From: "Bernard Normier" <bernard@zeroc.com>
To: <linux-kernel@vger.kernel.org>
Subject: Re: Concurrent access to /dev/urandom
Date: Tue, 7 Dec 2004 18:41:12 -0500 [thread overview]
Message-ID: <06a501c4dcb6$3cb80cf0$6401a8c0@centrino> (raw)
In-Reply-To: <02c001c4d58c$f6476bb0$6400a8c0@centrino>
Reading concurrently /proc/sys/kernel/random/uuid also returns duplicates
quite quickly ... which definitely looks like a bug. I included a small
python test-case below.
Can anybody suggest a work-around, for example a simple way to serialize
access to /dev/urandom from multiple threads/processes on the same box?
Thanks,
Bernard
#!/usr/bin/env python
import threading
from threading import Thread, Lock
map = {}
lock = Lock()
class MyThread(Thread):
def __init__(self):
Thread.__init__(self)
def run(self):
for i in range(500000):
f = open("/proc/sys/kernel/random/uuid")
k = f.readline().strip()
lock.acquire()
if map.has_key(k):
print "Error: duplicate key " + k
else:
map[k] = 1
lock.release()
f.close()
thread = {}
for i in range(4):
thread[i] = MyThread()
thread[i].start()
for i in range(4):
thread[i].join()
----- Original Message -----
From: "Bernard Normier" <bernard@zeroc.com>
To: "Jan Engelhardt" <jengelh@linux01.gwdg.de>
Cc: <linux-kernel@vger.kernel.org>
Sent: Sunday, November 28, 2004 3:58 PM
Subject: Re: Concurrent access to /dev/urandom
>>>> Rule of thumb: Post the smallest possible code that shows the problem.
>>>Will do next time!
>>
>> That would be great, because it could show that urandom is missing a lock
>> somewhere.
>
> Here is a smaller version (102 lines vs 173 before). It's difficult to get
> something very very small since I need to start a few threads.
>
> Bernard
>
> #include <pthread.h>
> #include <sys/types.h>
> #include <sys/stat.h>
> #include <fcntl.h>
> #include <unistd.h>
> #include <assert.h>
> #include <errno.h>
>
> #include <set>
> #include <iostream>
> using namespace std;
>
> // Each thread will generate keyCount keys
> static int threadCount = 3;
> static int keyCount = 1000000 / threadCount;
>
> // When not defined, all threads read /dev/urandom concurrently
> // #define SERIALIZE_READS 1
>
> struct Key
> {
> long long high;
> long long low;
>
> bool operator<(const Key& rhs) const
> {
> return high < rhs.high || (high == rhs.high && low < rhs.low);
> }
> };
>
> static set<Key> keySet;
> static pthread_mutex_t keySetMutex = PTHREAD_MUTEX_INITIALIZER;
>
> extern "C" void* readRandom(void*)
> {
> for(int i = 0; i < keyCount; ++i)
> {
> int fd = open("/dev/urandom", O_RDONLY);
> assert(fd != -1);
>
> #ifdef SERIALIZE_READS
> int err = pthread_mutex_lock(&keySetMutex);
> assert(err == 0);
> #endif
> size_t index = 0;
> char buffer[sizeof(Key)];
>
> while(index != sizeof(Key))
> {
> ssize_t bytesRead = read(fd, buffer + index, sizeof(Key) -
> index);
>
> if(bytesRead == -1)
> {
> if(errno != EINTR)
> {
> close(fd);
> return reinterpret_cast<void*>(-1);
> }
> }
> else
> {
> index += bytesRead;
> }
> }
>
> close(fd);
>
> #ifndef SERIALIZE_READS
> int err = pthread_mutex_lock(&keySetMutex);
> assert(err == 0);
> #endif
> pair<set<Key>::iterator, bool> result =
> keySet.insert(reinterpret_cast<Key&>(buffer));
> if(!result.second)
> {
> cerr << "Found duplicate!" << endl;
> }
> err = pthread_mutex_unlock(&keySetMutex);
> assert(err == 0);
> }
>
> return 0;
> }
>
> int main(int argc, char* argv[])
> {
> pthread_t* threads = new pthread_t[threadCount];
> for(int i = 0; i < threadCount; ++i)
> {
> int err = pthread_create(&threads[i], 0, readRandom, 0);
> assert(err == 0);
> }
> for(int i = 0; i < threadCount; ++i)
> {
> void* threadStatus;
> int err = pthread_join(threads[i], &threadStatus);
> assert(err == 0);
> assert(threadStatus == 0);
> }
>
> delete[] threads;
> return 0;
> }
>
> // build with g++ -D_REENTRANT -o utest utest.cpp -lpthread
>
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
next prev parent reply other threads:[~2004-12-07 23:41 UTC|newest]
Thread overview: 39+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-11-27 20:45 Bernard Normier
2004-11-27 20:56 ` Jan Engelhardt
2004-11-27 21:15 ` Bernard Normier
2004-11-27 21:22 ` Jan Engelhardt
2004-11-28 20:58 ` Bernard Normier
2004-12-07 23:41 ` Bernard Normier [this message]
2004-12-08 1:28 ` Theodore Ts'o
2004-12-08 1:56 ` Bernard Normier
2004-12-08 19:21 ` Theodore Ts'o
2004-12-08 20:15 ` Bernard Normier
2004-12-08 21:56 ` Matt Mackall
2004-12-09 1:57 ` Theodore Ts'o
2004-12-09 2:46 ` andyliu
2004-12-09 4:55 ` Matt Mackall
2004-12-09 2:58 ` Matt Mackall
2004-12-09 21:29 ` Matt Mackall
2004-12-10 4:47 ` Matt Mackall
2004-12-10 16:35 ` Theodore Ts'o
2004-12-10 18:28 ` Matt Mackall
2004-12-10 21:28 ` Theodore Ts'o
2004-12-10 22:23 ` Matt Mackall
2004-12-11 0:22 ` Adam Heath
2004-12-11 1:10 ` Matt Mackall
2004-12-11 17:33 ` Theodore Ts'o
2004-12-11 19:58 ` Adam Heath
2004-12-11 20:40 ` Matt Mackall
2004-12-12 16:19 ` Pavel Machek
2004-12-11 0:19 ` Adam Heath
2004-12-09 3:10 ` David Lang
2004-12-09 4:52 ` Matt Mackall
2004-12-09 6:36 ` Theodore Ts'o
2004-11-29 22:47 ` Jon Masters
2004-11-29 23:14 ` Bernard Normier
2004-11-29 23:43 ` Sven-Haegar Koch
2004-11-30 2:31 ` David Schwartz
2004-11-30 4:14 ` Kyle Moffett
2004-11-30 8:23 ` Jan Engelhardt
2004-11-30 18:50 ` David Schwartz
2004-11-29 23:42 ` David Wagner
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='06a501c4dcb6$3cb80cf0$6401a8c0@centrino' \
--to=bernard@zeroc.com \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®