Linus Torvalds wrote:
In article [EMAIL PROTECTED],
Ben Mansell [EMAIL PROTECTED] wrote:
On 14 Jan 2001, Linus Torvalds wrote:
And no, I don't actually hink that sendfile() is all that hot. It was
_very_ easy to implement, and can be considered a 5-minute hack to give
a feature that
Hi Rik,
On Wed, 17 Jan 2001, Rik van Riel wrote:
I don't even want to start thinking about how this would
screw up the (already fragile) page aging balance...
As of 2.4.1-pre we pin the pages by increasing the page count for
locked segments. No special list needed.
Greetings
On Wed, Jan 17, 2001 at 05:14:02PM -0800, you [Andre Hedrick] claimed:
On Thu, 18 Jan 2001, Tim Fletcher wrote:
Well that is useless test them because you can not test things completely.
I ment that if the partiton has no persient data on it then the test can
be run (the test wipes
Hello!
--- arch/arm/mach-sa1100/arch.c.origThu Jan 18 11:30:49 2001
+++ arch/arm/mach-sa1100/arch.c Thu Jan 18 11:30:57 2001
@@ -235,7 +235,7 @@
#ifdef CONFIG_SA1100_ITSY
MACHINE_START(ITSY, "Compaq Itsy")
BOOT_MEM(0xc000, 0x8000, 0xf800)
- BOOT_PARAMS(0xc100
But it works on all ATA disks? Does it work for SCSI as well?
The KIOBUFS version may, but not the taskfile version.
I think it would be cool if you'd make it available (on linux-ide.org?),
so that people install servers (and anybody who _cares_) could test their
hardware/driver
On Thu, Jan 18, 2001 at 12:45:13AM -0800, you [Andre Hedrick] claimed:
But it works on all ATA disks? Does it work for SCSI as well?
The KIOBUFS version may, but not the taskfile version.
Ok.
I think it would be cool if you'd make it available (on linux-ide.org?),
so that people
On Tue, Jan 16, 2001 at 06:39:46PM -0700, Duncan Laurie wrote:
There may be bogus data in the PIRQ table as well, which is why this
explicitly routes the interrupt sets the ELCR. If you enable DEBUG
in pci-i386.h and re-send the dmesg output I will look it over.
I tied your patch. dmesg
Paul Bristow a crit :
Hi,
I am trying to understand the new PCMCIA configuration. So far I am trying
to use the kernel PCMCIA driver. The yenta_socket driver is working fine.
Using 2.4.0, and cardmgr 3.1.23
When I insert my Iomega Clik drive, cardmgr correctly identifies it as an ATA
Ethernet is compiled into the kernel as is smbfs (not as modules). I've
compiled this kernel with 4GB bigmem support (otherwise I only get 8xxMB
total).
The smbfs cache code in 2.4.0 doesn't work with bigmem. For now disable
bigmem or don't use smbfs, it's oopsing all the time.
On Wed, 17 Jan 2001, Dan Hollis wrote:
They require not only an NDA, but that you also do all development on-site
at their santa clara HQ under their direct supervision.
I haven't went that far -- I'm not going to sign any NDA anytime soon, so
I haven't asked them for details. I recall
On 17 Jan 01 at 17:40, Pete Toscano wrote:
according to the linux-usb maintainers, it's a pci irq routing problem.
i've asked jeff garzik and martin mares if they'll look into it, but
they're pretty busy and i haven't heard anything back from them (not
that'd i'd expect to for quite a while,
On Wed, Jan 17, 2001 at 02:17:36PM -0800, Rick Jones wrote:
How does CORKing interact with ACK generation? In particular how it
might interact with (or rather possibly induce) standalone ACKs?
It doesn't change the ACK generation. If your cork'ed packets gets sent
before the delayed ack
Andreas Dilger wrote:
Ahh. What I was missing was that by specifying /dev/md0 as the root device,
not only do you get an identical map for the kernels, but the root device
remains /dev/md0 no matter which drive fails and LILO/kernel don't need to
do anything special to find it. This
On Thu, 18 Jan 2001, Petr Vandrovec wrote:
Rainer Mager reported the same thing yesterday ("Oops with 4GB memory
setting in 2.4.0 stable" if you want to read the thread).
I think that I found source of problem. I have no simple solution :-(
I think the source of the problem is that this
Roger Wolff writes:
I'd prefer an interface that says "copy this fd to that one, and
optimize that if you can".
For example, copying a file from one disk to another. I'm pretty sure
that some efficiency can be gained if you don't need to handle the
possibility of the userspace program
On Wed, 17 Jan 2001, Joel Franco Guzmn wrote:
With 128M the problem is not present, but with 192M it is. The only
difference is the memory quantity, or in other words, the additional slot
occupied by the new memory card.
- ASUS P299 (Chipset i440ZX). Note: the i440ZX don't support
On Thu, 18 Jan 2001, Maciej W. Rozycki wrote:
On Wed, 17 Jan 2001, Dan Hollis wrote:
They require not only an NDA, but that you also do all development on-site
at their santa clara HQ under their direct supervision.
I haven't went that far -- I'm not going to sign any NDA anytime
Andreas Dilger wrote:
David Balazic writes:
Andreas Dilger [EMAIL PROTECTED] wrote :
In the end I re-wrote most of the patch, so
that we resolve ROOT_DEV before calling mount_root(), just to be a bit
more consistent. I will release a new patch for 2.2.18 and 2.4.0 after
David
Tim Fletcher [EMAIL PROTECTED] wrote :
How does MD/RAID0 know which array should be /dev/md0? What if you had a
second array on /dev/hdb and /dev/hdd, would that become /dev/md0 (assuming
it had a kernel/boot sector)?
/etc/raidtab specifies which drives belong in which array, but
"copy this fd to that one, and optimize that if you can"
... isn't this Larry M's "splice" (http://www.bitmover.com/lm/papers/splice.ps)?
Andreas Dilger wrote:
Roger Wolff writes:
I'd prefer an interface that says "copy this fd to that one, and
optimize that if you can".
For example,
On 2001.01.17 Josh Myer wrote:
My guess is that JA's console player doesn't move around as much memory as
the gnome one (imagine that), therefore produces less memory noise.
No, that is the problem. It is just the same player, just launched from
the command line (also opens his gnome
On Thu, 18 Jan 2001, Andre Hedrick wrote:
I can get any info needed, you just have to define the scope.
Good.
Then will not can and will not give out details on a generic form.
Weird. Others somehow are able to provide specs. Documentation for the
entire line of Intel chipsets is
On Wed, 17 Jan 2001, Derek Wildstar wrote:
With 2.4.0 thru 2.4.1-pre8 (could possibly be sooner than 2.4.0)
PCMCIA_CONFIG_NETCARD is getting defined with CONFIG_PCMCIA, even when no
PCMCIA net cards are selected:
458 # PCMCIA network device support
459 #
460
How does MD/RAID0 know which array should be /dev/md0? What if you had a
second array on /dev/hdb and /dev/hdd, would that become /dev/md0 (assuming
it had a kernel/boot sector)?
/etc/raidtab specifies which drives belong in which array, but I only have
hda and hdc so I can't
[EMAIL PROTECTED] wrote:
In article 9463fj$gsq$[EMAIL PROTECTED] you write:
Basically, the _only_ think you should depend on is that st_size
contains:
- for regular files, the size of the file in bytes
- for symlinks, the length of the symlink.
I don't think this is right - for a
[Rogier Wolff]
I'd prefer an interface that says "copy this fd to that one, and
optimize that if you can".
So do exactly that in libc.
sendfile () {
if (sys_sendfile() == -1)
return (errno == EINVAL) ? do_slow_sendfile() : -1;
return 0;
}
Peter
-
To unsubscribe from this
Hi,
On Thu, Jan 18, 2001 at 01:10:11PM +0100, Simon Huggins wrote:
We have a server running 2.2.18 + RAID which has an eepro100 in it.
It's connected to a Dlink DFE 816 100 16port 100baseTX hub.
When the machine boots we get a whole series of timeout errors.
Jan 18 11:58:09 miguet
On Mon, 15 Jan 2001, H. Peter Anvin wrote:
"Albert D. Cahalan" wrote:
It looks like we let Microsoft fill the design guide void.
If you were to write "PC DESIGN GUIDE - For the Linux Operating
System" and a pile of test code, then there would be an
alternative to point people at.
Trond Myklebust wrote:
" " == Mogens Kjaer [EMAIL PROTECTED] writes:
getdents64(3, /* 6 entries */, 65536) = 160 lseek(3,
1547825467, SEEK_SET) = 1547825467 ... getdents64(3, /* 1
entries */, 65536) = 32
BTW: there does in any case seem to be a bug in your version of
Hiya Andrey,
On Thu, Jan 18, 2001 at 08:17:31PM +0800, Andrey Savochkin wrote:
On Thu, Jan 18, 2001 at 01:10:11PM +0100, Simon Huggins wrote:
We have a server running 2.2.18 + RAID which has an eepro100 in it.
It's connected to a Dlink DFE 816 100 16port 100baseTX hub.
When the machine
Thus spake Eric S. Raymond ([EMAIL PROTECTED]):
Here is what I think I know about stat(2) that isn't in the
Linux man pages:
* For a symlink (S_IFLNK) it reports the size of the link file, not the
size of the file the link points to.
I think you confuse stat and lstat here.
Felix
-
To
On Wed, 17 Jan 2001, dean gaudet wrote:
with the linux TCP_CORK API you only get one trailing small packet.
in case you haven't heard of TCP_CORK -- when the cork is set, the
kernel is free to send any maximum size packets it can form, but has
to hold on to the stragglers until userland
On 18 Jan 2001 11:35:57 +, Tim Fletcher wrote:
This is when devfs comes into its own, as the disks are refered to by
their device/controller id not by the /dev/sd{a,b,c,etc} numbering, hence
when one fails the others don't change. Also I think the kernel autodetect
code for scsi devices
On Thu, Jan 18, 2001 at 01:42:50PM +0100, Simon Huggins wrote:
On Thu, Jan 18, 2001 at 08:17:31PM +0800, Andrey Savochkin wrote:
[snip]
These two line are a workaround for the RxAddrLoad timing bug,
developed by Donald Becker. wait_for_cmd_done timeouts may be related
to this bug, too.
On Wed, 17 Jan 2001, Rick Jones wrote:
i'd heard interesting generalities but no specifics. for instance,
when the send is small, does TCP wait exclusively for the app to
flush, or is there an "if all else fails" sort of timer running?
yes there is a per-socket timer for this. According to
[James Bottomley]
The fundamental problem that we all agree on is that SCSI devices are
detected in the order that the mid-layer hosts.c file calls their
detect routines.
That was yesterday. Today they are detected in the order they are
linked into the kernel, cf. the Makefile. But yes, the
On Wed, 17 Jan 2001, Linus Torvalds wrote:
(I also had one person point out that BSD's have the notion of
TCP_NOPUSH, which does almost what TCP_CORK does under Linux, except
it doesn't seem to have the notion of uncorking - you can turn NOPUSH
off, but apparently it doesn't affect queued
sorry for reposting this message If some one has answered it, please
send me again...
I have lost some of the mails from the list
archan
-
I have compiled 2.4.0-test11 kernel for my Toshiba Laptop. Everything
works fine
Hello, all!
I am having some problems with de delivery of the USB irq.
Without the kernel-option "noapic", it seems, they just never make it to the
kernel.
I tried running with mps=1.1, mps=1.4, PNP OS=y/n, but that doesn't help...
I am also running windoze 2000 on this machine, and that sets
On Wed, 17 Jan 2001, Rick Jones wrote:
certainly, i see by your examples how cork can make life easier on the
developer - they can putc() the reply if they want. for a persistent
http connection, there would be the cork and uncork each time, for a
pipelined connection, it is basically a
Jakub Jelinek wrote:
This makes me wonder...
If the kernel only kept a queue of the three smallest unused fd's, and
when the queue emptied handed out whatever it liked, how many things
would break? I suspect this would cover a lot of bases...
First it would break Unix98 and
This is when devfs comes into its own, as the disks are refered to by
their device/controller id not by the /dev/sd{a,b,c,etc} numbering, hence
when one fails the others don't change. Also I think the kernel autodetect
code for scsi devices will deal with this case, but I'm not sure.
James Simmons wrote:
...
By you saying couldn't be acquired from interrupt context do you mean
from a process context or do you mean it failed to aquire it while in
the interrupt context?
Actually, printk() must always use __down_trylock().
- Get rid of console_tasklet. Do it in process
On Wed, 17 Jan 2001, Josh Myer wrote:
You're probably just hearing electrical noise from the memory buss; try
moving the 128 stick to the slot you put the 64 in (if you can) and run
that way -- you'll probably hear it there too.
i've exchanged the two memory slot, and the problem is still
Hi
when i use a PF_PACKET / raw socket when i read data from the
socket everything is fine apart from udp packets which are going to
the machine on port 2049 sometimes the checksums fail i belive that
the data might be getting changed on the udp pakcet somewhere inside the
kernel as with the
Good morning,
Perhaps this isn't the correct list to ask this, but
is there a site, does any one have, know of the existance
of a driver for the Efficient 3060 DSL modem? Even if the
driver in question is in beta, development, whatever, I
don't really care.
Thanks for your help in
Sorry if this has been asked of you before, but what happens if you put just
the 64 meg module in?
Bug Report
--
1. 128M memory OK, but with 192M the sound card generate a noise while
use the DSP.
2. i got the problem when I just put more 64M memory to the my machine.
With 128M the
On Thu, Jan 18, 2001 at 02:06:46PM +0100, Ingo Molnar wrote:
On Wed, 17 Jan 2001, Rick Jones wrote:
i'd heard interesting generalities but no specifics. for instance,
when the send is small, does TCP wait exclusively for the app to
flush, or is there an "if all else fails" sort of
Dear folks,
I use rsync to transfer my mail (including this list) from work to home
over ppp ussing OpenSSH 2.3.0. I have no problem transfering hundreds
of megabytes of my babies' photos from a non-raid partition (going to
work), but I get:
nsmail/Inbox
Write failed: Cannot allocate memory
On Thu, 18 Jan 2001, Mike Dresser wrote:
Sorry if this has been asked of you before, but what happens if you put just
the 64 meg module in?
Ok. i've not analised that possibility. I've tested that just now.
But, the system work perfect. Without the sound card problem.
I've exchanged the
On Thu, 18 Jan 2001, Stefan Ring wrote:
On Wed, 17 Jan 2001, Joel Franco Guzmn wrote:
With 128M the problem is not present, but with 192M it is. The only
difference is the memory quantity, or in other words, the additional slot
occupied by the new memory card.
- ASUS P299 (Chipset
This patch enables performance counters if RDTSC is not disabled (ie. it
assumes user wants to enable RDPMC if he wants to enable RDTSC too).
It is against 2.4.0.
To save me from typing, I'll quote Intel IA-32 manuals:
"PCEPerformance-Monitoring Counter Enable (bit 8 of CR4). Enables
On Thu, 18 Jan 2001, Nick Urbanik wrote:
Dear folks,
I use rsync to transfer my mail (including this list) from work to home
over ppp ussing OpenSSH 2.3.0. I have no problem transfering hundreds
of megabytes of my babies' photos from a non-raid partition (going to
work), but I get:
In article [EMAIL PROTECTED] you write:
Nope stat should return the details of the symlink
whereas lstat should return the details of the symlink target.
Not according to my manpages. From stat(2):
stat stats the file pointed to by file_name and fills in
buf.
lstat
I know that increasing the number of DIMMs on your board will require
speedier RAMs on ASUS boards with some sort of an i440 chipset. This may
well be the case for just about every other MB, it's only that I don't
know specifically about these other boards.
This means that a modules
Hi Urban.
I'm connecting to a Win2K server (same share between computers). A
slight change in my previous post: my dual P2 w/320MB appears to be
running 2.4.0-ac9 instead of 2.4.1-pre8. The bigmem machine (1gb mem)
had the oops on 2.4.0-ac9 but before reporting I thought I'd try
Hi all,
Testing a 2.4 kernel I have found something I think it's rather strange
Running a 2.4.0 kernel with APM enabled (not ACPI), it takes about ten minutes
to compile a kernel (Pentium III Coppermine 450 Mhz 128 MB)
[eugenio@gic12 linux]$ time make dep clean bzImage modules /dev/null
real
" " == Mogens Kjaer [EMAIL PROTECTED] writes:
This turned out to be more difficult than I thought...
I suspect glibc-2.2-12 being the reason, but I'm not quite sure
yet:
The problem is, that the 64-bit dirent's are converted to
32-bit dirent's and a sanity check
John O'Donnell wrote:
Roeland Th. Jansen wrote:
On Wed, Jan 17, 2001 at 03:11:52AM -0500, John O'Donnell wrote:
Please tell me I just didn't just see this message??!?!?!?!
Please??!?!?!? What are you doing?
I mean no one person here any disrespect - please do the same.
you just
When running 'quotaon -a' I get message 'quotaon: using /quota.user on
/dev/hdc3: Invalid argument'. This occurs at least with 2.4.0-ac4 and
2.4.0-ac9, but not with 2.4.0.
That's because 2.4.0-ac? supports new quota format which is incompatible
with the old one... You need also new
This morning my dual P2/300 with 320MB memory gave me:
NMI Watchdog detected LOCKUP on CPU0, registers:
CPU:0
EPI:0023:[08076580]
EFLAGS: 0292
eax: fb54ba8e ebx: bba692c5 ecx: b6fb4d53 edx: 562a18f5
esi: 216266e2 edi: 92811b25 ebp: 27cd9569 esp: 080f0148
ds: 002b es:
On Thu, 18 Jan 2001, Andreas Dilger wrote:
Actually, this is a great example, because at one point I was working
on a device interface which would offload all of the disk-disk copying
overhead to the disks themselves, and not involve the CPU/RAM at all.
It's a horrible example.
Although fs/udf's args to writepage() were updated in 2.4.0-test12,
its page unlocking was overlooked. udf_adinicb_writepage() should
now UnlockPage, udf_expand_file_adinicb() should not now UnlockPage
after udf_writepage i.e. block_write_full_page. Al Viro posted a
patch for the latter, still
All,
Here is a patch that adds the following to the pcnet32.c driver:
- adds dma_cache_*() calls to support ARM processors. Based on
Russell's changes to the tulip driver.
- add DXSUFLO to the AM79C973
- made DXSUFLO the default
- utilized the internal SRAM of the AM79C973
- added a 32 bit
On Thu, 18 Jan 2001, Ingo Molnar wrote:
[ BSD's TCP_NOPUSH ]
this is what MSG_MORE does.Basically i added MSG_MORE for the purpose of
getting perfect TUX packet boundaries (and was ignorant enough to not know
about BSD's NOPUSH), without an additional system-call overhead, and
without
On Thu, 18 Jan 2001, Ingo Molnar wrote:
Basically MSG_MORE is a simplified probability distribution of the next
SEND, and it already covers all the other (iovec, nagle, TCP_CORK)
mechanizm available, in a surprisingly easy way IMO. I believe MSG_MORE is
very robust from a theoreticaly
Trond Myklebust wrote:
It comes from the SGI. The NFS client just considers it all a cookie,
and passes it on to glibc. We probably shouldn't do that, as indeed
the cookie is not guaranteed to be 32-bit signed, but it's what we
always did for 2.2.x.
So what do I do to get it to work?
I
Matti Aarnio ([EMAIL PROTECTED]) wrote :
On Wed, Jan 17, 2001 at 08:22:22PM +0100, Werner Almesberger wrote:
The only cases when you really need to know the name of a disk is when
- doing disk-level management, e.g. partitioning or creating file
systems (*)
- adding a
On Thu, 18 Jan 2001, Linus Torvalds wrote:
Yeah, and how are you going to teach a perl CGI script that writes to
stdout to use it?
yep, correct. But you can have TCP_CORK behavior from user-space (by
setting the cork flag in user-space and writing it for all network
output), while you cannot
Just to follow up, on Vojtech's advice I installed kernel 2.4.0 instead(I
also moved to 7.0 RH to accomplish this) and everything seems to be working
fine(I did not apply your new patches Vojtech, as yet they have not proved
necessary). I have not done any exhaustive testing, however the failure
On Thu, 18 Jan 2001, Tuukka Toivonen wrote:
I suppose that if the CPU has MMX, then it's ok to write 1 in PCE
bit. This should be valid for Intel CPUs, but it's certainly possible that
there are non-Intel CPUs that have MMX but not RDPMC. However, I'd guess
that writing 1 to PCE bit doesn't
On Thu, 18 Jan 2001, Linus Torvalds wrote:
Remember the UNIX philosophy: everything is a file. MSG_MORE
completely breaks that, because the only way to use it is with
send[msg](). It's absolutely unusable with something like a
traditional UNIX "anonymous" application that doesn't know or
Ok, I tried the NFS mount command again but without dnetc running.
(BTW, I'm running rpm -Uhv to update some packages from another linux
machine; I download the updates on it and all other machines load the
updates from it.)
This time I got a stack dump to run through ksymoops:
ksymoops
On Thu, Jan 18, 2001 at 08:49:38AM -0800, Linus Torvalds wrote:
That has its advantages: it's a very local thing, and doesn't need any
state. However, the fact is that you _need_ the persistency of a socket
option if you want to take advantage of external programs etc getting good
behaviour
On Wed, Jan 17, 2001 at 05:13:31PM -0800, Linus Torvalds wrote:
On Thu, 18 Jan 2001, Christoph Hellwig wrote:
/*
* a simple page,offset,legth tuple like Linus wants it
*/
struct kiobuf2 {
struct page * page; /* The page itself */
u_int16_t
On Thu, 18 Jan 2001, Maciej W. Rozycki wrote:
Weird. Others somehow are able to provide specs. Documentation for the
entire line of Intel chipsets is available, for example.
This because the make chipsets to basically give away to sell processors.
That should be very obvious, they they
On Thu, 18 Jan 2001, Christoph Hellwig wrote:
Remove this. I don't think it's valid to lock the pages. Who wants to use
this anyway?
E.g. in the block IO pathes the pages have to be locked.
It's also used by free_kiovec to see wether to do unlock_kiovec before.
This is all MUCH
On Thu, Jan 18, 2001 at 02:04:08AM -0200, Rogerio Brito wrote:
On Jan 17 2001, David D.W. Downey wrote:
Could those that were involved in the VIA chipset discussion email me
privately at [EMAIL PROTECTED]?
Just to add a datapoint to the discussion, I'm using a VIA
chipset
On Thu, 18 Jan 2001, Stefan Ring wrote:
I know that increasing the number of DIMMs on your board will require
speedier RAMs on ASUS boards with some sort of an i440 chipset. This may
well be the case for just about every other MB, it's only that I don't
know specifically about these
Andi Kleen wrote:
On Wed, Jan 17, 2001 at 02:17:36PM -0800, Rick Jones wrote:
How does CORKing interact with ACK generation? In particular how it
might interact with (or rather possibly induce) standalone ACKs?
It doesn't change the ACK generation. If your cork'ed packets gets sent
Ingo Molnar wrote:
On Wed, 17 Jan 2001, Rick Jones wrote:
i'd heard interesting generalities but no specifics. for instance,
when the send is small, does TCP wait exclusively for the app to
flush, or is there an "if all else fails" sort of timer running?
yes there is a per-socket
Linus Torvalds wrote:
Remember the UNIX philosophy: everything is a file.
...and a file is simply a stream of bytes (iirc?)
rick jones
--
ftp://ftp.cup.hp.com/dist/networking/misc/rachel/
these opinions are mine, all mine; HP might not want them anyway... :)
feel free to email, OR post, but
Eli Carter wrote:
Here is a patch that adds the following to the pcnet32.c driver:
[snip]
- According to the Am79C973/Am79C975 docs from AMD, The collision bits
are only valid if ENP is set, so I added a check for that.
[snip]
@@ -1164,7 +1206,8 @@
}
#endif
On Thu, 18 Jan 2001, Rick Jones wrote:
Linus Torvalds wrote:
Remember the UNIX philosophy: everything is a file.
...and a file is simply a stream of bytes (iirc?)
Indeed.
And normal applications really shouldn't need to worry about things like
packetization etc.
Of course, many
Hi,
On my dbench runs I've noted a slowdown between pre4 and pre8 with 48
threads. (128MB, 2 CPU's machine)
pre4:
Throughput 7.05841 MB/sec (NB=8.82301 MB/sec 70.5841 MBit/sec)
70.94user 232.54system 15:17.39elapsed 33%CPU (0avgtext+0avgdata
0maxresident)k 0inputs+0outputs
Werner Almesberger [EMAIL PROTECTED] writes:
I agree writing the code to understand the table may be a significant
issue. On the other hand I still think it is worth a look, being
able to unify option parsing for multiple platforms is not a small
gain, nor is getting out from short
On Thu, 18 Jan 2001, Rik van Riel wrote:
On Fri, 12 Jan 2001, Vlad Bolkhovitine wrote:
You can see, mmap() read performance dropped significantly as
well as read() one raised. Plus, "interactivity" of 2.4.0 system
was much worse during mmap'ed test, than using read()
(everything was
Hi,
On Thu, 18 Jan 2001, Linus Torvalds wrote:
Actually, this is a great example, because at one point I was working
on a device interface which would offload all of the disk-disk copying
overhead to the disks themselves, and not involve the CPU/RAM at all.
It's a horrible example.
On Thu, 18 Jan 2001, Roman Zippel wrote:
Not going to happen.
device-to-device is not the same as disk-to-disk. A better example would
be a streaming file server.
No, it wouldn't be.
[ Crystal ball mode: ON ]
It's too damn device-dependent, and it's not worth it. There's no way to
On Thu, Jan 18, 2001 at 08:49:38AM -0800, Linus Torvalds wrote:
state. However, the fact is that you _need_ the persistency of a socket
option if you want to take advantage of external programs etc getting good
behaviour without having to know that they are talking to a socket.
I'm all for
Nope stat should return the details of the symlink
whereas lstat should return the details of the symlink target.
It's the other way around according to the manpage, and my code also says
it's the other way around.
It's logical the way it is..
I use lstat to check if a config file is a
On Thu, Jan 18, 2001 at 10:20:16AM -0800, Rick Jones wrote:
Andi Kleen wrote:
On Wed, Jan 17, 2001 at 02:17:36PM -0800, Rick Jones wrote:
How does CORKing interact with ACK generation? In particular how it
might interact with (or rather possibly induce) standalone ACKs?
It
On Thu, 18 Jan 2001, Andrea Arcangeli wrote:
I'm all for TCP_CORK but it has the disavantage of two syscalls for doing the
flush of the outgoing queue to the network. And one of those two syscalls is
spurious. Certainly it makes perfect sense that the uncork flushes the outgoing
queue,
What is the difference between physical and logical partitions ?
primary (what you call physical) partitions are partitions in their own
right logical partitions are partitions within a special partition
How does this solve the "I deleted hda5 and now the old hda6 became
hda5" problem ?
It
device-to-device is not the same as disk-to-disk. A better example would
be a streaming file server. Slowly the pci bus becomes a bottleneck, why
would you want to move the data twice over the pci bus if once is enough
and the data very likely not needed afterwards? Sure you can use a more
Hello!
I'm all for TCP_CORK but it has the disavantage of two syscalls for doing the
MSG_MORE was invented to allow to collapse this to 0 of syscalls. 8)
A new ioctl on the socket should be able to do that (and ioctl looks ligther
than a setsockopt, ok ignoring actually the VFS is grabbing
On Thu, 18 Jan 2001, Marcelo Tosatti wrote:
On my dbench runs I've noted a slowdown between pre4 and pre8 with 48
threads. (128MB, 2 CPU's machine)
Below some kernel compile numbers on a 32 MB RAM + 32 MB swap box. The
three lines mean compilation with the -j1, -j2 and -j4 option. Most of
Hi,
A patch for the via82cxxx_audio sound driver against 2.4.1-pre8.
It includes:
1. Support for variable fragment size and variable fragment number
2. Fixes for the SPEED, STEREO, CHANNELS, FMT ioctls when in read write
mode
2.1 Mmaped sound is now fully functional.
The patch is fairly
Hello!
Doing PUSH from setsockopt(TCP_CORK) looked obviously wrong because it isn't
setting any socket state,
? 8)
and also because the SIOCPUSH has nothing specific
with TCP_CORK, as said it can be useful also to flush the last fragment of data
pending in the send queue without having
On Thu, 18 Jan 2001, Kai Germaschewski wrote:
This looks more like a counterexample to what you're saying. Also, I don't
see how it could happen that CONFIG_PCMCIA_NETCARD=y without
CONFIG_NET_PCMCIA=y, from looking at the drivers/net/pcmcia/Config.in.
(It may be still possible somehow
1 - 100 of 433 matches
Mail list logo