rfc:preload

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision
Previous revision
Last revisionBoth sides next revision
rfc:preload [2018/10/19 08:04] dmitryrfc:preload [2018/11/21 22:35] – This RFC has already been implemented cmb
Line 3: Line 3:
   * Date: 2018-10-18   * Date: 2018-10-18
   * Author: Dmitry Stogov, dmitry@zend.com   * Author: Dmitry Stogov, dmitry@zend.com
-  * Status: Draft (or Under Discussion or Accepted or Declined)+  * Status: Implemented
   * First Published at: http://wiki.php.net/rfc/preload   * First Published at: http://wiki.php.net/rfc/preload
  
 ===== Introduction ===== ===== Introduction =====
-PHP has been using opcode caches for ages (APC, Turck MMCache, Zend Opcahce). They achieve significant performance boost by **ALMOST** completely eliminating the overhead of PHP code recompilation. With an opcode cache, files are compiled once (on the first request that uses them), and are there stored in shared memory. All the following HTTP requests use cached in shared memory representation.+PHP has been using opcode caches for ages (APC, Turck MMCache, Zend OpCache). They achieve significant performance boost by **ALMOST** completely eliminating the overhead of PHP code recompilation. With an opcode cache, files are compiled once (on the first request that uses them), and are then stored in shared memory. All the following HTTP requests use the representation cached in shared memory.
  
-This proposal is about the "**ALMOST**", mentioned above. Despite, while storing files in an opcode cache eliminates the compilation overhead - there is still cost associated with fetching a file from the cache and into a specific request's context.  We still have to check if the source file was modified, copy certain parts of classes and functions from the shared memory cache to the process memory, etc.  Notably, since each PHP file is compiled and cached completely independently from any other file  we can't resolve dependencies between classes stored in different files as we store the files in the opcode cache, and have to re-link the class dependencies at run-time on each request.+This proposal is about the "**ALMOST**", mentioned above. While storing files in an opcode cache eliminates the compilation overhead -- there is still cost associated with fetching a file from the cache and into a specific request's context.  We still have to check if the source file was modified, copy certain parts of classes and functions from the shared memory cache to the process memory, etc.  Notably, since each PHP file is compiled and cached completely independently from any other filewe can't resolve dependencies between classes stored in different files when we store the files in the opcode cache, and have to re-link the class dependencies at run-time on each request.
  
-This proposal is inspired by the "Class Data Sharing" technology designed for Java HotSpot VM. It aims to provide users with the ability to trade in some of the flexibility that the conventional PHP model provides them for increased performance.  On server startup - before any application code is run - we may load a certain PHP files into memory - and make their contents "permanently available" to all subsequent requests that will be served by that server.  All the functions and classes defined in these files will be available to requests out of the box, exactly like internal entities (e.g. strlen() or Exception). In this way, we may **preload** entire or partial frameworks, and even the entire application class library.  It will also allow for introducing "built-in" functions that will be written in PHP (similar to HHVM's sytemlib).  The traded-in flexibility would include the inability to update these files once the server has been started (updating these files on the filesystem will not do anything;  A server restart will be required to apply the changes);  And also, this approach will not be compatible with servers that host multiple applications, or multiple versions of applications - that would have different implementations for certain classes with the same name - if such classes are preloaded from the codebase of one app, it will conflict with loading the different class implementation from the other app(s).+This proposal is inspired by the "Class Data Sharing" technology designed for Java HotSpot VM. It aims to provide users with the ability to trade in some of the flexibility that the conventional PHP model provides themfor increased performance.  On server startup -- before any application code is run -- we may load a certain set of PHP files into memory - and make their contents "permanently available" to all subsequent requests that will be served by that server.  All the functions and classes defined in these files will be available to requests out of the box, exactly like internal entities (e.g. strlen() or Exception). In this way, we may **preload** entire or partial frameworks, and even the entire application class library.  It will also allow for introducing "built-in" functions that will be written in PHP (similar to HHVM's sytemlib).  The traded-in flexibility would include the inability to update these files once the server has been started (updating these files on the filesystem will not do anything;  A server restart will be required to apply the changes);  And also, this approach will not be compatible with servers that host multiple applications, or multiple versions of applications - that would have different implementations for certain classes with the same name - if such classes are preloaded from the codebase of one app, it will conflict with loading the different class implementation from the other app(s).
  
 ===== Proposal =====  ===== Proposal ===== 
-Preloading is going to be controlled by just a single new php.ini directive - **opcache.preload**. Using this directive we will specify a single PHP file - which will perform the preloading task.  Once loaded, this file is then fully executed - and may preload other files, either by including them or by using the opcache_compile_file() function. Previously, I tried to implement a reach DSL to specifywhich files to load, which ignore, using pattern matching etc, but then realized that writing the preloading scenarios in PHP itself was much more simple and much more flexible.+Preloading is going to be controlled by just a single new php.ini directive - **opcache.preload**. Using this directive we will specify a single PHP file - which will perform the preloading task.  Once loaded, this file is then fully executed - and may preload other files, either by including them or by using the opcache_compile_file() function. Previously, I tried to implement a rich DSL to specify which files to load, which to ignore, using pattern matching etc, but then realized that writing the preloading scenarios in PHP itself was much more simple and much more flexible.
  
 For example the following script introduces a helper function, and uses it to preload the whole Zend Framework. For example the following script introduces a helper function, and uses it to preload the whole Zend Framework.
Line 50: Line 50:
 </code> </code>
  
-As mentioned above, preloaded files remain cached in opcache memory forever. Modification of their corresponding source files won'make effect without another server restart. All functions and most classes defined in these files will be permanently loaded into PHP's function and class tables and become permanently available in the context of any future request. During preloading, PHP also resolves class dependencies and links with parent, interfaces and traits. It also removes unnecessary includes and performs some other optimizations.+As mentioned above, preloaded files remain cached in opcache memory forever. Modification of their corresponding source files won'have any effect without another server restart. All functions and most classes defined in these files will be permanently loaded into PHP's function and class tables and become permanently available in the context of any future request. During preloading, PHP also resolves class dependencies and links with parent, interfaces and traits. It also removes unnecessary includes and performs some other optimizations
 + 
 +''opcache_reset()'' is not going to reload preloaded files. It's just not possible using current opcache design, because during restart, they may be used by some process, and any modifications may lead to crash. 
 + 
 +''opcache_get_status'' is extended to provide information about preloaded functions, classes and scripts under the "preload_statistics" index. 
 + 
 +==== Static members and static variables ==== 
 +To avoid misunderstanding, it is clear stated that preloading doesn't change the behavior of static class members and static variables. Their values are not going to relive request boundary.
  
 ==== Preloading Limitation ==== ==== Preloading Limitation ====
 Only classes without unresolved parent, interfaces, traits and constant values may be preloaded.  If a class doesn't satisfy to this condition, it's stored in opcache SHM as a part of corresponding PHP script in the same way as without preloading. Only classes without unresolved parent, interfaces, traits and constant values may be preloaded.  If a class doesn't satisfy to this condition, it's stored in opcache SHM as a part of corresponding PHP script in the same way as without preloading.
-Also, only top-level entities that are not nested within control structures (e.g. if()...) may be preloaded.  +Also, only top-level entities that are not nested within control structures (e.g. if()...) may be preloaded. 
 + 
 +On Windows, it's also not possible to preload classes inherited from internal ones. Windows ASLR and absence of fork() don't allow to guarantee the same addresses of internal classes in different processes.
  
 ==== Implementation Details ==== ==== Implementation Details ====
-Preloading is implemented as a part of the opcache on top of another (already committed) patch that introduces "immutable" classes and functions. They assume that immutable part is stored in shared memory once (for all processes) and never copied to process memory, but variable part is specific for each process. The patch introduced the MAP_PTR pointer data structure, that allows pointers from SHM to process.+Preloading is implemented as a part of the opcache on top of another (already committed) patch that introduces "immutable" classes and functions. They assume that the immutable part is stored in shared memory once (for all processes) and never copied to process memory, but the variable part is specific for each process. The patch introduced the MAP_PTR pointer data structure, that allows pointers from SHM to process memory.
  
 ===== Backward Incompatible Changes ===== ===== Backward Incompatible Changes =====
Line 71: Line 80:
 ==== php.ini Defaults ==== ==== php.ini Defaults ====
   * opcache.preload - specifies a PHP script that is going to be compiled and executed at server start-up.   * opcache.preload - specifies a PHP script that is going to be compiled and executed at server start-up.
- 
-===== Open Issues ===== 
-  * preloading in ZTS build is not supported yet 
  
 ===== Performance ===== ===== Performance =====
Line 83: Line 89:
   * It might be possible to pre-compile the preload script and use a binary-form (may be even native .so or .dll) to speed-up server start-up.   * It might be possible to pre-compile the preload script and use a binary-form (may be even native .so or .dll) to speed-up server start-up.
   * In conjunction with ext/FFI (dangerous extension), we may allow FFI functionality only in preloaded PHP files, but not in regular ones   * In conjunction with ext/FFI (dangerous extension), we may allow FFI functionality only in preloaded PHP files, but not in regular ones
-  * It's possible to perform more aggressive optimizations and generate better JIT code for preloaded function and classes (similar to HHVM Repo Authoritative mode in HHVM)+  * It's possible to perform more aggressive optimizations and generate better JIT code for preloaded functions and classes (similar to HHVM Repo Authoritative mode in HHVM) 
 +  * It would be great, to extend preloading with some kind of deployment mechanism, to update preloaded bundle(s) without server restart. 
  
 ===== Proposed Voting Choices ===== ===== Proposed Voting Choices =====
-The RFC requires 50%+1 majority+The RFC requires 50%+1 majority. The voting started **2018-11-06** and will close on **2018-11-14** 
 + 
 +<doodle title="Include preloading ability into PHP-7.4" auth="dmitry" voteType="single" closed="true"> 
 +   * Yes 
 +   * No 
 +</doodle>
  
 ===== Patches and Tests ===== ===== Patches and Tests =====
Line 93: Line 105:
 ===== Implementation ===== ===== Implementation =====
 After the project is implemented, this section should contain  After the project is implemented, this section should contain 
-  - the version(s) it was merged into +  - merged into 7.4 
-  - a link to the git commit(s)+  - git commit [[https://github.com/php/php-src/commit/f6d227ed4f5d4c0276eb720806e808baceb37f10|f6d227ed4f5d4c0276eb720806e808baceb37f10]]
   - a link to the PHP manual entry for the feature   - a link to the PHP manual entry for the feature
-  - a link to the language specification section (if any) 
  
 ===== References ===== ===== References =====
Line 105: Line 116:
   * [[https://docs.hhvm.com/hhvm/advanced-usage/repo-authoritative|Repo Authoritative mode in HHVM]]   * [[https://docs.hhvm.com/hhvm/advanced-usage/repo-authoritative|Repo Authoritative mode in HHVM]]
   * [[https://github.com/php/php-src/commit/d57cd36e47b627dee5b825760163f8e62e23ab28|Immutable Classes implementation in PHP]]   * [[https://github.com/php/php-src/commit/d57cd36e47b627dee5b825760163f8e62e23ab28|Immutable Classes implementation in PHP]]
- 
-===== Rejected Features ===== 
-Keep this updated with features that were discussed on the mail lists. 
rfc/preload.txt · Last modified: 2019/01/21 18:20 by cmb