>hsu 2003/01/16 23:59:35 PST
>
> Modified files:
>sys/netinet if_ether.c
> Log:
> SMP locking for ARP.
I just got:
lock order reversal
1st 0xc0436078 arp mutex (arp mutex) @ /usr/src/sys/netinet/if_ether.c:151
2nd 0xf800a0668ef0 radix node head (radix node head) @
>Was /etc/protocols maybe simply forgotten in the 10/29/02 change?
Yes. Does changing it to 258 work?
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
>I meant sysinstall generating cons25 output. But there were recently a
>lot of terminfo changes that may have caused this.
Oh. sysinstall asked if I wanted ANSI, vt100, cons25, something else
related to FreeBSD console, or xterm. Most of those options were wrong,
which is the bug that I think
screen(1) says "Each virtual terminal provides the functions of a DEC
VT100 terminal and, in addition, several control functions from the ISO
6492 ... and ISO 2022 standards ...".
I took that to mean that it provides a vt100 interface. That's also
been my experience in the last 8 years of using
I installed 5.0-RC1 on a Sun Ultra-60 a couple of days ago.
The biggest problem that I ran into was terminal emulation
inside sysinstall.
I normally do most tasks inside screen, so I tried selecting
the vt100 emulation (since that's what screen tries to be).
However, it uses something that screen
One of the big problems is that install gives a bogus error
message when it can't unlink /usr/bin/make because it's non-root.
Since there's no way that I'm going to suggest changing
install's behavior this late in the release cycle, can we
at least make buildworld's "make" target ensure that you'r
>The sparc64 tinderbox is running a stale world (about 3 months old),
>so it's hitting the same problem.
I'm running a not-so-stale world (19 days old) and hitting the same
problem.
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the messa
>What on earth does this mean?
>
>>> mrtg-2.9.26b.tar.gz doesn't seem to exist in /tmp/distfiles/.
>>> Attempting to fetch from
>http://people.ee.ethz.ch/~oetiker/webtools/mrtg/pub/.
>fetch: mrtg-2.9.26b.tar.gz: Multiple Choices
It means the web server is behaving oddly. Here's a patch that give
I really don't understand why you keep claiming that the SYN cache
changes anything. Without the SYN cache, tcp_input() calls
sonewconn(so, 0) on receipt of a SYN; with the SYN cache, tcp_input()
calls some syncache function which calls sonewconn(so, SS_ISCONNECTED)
on receipt of a SYN/ACK. In e
Michal,
Alan Cox pointed out to me that backing out to using sodealloc()
instead of sotryfree() is probably a better fix anyway - it solves
the panic in more or less the same way as mine, but it backs the code
out to be the same as it's been for years -- a much more well-tested
fix =) He commi
Terry,
I think most of your 9k of reasoning is based on the thought that
soreserve() allocates memory. It doesn't, and never has.
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
>I think this can still crash (just like my patch); the problem is in
>what happens when it fails to allocate memory. Unless you set one of
>the flags, it's still going to panic in the same place, I think, when
>you run out of memory.
No. The flags are only checked when so_head is not NULL. so
Oops, you're right, I was looking too closely =)
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
BTW, /etc/rc.network never tried to save you from
rpcbind_enable=NO
nis_client_enable=YES
so it may be a mistake for /etc/rc.d/* to try to.
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
This is fixed in my WIP on rc.d . I'm more or less ready for wider
review; I especially need review of the atm and diskless changes.
Bill
http://people.freebsd.org/~fenner/rc.d.diff
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
sonewconn() hands sofree() a self-inconsistent socket -- so->so_head is
set, so so must be on a queue, but sonewconn() hasn't put it on a queue yet.
Please try this patch.
Bill
Index: uipc_socket2.c
===
RCS file: /home/ncvs/src/sys
/etc/rc runs /etc/rc.sysctl twice:
one "early", after mounting filesystems, reseeding the random number
generator and adding a swap file, and before running rc.serial, rc.pccard,
rc.network.
one "late", after network_pass4 but before raising the securelevel.
This was added in response to
http:/
I was working on (wlen == 0 && iov->iov_cnt != 0) for a while, thinking
that it would work in both cases, even though the logic is a little weird
in the writev case, but it would fail in the race where the connection
closed at the same time as the writev() with the zero length iov_len.
Bill
To
Turns out my writev patch for fetch broke SSL, since it could create
iov[0].iov_len = 0, which would cause SSL_write(..,0), which would
return 0, which would look like a short write and cause an error, which
then gets ignored by http.c . Ignoring the bigger picture of the error
checking, this fix
I think a lot of people would be happier if we could maintain backwards
compatability (and document the fact that they're extremely obsolete)
for a few more releases. Despite the fact that the main UNIX reference
that I use was published in 1984, I don't actually want everything to
stay the same
>It's not like people didn't have nine years' advance warning to fix
>their scripts.
When's the first time the FreeBSD sort(1) man page mentioned that this
syntax was deprecated? Can we at least start from there?
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-
>Until "sh", "make", "tar", and so on also drop behaviours that are
>not specified by POSIX, it's really silly to make "sort" drop them.
It's not that the +x/-y argument syntax is not specified - it's that
it's specifically disallowed. (I disagree with that restriction, but
let's at least have
>Please, no. They do the right thing.
I guess there are varying definitions of what the right thing is.
I don't think it's widely known that the +/- syntax was obsoleted.
I am vaguely a standards weenie and I didn't know.
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe
Here's my suggested fix:
stash% pwd
/usr/src/gnu/usr.bin/sort
stash% cvs diff -uN
cvs diff: Diffing .
Index: posixver.c
===
RCS file: posixver.c
diff -N posixver.c
--- /dev/null 1 Jan 1970 00:00:00 -
+++ posixver.c 24 Sep 200
I've seen the scrn_timer() one too:
Expensive timeout(9) function: 0xc049b3f0(0xc0657200) 0.003659743
c049b3f0 t scrn_timer
as well as uma_timeout().
Expensive timeout(9) function: 0xc047c6b0(0) 0.002146136
c047c6b0 t uma_timeout
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "u
And the answer is... it only uses the includes if you commit them
to the FreeBSD repository instead of your local repository.
Pointy hat to: fenner
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
Ok, so I broke world. What I don't get is how -- why doesn't the
build use the up to date includes?
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
I know that the specific mergemaster issues have been addressed, but I
thought this experience pointed out something subtly astonishing, so I
figured I'd point it out.
I ran mergemaster, and the perl wrapper started complaining that I
needed to install perl, so I did "pkg_add -r perl". The port
>Specifically what is the problem? Given the program below, take the
>ISO-C spec and explain the problem. Or even w/o the spec -- I haven't
>been reading this thread.
> > int
> > main()
> > {
> >unsigned char i = 127;
> >char j;
> >
> >printf("%d\n", ((char)(i << 1)));
This print
> So - yes - it seems gcc 3.1 does have a problem...
Indeed - easily determined by breaking down the expression.
So, who's gonna report it to gcc-bugs? knu?...
int
main()
{
unsigned char i = 127;
char j;
printf("%d\n", ((char)(i << 1)));
j = ((char)(i << 1)) / 2;
printf("%d\n"
Duh. Sometimes I wish I had the patience to wait for my tests to complete
before sharing my guesses. I jumped to a wildly incorrect conclusion; gcc
3.1 still defaults to signed chars. Sorry for the bizarre misdirection.
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe
gcc 3.1 simply defaults to unsigned chars. 127 << 1 = 254; 254 / 2 = 127.
My machine is too slow to test this expeditiously, but I'm trying
adding "#define DEFAULT_SIGNED_CHAR 1" into freebsd-native.h .
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" i
This is presumably fallout from the /usr/share/mk rearrangement, but
rev 1.306 of /usr/share/mk/bsd.port.mk should fix this.
Oh, hey, are the failing ports all ones that use bsd.port.pre.mk and
bsd.port.post.mk? I guess bsd.port.pre.mk needs the same fix as
bsd.port.mk does.
Bill
To Unsubsc
I had the same symptoms (drifting about 2 minutes an hour) on sources
before April 17 or so. Since then, ntpd has only logged 5 time updates,
as opposed to 3 per hour.
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
>Might, might not. ISTR strftime() can't correctly emulate ctime(),
>but some other format might be preferrable. Do you have a format
>string handy?
I'd think something like what "last" does would be good.
d_first = (*nl_langinfo(D_MD_ORDER) == 'd');
...
(void) strftime(ct,
While you're in here, does it make sense to use strftime() instead of
printing just a portion of what ctime returns? This would allow
i18n of the time format if desired (and if the locale is set this
early -- maybe this is just a can of worms =).
Bill
To Unsubscribe: send mail to [EMAIL PROT
Yes, I'm in favor of going back to the simple sequence number too.
I don't understand the advantage of the MD5.
While you're in there, could you put back minfree checking too?
That bit me pretty badly today, with savecore filling up my /var
because it doesn't care about minfree.
Bill
To Unsub
>Masquerading shouldn't need to be in submit.mc either -- the MTA should do
>any masquerading. The only catch here is for people who may be using
>limited masquerading and not masquerading the fully qualified hostname of
>submitted mail.
Some broken tests made me think that the MTA wouldn't mas
BTW, not that this is particularly related to startup scripts, but -- I
found it quite frustrating that due to a transient failure at submission
time, my mail went into /var/spool/clientmqueue, for which you have to
use "mailq -Ac", which is pretty much completely undocumented. It
turned out OK
> Although I am still getting the following lock problems when I shut
> the system down:
>
> lock order reversal
> 1st 0xc036afc0 allproc @ ../../../kern/vfs_syscalls.c:452
> 2nd 0xc7ecce34 filedesc structure @ ../../../kern/vfs_syscalls.c:457
I've been seeing this since Feb 4.
Bill
To Uns
>I don't know exactly what causes the b_to_q message. It is most likely
>a race in close. You can first-open tty's that are blocked in last-close,
>and having this open succeed is very important for unblocking the close
>usi9ng "comcontrol /dev/foo drainwait ", but the tty system doesn't
>seem
FreeBSD stash.attlabs.att.com 5.0-CURRENT FreeBSD 5.0-CURRENT #1: Fri Mar 8 18:16:53
PST 2002 [EMAIL PROTECTED]:/usr/obj/usr/src/sys/STASHNOV6 i386
responds to a break from a Cisco terminal server, invoked by
/usr/ports/comms/conserver:
FreeBSD/i386 (stash.attlabs.att.com) (ttyd0)
logi
>No. Leave it in, this will benifit us all in the long run.
Until we start hitting the broken/buggy warnings, which will cause
people to write more obfuscated or harder to maintain code in order
to avoid the warnings.
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe fre
Here's a patch for bind's port/freebsd/include/port_before.h .
--- port_before.h.orig Tue Feb 26 20:57:35 2002
+++ port_before.h Tue Feb 26 21:02:18 2002
@@ -15,6 +15,7 @@
#define SETPWENT_VOID
#endif
+#include
#include
#define GROUP_R_RETURN struct group *
@@ -26,8 +27,13 @@
#d
You could use http://people.freebsd.org/~fenner/__FreeBSD_version.html
to find an already-existing value of __FreeBSD_version to test based
on the date of the original change.
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
Can we move this multi-line stuff above the "Doing initial network
setup:" echo? It makes the echo -n stuff printed afterwards
look dumb -- right now, we see:
Doing initial network setup:
Generating /etc/host.conf for compatibility
hostname domain.
The "hostname domain" are somewhat orphaned.
awk does not copy input lines to its output unless asked; you can
ask with either an explicit "print" or an empty action.
Using an input file like:
gibberish
stuff
this doesn't match
here is some garbola
I don't want this file in the ouptut
here's some more stuff
and another line
and another
an
>2. cvsup to r1.96 of tty_cons.c, which should fix this, but due to lack
> of testers and the inability to reproduce it here, is unverified.
I've been testing it, and haven't had any panics, but since the panic
was irregular anyway it's hard to say that it's fixed.
Bill
To Unsubscribe: sen
I think this is a bug in the loadable VLAN code; there's a != which
should be an ==, which results in packets with known tags being
discarded and would result in packets with unknown tags causing a
null pointer dereference. I'll commit a fix soon as soon as I've
tested it more.
Bill
To Unsub
I also started getting this error with recent kernels (in the last
day or so).
Mounting root from ufs:/dev/ad0s1a
panic: lock (sleep mutex) vnode interlock not locked @
/usr/src/sys/kern/vfs_default.c:460
Debugger("panic")
Stopped at Debugger+0x44: pushl %ebx
db> t
Debugger(c03c5bbb) at
OOPS. That's what I get for just looking individual bits of the tree.
Of course, ru not only removed the code from the
compat/compat4x.i386/Makefile but also added some to compat/Makefile
to work [better] for everything.
Sorry for any confusion I introduced.
Bill
To Unsubscribe: send mail t
>Read UPDATING.
Having an old libc.so.4 in /usr/lib is not in UPDATING. (Perhaps it
should be...)
>You have an old libc.so.4. You can have it updated
>automatically by adding "COMPAT4X=yes" into /etc/make.conf. For now:
>
># echo "COMPAT4X=yes" >> /etc/make.conf
># cd /usr/src/lib/compat
>#
The testing I've done shows that postfix is buggy in two ways:
- The main() in inet_addr_local.c assumes that the addresses in
addr_list and mask_list are sockaddrs, but this is only true
when using IPv6. This only affects testing with -DTEST.
- inet_addr_local() calls inet_addr_list_append(..
>...since a lock order reversal means that you can get in a deadlock...
Argh, of course. It's only not problematic if it's a uniprocessor
and it doesn't take an interrupt at the wrong time. Sorry for being
dense, I'm still used to spl() =)
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED
This lock order reversal is not a problem. See
http://www.geocrawler.com/lists/3/FreeBSD/147/50/6267627/
for the meta-issue of witness being too verbose; I'd post URL's for
the followup discussion but there wasn't any.
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe fre
I'm curious what the long-term plan is for witness(4). For
example, it complains about BPF and device locks being reversed
when BPF takes the device out of promiscuous mode --
lock order reversal
1st 0xc04c1560 bpf global lock @ /usr/src/sys/net/bpf.c:365
2nd 0xc1302b88 dc1 @ /usr/src/sys/pci
You must have updated in the middle of the commit. Make sure you have
rev 1.38 of src/sys/net/route.h .
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
This should be fixed by rev 1.1.1.5 of src/contrib/libpcap/inet.c .
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
I guess this is
http://www.tcpdump.org/cgi-bin/cvsweb/libpcap/inet.c?r1=1.25&r2=1.26
The easiest thing to do is probably cvs import their rev 1.26 of
inet.c . Shall I do this?
Bill
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
>By "gratuituous arp" I was really saying "gratuitous arp reply".
>The machine needs to send a packet of the type
>
> arp reply 1.2.3.5 is-at 0:40:5:42:d6:de
The ARP processing specified in RFC 826 says that if you have an entry for
the source IP address you update the hardware address no matt
FreeBSD 4.0-RELEASE does the gratuitous ARP when ifconfig'ing an alias:
fenestro# ifconfig de1 1.2.3.5 alias
18:35:47.269471 0:40:5:42:d6:de Broadcast arp 42: arp who-has 1.2.3.5 tell 1.2.3.5
FreeBSD 3.4-STABLE also does:
mango# ifconfig xl0 135.197.2.250 netmask 255.255.255.255 alias
18:39:12
>That said, isn't there some way we could build it twice, once for the
>crypto dist and once for the bindist? That would mean that the crypto
>distribution copy simply blops over the bin distribution version if
>selected and POLA is fully obeyed.
Sounds perfectly reasonable to me. I'd be happy
Just out of curiosity, why is there an "AGAIN" in the subject line,
since this is the first email I've gotten on the subject?
tcpdump is capable of decrypting ESP, if you give it the key and if it's
linked with libcrypto. Since IPSEC is part of FreeBSD, and libcrypto
is part of FreeBSD, I figur
Bruce is right that machines expect to learn their prefixes from their
local router; however if you're just playing around you might want to
set it yourself. The easiest way I've found to do this is to say that
this machine is a router:
# sysctl -w net.inet6.ip6.forwarding=1
net.inet6.ip6.forwa
>Second, how does one specify options on the command line? In ssh
>1.2.x, I say ssh -o ForwardX11=yes, but that doesn't work in OpenSSH.
>Bug or feature?
Browsing the source, it looks like "ssh -o 'ForwardX11 yes'" should
work. Both ssh and openssh define -o as:
-o 'option'
>Right. I've seen this when I hit Enter rapidly twice at the first
>loader prompt. Doesn't ever happen if I wait for the second
>prompt.
That's my impression too -- I've seen it on my laptop when I do
that (sometimes), and I may have hit enter twice rapidly on this
reboot.
Bill
To Unsubsc
>In this case, I actually assume that the DHCP server will be providing
>the host name
In my case (which I think might actually be fairly common), the DHCP
server doesn't provide a hostname. In this case, I type in a hostname
which gets later ignored and sysinstall doesn't write anything to rc.
>Also, usual tools, rlogin, rlogind, rsh, rshd, telnet,
>telnetd, ftp, ftpd, and inetd are already IPv6 capable.
Hm. rlogin and rsh attempt to connect, but my inetd isn't listening;
do I have to update inetd.conf to get inetd to listen on IPv6 addresses?
telnet can't parse ::1:
emachine% teln
>Test IPv6 support! You've all been asking for it, test it.
Is there a quick primer on getting IPv6 up and running? I built a
kernel with INET6 and the ipsec stuff, and my interfaces now have IPv6
addresses, but no userland apps seem to be able to parse IPv6 addresses,
e.g. "ping ::1" says "no
Hi,
I installed 4.0 2125-CURRENT last night on a new box, and had
several problems that I wanted to share:
1. sysinstall forgot to write my hostname to /etc/rc.conf . I had gone
into the options menu and selected "DHCP"; when I picked my network
interface it looked for and found a DHCP s
I often want to know where in a "make world" I am, so I wrote this
quick little script. It breaks out the major sections, like
--
>>> Cleaning up the elf obj tree
--
and the
> - complain if a device is specified twice (eg: 2 x psm0)
Does this work for pseudo-devices also (i.e. can bin/9931 get closed)?
Bill
To Unsubscribe: send mail to majord...@freebsd.org
with "unsubscribe freebsd-current" in the body of the message
Isn't it easier to reclassify the bug as "uipc_send() wakes up the
reader before it's done appending the data from a write() to the socket
buffer" and use my patch? I don't think it makes sense for uipc_send()
to depend on sorwakeup() not actually waking up anyone in certain
situations.
Bill
T
How about a 1-line fix:
Index: uipc_usrreq.c
===
RCS file: /home/ncvs/src/sys/kern/uipc_usrreq.c,v
retrieving revision 1.37
diff -u -r1.37 uipc_usrreq.c
--- uipc_usrreq.c 1998/10/25 17:44:51 1.37
+++ uipc_usrreq.c 1999
Big Clue. Run "tcpdump -p" and see if the problem doesn't go away.
("tcpdump" puts the card in promiscuous mode, "tcpdump -p" does not).
Bill
To Unsubscribe: send mail to majord...@freebsd.org
with "unsubscribe freebsd-current" in the body of the message
Can you run a "tcpdump arp" on the machine that is having the problem,
as well? This could help to determine if it's a driver problem (e.g.
if the replies don't show up) or an ARP problem (e.g. if the replies
do show up but arp doesn't use them).
Bill
To Unsubscribe: send mail to majord...@fre
In message <19990129103757.a...@nagual.pp.ru>you write:
>I saw it several times with very recent -current
Is your machine getting a lot of incoming connections? I'd like to
try to replicate this. If it's a problem for you you can try
reverting rev 1.52 of /sys/kern/uipc_socket.c .
Thanks,
Bill
In message <36a85ef1.444a...@urc.ac.ru> you write:
>Why I'm asking about this, is because I recently read an advice in one
>of the FreeBSD mailing lists,
>about "Why my dial-up PPP connection from a FreeBSD box is so slow
>comparing with Windows NT
>(about ten times slower)?"
>
>And the advice was
The one that failed:
cc -c -nostdinc -O -pipe -DVERSION=\"1.03\" -DXS_VERSION=\"1.03\" -DPIC -
fpic -I/usr/obj/usr/src/gnu/usr.bin/perl/perl -DPERL_CORE -DLIBC="" DynaLoader.c
The one that succeeded:
cc -c-DVERSION=\"1.03\" -DXS_VERSION=\"1.03\" -DPIC -fpic -I/usr/obj/us
r/src/gn
In message <199901171000.maa16...@greenpeace.grondar.za> Mark Murray wrote:
>OK - but losing stdio.h?? :-) That confuses the crap out of me. Anyone
>been screwing with build tools?
Uh, the failing command in my log file included "-nostdinc"; that
does a good job of hiding friends like stdio.h...
80 matches
Mail list logo
