Documentation update

git-svn-id: svn://svn.code.sf.net/p/nuttx/code/trunk@5493 42af7a65-404d-4744-a932-0658087f49c3
This commit is contained in:
patacongo 2013-01-08 16:51:22 +00:00
parent c76795d32b
commit ddb0ad618e
3 changed files with 135 additions and 8 deletions

View File

@ -13,7 +13,7 @@
<h1><big><font color="#3c34ec"><i>NuttX Operating System<p>User's Manual</i></font></big></h1>
<p><small>by</small></p>
<p>Gregory Nutt<p>
<p>Last Updated: January 7, 2013</p>
<p>Last Updated: January 8, 2013</p>
</td>
</tr>
</table>
@ -651,7 +651,134 @@ pid_t vfork(void);
</p>
<H3><a name="execv">2.1.9 execv</a></H3>
<p>
<b>Function Prototype:</b>
</p>
<ul><pre>
#include &lt;unistd.h&gt;
#ifdef CONFIG_LIBC_EXECFUNCS
int execv(FAR const char *path, FAR char *const argv[]);
#endif
</pre></ul>
<p>
<b>Description:</b>
The standard <code>exec</code> family of functions will replace the current process image with a new process image.
The new image will be constructed from a regular, executable file called the new process image file.
There will be no return from a successful <code>exec</code>, because the calling process image is overlaid by the new process image.
</p>
<p>
Simplified <code>execl()</code> and <code>execv()</code> functions are provided by NuttX for compatibility.
NuttX is a tiny embedded RTOS that does not support processes and hence the concept of overlaying a tasks process image with a new process image does not make any sense.
In NuttX, these functions are wrapper functions that:
</p>
<ol>
<li>
Call the non-standard <code>binfmt</code> function <code>exec()</code>, and then
</li>
<li>
<code>exit(0)</code>.
</li>
</ol>
<p>
Note the inefficiency when <code>execv()</code> or <code>execl()</code> is called in the normal, two-step process:
(1) first call <code>vfork()</code> to create a new thread, then (2) call <code>execv()</code> or <code>execl()</code> to replace the new thread with a program from the file system.
Since the new thread will be terminated by the <code>execv()</code> or <code>execl()</code> call, it really served no purpose other than to support Unix compatility.
</p>
<p>
The non-standard binfmt function <code>exec()</code> needs to have (1) a symbol table that provides the list of symbols exported by the base code, and (2) the number of symbols in that table.
This information is currently provided to <code>exec()</code> from <code>execv()</code> or <code>execl()</code> via NuttX configuration settings:
</p>
<ul>
<li>
<code>CONFIG_LIBC_EXECFUNCS</code>:
Enable <code>execv()</code> and <code>execl()</code> support
</li>
<li>
<code>CONFIG_EXECFUNCS_SYMTAB</code>:
Symbol table used by <code>execv()</code> or <code>execl()</code>.
</li>
<li>
<code>CONFIG_EXECFUNCS_NSYMBOLS</code>:
Number of symbols in the symbol table
</li>
</ul>
<p>
As a result of the above, the current implementations of <code>execl()</code> and <code>execv()</code> suffer from some incompatibilities that may or may not be addressed in a future version of NuttX.
Other than just being an inefficient use of MCU resource, the most serious of these is that
the <code>exec</code>'ed task will not have the same task ID as the <code>vfork</code>'ed function.
So the parent function cannot know the ID of the <code>exec</code>'ed task.<p>
<p>
<b>Input Parameters:</b>
</p>
<ul>
<li>
<code>path</code>:
The path to the program to be executed.
If <code>CONFIG_BINFMT_EXEPATH</code> is defined in the configuration, then this may be a relative path from the current working directory.
Otherwise, <code>path</code> must be the absolute path to the program.
<li>
</li>
<code>argv</code>:
A pointer to an array of string arguments.
The end of the array is indicated with a NULL entry.
</li>
</ul>
<p>
<b>Returned Value:</b>
This function does not return on success.
On failure, it will return -1 (<code>ERROR</code>) and will set the <code>errno</code> value appropriately.
<p>
<b>Assumptions/Limitations:</b>
</p>
<p>
<b>POSIX Compatibility:</b>
Similar with the Unix interface of the same name.
There are, however, several compatibility issues as detailed in the description above.
</p>
<H3><a name="execl">2.1.10 execl</a></H3>
<p>
<b>Function Prototype:</b>
</p>
<ul><pre>
#include &lt;unistd.h&gt;
#ifdef CONFIG_LIBC_EXECFUNCS
int execl(FAR const char *path, ...);
#endif
</pre></ul>
<p>
<b>Description:</b>
<code>execl()</code> is functionally equivalent to <a href="#execv">execv()</a>, differing only in the form of its input parameters.
See the decription of <a href="#execv">execv()</a> for additional information.
<p>
<p>
<b>Input Parameters:</b>
</p>
<ul>
<li>
<code>path</code>:
The path to the program to be executed.
If <code>CONFIG_BINFMT_EXEPATH</code> is defined in the configuration, then this may be a relative path from the current working directory.
Otherwise, <code>path</code> must be the absolute path to the program.
<li>
</li>
<code>...</code>:
A list of the string arguments to be recevied by the program.
Zero indicates the end of the list.
</li>
</ul>
<p>
<b>Returned Value:</b>
This function does not return on success.
On failure, it will return -1 (<code>ERROR</code>) and will set the <code>errno</code> value appropriately.
<p>
<b>Assumptions/Limitations:</b>
</p>
<p>
<b>POSIX Compatibility:</b>
Similar with the Unix interface of the same name.
There are, however, several compatibility issues as detailed in the description of <a href="#execv">execv()</a>.
</p>
<table width ="100%">
<tr bgcolor="#e4e4e4">

View File

@ -83,7 +83,7 @@
* step process: (1) first call vfork() to create a new thread, then (2)
* call 'exec[l|v]()' to replace the new thread with a program from the
* file system. Since the new thread will be terminated by the
* 'exec[l|v]()' call, it really served no purpose other than to suport
* 'exec[l|v]()' call, it really served no purpose other than to support
* Unix compatility.
*
* The non-standard binfmt function 'exec()' needs to have (1) a symbol
@ -91,9 +91,9 @@
* (2) the number of symbols in that table. This information is currently
* provided to 'exec()' from 'exec[l|v]()' via NuttX configuration settings:
*
* CONFIG_LIBC_EXECFUNCS : Enable exec[l|v] Support
* CONFIG_LIBC_EXECFUNCS : Enable exec[l|v] support
* CONFIG_EXECFUNCS_SYMTAB : Symbol table used by exec[l|v]
* CONFIG_EXECFUNCS_NSYMBOLS : Number of Symbols in the Table
* CONFIG_EXECFUNCS_NSYMBOLS : Number of symbols in the table
*
* As a result of the above, the current implementations of 'execl()' and
* 'execv()' suffer from some incompatibilities that may or may not be
@ -108,7 +108,7 @@
* is defined in the configuration, then this may be a relative path
* from the current working directory. Otherwise, path must be the
* absolute path to the program.
* arg0,... - A list of the string arguments to be recevied by the
* ... - A list of the string arguments to be recevied by the
* program. Zero indicates the end of the list.
*
* Returned Value:

View File

@ -104,7 +104,7 @@ extern struct symtab_s CONFIG_EXECFUNCS_SYMTAB;
* step process: (1) first call vfork() to create a new thread, then (2)
* call 'exec[l|v]()' to replace the new thread with a program from the
* file system. Since the new thread will be terminated by the
* 'exec[l|v]()' call, it really served no purpose other than to suport
* 'exec[l|v]()' call, it really served no purpose other than to support
* Unix compatility.
*
* The non-standard binfmt function 'exec()' needs to have (1) a symbol
@ -112,9 +112,9 @@ extern struct symtab_s CONFIG_EXECFUNCS_SYMTAB;
* (2) the number of symbols in that table. This information is currently
* provided to 'exec()' from 'exec[l|v]()' via NuttX configuration settings:
*
* CONFIG_LIBC_EXECFUNCS : Enable exec[l|v] Support
* CONFIG_LIBC_EXECFUNCS : Enable exec[l|v] support
* CONFIG_EXECFUNCS_SYMTAB : Symbol table used by exec[l|v]
* CONFIG_EXECFUNCS_NSYMBOLS : Number of Symbols in the Table
* CONFIG_EXECFUNCS_NSYMBOLS : Number of symbols in the table
*
* As a result of the above, the current implementations of 'execl()' and
* 'execv()' suffer from some incompatibilities that may or may not be