an I/O error on the physical media.
For example, swap-backed VN devices have a sector size of one page,
i.e. 4K.
-Matt
Matthew Dillon
The first part of this patch is not yet part of my multipatch at
http://www.backplane.com/FreeBSD4/
This is a tentative fix, but I believe it to be correct. Until
yesterday I was testing swap-backed VN with only one swap partition,
otherwise this would have been found and fi
That fixes a problem with ccd, but not the one causing John's failures.
You will note that with John's failure's the I/O is properly page-aligned.
The fix to ccd deals with a misalignment problem.
But I believe I have found the problem... it is a bug in vm/swap_pager.c
that o
:swapping on a vn device?
:the 'size' and 'resid' of 0 looks suspicious.
:
:a fix was just committed to teh vn code that may fix this if that's your
:problem.
A fix was? When? Where? What?
-Matt
To Unsubscribe: send mail to [EMAIL PROTECT
'4.2BSD'. If we got rid of that requirement then you
could run it on top of a 'raw' VN device (without -s labels). If
the ccd device didn't try to fake up its label then newfs would
not attempt to rewrite it and thus would not complain.
:Since I am currently experimenting with BOOTP in -current, I patched ONLY
:vfs_conf.c and vfs_subr.c with your rootfsid changes. Works fine in a BOOTP
:configuration. -current sources are as of last night.
:
:I can now boot a -current system from only a floppy disk with a (kgzipped) kernel
:on
A new combo patch is now avaiable:
http://www.backplane.com/FreeBSD4/
first section, multipatch-2.diff hotlink.
The fixes are too numerous to mention. Portions of the patch may be
committed by Alan or dg at any time so anyone who uses the above should
beware. Al
I have submitted a patch to DG that fixes VN for CURRENT. Since VN is
currently broken on CURRENT, the patch is on a fast-track for commit.
I've include the patch below, but beware that it may be committed
quickly. This also fixes swap-backed VN. The -S option can be applied
empty.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
t fix.
:
:Doug
Why don't you ktrace the program to see what it does? On the face of it
it sounds like the program may be doing it on purpose.
-Matt
Matthew Dillon
<[EM
. The results of the trace should be of great benefit.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe fr
:linux> cp multipatch-1.diff test1
:fbsd> cat test1
:cat: test1: RPC struct is bad
:fbsd> cp test1 /tmp
:fbsd> cat test1
:
:+ (vp->v_object->flags & OBJ_MIGHTBEDIRTY)) {
:
:^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@
:^@^@^@^@^@^@^@^@^@^@^@^@^@
: fbsd> cat /test1 local1
Oops. I obviously meant to redirect the output for the cat's:
fbsd> cat /test1 > local1
.. etc...
-Matt
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-curren
is
to try to determine whether the bug is in the linux server or the freebsd
client. I kinda suspect the linux box because there are no cache
interaction issues w/ the freebsd box if the client is simply reading the
file.
-Matt
also
had to be made to getnewbuf(), the syncer (Kirk's circle queue), and
other places.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send m
code to follow to get
things in the right order.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
require softupdates-related buffer rewriting. Kirk has been looking for
a solution.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PR
almost guarentee crashes when kld's are used.
I run all my kernels with INVARIANTS and INVARIANT_SUPPORT, and all
modules are built into the kernel. That works for me.
-Matt
Matthew Dillon
ver in
half for operations in question.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
I forgot to mention: It's the 'Multi-patch #1' section of the web page,
the section at the top. Ignore all the stuff that comes after that -
they've already been committed, as have many other things not listed.
http://www.backplane.com/FreeBSD4/
Ok, I have put up the current patch set -- for patching against CURRENT
only - on my web site.
http://www.backplane.com/FreeBSD4/
They contain a bit more then the NFS stuff, but it's all related to
performance. It would take me too long to try to separate them out
(t
:Hello,
:
:I just got a panic when trying to play with NFSv3. The system is a dual
:PentiumPro system, running CURRENT/SMP. I use NFS as a kld.
The very first thing I would do is compile NFS into the kernel and not
use it as a kld. Then see if you can repeat the problem.
10 MBytes/sec to the platter on the server, but the cpu required on both
the client and server to do that is well over 60% of a Pentium III-450.
-Matt
Matthew Dillon
<[EMAIL P
eifer/
I've had fixes for CURRENT in the commit queue for a week but they haven't
gone in yet. Once they get put in we will be able to backport the
interesting parts to STABLE. There isn't much I can do about it without
commit privs.
files. Apart from writing the mkisofs output to a file,
the easiest solution is to tell cdrecord to burn at a slower rate - e.g.
1x or 2x instead of 4x or 6x.
-Matt
Matthew Dillon
ke NFS more readable, it just isn't worth the effort.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
o I think we are in the clear.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
:Hi,
:
:i wonder if some form of pageable memory is available (or will be, or
:can be done with relatively little effort) to pieces of the lower half
:of the kernel. The reason is, the PGM implementation i am working on
:might need to work with really huge windows (megabytes) and on the
:sender si
ttles down. But in
a year or two when the DVD-writers come down in price I'll probably
start using them.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsu
no problem with the structural
field itself.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
sck' had overflow
problems, but those were fixed.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
:The device is probably dev=0,1
I'm sorry, I meant 1,0. Also 0,1,0 should work.
Try:
cdrecord dev=1,0 -inq
You should get:
Cdrecord release 1.8a22 Copyright (C) 1995-1999 Jörg Schilling
scsidev: '0,3,0'
scsibus: 0 target: 3 lun: 0
Devic
there is little
chance of mkisofs getting behind so I then 'mkisofs -R' that directory
and pipe it to cdrecord to do the actual backup.
-Matt
Matthew Dillon
&l
:Hate to ask, do we support CD-RW?
Yup! In fact, I recommend that you get a 5-pack of CD-RW disks
so you don't turn your CD-R's into scrap while playing with the
unit.
-Matt
:And I will probably bug next the multimedia group;however,
:those
One thing www.buy.com is really good for, even if you don't buy from them,
is to get pricing information on a general search. You can then use it
to do comparison shopping.
http://www.buy.com/
click on "comp"
Search for "rewritable"
And there you have it.
:Where did you get your Yamaha cd writer ?
:
:Tnks
:
:--
:
: Amancio Hasty
I just picked it up at CompUSA. You can probably also get it via
buy.com but I haven't looked.
-Matt
Matthew D
16S today and it "just works" as well.
Man, CD-R disks are getting cheap!
I'm using it mainly for backups. I got tired of my tape drive dying.
Does anyone know how to create a bootable CDRom?
-Matt
Linux for the most part.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
:
:On Sun, 15 Aug 1999, Matthew Dillon wrote:
:
:> :>2) under "kern.emu.linux"
:> :>3) under "linux"
:> :
:> :I vote for 3.
:> :
:> :--
:> :Poul-Henning Kamp FreeBSD coreteam member
:> :[EMAIL PROTECTED] "Real h
ess goes too far!
Ditto. Linux emulation is going to become increasingly important,
burying deep would just make everyone's life more difficult.
-Matt
Matthew Dillon
<[E
our source tree.
I don't know about the first problem... maybe compile up your kernel
DDB and do a 'trace' when it panics - record the original panic
message as well as the trace.
-Matt
0 12960 - R ??0:00.00 (sshd1)
:0 4672 199 1502 105 0 8680 - Rs??0:00.00 (rpc.rstatd
:#
:...
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
:On my home network, I make world on one machine and install it (from an NFS
:mounted directory) on another. Some time ago
:As I recall, Matthew Dillon wrote:
:> I would recommend simply having a 'FreeBSD' T-Shirt. None of this
:> 'committers' or 'core' stuff.
:>
:> -Matt
:>
I would recommend simply having a 'FreeBSD' T-Shirt. None of this
'committers' or 'core' stuff.
-Matt
Matthew Dillon
<[EMAIL PROTECT
be symlinking /dev/tty4 to
/dev/ttyv4.
I'm not up on the linux emulation so the best I can do is provide
general ideas.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]&g
:--0-169768575-933691257=:71237
:Content-Type: TEXT/PLAIN; charset=US-ASCII
:
:Hello
: I have been using FreeBSD under vmware for a few weeks now,
:running 3.2R. Everything has been running good including X11. I upgraded
:to -current and the linux XF86_vmware server stopped working. When
)
:
:Kevin
I've never in my life tried this - it probably won't work, but ...
use the null device maybe to create a mount point for each home
dir and then export that?
-Matt
sr m2
mount apollo:/usr m3
mount apollo:/usr m3
mount apollo:/usr m3
I can import a filesystem as many times as I want, and even overlay mount
points.
-Matt
are how many times a
directory is imported or exported.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe free
us . $2 > $3
R$* < @ pplus . $=w . > $* $#popplus $: $1 < @ pplus . $2 > $3
And then add the new mailer:
Mpopplus, P=/usr/local/bin/dpopmail, F=SDEFhlMsu, S=10/30, R=20/40,
U=dpop, A=dpopmail $u
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
code,
without my commit access there isn't going to be.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubs
ing with any locks.
vput() will release and unlock a vnode.
Hence the complexity: if (newvp == vp) vrele(newvp); else; vput(newvp);
-Matt
Matthew Dillon
<[EMAIL
reproduce the problem with a du in gs5.50 (and also see if my patch
fixes it).
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTE
s support: the behavior with the
:...
:was just a consequence of the filesystems being laid out differently.
:The patched FreeBSD client works fine now with the SGI server.
:
:-Bill
Ah, ok. I mistook the SGI for being a client.
-Matt
ient & server implementation of the protocol in FreeBSD.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
Index: nfs_vnops.c
:
:Okay. I committed the fix to the length calculation to -current and
:-stable (I just love one-line patches that stop panics). I just got
:done patching my NFS server machines and they all seem to get along
:nicely with the SGI now. Now I can upgrade the other SGIs without
:worrying about them
(6 * NFSX_UNSIGNED) is basically a NFS constant.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
nfsm_clget;
if ((bp + xfer) > be)
tsiz = be - bp;
else
tsiz = xfer;
bcopy(cp, bp, tsiz);
bp += tsiz;
xfer -= tsiz;
if (xfer > 0)
cp += tsiz;
}
:Yes, we do. I've run into this problem elsewhere but a quick fix was needed
:so it just got hacked. NT NFS clients tend to trigger it too.
:
:The problem is that the sanity check is a fair way away from where the problem
:packet is generated. The bad reply is generated in the readdirplus routi
test3:/usr/src/sys/compile/ALPHA# make
cc -c -Os -pipe -Wall -Wredundant-decls -Wnested-externs -Wstrict-prototypes
-Wmissing-prototypes -Wpointer-arith -Winline -Wcast-qual -fformat-extensions -ansi
-g -nostdinc -I- -I. -I../.. -I../../../include -DKERNEL -include opt_global.h -elf
../../i
e ldconfig_paths in
/etc/rc.conf (defaults in /etc/defaults/rc.conf) are wrong, or
you do not have them installed.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Un
ght help to get a backtrace if you have a kernel set to drop
into DDB on a panic. This would be the 'trace' command from the DDB
prompt. I could try to locate where the limit is being smashed.
-Matt
Ah ha! Yes, 32K packets will certainly screw up NFS under FreeBSD.
We need to fix that panic to have it simply drop the packet, I guess.
-Matt
Matthew Dillon
<[EM
ance of reproducing the problem.
The SGI is probably tickling something that FreeBSD doesn't or I
would have caught the problem earlier.
Your bug really sounds like a packet realignment bug, but I was
sure I fixed those! So there may be a new one.
: >
: > We 32 really ought to be enough. Looking at your PR I think
:
:It's not my PR. I am just asking for opinions on his proposal.
Oops, sorry! s/you/him/g
: > dealing with loops, I would hate to change it based on this particular
: > setup because I think this set
cept, as you say, when
dealing with loops, I would hate to change it based on this particular
setup because I think this setup could be optimized considerably to
get well under the current limit of 32.
-Matt
failure.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
I'm getting really annoyed with this dev_t stuff. I wish you people
would have tested a bit more before committing it all.
MFS is badly broken when used in a diskless configuration. I am trying
to track it all down but it is very, very frustrating.
Also BOOTP seems to be
:On Tue, 20 Jul 1999, Matthew Dillon wrote:
:
:>
:> How old is your -CURRENT source tree ?Fixes to the pipe code
:> have been made, though they are not 100% complete.
:>
:...
:
:What's happing with regard to your pipe write locking fixes? I still
:apply th
:On Tue, 20 Jul 1999, Matthew Dillon wrote:
:
:>
:> * make sure your swap partition is large enough to hold the crash
:> dump. If you have 256MB of ram, your swap partition must be
:> at least 256MB in size.
:
:Is there any reason that savecore(8) can't
:And how do you create dumps from a kernel that hasn't finished booting
:(not gotten to the stage of reading rc.conf)? 'dumps on' in kernel
:config does not seem to do the job.
:
:Nick
You can do it manually from /etc/rc. If it doesn't even get that far,
you used to be able to specify i
ry you have),
*and* the kernel image. Normally you give it a lot more space so
you store several crash dumps in it at once.
-Matt
Matthew Dillon
<[EMA
:Hi there,
:
: I recently had a crash on my SMP system. Actually, I have had a
:number of crashes over the past six months, but have just recently had
:time to configure the system to get the crash dumps. Like a good
:citizen, I filed a problem report (kern/12127). I have now gotten
:another cra
:I have tested out of swap handling in current. The situation is that the
:
:system just lockes up, "looping" in vm_pageout_scan.
Yah, this is a known problem. I'll take a look at your patch but I
was thinking of something a little simpler. But I wasn't planning on
fixing it for a f
OSTIC became due to being overly intrusive.
INVARIANTS should simply assert conditions that it expects to be
true and panic if they aren't.
-Matt
Matthew Dillon
<[EMAIL
Brian Fundakowski Feldman _ __ ___ ___ ___ ___
Wow, now that *is* expensive! The K6 must be implementing it in
microcode for it to be that bad.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]&
fely initiate DMA without having to flush your caches. This
is not true of all processors. I'm not sure about the Pentium's.
-Matt
Matthew Dillon
<[EMAIL P
as a complete cache-miss
case. The cost of snooping another cpu's L2 cache is much less then
the cost of going to main memory.
-Matt
Matthew Dillon
<[EMAIL PROTECTED
tle special case for the stack. Only the L1 cache is coherent. If
you wanted you could implement multiple L1 data caches on-chip - that
might be of some benefit, but otherwise branch-and-link is the better
way to do it.
-Matt
where a routine is called which
calculates the method address and then calls it, the branch prediction
cache will tend to fail because a different address will tend to be
called at the PC of the call.
-Matt
the push/pop, and potential arguments, but it is very fast.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe fr
isor code (e.g. network interrupt and network
stack ).
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
dure calls
are very cheap on intel cpu's, so it will probably have an effect
similar to the global assignments I added in the second half of the
test.
-Matt
Matthew Dillon
How did that happen?!!! Argg..I didn't save the email.
Here are better results:
Empty loop:
mode 09.21 ns/loop nproc=1 lcks=EMPTY
With and without lock prefix, one and two processes
mode 1 16.48 ns/loop nproc=1 lcks=no
mode 2 23.65 ns/loop
Here we are:
Empty loop
mode 09.21 ns/loop nproc=1 lcks=EMPTY
Tight loop, 1 and 2 processes, with and without lock prefix
mode 1 16.48 ns/loop nproc=1 lcks=no
mode 2 23.65 ns/loop nproc=2 lcks=no
mode 3 93.02 ns/loop nproc=1 lcks=yes
mod
:actually, I'm not so sure, it guarantees that NO other bus operation
:will succeed while this is happening... what happens if a pci bus
:mastering card makes a modification to this value? sure, it normally
:won't happen, but it can... and w/o the lock prefix, this CAN happen
:from what I unders
Ok, here we are. The lock instruction overhead is significant relative
the execution time of the instruction, though on a 450 MHz P-III it is
still not a huge problem. Basically 12 vs 89 ns for the non competitive
case, and 26 vs 161 ns for the competitive case.
empty loop
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
:
:That said, it should be fairly simple to change Matt's new in-line
:assembler versions to insert LOCK prefixes when building an SMP
:kernel. (Although I don't know that this is necessary yet, given
:the `Big Giant Lock').
:
:There remains the problem of locating all the operations in the kerne
dn't hurt if someone with more assembly experience could audit
all of the inline assembly present in the source base.
I am re-running the 16MB/make -j5 buildworld test to see if this has
really fixed the problem.
-Matt
Yahhh. I was finally able to reproduce the problem - running Stephen's
16MB make -j5 buildworld test overnight.
I haven't found the exact cause yet, and I suspect that Alan's patch will
not fix it (but I'll try it if I exhaust other possibilities). There is
plenty of free me
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
t.
-Matt
Matthew Dillon
<[EMAIL PROTECTED]>
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
:Hi,
:
:Any current's more recent than about a month ago on my main system seem to
:'hang' on biord whenever they access the IDE drives...
:...
Make sure you have the absolute latest CURRENT. There was a situation
that broken current a week or so ago for 2 days that could result
in p
801 - 900 of 1252 matches
Mail list logo
