[#80974] [Ruby trunk Feature#13517] [PATCH] reduce rb_mutex_t size from 160 to 80 bytes on 64-bit — ko1@...
Issue #13517 has been updated by ko1 (Koichi Sasada).
4 messages
2017/05/02
[#81024] Re: [Ruby trunk Feature#13517] [PATCH] reduce rb_mutex_t size from 160 to 80 bytes on 64-bit
— SASADA Koichi <ko1@...>
2017/05/07
sorry for late response.
[#80996] [Ruby trunk Feature#13544] Allow loading an ISeqs sequence directly from a C extension without requiring buffer is in an RVALUE — sam.saffron@...
Issue #13544 has been reported by sam.saffron (Sam Saffron).
3 messages
2017/05/04
[#81016] [Ruby trunk Bug#13526] Segmentation fault at 0x0055c2e58e8920 ruby 2.3.1p112 (2016-04-26 revision 54768) [x86_64-linux] — s.wanabe@...
Issue #13526 has been updated by wanabe (_ wanabe).
3 messages
2017/05/07
[#81048] Re: [ruby-cvs:65788] normal:r58614 (trunk): rb_execution_context_t: move stack, stack_size and cfp from rb_thread_t — SASADA Koichi <ko1@...>
It causes compile error on raspi 3.
3 messages
2017/05/09
[#81201] Re: [ruby-cvs:65935] normal:r58761 (trunk): test/test_extilibs.rb: do not check the existence of fiddle — "U.NAKAMURA" <usa@...>
Hi, Eric
4 messages
2017/05/16
[#81202] Re: [ruby-cvs:65935] normal:r58761 (trunk): test/test_extilibs.rb: do not check the existence of fiddle
— Eric Wong <normalperson@...>
2017/05/16
"U.NAKAMURA" <usa@garbagecollect.jp> wrote:
[#81427] Fwd: [ruby-changes:46809] normal:r58924 (trunk): test for IO.copy_stream CPU usage (r58534) — SASADA Koichi <ko1@...>
Hi,
6 messages
2017/05/28
[#81428] Re: Fwd: [ruby-changes:46809] normal:r58924 (trunk): test for IO.copy_stream CPU usage (r58534)
— Eric Wong <normalperson@...>
2017/05/28
SASADA Koichi <ko1@atdot.net> wrote:
[ruby-core:81460] [Ruby trunk Feature#13378] Eliminate 4 of 8 syscalls when requiring file by absolute path
From:
burke@...
Date:
2017-05-29 18:56:06 UTC
List:
ruby-core #81460
Issue #13378 has been updated by burke (Burke Libbey).
File 0001-reduce-syscalls-on-require-v2.patch added
Thank you for the feedback! I've attached an updated patch to address the issues.
As for testing it, I haven't been able to think of a reasonable method to verify the behaviour without using dtrace/strace, since the only observable effect without system-level instrumentation should be a slight reduction in runtime. I *could* add a case to `DTrace::TestRequire`, but it feels wrong, since this file is concerned with testing the dtrace probes implemented by ruby, not testing ruby using built-in dtrace probes.
Suggestions?
----------------------------------------
Feature #13378: Eliminate 4 of 8 syscalls when requiring file by absolute path
https://bugs.ruby-lang.org/issues/13378#change-65163
* Author: burke (Burke Libbey)
* Status: Open
* Priority: Normal
* Assignee:
* Target version:
----------------------------------------
Don't open file twice when specified by absolute path.
When invoking `require '/a.rb'` (i.e. via an absolute path), ruby generates this sequence of syscalls:
open /a.rb
fstat64 /a.rb
close /a.rb
open /a.rb
fstat64 /a.rb
fstat64 /a.rb
read /a.rb
close /a.rb
It is apparent that the only inherently necessary members of this sequence are:
open /a.rb
fstat64 /a.rb
read /a.rb
close /a.rb
(the fstat64 isn't *obviously* necessary, but it does serve a purpose and probably shouldn't be removed).
The first open/fstat64/close is used to check whether the file is loadable. This is important when scanning the `$LOAD_PATH`, since it is used to determine when a file has been found. However, when we've already unambiguously identified a file before invoking `require`, this serves no inherent purpose, since we can move whatever work is happening as a result of that `fstat64` into the second open/close sequence.
This change bypasses the first open/fstat64/close in the case of an absolute path to `require`. It also removes one of the doubled-up `fstat64` calls later in the sequence. As a result, the number of syscalls to require a file changes:
* From 8 to 4 when specified by absolute path;
* From 5+3n to 4+3n otherwise *(where n is the number of `$LOAD_PATH` items scanned)*.
In future work, it would be possible to re-use the file descriptor opened while searching the `$LOAD_PATH` without the close/open sequence, but this would cause some ugly layering issues.
---
*We intend to use this in conjunction with something like https://github.com/shopify/bootscale, which pre-resolves required features to absolute paths before calling `require`. This change reduces our total number of filesystem accesses by 13% during application boot.*
*Various notes and rationale at http://notes.burke.libbey.me/ruby-require-optimization*
---Files--------------------------------
0001-reduce-syscalls-on-require.patch (7.56 KB)
0001-reduce-syscalls-on-require-fixed.patch (6.94 KB)
0001-reduce-syscalls-on-require-v2.patch (6.44 KB)
--
https://bugs.ruby-lang.org/
Unsubscribe: <mailto:ruby-core-request@ruby-lang.org?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>