> >
> > When you say it reloads it's VM, you mean it reloads the CR3 register?
>
> Yes.
>
> No. In 2.4 you could probably use the on demand lazy vm mechanism ingo described for
> the nfsd processes. In 2.2 it is a bit more tricky, if I remember right lazy mm
>needed
> quite a few changes.
>
On Mon, Oct 30, 2000 at 09:39:37AM +0100, Ingo Molnar wrote:
>
> On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
>
> > Is there an option to map Linux into a flat address space [...]
>
> nope, Linux is fundamentally multitasked.
>
> what you can do to hack around this is to not switch to the idle t
On Mon, Oct 30, 2000 at 01:04:34AM -0700, Jeff V. Merkey wrote:
> > >
> > > When you say it reloads it's VM, you mean it reloads the CR3 register?
> >
> > Yes.
> >
> > No. In 2.4 you could probably use the on demand lazy vm mechanism ingo described
>for
> > the nfsd processes. In 2.2 it is a
I'm posting this hoping that someone will fix this soon ...
On Sat, 28 Oct 2000, Jeff Chua wrote:
> freeram /dev/ram0
> File freeram.c: Line 41: ioctl /dev/ram0: Error 16: Device or resource
busy
this should not happen.
> I cannot free the old initrd root partition under 2.4.0-testx, but it
wo
> >
> > I just guess the end result will be as crash prone as Netware when you install any
>third
> > party software ;)
> >
> > lazy mm is probably a better path, as long as you stay in kernel threads and a
>single user mm
> > it'll never switch VMs.
>
> It's not that bad. Some code has bee
On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
> [...] All protection has to go away in all LAN paths for this to
> happen, and user space apps set to ring 0. [...]
i found that this is not a requirement for good network scalability. We do
not do a syscall for every packet, so the cost evens out. S
On Mon, Oct 30, 2000 at 10:52:08AM +0100, Ingo Molnar wrote:
>
> On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
>
> > [...] All protection has to go away in all LAN paths for this to
> > happen, and user space apps set to ring 0. [...]
>
> i found that this is not a requirement for good network sca
On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
> No argument here, but the overhead of reloading CR3 period will kill
> performance. [...]
2.4 does not reload CR3, unless you are using multiple user-space
processes.
> 2.4 does not beat NetWare, BTW, it gets a little further, but still
> hits the w
Ages ago, I modularized the partition parsing code and
posted the patch to linux-kernel. I received a few suggestions
for some small improvements, and I confess it took me a while
to get around to producing a new patch. Anyhow, here is the
new partitioning modularization patch. It now l
On Mon, Oct 30, 2000 at 11:13:58AM +0100, Ingo Molnar wrote:
>
> On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
>
> > No argument here, but the overhead of reloading CR3 period will kill
> > performance. [...]
>
> 2.4 does not reload CR3, unless you are using multiple user-space
> processes.
>
> >
On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
> For example, if you put a MOV EAX, CR3; MOV CR3, EAX; in a context
> switching path, on a PPro 200, you can do about 35,000 context
> switches/second
in 2.4 & Xeons we can do more than 100,000 context switches/second, and
that is more than enough. B
On Mon, Oct 30, 2000 at 11:27:04AM +0100, Ingo Molnar wrote:
>
> On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
>
> > For example, if you put a MOV EAX, CR3; MOV CR3, EAX; in a context
> > switching path, on a PPro 200, you can do about 35,000 context
> > switches/second
>
> in 2.4 & Xeons we can
Jeff Garzik wrote:
>
> pavel rabel wrote:
> > help. So I removed PCI code from ne.c to have ISA only driver. It
>
> This change sounds ok to me, if noone else objects. (I added to the CC
> a bit) I saw that code, and was thinking about doing the same thing
> myself. ne2k-pci.c definitely has
On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
> [...] you've got the web server covered. What about file and print.
a web server, as you probably know, is a read-mostly fileserver that
serves files via the HTTP protocol. The rest is only protocol fluff.
> I think this is great, but most web serv
On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
> This is putrid. NetWare does 353,00,000/second on a Xenon, pumping out
> gobs of packets in between them. MANOS does 857,000,000/second. This
> is terrible. No wonder it's so f_cking slow!!!
(no need to get emotional.) And please check your numbers,
On Mon, Oct 30, 2000 at 11:41:35AM +0100, Ingo Molnar wrote:
>
> On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
>
> > [...] you've got the web server covered. What about file and print.
>
> a web server, as you probably know, is a read-mostly fileserver that
> serves files via the HTTP protocol. T
On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
> ds: and es: are both used in copy-to-user and copy-from-user and they
> get reloaded.
And they all share the same segment descriptor. Whats your point? ES is
the default target segment for string operations. DS is the default data
segment. Have you e
On Mon, Oct 30, 2000 at 11:44:26AM +0100, Ingo Molnar wrote:
>
> On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
>
> > This is putrid. NetWare does 353,00,000/second on a Xenon, pumping out
> > gobs of packets in between them. MANOS does 857,000,000/second. This
> > is terrible. No wonder it's so f_c
On Mon, Oct 30, 2000 at 11:50:24AM +0100, Ingo Molnar wrote:
>
> On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
>
> > ds: and es: are both used in copy-to-user and copy-from-user and they
> > get reloaded.
>
> And they all share the same segment descriptor. Whats your point? ES is
> the default tar
On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
> > reads dominate writes in almost all workloads, thats common wisdom. Why
> > write if nobody reads the data? And while web servers are mostly read only
> > data, they can write data as well, see POST and PUT. The fact that
> > incoming writes are har
On Mon, Oct 30, 2000 at 11:56:06AM +0100, Ingo Molnar wrote:
>
> On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
>
> > > reads dominate writes in almost all workloads, thats common wisdom. Why
> > > write if nobody reads the data? And while web servers are mostly read only
> > > data, they can write
On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
> > And please check your numbers, 857 million
> > context switches per second means that on a 1 GHZ CPU you do one context
> > switch per 1.16 clock cycles. Wow!
>
> Excuse me, 857,000,000 instructions executed and 460,000,000 context switches
> a sec
On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
> It's not curious, it's not about bandwidth, it's about latency, and
> getting packets in and out of the server as fast as possible, and
> ahead of everything else. [...]
TUX prepares a HTTP reply in about 30 microseconds (plus network latency),
good
On Mon, Oct 30, 2000 at 12:01:08PM +0100, Ingo Molnar wrote:
>
> On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
>
> > > And please check your numbers, 857 million
> > > context switches per second means that on a 1 GHZ CPU you do one context
> > > switch per 1.16 clock cycles. Wow!
> >
> > Excuse m
On Mon, Oct 30, 2000 at 12:04:43PM +0100, Ingo Molnar wrote:
>
> On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
>
> > It's not curious, it's not about bandwidth, it's about latency, and
> > getting packets in and out of the server as fast as possible, and
> > ahead of everything else. [...]
>
> TUX
On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
> > > Excuse me, 857,000,000 instructions executed and 460,000,000 context switches
> > > a second -- on a PII system at 350 Mhz. [...]
> Go download it and try it, then come back with that smirk wiped off
> your face. I'll enjoy it.
so in 0.53 cl
On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
> > TUX prepares a HTTP reply in about 30 microseconds (plus network latency),
> > good enough? Network latency is the limit, even on gigabit - not to talk
> > about T1 lines.
>
> Great. Now how do we get the smae numbers on SAMBA and MARS-NWE? [...]
On Mon, Oct 30, 2000 at 12:12:44PM +0100, Ingo Molnar wrote:
>
> On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
>
> > > > Excuse me, 857,000,000 instructions executed and 460,000,000 context switches
> > > > a second -- on a PII system at 350 Mhz. [...]
>
> > Go download it and try it, then come ba
On Mon, Oct 30, 2000 at 12:13:52PM +0100, Ingo Molnar wrote:
>
> On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
>
> > > TUX prepares a HTTP reply in about 30 microseconds (plus network latency),
> > > good enough? Network latency is the limit, even on gigabit - not to talk
> > > about T1 lines.
> >
I am trying to determine how I should try to fix the
problem of my notebook computer not receiving USB interrupts.
I seem to have gotten it to sort of work by kludging to the USB UHCI
driver's millisecond timer routine to also invoke the interrupt
handler, but I hope I do not have to resor
On Mon, 30 Oct 2000, Jeff V. Merkey wrote:
> Ingo, This original thread was regarding Linux vs. NetWare 5.x
> performance metrics and responses from Linux folks about how to affect
> and improve them, not a diatribe on the features of TUX.
oh, i believe you misunderstand. TUX itself is quite si
Alan,
student:/susr/ /usr
nfs defaults,rsize=1024,wsize=1024
0 0
Thanks
Graham Jordan
Alan Cox wrote:
> By the evidence that we have gathered it seems
that the Server is not
> taxed too much as samba users are getting files OK etc. The
can't get
> request slo
ray_cs was imported to the kernel as one of the first test drivers by
Linus. At the time, he wanted to get rid of the CardServices(function...)
interface and replace it with pcmcia_function calls. It seems that
the correct fix would be to change pcmcia_request_irq to RequestIRQ
like all the oth
Hi :)
I'd like to report some strange things happening to me since
using kernel 2.2.17 (2.2.15-4mdk was better instead).
*** MOST STRANGE ***
The following are strange behaviours, because they only happen
"sometimes" when booting or shutting down!
Sorry, I was not able to determine *when* they
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
SUBTERFUGUE 0.1.99 is available. It's been updated to work with the latest
kernel (linux-2.4.0-test9), and there are new tricks, documentation, and a
Debian package. (see below)
As always, feedback is welcome.
Have fun!
- --Mike
On Mon, Oct 30, 2000 at 03:06:25AM -0700, Jeff V. Merkey wrote:
> Ingo, This original thread was regarding Linux vs. NetWare 5.x performance
> metrics and responses from Linux folks about how to affect and
> improve them, not a diatribe on the features of TUX.
while beating netware in certain a
[rth]
> > Which was a nice idea, but it doesn't actually work. Changes
> > in spec file format between versions makes this fall over.
[Dominik Kubla]
> Wow. So much for reading the manual... well, that's considered
> cheating anyway, isn't it?
I know this was true at one time -- egcs couldn'
[Rusty]
> > CC=gcc-2723 make 2.0 kernel
> > CC=gcc-2723 make 2.2 kernel
> > CC=egcs make 2.4 kernel
>
> No, environment doesn't override make variables by default. This
> works on any shell:
>
> make CC=egcs
If you're going to get pedantic, that won't work either -- since the
makefiles
Hello,
I experienced problems in programs which used priorities under FIFO-Scheduling
with Kernel 2.2 although these programs worked well with Kernel 2.0.
I read in this list that other people had the same problems.
Is any solution (a newer kernel version or a special patch) available for tha
On Mon, 30 Oct 2000, Adam J. Richter wrote:
> My question is: could I get a slightly more detailed
> explanation of what exactly the quirk_piix3usb routine was
> trying to fix so I can better understand if I am bumping into
> the same problem? Do I understand correctly that the piix3 fixup
Hi all,
After a delay on some other project I've again started up the process of
fixing up the Linux headers, i.e. removing the use of C++ keywords as
variable and stuff. I have questions on the use of three datastructures
which happen to use the C++ keyword new but first a couple of things.
So which is the recommended compiler for each kernel version 2.2.x,
2.4.x(pre?) nowadays? I've pretty much kept gcc 2.7.2.3 around just for
compiling the kernel however now I hear you need egcs to compile 2.4? I
don't mind keeping 2.7.2.3 around in its own installation directory just for
the pur
On Fri, Oct 27, 2000 at 02:23:04PM -0700, Linus Torvalds wrote:
>
> [...]
>
> That solution, btw, might be as simple as just saying:
>
> - raw IO is based on physical pages, and the COW mapping crated by
>fork() may cause the changes to be visibile to either child or parent
>or both, dep
> So which is the recommended compiler for each kernel version 2.2.x,
> 2.4.x(pre?) nowadays?
* 2.91.66 aka egcs 1.1.2. It has been officially blessed for 2.4 and
has been given an informal thumbs-up by Alan for 2.2. (It does NOT
work for 2.0, if you still care about that.)
* 2.7.2.3 work
On Mon, 30 Oct 2000 12:09:49 Linux Kernel Developer wrote:
> The goal of this project is to clean up the kernel headers so as they are
> useable/compatible with those who wish to program their kernel modules in
> C++. It is not my goal to rewrite the kernel in C++ or anything like that,
Fist of
Le 30-Oct-2000, Mohammad A. Haque écrivait :
> Could someone who knows ide and drive inside and out (Andre?) please
> take a look at these figures? Am I forgetting to do something (or doing
> something I'm not suposed to) to get the best numbers? I thought I'd be
> able to get more than ~4MB/sec
Alan Cox wrote:
>
> > If the answer to this is "yes", then Microsoft should own some rights to
> > every piece of software that uses the Windows API.
>
> As US copyright law stands of the last few days Microsoft are entitled to
> require a magic constant is passed in one register to 'unloc
[EMAIL PROTECTED] (Eric W. Biederman) writes:
> The general consensus is that if we have a major reorganization, in proc
> the rule will be one value per file. And let directories do the grouping.
IIRC some time ago somebody suggested to rename 'proc' to something
like 'sys' or 'system', becaus
> I want my / to be a ramfs filesystem. I intend to populate it from an
> initrd image, and then remount / as the ramfs filesystem. Is that at
> all possible? The way I see it the kernel requires / on a device
> (major,minor) or nfs.
>
> Am I out of luck using ramfs as /? If it's easy to fix,
On Mon, 30 Oct 2000 13:00:06 +0100,
"J . A . Magallon" <[EMAIL PROTECTED]> wrote:
>And what about struct fields ? It is the same. If you change the name of a field
>permanently, you have to modify the C source that uses it. But names are not
>important for binary compatability, so you can make th
> We will never beat NetWare on scaling if this is the case, even in 2.4.
> Andre and my first job will be to create an arch port with MANOS that
> disables this and restructures the VM.
In the 2.4 case if you are just running NFS daemons then there are no tlb
reloads going on at all. Whats m
On Mon, Oct 30, 2000 at 12:47:10PM +, Alan Cox wrote:
> > We will never beat NetWare on scaling if this is the case, even in 2.4.
> > Andre and my first job will be to create an arch port with MANOS that
> > disables this and restructures the VM.
>
> In the 2.4 case if you are just runnin
Peter Samuelson wrote:
>
> > So which is the recommended compiler for each kernel version 2.2.x,
> > 2.4.x(pre?) nowadays?
>
> * 2.91.66 aka egcs 1.1.2. It has been officially blessed for 2.4 and
> has been given an informal thumbs-up by Alan for 2.2. (It does NOT
> work for 2.0, if you st
> one -- I've got extra licensed copies), install it, put a load of 5000
> connections on it, with 4 adapters. Dual boot Linux on it, and attempt
> the same with SAMBA or MARS-NWE, and watch it oink.
SAMBA and Mars-nwe are running user space thats why. They have flexibility,
protection and ca
> js_dev::new. My questions are basically this. If I update these data
> structure members' names along with the references to them in various C
> files in the kernel will all be happy in Linuxland. Can any external
That may well be a problem. Also the use of private.
> utilities be broken or
Horst von Brand wrote:
>
> Red Hat 7.0, i686, gcc-20001027 (Yes, I know. Just to flush out bugs on
> both sides).
>
> abs() is used at least in:
>
> arch/i386/kernel/time.c
> drivers/md/raid1.c
> drivers/sound/sb_ess.c
>
> gcc warns about use of a non-declared function each time.
>
> No defin
On Mon, Oct 30, 2000 at 03:01:16PM +0100, Martin Dalecki wrote:
> Horst von Brand wrote:
> >
> > Red Hat 7.0, i686, gcc-20001027 (Yes, I know. Just to flush out bugs on
> > both sides).
> >
> > abs() is used at least in:
> >
> > arch/i386/kernel/time.c
> > drivers/md/raid1.c
> > drivers/sound/s
On Mon, 30 Oct 2000 13:04:06 + (GMT),
Alan Cox <[EMAIL PROTECTED]> wrote:
>You may find that creating your own wrappers for these files that do
>
>extern "C" {
>#define new new_
>#define private private_
>#include
>#undef new
>#undef private
>}
>
>safer, since you won't break anything
It br
Linus,
Rusty Russell suggested I send the following patch to you. It redirects
to the log file those annoying dropping untracked packet messages. This
patch works agains any recent 2.4.0-test kernels.
diff -urN linux.new1/net/ipv4/netfilter/ip_nat_standalone.c
linux.new/net/ipv4/netfilter/ip_n
On Sun, Oct 29, 2000 at 09:01:21PM -0500, vwbug wrote:
> it look like you didn't add the lines to .config the stupid bttv drivers
> that's why mine didnt work with bttv card i have read the .config under
> video for linux and found no refernce to bttv???
I'm not sure if this is what you mean, bu
> >You may find that creating your own wrappers for these files that do
> >
> >extern "C" {
> >#define new new_
> >#define private private_
> >#include
> >#undef new
> >#undef private
> >}
> >
> >safer, since you won't break anything
>
> It breaks module symbol versions, see earlier mail to l-k.
On Sat, 28 Oct 2000, Alan Cox wrote:
> > I have a question - Why does the PLIP driver does consume so much CPU ?
> > I tried it today, and when i did ping -s 16000 dst_ip, the kernel consumed
> > about 50% of the CPU time ( /proc/cpuinfo and /proc/interrupts follow).
> > Any ideas ?
>
> It h
Pierre Etchemaite wrote:
>
> Le 30-Oct-2000, Mohammad A. Haque écrivait :
> > Could someone who knows ide and drive inside and out (Andre?) please
> > take a look at these figures? Am I forgetting to do something (or doing
> > something I'm not suposed to) to get the best numbers? I thought I'd b
On Mon, 30 Oct 2000 13:41:40 + (GMT),
Alan Cox <[EMAIL PROTECTED]> wrote:
>Keith Owens wrote
>> >You may find that creating your own wrappers for these files that do
>> >
>> >extern "C" {
>> >#define new new_
>> >#define private private_
>> >#include
>> >#undef new
>> >#undef private
>> >}
>
> As part of the 2.5 kbuild redesign, symbol versions will be completely
> redone. One of the things on my todo list is to detect this mismatch.
> There are some problems in doing that which I may or may not be able to
> overcome, but if the field names are different between C and C++ then I
> ca
> > It has to bang on the parallel port controller the hard way, there is no
> > useful hardware support on a basic parallel port for the kind of abuse needed
> > for PLIP
> >
> (sorry for the late reply)
> I used plip with kernel 1.2.8 and had no problem with it...The machines
> that I'm using
Linus,
This patch fixes the bogus value of the TracerPid field in /proc//status.
(I thought was patched several months back, but I guess it wasn't, or it got
mistakenly backed out.)
--Mike
--- fs/proc/array.c-distFri Sep 1 16:32:17 2000
+++ fs/proc/array.c Mon Oct 30 08:02:35 200
- Received message begins Here -
Jesse Pollard <[EMAIL PROTECTED]>:
> On Sun, 29 Oct 2000, Stephen Harris wrote:
> >Horst von Brand wrote:
> >
> >> > > If you send SIGSTOP to syslogd on a Red Hat 6.2 system (glibc 2.1.3,
> >> > > kernel 2.2.x), within a few minutes you will find
On Mon, 30 Oct 2000 14:02:38 + (GMT),
Alan Cox <[EMAIL PROTECTED]> wrote:
>> As part of the 2.5 kbuild redesign, symbol versions will be completely
>> redone. One of the things on my todo list is to detect this mismatch.
>> There are some problems in doing that which I may or may not be able
Hello,
We are trying to port a network driver from 2.2.x to 2.4.x and have some
question regarding locks.
According to the kernel locking HOWTO, we have to take extra care when
locking between user context threads and BH/tasklet/softIRQ,
so we learned (the hard way ;-) that when running the ioctl
Stephen Harris <[EMAIL PROTECTED]>:
> > It was NOT ignored. If syslogd dies, then the system SHOULD stop, after a
>
> Huh? "SHOULD"? Why? If syslog dies for any reason (bug, DOS, hack,
> admin stupidity) then I sure don't want the system freezing up.
Should because this is the only audit log
I've recently updated the SysRQ registration patch, which is still available at:
http://bama.ua.edu/~dunna001/sysrq-register/
to have a 2.4.0-test10-pre6 incarnation. I've also added a more complex module
example, showing some of the things that this patch makes possible, that is
available at the
On Mon, 30 Oct 2000, Paul Gortmaker wrote:
> There is no urgency in trying to squeeze a patch like this in the back
> door of a 2.4.0 release. For example, there are people out there now
> who are using the ne.c driver to run both ISA and PCI cards in the same
> box without having to use 2 di
On Sat, Oct 28, 2000 at 11:50:22AM -0400, Brian Gerst wrote:
> Philipp Rumpf wrote:
> >
> > On Sat, Oct 28, 2000 at 09:55:21AM -0400, Brian Gerst wrote:
> > > Yes, but they can be called (and sleep) with module refcount == 0. This
> > > is because the file descripter used to perform the ioctl is
Jakub Jelinek wrote:
>
> On Mon, Oct 30, 2000 at 03:01:16PM +0100, Martin Dalecki wrote:
> > Horst von Brand wrote:
> > >
> > > Red Hat 7.0, i686, gcc-20001027 (Yes, I know. Just to flush out bugs on
> > > both sides).
> > >
> > > abs() is used at least in:
> > >
> > > arch/i386/kernel/time.c
> >
On Mon, Oct 30, 2000 at 11:36:10AM +, Mark Cooke wrote:
> As a followup - I modified your reset patch to work with 2.2.x, and in
> doing so, I noticed that it is only the slow get time of day (not
> using TSC) that is affected. I have my kernel compiled with TSC, so
> my apparent clock jump
On Mon, 30 Oct 2000, Vojtech Pavlik wrote:
> I don't think so. If you check the code paths more closely, you'll see
> that the timer is used even in the fast TSC case, the error causes by
> too big 'count' variable propagates up to the TSC routine. That's where
> I started hunting for the bug.
>
Linus Torvalds wrote:
>
> On Wed, 25 Oct 2000, Rik van Riel wrote:
> >
> > OTOH, block-dev readahead makes sense for filesystems where
> > the packing locality is close to the access pattern BUT NOT
> > close to anything the page cache would recognise as being
> > close.
>
> I dunno. The main re
Interesting theory, but no go here.
I just remembered one other thing that may have an impact. I have a
IBM-DTLA-307045 (~45 GB i think) hanging off the same channel as slave.
The thing with that one is that if I try to do a lot of i/o on the
disk..my machine locks up. Hard. alt-sysReq doesn't ev
Hello,
Compaq has a dual-port ethernet adapter which is based on the Intel 82559
(the chip used in the Intel Etherexpress Pro/100) as well as an addition
module to this adapter with two extra ports usign the same chip. Any ideas
what does the current kernel module say about this card (w/ and w/o
At the bottom is the IDE/ATA part of my .config. let me knwo if I am
missing something. Should I worry about the Multi_Mode configuration?
Pierre Etchemaite wrote:
> As a side note, I usually get better results tweaking disks using kernel
> compiling options ("default to DMA", "autotune chipset",
Hello,
How much memory would it be reasonable for kmalloc() to be able
to allocate to a module?
Oct 30 10:48:31 chaos kernel: kmalloc: Size (524288) too large
Using Version 2.2.17, I can't allocate more than 64k! I need
to allocate at least 1/2 megabyte and preferably more (like 2 megabytes).
Hi Dick,
Sorry, I thought you knew this already :) The maximum for kmalloc is 128K
and is defined in mm/slab.c. It is trivial to "enhance" slab.c to support
more but it is in practice not very useful because requesting too much
physically-contiguous (which kmalloc is all about) memory is impossib
On Mon, 30 Oct 2000, Richard B. Johnson wrote:
>
> Hello,
> How much memory would it be reasonable for kmalloc() to be able
> to allocate to a module?
>
> Oct 30 10:48:31 chaos kernel: kmalloc: Size (524288) too large
>
> Using Version 2.2.17, I can't allocate more than 64k! I need
> to allo
On Mon, 30 Oct 2000, Tigran Aivazian wrote:
> Hi Dick,
>
> Sorry, I thought you knew this already :) The maximum for kmalloc is 128K
> and is defined in mm/slab.c. It is trivial to "enhance" slab.c to support
> more but it is in practice not very useful because requesting too much
> physically-c
On Mon, 30 Oct 2000, Richard B. Johnson wrote:
> > So, if you don't need physically contiguous (and fast) allocations perhaps
> > you could make use of vmalloc()/vfree() instead? There must be also some
> > "exotic" allocation APIs like bootmem but I know nothing of them so I stop
> > here.
>
> O
On Mon, 30 Oct 2000, Richard B. Johnson wrote:
> How much memory would it be reasonable for kmalloc() to be able
> to allocate to a module?
> There are 256 megabytes of SDRAM available. I don't think it's
> reasonable that a 1/2 megabyte allocation would fail, especially
> since it's the first m
On Mon, 30 Oct 2000, John Levon wrote:
> On Mon, 30 Oct 2000, Richard B. Johnson wrote:
>
> >
> > Hello,
> > How much memory would it be reasonable for kmalloc() to be able
> > to allocate to a module?
> >
> > Oct 30 10:48:31 chaos kernel: kmalloc: Size (524288) too large
> >
> > Using Versi
On Sat, 28 Oct 2000, Mark Spencer wrote:
> Now firstly, let's eliminate the ISDN red-herring from consideration
> because the authors of the code do not place any additional restrictions
> on the GPL whatsoever, they simply bring it to your attention that using
> an un-certified ISDN stack may be
On Mon, 30 Oct 2000, J{rvensivu Riku wrote:
> Useless to mention, Compaq doesn't have linux drivers for this board.
Why useless? It is the vendor's responsibility to write most efficient
drivers for their hardware, at least that was the case with the previously
popular systems like Windows and I
"Richard B. Johnson" wrote:
> Now, I could set up a linked-list of buffers and use vmalloc()
> if the buffers were allocated from non-paged RAM. I don't think
> they are. These buffers must be present during an interrupt.
Non-paged RAM? I'm not sure what you mean by that.
Both kmalloc and vmall
On Mon, 30 Oct 2000, Jeff Garzik wrote:
> "Richard B. Johnson" wrote:
> > Now, I could set up a linked-list of buffers and use vmalloc()
> > if the buffers were allocated from non-paged RAM. I don't think
> > they are. These buffers must be present during an interrupt.
>
> Non-paged RAM? I'm no
On Sat, 28 Oct 2000, Gregory Maxwell wrote:
> See section 7 of the GPL.
"If you cannot distribute so as to satisfy simultaneously your obligations
under this License and any other pertinent obligations, then as a
consequence you may not distribute the Program at all."
But I can, so I may. See t
Tigran Aivazian wrote:
>
> On Mon, 30 Oct 2000, Jeff Garzik wrote:
>
> > "Richard B. Johnson" wrote:
> > > Now, I could set up a linked-list of buffers and use vmalloc()
> > > if the buffers were allocated from non-paged RAM. I don't think
> > > they are. These buffers must be present during an
Is there any known bug in 2.2.18pre15 that could cause the
following oops in fs/buffer.c:find_buffer() ?
=== snip ===
ksymoops 0.7c on i586 2.2.18pre15. Options used
-V (default)
-k /proc/ksyms (default)
-l /proc/modules (default)
-o /lib/modules/2.2.18pre15/ (default)
-
Christoph Hellwig wrote:
> +Locking down user memory and doing mass storage device IO with it is not
> +the only purpose of kiobufs. Another use for kiobufs is allowing
> +user-space mmaping dma memory, e.g in sound drivers. To do so you
> +need to lock-down kernel virtual memory and refernece i
We have written a char driver for our proprietary h/w. This driver
handles a
multitude of interrupts from the h/w in the following fashion: The ISR
reads/saves the status register (indication of which int was hit) in
global, and the marks the BH to run. The BH uses the global to call one
of 32
On Mon, Oct 30, 2000 at 12:13:52PM +0100, Ingo Molnar wrote:
> simple, write a TUX protocol module for it. FTP protocol module is on its
> way. Stay tuned.
TUX modules are kernel modules (I mean you have to write kernel space code for
doing TUX ftp). Don't you agree that zero-copy sendfile like f
I also just got this Oops, but with 2.2.18pre18.
Note: vmware was not loaded :)
Unable to handle kernel paging request at virtual address 009e
current->tss.cr3 = 108e5000, %cr3 = 108e5000
*pde =
Oops:
CPU:0
EIP:0010:[]
EFLAGS: 00010206
eax: 009e ebx: 0007
On Mon, 30 Oct 2000, Andrea Arcangeli wrote:
> functionality that needs high performance completly in kernel? People
> may need to write high performance network code for custom protocols,
> this way they will end creating kernel modules with system-crashing
> bugs, memory leaks and kernel buffe
1 - 100 of 227 matches
Mail list logo
