* Semaphore and Shared memory questions...
@ 2002-11-20 21:30 Richard Whittaker
2002-11-20 22:04 ` Alan Cox
[not found] ` <sddb931a.041@nwtel.ca>
0 siblings, 2 replies; 6+ messages in thread
From: Richard Whittaker @ 2002-11-20 21:30 UTC (permalink / raw)
To: linux-kernel
Greetings:
I'm building a Linux box up for use with Oracle 8i.
The machine I'm using is a Dell Poweredge 2650 server with a pair of Xeon
Processors, 2GB of RAM, a disk array, etc, etc...
Part of building this machine up is tuning the kernel parameters for
shmmax, semmni, semmsl, semmns, etc...
In reading, this should be fairly ieasy... Make some changes to
/usr/include/linux/sem.h and /usr/include/linux/shm.h, recompile, install,
LILO, reboot, and yer done... Well, my experience wasn't that simple...
Here's a clip of my /usr/include/linux/shm.h
#define SHMMAX 0x3E00000 /* max shared seg size (bytes) */
#define SHMMIN 1 /* min shared seg size (bytes) */
#define SHMMNI 4096 /* max num of segs system wide */
#define SHMALL (SHMMAX/PAGE_SIZE*(SHMMNI/16)) /* max shm system wide (pages) */
#define SHMSEG SHMMNI /* max shared segs per process */
...and what sysctl shows...
root@tom:/usr/include/linux# sysctl -a | grep shmmax
kernel/shmmax = 33554432
root@tom:/usr/include/linux#
I can change this value with sysctl -p at boot time, but I would really
prefer it to be built into the kernel...
...now for semaphores...
/usr/include/linux/sem.h
#define SEMMNI 128 /* <= IPCMNI max # of semaphore identifiers */
#define SEMMSL 450 /* <= 8 000 max num of semaphores per id */
#define SEMMNS 1000 /* <= INT_MAX max # of semaphores in system */
#define SEMOPM 100 /* <= 1 000 max num of ops per semop call */
#define SEMVMX 32767 /* <= 32767 semaphore maximum value */
#define SEMAEM SEMVMX /* adjust on exit max value */
The only values I've changed are SEMMSL, and SEMMNS...
Looking at the semaphores, here's what I see...
root@tom:/usr/include/linux# cat /proc/sys/kernel/sem
250 32000 32 128
root@tom:/usr/include/linux#
How would I sysctl these values?...
I would prefer them to be statically built into the kernel...
I know that the kernel SHOULD be the right version, and I've done a make
dep ; make clean ; make bzImage
My kernel version info...
root@tom:/usr/include/linux# cat /proc/sys/kernel/osrelease
2.4.19
root@tom:/usr/include/linux# cat /proc/sys/kernel/version
#4 SMP Tue Nov 19 16:21:21 PST 2002
Thanks muchly in advance!!!
Regards,
Richard.
---
Richard Whittaker,
System Manager,
NorthwesTel Inc.
Whitehorse, Yukon, Canada.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Semaphore and Shared memory questions...
2002-11-20 22:04 ` Alan Cox
@ 2002-11-20 21:42 ` Richard Whittaker
2002-11-20 22:16 ` Alan Cox
0 siblings, 1 reply; 6+ messages in thread
From: Richard Whittaker @ 2002-11-20 21:42 UTC (permalink / raw)
To: Alan Cox; +Cc: Linux Kernel Mailing List
At 10:04 PM 11/20/2002 +0000, Alan Cox wrote:
>The sysctl interface and sysctl setting tools on boot exist precisely so
>you dont have to hack these things around
Allright... This is fine by me... One of my questions was (and still is),
what do I pass to sysctl for setting /proc/sys/kernel/sem?...
Setting shmmax was easy, since it's only a single hex value, but sem is
puzzling me... Is a space separated list of values going to work?...
Thanks,
Richard.
---
Richard Whittaker,
System Manager,
NorthwesTel Inc.
Whitehorse, Yukon, Canada.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Semaphore and Shared memory questions...
2002-11-20 21:30 Semaphore and Shared memory questions Richard Whittaker
@ 2002-11-20 22:04 ` Alan Cox
2002-11-20 21:42 ` Richard Whittaker
[not found] ` <sddb931a.041@nwtel.ca>
1 sibling, 1 reply; 6+ messages in thread
From: Alan Cox @ 2002-11-20 22:04 UTC (permalink / raw)
To: Richard Whittaker; +Cc: Linux Kernel Mailing List
On Wed, 2002-11-20 at 21:30, Richard Whittaker wrote:
> Part of building this machine up is tuning the kernel parameters for
> shmmax, semmni, semmsl, semmns, etc...
> In reading, this should be fairly ieasy... Make some changes to
> /usr/include/linux/sem.h and /usr/include/linux/shm.h, recompile, install,
> LILO, reboot, and yer done... Well, my experience wasn't that simple...
Well if you insist on hacking code paths not using /proc Im not
suprised. You also probably ended up with a non oracle certified
platform by doing so (if that matters)
The sysctl interface and sysctl setting tools on boot exist precisely so
you dont have to hack these things around
Alan
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Semaphore and Shared memory questions...
2002-11-20 21:42 ` Richard Whittaker
@ 2002-11-20 22:16 ` Alan Cox
0 siblings, 0 replies; 6+ messages in thread
From: Alan Cox @ 2002-11-20 22:16 UTC (permalink / raw)
To: Richard Whittaker; +Cc: Linux Kernel Mailing List
On Wed, 2002-11-20 at 21:42, Richard Whittaker wrote:
> At 10:04 PM 11/20/2002 +0000, Alan Cox wrote:
>
> >The sysctl interface and sysctl setting tools on boot exist precisely so
> >you dont have to hack these things around
>
> Allright... This is fine by me... One of my questions was (and still is),
> what do I pass to sysctl for setting /proc/sys/kernel/sem?...
>
> Setting shmmax was easy, since it's only a single hex value, but sem is
> puzzling me... Is a space separated list of values going to work?...
Right first guess. Its mostly historical that it isnt 4 values.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Semaphore and Shared memory questions...
[not found] ` <sddb931a.041@nwtel.ca>
@ 2002-11-20 22:17 ` Richard Whittaker
2002-11-20 22:33 ` Randy.Dunlap
0 siblings, 1 reply; 6+ messages in thread
From: Richard Whittaker @ 2002-11-20 22:17 UTC (permalink / raw)
To: Alan Cox; +Cc: linux-kernel
At 02:16 PM 11/20/2002 -0800, Alan Cox wrote:
>On Wed, 2002-11-20 at 21:42, Richard Whittaker wrote:
> > At 10:04 PM 11/20/2002 +0000, Alan Cox wrote:
>
>Right first guess. Its mostly historical that it isnt 4 values.
What's the order of the values?...
I can take a wild stab that the first value is semmsl, or is it semmni,
then the subsequent values?...
Are there plans to split them out into distinct values, or is that a
massive rewrite?...
Thanks!
Richard.
---
Richard Whittaker,
System Manager,
NorthwesTel Inc.
Whitehorse, Yukon, Canada.
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Semaphore and Shared memory questions...
2002-11-20 22:17 ` Richard Whittaker
@ 2002-11-20 22:33 ` Randy.Dunlap
0 siblings, 0 replies; 6+ messages in thread
From: Randy.Dunlap @ 2002-11-20 22:33 UTC (permalink / raw)
To: Richard Whittaker; +Cc: Alan Cox, linux-kernel
On Wed, 20 Nov 2002, Richard Whittaker wrote:
| >Right first guess. Its mostly historical that it isnt 4 values.
|
| What's the order of the values?...
|
| I can take a wild stab that the first value is semmsl, or is it semmni,
| then the subsequent values?...
|
| Are there plans to split them out into distinct values, or is that a
| massive rewrite?...
I haven't seen a kernel version that you are working with,
but hopefully that won't matter. This is from 2.4.19:
./ipc/sem.c:96:int sem_ctls[4] = {SEMMSL, SEMMNS, SEMOPM, SEMMNI};
./ipc/sem.c:97:#define sc_semmsl (sem_ctls[0])
./ipc/sem.c:98:#define sc_semmns (sem_ctls[1])
./ipc/sem.c:99:#define sc_semopm (sem_ctls[2])
./ipc/sem.c:100:#define sc_semmni (sem_ctls[3])
so that's the order. If Alan replies, his matching reply can confirm
this. :)
--
~Randy
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2002-11-20 22:27 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2002-11-20 21:30 Semaphore and Shared memory questions Richard Whittaker
2002-11-20 22:04 ` Alan Cox
2002-11-20 21:42 ` Richard Whittaker
2002-11-20 22:16 ` Alan Cox
[not found] ` <sddb931a.041@nwtel.ca>
2002-11-20 22:17 ` Richard Whittaker
2002-11-20 22:33 ` Randy.Dunlap
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®