591 lines
24 KiB
Plaintext
591 lines
24 KiB
Plaintext
Subject: Linux-Development Digest #574
|
|
From: Digestifier <Linux-Development-Request@senator-bedfellow.MIT.EDU>
|
|
To: Linux-Development@senator-bedfellow.MIT.EDU
|
|
Reply-To: Linux-Development@senator-bedfellow.MIT.EDU
|
|
Date: Wed, 23 Mar 94 14:13:09 EST
|
|
|
|
Linux-Development Digest #574, Volume #1 Wed, 23 Mar 94 14:13:09 EST
|
|
|
|
Contents:
|
|
crash hosed my superblock...can it be saved? (Jerod Tufte)
|
|
Re: 'format' program (Sun style) (Keith Medcalf)
|
|
Re: IPX compliancy? (Alan Cox)
|
|
PPP defroute problem - hosts, rc.inet1 config? (Eric Kimminau)
|
|
Kernel don't recognizes HP PC LAN Ethernet Card ! (Thomas Heiling)
|
|
Hanging CSLIP with ftp (Nick Vargish)
|
|
Re: Real-Time Linux and a/d device drivers (Matthew Donadio)
|
|
Re: <stddef.h> in libc-4.5.21? (Dwayne Springfield)
|
|
plotter works with 0.99pl1[4,5], but not 1.0 (Denis Endisch)
|
|
bug? or don't I get something? (Filip M Gieszczykiewicz)
|
|
|
|
----------------------------------------------------------------------------
|
|
|
|
From: jet5@pyrite.SOM.CWRU.Edu (Jerod Tufte)
|
|
Subject: crash hosed my superblock...can it be saved?
|
|
Date: 23 Mar 1994 10:40:15 GMT
|
|
|
|
I had my first crash in many weeks a few hours ago, and I can't seem to
|
|
be able to do anything to recover. I was shutting down X, when the
|
|
system just suddenly seized. I rebooted to get this message:
|
|
extra data not valid Current error sd801 sense Key Illegal request
|
|
Additional sense indicates Logical block address out of range
|
|
|
|
and when I booted with an old recovery disk, I got:
|
|
in2000: 01 80 4b 36 00
|
|
Dumping sense buffer: 112 0 5 0 0 0 0 10 0 0
|
|
host 0 id 0 lun 0 return code 28000000
|
|
Sense class 7 sense error 0 extended sense 5
|
|
scsidisk I/O error dev 0801, sector 2
|
|
EXT2-fs unable to read superblock.
|
|
|
|
My system is a dx2/66 VESA m/b w/ IN2000 scsi and S3-805VLB video.
|
|
someone please tell me there's a tool that can help me recover some/all
|
|
of my files. thanks.
|
|
|
|
Jerod Tufte
|
|
jet5@po.cwru.edu
|
|
--
|
|
jet5@po.cwru.edu Case Western Reserve University Computer Engineering
|
|
jet@b62528.student.cwru.edu <<< finger me for PGP2.3a public key
|
|
|
|
------------------------------
|
|
|
|
From: kmedcalf@zealot.uucp (Keith Medcalf)
|
|
Subject: Re: 'format' program (Sun style)
|
|
Date: 23 Mar 1994 06:40:23 GMT
|
|
Reply-To: kmedcalf%zealot@io.org
|
|
|
|
W. Tait Cyrus (cyrus@jemez.eece.unm.edu) wrote:
|
|
|
|
: Last week I purchased a new SCSI disk and after only 6 days it quit
|
|
: working. Fortunately I had made a backup off all my stuff the day
|
|
: before it died so I didn't loose a lot of data.
|
|
|
|
If the SCSI disk is properly designed, and properly implements the SCSI
|
|
spec, all that is required is to set the appropriate control pages, issue a
|
|
mode select, the issue a format unit with the certification bit set (which
|
|
is SUPPOSED to be the default). Many non-SCSI compliant drive manufacturers
|
|
(particularly of low-cost low-reliability low-stability drives) do not
|
|
implement the SCSI spec correctly (most notable is Maxtor).
|
|
|
|
If you set the control mode pages, do a mode select and a default format
|
|
unit, and the drive develops an error within 2 years, you should send it
|
|
back to the manufacturer for replacement as it has a manufacturing or design
|
|
defect.
|
|
|
|
Manufacturers who create high quality drives will take them back for free
|
|
re-conditioning under warranty on a yearly basis (CDC, Fujitsu), and will
|
|
last for years of continuous heavy service with no problems.
|
|
|
|
Other drives (not to mention Maxtor again, but they make the most unreliable
|
|
and feature-free drives you can waste your money on) will last barely any
|
|
time at all before developing errors.
|
|
|
|
This is my opinion and experience only, and your mileage may, of course,
|
|
vary. You will find however that a bottom-of-the-line Fujitsu drive will
|
|
operate and last much longer than a top-of-the-line Maxtor, and that the
|
|
Maxtor does not implement most of the features in SCSI that are most useful.
|
|
|
|
I have a CDC (real CDC, not ScuzGate) drive that is now coming up on 8 years
|
|
of continuous heavy service, almost 24 hours a day, 366 days a year, with
|
|
nary a hitch. I have yet to see any maxtor drive last more than a year
|
|
under the same load ...
|
|
|
|
--
|
|
Keith Medcalf | Internet: kmedcalf@zealot.uucp kmedcalf@io.org
|
|
Keith Medcalf and Associates | UUCP: ... uunet.ca!io.org!zealot!kmedcalf
|
|
Database Systems Engineer | Anonymous UUCP: (416) 465-4370 in: nuucp
|
|
Toronto/Canada (416)465-9578 | Fidonet: 1:250/750 - IBMNet: 40:6482/303
|
|
|
|
------------------------------
|
|
|
|
From: iiitac@uk.ac.swan.pyr (Alan Cox)
|
|
Subject: Re: IPX compliancy?
|
|
Date: Tue, 22 Mar 1994 14:55:03 GMT
|
|
|
|
In article <1994Mar14.185508.46244@ucl.ac.uk> zceed04@ucl.ac.uk (Mr Ivan Alastair Beveridge) writes:
|
|
>Basically, I was wondering if anyone has made Linux compliant with Netware
|
|
>at all. As I do not really know much about protocols, I cannot really ask
|
|
>much more than this.
|
|
|
|
There is a beta test IPX layer for Linux, but no netware support. Novell
|
|
guards its netware details with lawyers and complex licensing agreements
|
|
involving thousands of dollars. So forget it - Linux does Lan manager and NFS
|
|
|
|
|
|
------------------------------
|
|
|
|
From: ekimmina@pms709.pms.ford.com (Eric Kimminau)
|
|
Subject: PPP defroute problem - hosts, rc.inet1 config?
|
|
Date: 22 Mar 1994 15:42:00 GMT
|
|
|
|
I have just completed upgrading to LINUX 1.0, GCC 2.5.8 & LIBC4.5.21.
|
|
|
|
PPP was working beautifully on 99.14 before the upgrade. I have a
|
|
static hostname and IP address on every connect, so I want to hardcode
|
|
my system files to use it. I am still unable to get a stable chat
|
|
script to work consistantly, but manually connecting via kermit will
|
|
work with an additional step. Here lies my problem:
|
|
|
|
Now, after connectiong to my ppp server and escaping back to the
|
|
kermit prompt and running pppd, I now have to manually add the default
|
|
gw route to get things working. Under 99.14, that route was added
|
|
automagically. I have tried with and without the defaultroute
|
|
command.
|
|
|
|
Could someone who has ppp working in that way under 1.0 send me a
|
|
couple things:
|
|
|
|
#1 A copy of your /etc/hosts with the significant portions (routeer,
|
|
net, localhost, etc.
|
|
|
|
#2 A copy of your rc.inet1 or wherever you happen to configure lo
|
|
and/or ppp0 with your hostname, ip, default raoute, net and gateway.
|
|
|
|
#3 A copy of your chat (or kermit) startup script if you have it working.
|
|
|
|
Thanks in advance!
|
|
|
|
--
|
|
Eric Kimminau Workstation Systems Department
|
|
313-322-3431 Product & Manufacturing Systems
|
|
ekimmina@pms709.pms.ford.com Ford Motor Co.
|
|
Planning and Implementation "Not an official Ford Spokesperson"
|
|
TIP#111
|
|
|
|
------------------------------
|
|
|
|
From: tom@wpzd07.pzlc.uni-wuerzburg.de (Thomas Heiling)
|
|
Subject: Kernel don't recognizes HP PC LAN Ethernet Card !
|
|
Date: 22 Mar 1994 15:57:30 GMT
|
|
|
|
Hello all!
|
|
Finally i got a new machine and tried to install
|
|
Linux. Everything fine, some fiddling with XFree 2.1,
|
|
and no the Problem:
|
|
I have a HP PC LAN Adapter / 16 TL+ (aka HP 27252A )
|
|
and compiled the Kernel ( Version 1.0 + patch 1 + 2 )
|
|
with CONFIG HP PCLAN. But after compiling, the Kernel
|
|
don't recognizes my Card.
|
|
Here is my Machine:
|
|
HP Vectra 486/66 MHZ, 8 MB,
|
|
integrated VGA Card Localbus with 512 KB
|
|
HP 27252A Ethernet Card /* This Card is mentioned in the
|
|
Ethernet-HOWTO as one of the best Cards, but there are no
|
|
hints / tips , why the Kernel don't recognizes it */
|
|
|
|
What i have done this weekend :
|
|
HP has a nice program for setting up the card,
|
|
it is called hplanset.
|
|
I have done this :
|
|
Set the IO to 300-31fh
|
|
IRQ to 10
|
|
Station Adress is 080009-530D4C
|
|
compiled with CONFIG HP PCLAN -> no chance
|
|
compiled with CONFIG_LANCE -> no chance
|
|
compiled with every available Ethernet card -> no chance
|
|
Tried it with LILO : ether=10,0x300,0,0,eth0 -> no chance
|
|
Copied a kernel from this machine ( This is a 386/25 MHz
|
|
HP Vectra with only 4 MB and a HP LAN Card, the older model
|
|
with only 8 bit ) to the new machine, but it fails again.
|
|
|
|
And now i have absolutly no idea, what to do next.
|
|
Any help would be very nice.
|
|
Thanks Thomas
|
|
|
|
|
|
|
|
--
|
|
===================================================
|
|
Thomas Heiling Pharmacist & Doctorate at
|
|
Pharmazeutisches Institut Uni Wuerzburg - Germany
|
|
Email phar006@wrzx12.rz.uni-wuerzburg.de (HP-UX)
|
|
tom@wpzd07.pzlc.uni-wuerzburg.de (Linux)
|
|
or phar006@vax.rz.uni-wuerzburg.de ( VAX )
|
|
===================================================
|
|
|
|
--
|
|
===================================================
|
|
Thomas Heiling Pharmacist & Doctorate at
|
|
Pharmazeutisches Institut Uni Wuerzburg - Germany
|
|
Email phar006@wrzx12.rz.uni-wuerzburg.de (HP-UX)
|
|
tom@wpzd07.pzlc.uni-wuerzburg.de (Linux)
|
|
or phar006@vax.rz.uni-wuerzburg.de ( VAX )
|
|
===================================================
|
|
|
|
------------------------------
|
|
|
|
From: vargish@yx.sura.net (Nick Vargish)
|
|
Crossposted-To: comp.os.linux.misc
|
|
Subject: Hanging CSLIP with ftp
|
|
Date: 23 Mar 1994 14:38:41 GMT
|
|
|
|
Folks,
|
|
|
|
Here's the scoop... All I have to do to hang my CSLIP connection is
|
|
try to ftp an uncompressed xpilot map across the connection. Once I
|
|
compress the map (which I will include here), the transfer goes
|
|
smoothly.
|
|
|
|
I suspect it has something to do with long strings of identical
|
|
characters and the compression algorithm(s), but I'm basically taking
|
|
an educated guess (no facts or code-grazing to back me up).
|
|
|
|
I thought someone might be interested in this, assuming it's not some
|
|
boneheaded mistake on my part.
|
|
|
|
And by the way, I meant _hung_, so don't try this unless you're
|
|
willing to restart your CSLIP connection...
|
|
|
|
Nick Vargish
|
|
|
|
p.s. This is the map I was talking about, gzipped and uuencoded. It
|
|
should work gzipped, and fail ungzipped...
|
|
|
|
============================================ CUT HERE
|
|
begin 644 13.map.gz
|
|
M'XL(",!D'RP S$S+FUA< #MW$]+PS 8!O![/L4+\: @TSJ5V9,#!07_X4$O
|
|
M@G1KW")M.K(ZMV]OIB"K;23.S:7S>=@I[_NVOQ6R0QCE=!D-J)NID="YB*DS
|
|
MH30:F,] :,;I>)-O;9S'(07-AEG<'E'0V*/@Z*BYL]O<V6M1<!CNM\*#%G6>
|
|
M,Z'5D$[' ]J@.Z&',E.F>Y>9N7L9Y_V09A+L!]/"F9"]?AZ6"U=1*@H#1C!=
|
|
M;[_D_4S/5FXSI29TUJ"VEDG,<A&E-TDT*0RKC.4RE:I7O.1T/3'KYHM?R)$8
|
|
M?E8G8LA>,YU\629Z-YQ$>1320_J2Y#*1RCA/57S]9!YD;"IL[%/8F,JI:JQH
|
|
M6WP?--"LJ695@<8>:.RI@\:O'0X--'73/+I-E]L<[_+HWL<*Q:JQJHLLJP\:
|
|
M:*"!!AJ':?[1QQ?5MT[/!AIHH('F%QHJQSXV7URO!PTTZZ_AWT[_6,/GT:PJ
|
|
MT-@#C3UUT/BUPWW]]8,&&IO&\;QXZ9J?G!<O6E-Y7VB@@>8]6FM_-#'G_,DC
|
|
MS=@KC4_/AKK=[G?E?[VGH($&&FB<-([3B]94]D$#S=IK?#A-6E6@L0<:>^J@
|
|
M\6N'^_KK!PTT-DT=SXO_I \::*"!!AJ':?R_&!IHH(%F*1HJQSZVY#YHH('F
|
|
E#S2K"C3V0&-/'31^[7!HH*F;QL/W4?@15GB%W!OP"[TWED\ #ZV
|
|
|
|
end
|
|
============================================ CUT HERE
|
|
|
|
|
|
--
|
|
---------------------- ---------------------- -----------------------
|
|
|. Nick Vargish .| |. SURAnet .| |. O: (301) 982-4600 .|
|
|
|. vargish@sura.net .| |. systems engineer .| |. rust never sleeps .|
|
|
---------------------- ---------------------- -----------------------
|
|
|
|
------------------------------
|
|
|
|
From: donadio@mxd120.rh.psu.edu (Matthew Donadio)
|
|
Subject: Re: Real-Time Linux and a/d device drivers
|
|
Date: 22 Mar 1994 16:42:57 GMT
|
|
|
|
Kevin Brown (kevin@frobozz.sccsi.com) wrote:
|
|
: There may be something I'm overlooking, of course (which wouldn't be
|
|
: surprising), but I'm somewhat convinced that between the above and some
|
|
: kind of semaphore implementation, you should be able to do pretty much
|
|
: everything you'd expect out of a "real" threads implementation.
|
|
|
|
You can emulate a lot with fork(), but in my experience the
|
|
applications I have written using pthreads() seem to behave better.
|
|
Mutexes can be done with semaphores, but conditions (as in
|
|
pthread_condition) get a little hairy when you have a master process
|
|
controlling a bunch of slave processes. Plus, pthreads lets you give
|
|
up the processor with pthread_yeild() which can make threaded
|
|
applicaitons more effieient.
|
|
|
|
--
|
|
Beaker aka Matt Donadio | Life is short, --- __ o __~o __ o
|
|
donadio@mxd120.rh.psu.edu | ride like ---- _`\<, _`\<, _`\<,
|
|
--- Penn State Cycling ---| the wind. --- ( )/( ) ( )/( ) ( )/( )
|
|
|
|
------------------------------
|
|
|
|
From: dwaynes@netcom.com (Dwayne Springfield)
|
|
Subject: Re: <stddef.h> in libc-4.5.21?
|
|
Date: 20 Mar 94 18:54:06 GMT
|
|
|
|
Joseph Toman (toman@darkwing.uoregon.edu) wrote:
|
|
: Hi, I am trying to compile various source code packages for "Lee-noocks" :)
|
|
: and I don't seem to have the ANSI C standard include file <stddef.h>. It is
|
|
: neither in Slackware 1.1.2 nor in libc-4.5.21 on tsx-11. Where can I find it?
|
|
|
|
: Thanks, Johannes
|
|
sometimes becuase of the gcc installation that file gets moved to
|
|
a relatively obscure spot in the directory hierarchy....
|
|
if you have not tried using the find command to locate the file
|
|
(find / -name stddef.h -print) then you just might find the file
|
|
in a directory path resembling but not necessarily exactly like --->
|
|
/usr/lib/gcc-lib/i486-linux/2.5.7/include/stddef.h
|
|
Like I said... obscure. You will of course adjust OS and version numbers
|
|
according to your own installation :)
|
|
|
|
hope this helps.
|
|
|
|
l8r,
|
|
dwayne
|
|
|
|
|
|
--
|
|
+------------------------------+-------------------------------------------+
|
|
| Dwayne Springfield | Time flies like an arrow... |
|
|
| dwaynes@netcom.com | Fruit flies like a banana. |
|
|
| dwayne.springfield@octel.com | Groucho (Karlo's Brother) Marx |
|
|
+------------------------------+-------------------------------------------+
|
|
|
|
------------------------------
|
|
|
|
From: dendisch@cat..physics.uwo.ca (Denis Endisch)
|
|
Subject: plotter works with 0.99pl1[4,5], but not 1.0
|
|
Date: 22 Mar 94 21:07:49 GMT
|
|
|
|
Hi there
|
|
|
|
I have a plotter (HP ColorPro) attached to /dev/cua1.
|
|
It worked fine with plain 0.99pl14 and 0.99pl15, but with 1.0
|
|
it seem to loose characters from time to time. Sometimes the plots
|
|
are ok, but most of the time the plotter starts going crazy in
|
|
the middle of the plot, obviosly getting some wrong vector values.
|
|
|
|
For the older kernels (pl14,pl15) I have NOT used the kernel patch
|
|
mentioned in the Printing-HOWTO (Dec 1 93, included at the end of
|
|
this posting) for setting up a serial printer at that time, since
|
|
it worked fine without it.
|
|
|
|
In the current Printing-HOWTO (Feb 22 94) the patch is not mentioned any
|
|
more. Also the relevant lines in /usr/src/linux/kernel/chr_drv/tty_ioctl.c
|
|
have changed, so that the older patch can not be applied anyway. It
|
|
looks as if some kind of bug was implemented by this recent changes.
|
|
|
|
Does anybody has similar problems or a idea what's going wrong?
|
|
Or a idea for solving the problem?
|
|
My current solution is not to use 1.0 :-(
|
|
|
|
Thanks Denis
|
|
|
|
====================================================================
|
|
Some more information:
|
|
I have the plotter running with lpd connected to a dummy device, as
|
|
it is described in the Printing-HOWTO. The setting for the serial
|
|
port are made at boottime with 'stty'. I have set up these settings
|
|
with the 0.99pl14 kernel and have nothing changed since then.
|
|
|
|
====================================================================
|
|
My stty settings are:
|
|
cat:~$ stty -a < /dev/cua1
|
|
speed 9600 baud; rows 0; columns 0; line = 0;
|
|
intr = ^C; quit = ^\; erase = ^?; kill = ^U; eof = ^D; eol = <undef>;
|
|
eol2 = <undef>; start = ^Q; stop = ^S; susp = ^Z; rprnt = ^R; werase = ^W;
|
|
lnext = ^V; flush = ^O; min = 1; time = 0;
|
|
-parenb -parodd cs8 hupcl -cstopb cread -clocal crtscts
|
|
-ignbrk -brkint ignpar -parmrk -inpck -istrip -inlcr -igncr -icrnl ixon -ixoff
|
|
-iuclc -ixany -imaxbel
|
|
-opost -olcuc -ocrnl -onlcr -onocr -onlret -ofill -ofdel nl0 cr0 tab3 bs0 vt0
|
|
ff0
|
|
-isig -icanon -iexten -echo -echoe -echok -echonl -noflsh -xcase -tostop
|
|
-echoprt -echoctl -echoke
|
|
|
|
====================================================================
|
|
From the older Printing-HOWTO (1 Dec 93) about setting up a serial printer:
|
|
...
|
|
First, there is a kernel patch you need to make. This applies
|
|
to pre-net-2 lpd as well (actually, the use of this should
|
|
probably just be commented out of the lpd source code). The
|
|
change is for at least patchlevel 10, but probably all others
|
|
as well.
|
|
|
|
in /usr/src/linux/kernel/chr_drv/tty_ioctl.c, around line
|
|
390, you will see:
|
|
|
|
case TIOCEXCL:
|
|
return -EINVAL; /* not implemented */
|
|
|
|
Change this to return 0, recompile the kernel, and you'll be
|
|
all set.
|
|
...
|
|
====================================================================
|
|
|
|
I can give more info about system configuration, if this is important.
|
|
|
|
--
|
|
Denis Endisch Phone: (519) 661 - 2111x6413
|
|
Department of Physics Fax: (519) 661 - 2033
|
|
Univ. of Western Ontario Email: dendisch@uwo.ca
|
|
London, Ontario, Canada N6A 3K7
|
|
|
|
------------------------------
|
|
|
|
From: filip@alpha.smi.med.pitt.edu (Filip M Gieszczykiewicz)
|
|
Subject: bug? or don't I get something?
|
|
Date: 23 Mar 94 16:45:10 GMT
|
|
|
|
|
|
Greetings. I ran a butchered version of iozone 1.6 (old) with
|
|
mods to allow for concurrent execution (to test just to what
|
|
extent IDE sucks in multi-tasking applications - oh, btw, it
|
|
sucks the chrome off the bumpers... more later)
|
|
|
|
Anyway, for this little test to be "fair", I had to shoot
|
|
Linux buffers in the foot (by allocating most of my memory
|
|
with an "eatmem" program). I left about 1.5MB for the
|
|
multiple instances (up to 20) of the 23K executable.
|
|
|
|
What I noticed was that linux was groping the swap even
|
|
though there was ~1MB in the buffers. After a few [dozen]
|
|
trials, it was REALLY eating swap like there was no tomorrow!
|
|
|
|
I'd like to know something before I go any further: What does
|
|
linux prefer to do a) write/read directly to disk or b) write/
|
|
read to buffers which are really swap [sigh]. Also, can it
|
|
tell the difference? Further, is there anything that can be
|
|
done to help this?
|
|
|
|
Also, my system is as follows: Linux 1.0 kernel (debugging on,
|
|
SCSI compiled in (not used)), xiafs FS on a 200MB Maxtor IDE
|
|
(17 ms), 486DX 2/66 VLB, 16MB ram, 13MB swap (partition),
|
|
256K cache, 11MHz ISA buss, 33MHz VLBuss. (note that eatmem
|
|
program "ate" 11MB)
|
|
|
|
I include a list of "free" after it started looking fishy:
|
|
|
|
=================chop=with=axe================chop=with=axe=====================
|
|
total used free shared buffers
|
|
Mem: 13748 13668 80 1268 1536
|
|
Swap: 13368 3096 10272
|
|
|
|
total used free shared buffers
|
|
Mem: 13748 13432 316 976 1864
|
|
Swap: 13368 3508 9860
|
|
|
|
total used free shared buffers
|
|
Mem: 13748 13668 80 1336 2212
|
|
Swap: 13368 3768 9600
|
|
|
|
total used free shared buffers
|
|
Mem: 13748 13668 80 1624 2480
|
|
Swap: 13368 4488 8880
|
|
|
|
total used free shared buffers
|
|
Mem: 13748 13164 584 912 2832
|
|
Swap: 13368 4608 8760
|
|
|
|
total used free shared buffers
|
|
Mem: 13748 13668 80 1408 3048
|
|
Swap: 13368 4728 8640
|
|
|
|
total used free shared buffers
|
|
Mem: 13748 13400 348 1000 3180
|
|
Swap: 13368 4856 8512
|
|
total used free shared buffers
|
|
Mem: 13748 13384 364 1000 3684
|
|
Swap: 13368 5376 7992
|
|
|
|
total used free shared buffers
|
|
Mem: 13748 13384 364 1000 3752
|
|
Swap: 13368 5444 7924
|
|
|
|
total used free shared buffers
|
|
Mem: 13748 13624 124 1008 4012
|
|
Swap: 13368 5464 7904
|
|
|
|
total used free shared buffers
|
|
Mem: 13748 13584 164 836 5420
|
|
Swap: 13368 6640 6728
|
|
|
|
total used free shared buffers
|
|
Mem: 13748 13668 80 1320 4852
|
|
Swap: 13368 6712 6656
|
|
=================chop=with=axe================chop=with=axe=====================
|
|
|
|
By this time ~5 hours & maybe 30-40 trials had been run. Right
|
|
now, as I write this, "free" says:
|
|
|
|
total used free shared buffers
|
|
Mem: 13748 13012 736 1056 4860
|
|
Swap: 13368 6740 6628
|
|
|
|
And, let me exit the eatmem,
|
|
|
|
total used free shared buffers
|
|
Mem: 13748 7268 6480 1044 4864
|
|
Swap: 13368 1120 12248
|
|
|
|
Well, the swap looks better but... why was I loosing memory
|
|
above... Is that a "fact of life", is it a bug, is it a leak,
|
|
is it paranoia, what is it?
|
|
|
|
Also, why does linux keep buffers @ 4MB even though it [should]
|
|
know that they are really swap. Or is that the problem... it
|
|
doesn't know the buffers are swap... eh?
|
|
|
|
BTW, I case anyone is wondering what kind of #'s I got from
|
|
my benchmarks... If you're planning to get an IDE, you
|
|
DEFINITELY want to read this (hopefully you'll change your
|
|
mind):
|
|
|
|
Well, the best (or worst, depending on your point of view)
|
|
test was 20 instances of iozone writing/reading a 1MB file.
|
|
|
|
RESULTS:
|
|
|
|
20 : concurrent jobs
|
|
1 : MB (file size)
|
|
10425 : bytes/second READ
|
|
89767 : bytes/second WRITE
|
|
|
|
[sniff..sniff..Whhhaaaaaa....]
|
|
|
|
Right behind (or in front :-) was 10 instances of 2MB iozone.
|
|
|
|
RESULTS:
|
|
|
|
10 : concurrent jobs
|
|
2 : MB (file size)
|
|
20274 : bytes/second READ
|
|
76145 : bytes/second WRITE
|
|
|
|
It got better with less instances... 1 instance of 20MB was
|
|
|
|
RESULTS:
|
|
|
|
1 : concurrent jobs
|
|
20 : MB (file size)
|
|
302009 : bytes/second READ
|
|
552318 : bytes/second WRITE
|
|
|
|
which just goes to prove that while IDE is "fast" in a single
|
|
task environment like MS-DOG or, to a lesser extent, MS-WINDOZE,
|
|
it really reeks for anything that has 2 or more processes writing
|
|
to the disk. And, God forbid, you have 20 users on line...
|
|
|
|
This is my BEFORE segment... I'll rerun all these tests after
|
|
I get my BT445S & 1.8GB :-) and post the AFTER...
|
|
|
|
I can upload the kludged iozone to sunsite so others can have
|
|
a go at it (if you don't believe me). A more complete version
|
|
of this article will be posted to comp.periphs.scsi to stop
|
|
silly flame-fest on IDEvsSCSI speed...
|
|
|
|
Take care (sorry this is so long and I hope the first item can
|
|
solved. Thanks!)
|
|
--
|
|
+-->Filip "I'll buy a vowel" Gieszczykiewicz | E-mail: filip@alpha.med.pitt.edu
|
|
| ftp to bode.ee.ualberta.ca and get OLD electonics faq from /pub/cookbook/faq
|
|
| Checkout: ftp bode.ee.ualberta.ca and get file /pub/cookbook/contents. Neat.
|
|
| Making money with CS and spending it on EE, robotics, windsurfing, & dreams.
|
|
|
|
------------------------------
|
|
|
|
|
|
** FOR YOUR REFERENCE **
|
|
|
|
The service address, to which questions about the list itself and requests
|
|
to be added to or deleted from it should be directed, is:
|
|
|
|
Internet: Linux-Development-Request@NEWS-DIGESTS.MIT.EDU
|
|
|
|
You can send mail to the entire list (and comp.os.linux.development) via:
|
|
|
|
Internet: Linux-Development@NEWS-DIGESTS.MIT.EDU
|
|
|
|
Linux may be obtained via one of these FTP sites:
|
|
nic.funet.fi pub/OS/Linux
|
|
tsx-11.mit.edu pub/linux
|
|
sunsite.unc.edu pub/Linux
|
|
|
|
End of Linux-Development Digest
|
|
******************************
|