Copyright (C) 2002, Sun Microsystems, Inc. All Rights Reserved
The contents of this file are subject to the terms of the Common Development and Distribution License (the "License"). You may not use this file except in compliance with the License.
You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE or http://www.opensolaris.org/os/licensing. See the License for the specific language governing permissions and limitations under the License.
When distributing Covered Code, include this CDDL HEADER in each file and include the License file at usr/src/OPENSOLARIS.LICENSE. If applicable, add the following below this CDDL HEADER, with the fields enclosed by brackets "[]" replaced with your own identifying information: Portions Copyright [yyyy] [name of copyright owner]
fx_dptbl
The process scheduler or dispatcher is the portion of the kernel that controls allocation of the CPU to processes. The scheduler supports the notion of scheduling classes, where each class defines a scheduling policy used to schedule processes within that class. Associated with each scheduling class is a set of priority queues on which ready-to-run processes are linked. These priority queues are mapped by the system configuration into a set of global scheduling priorities, which are available to processes within the class. The dispatcher always selects for execution the process with the highest global scheduling priority in the system. The priority queues associated with a given class are viewed by that class as a contiguous set of priority levels numbered from 0 (lowest priority) to n (highest priority\(ema configuration-dependent value). The set of global scheduling priorities that the queues for a given class are mapped into might not start at zero and might not be contiguous, depending on the configuration.
Processes in the fixed priority class are scheduled according to the parameters in a fixed-priority dispatcher parameter table (fx_dptbl). The fx_dptbl table consists of an array (config_fx_dptbl[]) of parameter structures (struct fxdpent_t), one for each of the n priority levels used by fixed priority processes in user mode. The structures are accessed by way of a pointer, (fx_dptbl), to the array. The properties of a given priority level i are specified by the ith parameter structure in this array (fx_dptbl[i]).
A parameter structure consists of the following members. These are also described in the /usr/include/sys/fx.h header.
fx_globpri
The global scheduling priority associated with this priority level. The mapping between fixed-priority priority levels and global scheduling priorities is determined at boot time by the system configuration. fx_globpri can not be changed with dispadmin(1M).
fx_quantum
The length of the time quantum allocated to processes at this level in ticks (hz). The time quantum value is only a default or starting value for processes at a particular level, as the time quantum of a fixed priority process can be changed by the user with the priocntl(1) command or the priocntl(2) system call. In the high resolution clock mode (hires_tick set to 1), the value of hz is set to 1000. Increase quantums to maintain the same absolute time quantums. An administrator can affect the behavior of the fixed priority portion of the scheduler by reconfiguring the fx_dptbl. There are two methods available for doing this: reconfigure with a loadable module at boot-time or by using dispadmin(1M) at run-time.
The fx_dptbl can be reconfigured with a loadable module that contains a new fixed priority dispatch table. The module containing the dispatch table is separate from the FX loadable module, which contains the rest of the fixed priority software. This is the only method that can be used to change the number of fixed priority priority levels or the set of global scheduling priorities used by the fixed priority class. The relevant procedure and source code is described in Replacing the fx_dptbl Loadable Module below.
The fx_quantum values in the fx_dptbl can be examined and modified on a running system using the dispadmin(1M) command. Invoking dispadmin for the fixed-priority class allows the administrator to retrieve the current fx_dptbl configuration from the kernel's in-core table or overwrite the in-core table with values from a configuration file. The configuration file used for input to dispadmin must conform to the specific format described as follows:
Blank lines are ignored and any part of a line to the right of a # symbol is treated as a comment.
The first non-blank, non-comment line must indicate the resolution to be used for interpreting the time quantum values. The resolution is specified as:
RES=res
where res is a positive integer between 1 and 1,000,000,000 inclusive and
the resolution used is the reciprocal of res in seconds (for example,
RES=1000 specifies millisecond resolution). Although you can specify very
fine (nanosecond) resolution, the time quantum lengths are rounded up to the
next integral multiple of the system clock's resolution.
The remaining lines in the file are used to specify the fx_quantum values for each of the fixed-priority priority levels. The first line specifies the quantum for fixed-priority level 0, the second line specifies the quantum for fixed-priority level 1, and so forth. There must be exactly one line for each configured fixed priority priority level. Each fx_quantum entry must be a positive integer specifying the desired time quantum in the resolution given by res.
See Examples for an example of an excerpt of a dispadmin configuration file.
To change the size of the fixed priority dispatch table, you must build the loadable module that contains the dispatch table information. Save the existing module before using the following procedure.
1. Place the dispatch table code shown below in a file called fx_dptbl.c. See EXAMPLES, below, for an example of this file.
2. Compile the code using the given compilation and link lines supplied:
cc -c -0 -D_KERNEL fx_dptbl.c ld -r -o FX_DPTBL fx_dptbl.o
5. Make changes in the /etc/system file to reflect the changes to the sizes of the tables. See system(4). The variables affected is fx_maxupri. The syntax for setting this is as follows:
set FX:fx_maxupri=(value for max fixed-priority user priority)
6. Reboot the system to use the new dispatch table.
Exercise great care in using the preceding method to replace the dispatch table. A mistake can result in panics, thus making the system unusable.
Example 1 Configuration File Excerpt
The following excerpt from a dispadmin configuration file illustrates the correct format. Note that, for each line specifying a set of parameters, there is a comment indicating the corresponding priority level. These level numbers indicate priority within the fixed priority class; the mapping between these fixed-priority priorities and the corresponding global scheduling priorities is determined by the configuration specified in the FX_DPTBL loadable module. The level numbers are strictly for the convenience of the administrator reading the file and, as with any comment, they are ignored by dispadmin. The dispadmin command assumes that the lines in the file are ordered by consecutive, increasing priority level (from 0 to the maximum configured fixed-priority priority). For the sake of someone reading the file, the level numbers in the comments should agree with this ordering. If for some reason they do not, dispadmin is unaffected.
# Fixed Priority Dispatcher Configuration File RES=1000 RES=1000 # TIME QUANTUM PRIORITY # (fx_quantum) LEVEL 200 # 0 200 # 1 200 # 2 200 # 3 200 # 4 200 # 5 200 # 6 200 # 7 . . . . . . . . . 20 # 58 20 # 59 20 # 60
Example 2 fx_dptbl.c File Used for Building the New fx_dptbl
The following is an example of a fx_dptbl.c file used for building the new fx_dptbl.
/* BEGIN fx_dptbl.c */ #include <sys/proc.h> #include <sys/priocntl.h> #include <sys/class.h> #include <sys/disp.h> #include <sys/fx.h> #include <sys/fxpriocntl.h> /* * This is the loadable module wrapper. */ #include <sys/modctl.h> extern struct mod_ops mod_miscops; /* * Module linkage information for the kernel. */ static struct modlmisc modlmisc = { &mod_miscops, "Fixed priority dispatch table" }; static struct modlinkage modlinkage = { MODREV_1, &modlmisc, 0 }; _init() { return (mod_install(&modlinkage)); } _info(modinfop) struct modinfo *modinfop; { return (mod_info(&modlinkage, modinfop)); } #define FXGPUP0 0 /* Global priority for FX user priority 0 */ fxdpent_t config_fx_dptbl[] = { /* glbpri qntm */ FXGPUP0+0, 20, FXGPUP0+1, 20, FXGPUP0+2, 20, FXGPUP0+3, 20, FXGPUP0+4, 20, FXGPUP0+5, 20, FXGPUP0+6, 20, FXGPUP0+7, 20, FXGPUP0+8, 20, FXGPUP0+9, 20, FXGPUP0+10, 16, FXGPUP0+11, 16, FXGPUP0+12, 16, FXGPUP0+13, 16, FXGPUP0+14, 16, FXGPUP0+15, 16, FXGPUP0+16, 16, FXGPUP0+17, 16, FXGPUP0+18, 16, FXGPUP0+19, 16, FXGPUP0+20, 12, FXGPUP0+21, 12, FXGPUP0+22, 12, FXGPUP0+23, 12, FXGPUP0+24, 12, FXGPUP0+25, 12, FXGPUP0+26, 12, FXGPUP0+27, 12, FXGPUP0+28, 12, FXGPUP0+29, 12, FXGPUP0+30, 8, FXGPUP0+31, 8, FXGPUP0+32, 8, FXGPUP0+33, 8, FXGPUP0+34, 8, FXGPUP0+35, 8, FXGPUP0+36, 8, FXGPUP0+37, 8, FXGPUP0+38, 8, FXGPUP0+39, 8, FXGPUP0+40, 4, FXGPUP0+41, 4, FXGPUP0+42, 4, FXGPUP0+43, 4, FXGPUP0+44, 4, FXGPUP0+45, 4, FXGPUP0+46, 4, FXGPUP0+47, 4, FXGPUP0+48, 4, FXGPUP0+49, 4, FXGPUP0+50, 4, FXGPUP0+51, 4, FXGPUP0+52, 4, FXGPUP0+53, 4, FXGPUP0+54, 4, FXGPUP0+55, 4, FXGPUP0+56, 4, FXGPUP0+57, 4, FXGPUP0+58, 4, FXGPUP0+59, 2, FXGPUP0+60 2, }; pri_t config_fx_maxumdpri = sizeof (config_fx_dptbl) / sizeof (fxdpent_t) - 1; /* * Return the address of config_fx_dptbl */ fxdpent_t * fx_getdptbl() { return (config_fx_dptbl); } /* * Return the address of fx_maxumdpri */ pri_t fx_getmaxumdpri() { /* * the config_fx_dptbl table. */ return (config_fx_maxumdpri); }
priocntl(1), dispadmin(1M), priocntl(2), system(4)
System Administration Guide, Volume 1, System Interface Guide
In order to improve performance under heavy system load, both the nfsd daemon and the lockd daemon utilize the maximum priority in the FX class. Unusual fx_dptbl configurations may have significant negative impact on the performance of the nfsd and lockd daemons.