From: Greg.mpls@... Date: 2017-10-17T15:49:16+00:00 Subject: [ruby-core:83322] [Ruby trunk Bug#14013] [PATCH] Webrick 60172 fix Issue #14013 has been updated by MSP-Greg (Greg L). File webrick.patch added Eric, thanks. I was working some read_nonblock code at the same time... > There's also no need to call .to_io with IO.select, as it's already done transparently. Aware of that, it was used in several ruby doc examples... Anyway, since you mentioned 'optimized under Linux', I worked with that; it seems that on Windows, after the `select` or `sock.to_io.__send__(ret)` calls complete, it cannot have `sock.accept_nonblock(exception: false)` called again. Hence, I added the additional `break`, but only on windows. Passes locally and on [Appveyor CI](https://ci.appveyor.com/project/MSP-Greg/trunk-mingw-testing/build/2). Patch attached and below: ```diff diff --git a/lib/webrick/server.rb b/lib/webrick/server.rb index 2d678273e5..57ffe5a48b 100644 --- a/lib/webrick/server.rb +++ b/lib/webrick/server.rb @@ -295,12 +295,14 @@ def start_thread(sock, &block) end if sock.respond_to?(:sync_close=) && @config[:SSLStartImmediately] WEBrick::Utils.timeout(@config[:RequestTimeout]) do - - # we must call OpenSSL::SSL::SSLSocket#accept_nonblock until - # it stop returning wait_* symbols: + + # we must call OpenSSL::SSL::SSLSocket#accept_nonblock until it + # stops returning wait_* symbols + # accept_nonblock can only be called once on Windows case ret = sock.accept_nonblock(exception: false) when :wait_readable, :wait_writable sock.to_io.__send__(ret) + break if /mingw|mswin/ =~ RUBY_PLATFORM else break end while true ``` ---------------------------------------- Bug #14013: [PATCH] Webrick 60172 fix https://bugs.ruby-lang.org/issues/14013#change-67280 * Author: MSP-Greg (Greg L) * Status: Closed * Priority: Normal * Assignee: * Target version: * ruby -v: ruby 2.5.0dev (2017-10-13 trunk 60176) [x64-mingw32] * Backport: 2.3: UNKNOWN, 2.4: UNKNOWN ---------------------------------------- I was looking at the failure from 60172, and just tried changing some code. Low and behold, it passed all tests. But, since I'm not that familiar with 'nonblock' issues, I thought asking someone with more knowledge would be appropriate. The patch also included a patch to [TestNetHTTPS.test_verify](https://github.com/ruby/ruby/blob/41910c0d0e989c368c6e08690df8171589e52f76/test/net/http/test_https.rb#L119_L128), which was bypassed (not skipped) based on an ENV setting. I changed it to skip based on what I think is a sensible criteria. I'm somewhat concerned about the test times under windows. Since this is a 'ruby only' patch, maybe someone can test under OSX or *nix? After the patch, I have the following test results: ``` E:\GitHub\ruby\test>ruby -v --disable-gems runner.rb -I./lib -v --show-skip net/http/test_https.rb ruby 2.5.0dev (2017-10-13 trunk 60176) [x64-mingw32] Run options: -I./lib -v --show-skip # Running tests: [1/9] TestNetHTTPS#test_certificate_verify_failure = 1.08 s [2/9] TestNetHTTPS#test_get = 1.14 s [3/9] TestNetHTTPS#test_identity_verify_failure = 0.08 s [4/9] TestNetHTTPS#test_post = 1.13 s [5/9] TestNetHTTPS#test_session_reuse = 3.29 s [6/9] TestNetHTTPS#test_session_reuse_but_expire = 2.25 s [7/9] TestNetHTTPS#test_timeout_during_SSL_handshake = 0.05 s [8/9] TestNetHTTPS#test_verify = 0.76 s [9/9] TestNetHTTPS#test_verify_none = 1.14 s Finished tests in 10.919400s, 0.8242 tests/s, 4.9453 assertions/s. 9 tests, 54 assertions, 0 failures, 0 errors, 0 skips E:\GitHub\ruby\test>ruby -v --disable-gems runner.rb -I./lib -v --show-skip webrick/test_ssl_server.rb ruby 2.5.0dev (2017-10-13 trunk 60176) [x64-mingw32] Run options: -I./lib -v --show-skip # Running tests: [1/3] TestWEBrickSSLServer#test_self_signed_cert_server = 0.14 s [2/3] TestWEBrickSSLServer#test_self_signed_cert_server_with_string = 0.19 s [3/3] TestWEBrickSSLServer#test_slow_connect = 0.30 s Finished tests in 0.639600s, 4.6904 tests/s, 20.3252 assertions/s. 3 tests, 13 assertions, 0 failures, 0 errors, 0 skips ``` Thanks, Greg ---Files-------------------------------- webrick_60172_fix.patch (1.8 KB) webrick_60172_fix.patch (1.81 KB) webrick.patch (990 Bytes) -- https://bugs.ruby-lang.org/ Unsubscribe: