Kernel Stack vs User Mode Stack Post: 302586912

Sponsored Content

Top Forums UNIX for Dummies Questions & Answers Kernel Stack vs User Mode Stack Post 302586912 by ygemici on Tuesday 3rd of January 2012 03:54:52 PM

01-03-2012

Registered User

Quote:

Originally Posted by saurabhkoar

@fpmurphy,

That means in Linux each process has a user mode stack and a corresponding Kernel mode stack? So, there is two stacks per process?

actually they are used on the separate [cpu processing] modes.
and in this state lets we split to as modes(user/kernel) for better understanding of its.
shortly,linux process can operates in two modes:

* in user_mode
any process which runs the in user mode refers to private(own) stack area that is used for holds local variables,parameters,frame pointer(very useful while detect relative addres of any func local variables after pushes,pops especially in debug processing[kernel oops]),temporary(as a cpy) values(like in c,c++,java),return address from funcs and automatic variables within functions.

* in kernel mode (after switching to kernel mode by [switch_to] ), portion of the process (code areas),its addresses and core kernel data, starts to use the kernel stack.
[as additonal infos; hardware context informations(like contents of all registers in user mode) also are saved in the kernel stack while the hardware context processing]

linux kernel has many complex structures like kernel control path that is sequence of instructions(* executed several kernel functions) executed by a kernel for an interrupt,system call or exception handling.
so these instructions are belonging to different processes.therefore each kernel path must refers to its kernel stack(8k)
[actually 8140 bytes 52 byte save for thread_info structure]
you can define the size of kernel stacks while compaling if you see the like this message

Code:

Use 4Kb for kernel stacks instead of 8Kb (4KSTACKS) [N/y/?] n

and user mode process can not know where is the address of own kernel stack memory area; (pointer) address, therefore cannot accessed to kernel stack area in the user mode.

while switching to kernel mode with mode switch,process will continue to work in the reentrance kernels.
so it is a reentrance kernel! user mode process executions can be resume in kernel mode and continues processing for new processes by kernel schedulers.[assisgns the another Cpu to the process with schedule()]
but on a single processor system,one process will be process in the Cpu and new processes blocked until the current process turn.

I would recommend for more information below usefull links..
Understanding the Linux Kernel, 3rd Edition
4K Stacks in 2.6 [LWN.net]

regards
ygemici

ygemici

View Public Profile for ygemici

Find all posts by ygemici

7 More Discussions You Might Find Interesting

1. Programming

User Frame & Stack

Can anybody tell me what a user frame & stack looks like on a sun, hp, powerpc system?

2. Programming

what is stack winding and stack unwinding

helo can u tell me what do you mean by stack winding and stack unwinding Regards, Amit

3. Solaris

How to set stack for Oracle's user?

Hi all! I've got a problem. Primary: Sun Sparc V890, 64bit Standby: HP-Proliant ML 350, 64bit In the machine that I've got configured physical standby database, the alert_log_file point me one error: Sat Feb 28 00:40:08 2009 ORA-00202: control file:...

4. SCO

SCO Openserver Kernel and network stack

Hi everybody My SCO 5.0.2 OpenServer got its kernel corruptions and network stack was accicentially deleted. I would like to know if we have any chance to rebuild its kernel and network stack. Any help will be great thankfully appreciate TN

5. Programming

which function copies data from user to kernel mode

when transitionaning from user to kernel mode which function copies data from user mode buffer to kernel mode?

6. UNIX for Dummies Questions & Answers

View User Mode Call Stack of Hung Process

I have a multithreaded usermode program(actually a daemon) which is in hanged state. To debug it I tried attaching the process to gdb, but the gdb hangs. gstack also gets hanged. I peeped into the proc file system and saw the process to be in sleeping state. /proc/sysrq-trigger I guess...

7. Red Hat

Cannot set 'soft limits' for 'maximum stack size' for a standard user

Hi Guys, I'm trying to install Oracle Database on to Oracle Linux 7.6 but when the database install package checks the OS set-up, it keeps on failing on the soft limits for the stack. It's default value is 8192 but I'm trying to set it to 10240. This is what I added to...

LEARN ABOUT HPUX

maxrsessiz_64bit

maxrsessiz(5)							File Formats Manual						     maxrsessiz(5)

NAME

       maxrsessiz, maxrsessiz_64bit - maximum size (in bytes) of the RSE stack for any user process on the Itanium-based platform

VALUES

   Default
   Allowed values
DESCRIPTION

       Itanium-based  systems  utilize a per-process register stack in main memory (for more information see the Intel IA-64 Architecture Software
       Developer's Manual, Volume 2, Chapter 6). This stack moves registers between the physical registers and main memory and	is  maintained	by
       the Register Stack Engine (RSE).  determines the size of this stack.

   Who is Expected to Change This Tunable?
       Anyone.

   Restrictions on Changing
       Changes to this tunable only take effect after a reboot of the system.

   When Should the Value of This Tunable Be Raised?
       should be raised if user processes are being terminated with the error due to overflow of the RSE stack.

   What Are the Side Effects of Raising the Value?
       User processes will dedicate more virtual memory to the RSE stack. This will result in less virtual memory being available for data alloca-
       tion.

   When Should the Value of This Tunable Be Lowered?
       This tunable should be lowered only if swap space is at a premium on the system, since virtual memory for  the  RSE  stack  still  requires
       matching  swap  reservation  or	allocation.  Serious consideration should be given if is being lowered below the default as this may cause
       unexpected failures of user applications on RSE stack overflow.

       This tunable is best left unchanged without complete and detailed knowledge of the RSE stack usage of the machine workload.

   What Are the Side Effects of Lowering the Value?
       Lowering this tunable will limit the amount of memory available for RSE stack usage per process.  This could cause processes with large RSE
       stack requirements to terminate with a error.

   What Other Tunable Values Should Be Changed at the Same Time?
       None.

WARNINGS

       All  HP-UX kernel tunable parameters are release specific.  This parameter may be removed or have its meaning changed in future releases of
       HP-UX.

       Installation of optional kernel software, from HP or other vendors, may cause changes to tunable  parameter  values.   After  installation,
       some  tunable  parameters may no longer be at the default or recommended values.  For information about the effects of installation on tun-
       able values, consult the documentation for the kernel software being installed.	For information about optional kernel  software  that  was
       factory installed on your system, see at

AUTHOR

       was developed by HP.

SEE ALSO

       maxssiz(5), maxdsiz(5), maxtsiz(5).

Itanium(R)-Based System Only				     Tunable Kernel Parameters						     maxrsessiz(5)