted by: Amancio Hasty <hasty@rah.star-gate.com>
Increase the delay in read_eeprom_data() by two orders of magnitude.
> A better fix would be to make read_eeprom_data() call
> f_is_eeprom_busy() after the DELAY().
Submitted by: Samuel Lam <skl@ScalableNetwork.com>
