Bug #62658 [Fbk->NoF]: When enable –with-curlwrappers option, stream_get_contents() may cpu 100%

Edit report at https://bugs.php.net/bug.php?id=62658&edit=1 ID: 62658 Updated by: php-bugs@lists.php.net Reported by: sopl dot wang at gmail dot com Summary: When enable –with-curlwrappers option, stream_get_contents() may cpu 100% -Status: Feedback +Status: No Feedback Type: Bug Package: cURL related Operating System: ALL PHP Version: master-Git-2012-07-25 (Git) Assigned To: pierrick New Comment: No feedback was provided. The bug is being suspended because we assume that you are no longer experiencing the problem. If this is not the case and you are able to provide the information that was requested earlier, please do so and change the status of the bug back to "Open". Thank you. Previous Comments: ------------------------------------------------------------------------ [2012-11-17 00:53:40] pierrick@php.net Here's what I have on my side with today's version of the master branch : select(4, [3], [], [], {15, 0}) = 1 (in [3], left {4, 999876}) poll([{fd=3, events=POLLIN|POLLPRI}], 1, 0) = 1 ([{fd=3, revents=POLLIN}]) recvfrom(3, "1\r\n1\r\n", 16384, 0, NULL, NULL) = 6 select(4, [3], [], [], {15, 0}) = 1 (in [3], left {5, 182}) poll([{fd=3, events=POLLIN|POLLPRI}], 1, 0) = 1 ([{fd=3, revents=POLLIN}]) recvfrom(3, "1\r\n1\r\n", 16384, 0, NULL, NULL) = 6 I compiled it with the same curl version 7.21.6. As you can see I can't reproduce it. My select don't have anything in the Output file descriptor. Can you test if you still have this different output in your strace ? Do you use the exact same script you sent in your previous comment ? ------------------------------------------------------------------------ [2012-10-07 13:59:43] sopl dot wang at gmail dot com You may try use file_get_contents to open a very slow website to help reproduce this problem. When open a very very slow website, this issue will be more evident. ------------------------------------------------------------------------ [2012-10-07 13:49:51] sopl dot wang at gmail dot com Still have issue (a bit different of output, but still cpu 100%) when I build and test on php-5.4.7 GA release. Follow is my build config: ./configure --prefix=/usr/local/webserver/php5.4.7 --with-config-file- path=/usr/local/webserver/php5.4.7/etc --with-mysql=/usr/local/webserver/mysql5 --with-mysqli=/usr/local/webserver/mysql5/bin/mysql_config --with-iconv-dir=/usr --with-freetype-dir --with-jpeg-dir --with-png-dir --with-zlib --with-libxml- dir=/usr --enable-xml --disable-rpath --enable-bcmath --enable-shmop --enable- sysvsem --enable-inline-optimization --with-curl=/usr --with-curlwrappers -- enable-ftp --enable-mbregex --enable-fpm --with-fpm-user=www --with-fpm- group=www --disable-ipv6 --enable-mbstring --with-mcrypt --with-gd=/usr -- enable-gd-native-ttf --with-openssl --with-mhash --enable-pcntl --enable-sockets --enable-gd-jis-conv --disable-debug --enable-sysvshm --enable-soap --with-pdo- mysql=shared,/usr/local/webserver/mysql5 --with-pdo-sqlite=shared --with- sqlite=shared --enable-pdo=shared Test script ----------- ./sapi/cli/php <<<'<?php while(1) file_get_contents("http://www.l4ka.org");' Output ------ poll([{fd=3, events=POLLIN|POLLPRI}], 1, 0) = 0 (Timeout) clock_gettime(CLOCK_MONOTONIC, {26871452, 912511643}) = 0 clock_gettime(CLOCK_MONOTONIC, {26871452, 912536643}) = 0 clock_gettime(CLOCK_MONOTONIC, {26871452, 912553643}) = 0 select(4, [3], [3], [], {15, 0}) = 1 (out [3], left {15, 0}) clock_gettime(CLOCK_MONOTONIC, {26871452, 912606643}) = 0 poll([{fd=3, events=POLLIN|POLLPRI}], 1, 0) = 0 (Timeout) clock_gettime(CLOCK_MONOTONIC, {26871452, 912643643}) = 0 clock_gettime(CLOCK_MONOTONIC, {26871452, 912668643}) = 0 clock_gettime(CLOCK_MONOTONIC, {26871452, 912685643}) = 0 select(4, [3], [3], [], {15, 0}) = 1 (out [3], left {15, 0}) clock_gettime(CLOCK_MONOTONIC, {26871452, 912738643}) = 0 poll([{fd=3, events=POLLIN|POLLPRI}], 1, 0) = 0 (Timeout) clock_gettime(CLOCK_MONOTONIC, {26871452, 912775643}) = 0 clock_gettime(CLOCK_MONOTONIC, {26871452, 912800643}) = 0 clock_gettime(CLOCK_MONOTONIC, {26871452, 912817643}) = 0 select(4, [3], [3], [], {15, 0}) = 1 (out [3], left {15, 0}) System version -------------- LSB Version: :core-4.0-amd64:core-4.0-ia32:core-4.0-noarch:graphics-4.0- amd64:graphics-4.0-ia32:graphics-4.0-noarch:printing-4.0-amd64:printing-4.0- ia32:printing-4.0-noarch Distributor ID: CentOS Description: CentOS release 5.6 (Final) Release: 5.6 Codename: Final Curl version ------------ curl 7.21.6 (x86_64-unknown-linux-gnu) libcurl/7.21.6 OpenSSL/0.9.8b zlib/1.2.3 libidn/0.6.5 Protocols: dict file ftp ftps gopher http https imap imaps ldap ldaps pop3 pop3s rtsp smtp smtps telnet tftp Features: IDN Largefile NTLM SSL libz ------------------------------------------------------------------------ [2012-09-28 19:44:39] pierrick@php.net Do you still have this issue ? I tried on my local build and everything work as expected. If it doesn't work could you give me more details about your build ? libcurl version ? ------------------------------------------------------------------------ [2012-07-25 10:26:16] sopl dot wang at gmail dot com Description: ------------ When enable –with-curlwrappers option, with http(s)/ftp(s)/other curl-wrapped stream protocols, file_get_contents(), stream_get_contents() and alike functions may lead consume 100% cpu cycle on waiting for receive. When strace problem php process, following will be develop: <pre> select(7, [6], [6], [], {15, 0}) = 1 (out [6], left {15, 0}) poll([{fd=6, events=POLLIN}], 1, 0) = 0 (Timeout) select(7, [6], [6], [], {15, 0}) = 1 (out [6], left {15, 0}) poll([{fd=6, events=POLLIN}], 1, 0) = 0 (Timeout) select(7, [6], [6], [], {15, 0}) = 1 (out [6], left {15, 0}) poll([{fd=6, events=POLLIN}], 1, 0) = 0 (Timeout) select(7, [6], [6], [], {15, 0}) = 1 (out [6], left {15, 0}) poll([{fd=6, events=POLLIN}], 1, 0) = 0 (Timeout) select(7, [6], [6], [], {15, 0}) = 1 (out [6], left {15, 0}) poll([{fd=6, events=POLLIN}], 1, 0) = 0 (Timeout) select(7, [6], [6], [], {15, 0}) = 1 (out [6], left {15, 0}) poll([{fd=6, events=POLLIN}], 1, 0) = 0 (Timeout) select(7, [6], [6], [], {15, 0}) = 1 (out [6], left {15, 0}) poll([{fd=6, events=POLLIN}], 1, 0) = 0 (Timeout) select(7, [6], [6], [], {15, 0}) = 1 (out [6], left {15, 0}) poll([{fd=6, events=POLLIN}], 1, 0) = 0 (Timeout) select(7, [6], [6], [], {15, 0}) = 1 (out [6], left {15, 0}) poll([{fd=6, events=POLLIN}], 1, 0) = 0 (Timeout) select(7, [6], [6], [], {15, 0}) = 1 (out [6], left {15, 0}) poll([{fd=6, events=POLLIN}], 1, 0) = 0 (Timeout) </pre> The reason we found is in 'ext/curl/streams.c', function php_curl_stream_read(), line 174: <code> select(curlstream->maxfd + 1, &curlstream->readfds, &curlstream->writefds, &curlstream->excfds, &tv) </code> When select(), writefds always can write and immediate return, no block at here and, upper module will detect that no data could read and re-invoke this function (php_curl_stream_read()), a nonblock loop, lead this consume 100% cpu bug. If the link very slow, this 100% cpu problem may take the system load very high. Test script: --------------- Enable –with-curlwrappers and test: <code> <?php while (true) file_get_contents('http://soplwang.com'); </code> Expected result: ---------------- Expect the behavior like native php does… When strace, native php generate: <pre> poll([{fd=3, events=POLLIN|POLLERR|POLLHUP}], 1, 60000) = 1 ([{fd=3, revents=POLLIN}]) </pre> Actual result: -------------- When strace, generate: <pre> select(4, [3], [3], [], {15, 0}) = 1 (out [3], left {15, 0}) poll([{fd=3, events=POLLIN}], 1, 0) = 0 (Timeout) select(4, [3], [3], [], {15, 0}) = 1 (out [3], left {15, 0}) poll([{fd=3, events=POLLIN}], 1, 0) = 0 (Timeout) select(4, [3], [3], [], {15, 0}) = 1 (out [3], left {15, 0}) poll([{fd=3, events=POLLIN}], 1, 0) = 0 (Timeout) select(4, [3], [3], [], {15, 0}) = 1 (out [3], left {15, 0}) poll([{fd=3, events=POLLIN}], 1, 0) = 0 (Timeout) select(4, [3], [3], [], {15, 0}) = 1 (out [3], left {15, 0}) poll([{fd=3, events=POLLIN}], 1, 0) = 0 (Timeout) ... </pre> ------------------------------------------------------------------------ -- Edit this bug report at https://bugs.php.net/bug.php?id=62658&edit=1