alfred [Tue, 5 Feb 2002 21:23:05 +0000 (21:23 +0000)]
Fix a race with free'ing vmspaces at process exit when vmspaces are
shared.
Also introduce vm_endcopy instead of using pointer tricks when
initializing new vmspaces.
The race occured because of how the reference was utilized:
test vmspace reference,
possibly block,
decrement reference
When sharing a vmspace between multiple processes it was possible
for two processes exiting at the same time to test the reference
count, possibly block and neither one free because they wouldn't
see the other's update.
imp [Tue, 5 Feb 2002 21:06:56 +0000 (21:06 +0000)]
o Move externs to extern.h
o Use new-style prototypes exclusively rather than the old foo() style.
o Use new-style function definitions.
o remove register
o make functions passed to signal have the right signature.
o do minor const poisoning.
phk [Tue, 5 Feb 2002 20:44:56 +0000 (20:44 +0000)]
Let the number of timecounters follow hz, otherwise people with
HZ=BIGNUM will strain the assumptions behind timecounters to the
point where they break.
This may or may not help people seeing microuptime() backwards messages.
Make the global timecounter variable volatile, it makes no difference in
the code GCC generates, but it makes represents the intent correctly.
alfred [Tue, 5 Feb 2002 19:30:30 +0000 (19:30 +0000)]
Fix the credential handling code.
In NetBSD, Solaris, xprt->xp_p2 pointed directly to the credentials,
in FreeBSD xprt->xp_verf.oa_base was a pointer to a struct cmessage,
which is defined as follow:
The credentials were submitted the right way and xprt->xp_p2 pointed to them.
But cb_verf.oa_flavor was still empty. There was an assignment missing
in svc_recv() in svc_vc.c:
alfred [Tue, 5 Feb 2002 19:26:04 +0000 (19:26 +0000)]
The clnt_create(KEYSERVSOCK, CRYPT_PROG, CRYPT_VERS, "unix") hacks
were removed and replaced them with clnt_tp_create, now the af_local
support is fixed.
I also removed the hack how rpcinfo contacted rpcbind, now we can
relay on clnt_tp_create create the client-handle for us. Only
rpcbind itself needs a hardcoded socket-path.
alfred [Tue, 5 Feb 2002 18:51:13 +0000 (18:51 +0000)]
Fix the breakage in rpc.yppasswd. Readded the svc_create() and
the registering of the "unix" transport, now it is fixed.
Everywhere, rq_cred is taken to look what authentification we have.
We can not be sure that transp>xp_verf.oa_flavor is also filled in.
This seems to be the same for all sun source. they take the flavor
of rq_cred, instead of transp.
archie [Tue, 5 Feb 2002 18:27:30 +0000 (18:27 +0000)]
The hardware checksum code makes the assumption that a packet routed out
a particular Ethernet interface will actually be delivered by (only) that
device driver. This is not necessarily true when ng_ether(4) is used.
To word around this, while a ng_ether(4)'s "upper" hook is connected,
turn off all hardware checksum, fragmentation, etc., features for that
interface.
sheldonh [Tue, 5 Feb 2002 09:17:24 +0000 (09:17 +0000)]
Fix obscure bug in code loading support that could cause parsing of
functions in multi-line files to fail, depending on the state of the
stack. The fix is sanctioned by the vendor and will appear in bc-1.07.
PR: bin/34601
Reported by: mi
Obtained from: Phil Nelson <phil@cs.wwu.edu>
bbraun [Tue, 5 Feb 2002 06:21:34 +0000 (06:21 +0000)]
Makes __istype() an inline (and a non inlined) function to avoid C++
compile issues. std::isspace(' ') was expanding to std::(!!_maskrune...)
which would cause a C++ compile error. Making __istype() an inline
causes the expansion to be std::__istype() instead, which is valid.
mjacob [Mon, 4 Feb 2002 21:04:25 +0000 (21:04 +0000)]
+ A variety of 23XX changes:
disable MWI on 2300
based on function code, set an 'isp_port' for the 2312- it's a
separate instance, but the NVRAM is shared, and the second port's
NVRAM is at offset 256.
+ Enable RIO operation for LVD SCSI cards. This makes a *big* difference
as even under reasonable load we get batched completions of about 30
commands at a time on, say, an ISP1080.
+ Do 'continuation' mailbox commands- this allows us to specify a work
area within the softc and 'continue' repeated mailbox commands. This is
more or less on an ad hoc basis and is currently only used for firmware
loading (which f/w now loads substantially faster becuase the calling
thread is only woken when all the f/w words are loaded- not for each
one of the 40000 f/w words that gets loaded).
+ If we're about to return from isp_intr with a 'bogus interrupt' indication,
and we're not a 23XX card, check to see whether the semaphore register is
currently *2* (not *1* as it should be) and whether there's an async completion
sitting in outgoing mailbox0. This seems to capture cases of lost fast posting
and RIO interrupts that the 12160 && 1080 have been known to pump out under
extreme load (extreme, as in > 250 active commands).
+ FC_SCRATCH_ACQUIRE/FC_SCRATCH_RELEASE macros.
+ Endian correct swizzle/unswizzle of an ATIO2 that has a WWPN in it.
des [Mon, 4 Feb 2002 20:16:50 +0000 (20:16 +0000)]
Move the pseudofs, procfs and linprocfs modules out from the fs directory.
Keeping them there seemed like a good idea at the time, but it annoys bde
and confuses people who do not understand how MODULES_OVERRIDE works.
sos [Mon, 4 Feb 2002 19:23:40 +0000 (19:23 +0000)]
Major update of the ATA RAID code, part 1:
Overhaul of the attach/detach code and structures, there were some nasty
bugs in the old implementation. This made it possible to collapse the
ATA/ATAPI device control structures into one generic structure.
A note here, the kernel is NOT ready for detach of active devices,
it fails all over in random places, but for inactive devices it works.
However for ATA RAID this works, since the RAID abstration layer
insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the
physical disks.
Proberly detect the RAID's from the BIOS, and mark critical RAID1
arrays as such, but continue if there is enough of the mirror left
to do so.
Properly fail arrays on a live system. For RAID0 that means return EIO,
and for RAID1 it means continue on the still working part of the mirror
if possible, else return EIO.
If the state changes, log this to the console.
Allow for Promise & Highpoint controllers/arrays to coexist on the
same machine. It is not possible to distribute arrays over different
makes of controllers though.
If Promise SuperSwap enclosures are used, signal disk state on the
status LED on the front.
Misc fixes that I had lying around for various minor bugs.
rwatson [Mon, 4 Feb 2002 16:41:59 +0000 (16:41 +0000)]
Note that Kirk apparently missed adding vn_start_write() and friends
to kern_acl.c when he added snapshotting. This will need to be added
at some point.
markm [Mon, 4 Feb 2002 00:28:54 +0000 (00:28 +0000)]
Add the other half of the salt-generating code. No functional
difference except that the salt is slightly harder to build
dictionaries against, and the code does not use srandom[dev]().
markm [Sun, 3 Feb 2002 22:42:46 +0000 (22:42 +0000)]
Forced commit.
Previous commit included a large improvement to the salt generation to
help obfuscate information that is otherwise leaked. No real risks
involved, but this feels better, and does not use srandom[dev]().
markm [Sun, 3 Feb 2002 11:36:59 +0000 (11:36 +0000)]
Zero functional difference; make some integer constants unsigned, as
they are used in unsigned context. This shuts lint(1) up in a few
significant ways with "signed/unsigned" arithmetic warnings.
knu [Sun, 3 Feb 2002 10:01:49 +0000 (10:01 +0000)]
Work around a buffer overflow problem on argv that has been exposed
after making test(1) a sh(1) builtin; sh(1) coredumps when you run
something like this:
sh -c 'test ! `true 1`'
The cause is that the test(1) code totally depends on the presence of
two extra cells at the end of argv that are filled with NULL's. The
reason why the bug hasn't been exposed would be because the C startup
code kindly prepares argv with some extra zeroed cells for a program.
I know this is not the best fix, but since there are argv++'s without
boundary checks everywhere, I'd rather patch it up like this
(preparing a copy of argv with extra NULL's) for the moment.
bde [Sun, 3 Feb 2002 09:13:58 +0000 (09:13 +0000)]
Use osigreturn(2) instead of sigreturn(2) plus broken magic for returning
from old signal handlers. This is simpler and faster, and fixes (new)
sigreturn(2) when %eip in the new signal context happens to match the
magic value (0x1d516). 0x1d516 is below the default ELF text section,
so this probably never broken anything in practice.
locore.s:
In addition, don't build the signal trampoline for old signal handlers
when it is not used.
alpha:
Not fixed, but seems to be even less broken in practice due to more
advanced magic. A false match occurs for register #32 in mc_regs[].
Since there is no hardware register #32, a false match is only possible
for direct calls to sigreturn(2) that happen to have the magic number
in the spare mc_regs[32] field.
matusita [Sun, 3 Feb 2002 07:00:37 +0000 (07:00 +0000)]
Change variable name from DOCDISTFILES to RELEASEDISTFILES. Respect
DOCDISTFILES if defined. No impact should happen with this change.
DOCDISTFILES is introduced by rev. 1.541, to point the directory that
contains the distfiles for ports/textproc/docproj. It's good, but
we also build ports/sysutils/mkisofs during the release (if MAKE_ISOS
is defined). It is natural that cdrtools tarball is prefetched, then
copy it under ${CHROOTDIR}/usr/ports/distfiles.
As a result, DOCDISTFILES variable is somewhat abused for non-doc
tarballs; *DOC*DISTFILES is no longger appropriate name for that.